[CI] (Workflows): run the test and apk lanes on every push to feat/disaggregated

- the MGPipe phases land as a series of pushes and each one needs the full lane; dispatching
  by hand after every merge is a step that gets forgotten
- both entries carry a remove-before-merging-to-dev note: dev's trigger set is what ships

.github/scripts/validate-plugin-apks.sh

@@ -44,12 +44,12 @@ require 'key:MOBILEGL_BACKEND_TYPE' "$plugin_resource_text" 'V2 backend variable
 require 'defaultValue:DirectGLES' "$plugin_resource_text" 'V2 DirectGLES default'
 require 'DirectVulkan' "$plugin_resource_text" 'V2 DirectVulkan option'
 require 'key:MOBILEGL_DISABLE_TIMERQUERY' "$plugin_resource_text" 'V2 timer-query toggle'
-require 'key:MOBILEGL_DISABLE_SUBGROUP' "$plugin_resource_text" 'V2 Vulkan subgroup toggle'
+require 'key:MOBILEGL_MAGMA_DISABLE_SUBGROUP' "$plugin_resource_text" 'V2 Vulkan subgroup toggle'
 require 'key:MOBILEGL_MAGMA_R11G11B10F_FALLBACK' "$plugin_resource_text" 'V2 Magma format fallback toggle'
 require 'key:MOBILEGL_MAGMA_FRAMESINFLIGHT' "$plugin_resource_text" 'V2 Magma frames-in-flight setting'
-require 'key:MOBILEGL_AVOID_SAMPLER_MIPMAP_MIN_FILTER' "$plugin_resource_text" 'V2 sampler workaround toggle'
+require 'key:MOBILEGL_ESPRYT_AVOID_SAMPLER_MIPMAP_MIN_FILTER' "$plugin_resource_text" 'V2 sampler workaround toggle'
 require 'key:MOBILEGL_COHERENT_AS_FLUSH' "$plugin_resource_text" 'V2 coherent-as-flush toggle'
-require 'key:MOBILEGL_USE_ANGLE' "$plugin_resource_text" 'V2 ANGLE toggle'
+require 'key:MOBILEGL_ESPRYT_USE_ANGLE' "$plugin_resource_text" 'V2 ANGLE toggle'
 
 if [[ $(grep -Fc 'fclPlugin_V2' <<<"$plugin_manifest") -ne 1 ]]; then
   echo '::error::Plugin manifest must expose exactly one V2 descriptor' >&2

.github/workflows/apk.yml

@@ -6,6 +6,10 @@ on:
       - dev
       - Feat/Backend-Direct-GLES
       - Feat/Backend-Direct-Vulkan
+      # TEMPORARY, remove before merging the MGPipe work into dev: the disaggregation
+      # branch runs the full lane on every push so a phase's landing is not gated on
+      # someone remembering to dispatch the workflow by hand.
+      - feat/disaggregated
   workflow_dispatch:
 
 jobs:
@@ -209,7 +213,7 @@ jobs:
       - name: Load trace cases
         id: trace-cases
         run: |
-          echo "android=$(python3 tools/trace_replay/trace_cases.py --ci --format github-apk)" >> "$GITHUB_OUTPUT"
+          echo "android=$(python3 tools/trace_replay/trace_cases.py --ci --format github-apk-matrix)" >> "$GITHUB_OUTPUT"
           echo "names=$(python3 tools/trace_replay/trace_cases.py --ci --format names)" >> "$GITHUB_OUTPUT"
 
   trace-fixtures:
@@ -337,13 +341,7 @@ jobs:
     strategy:
       fail-fast: false
       max-parallel: 4
-      matrix:
-        backend:
-          - name: DirectGLES
-            gpu: software
-          - name: DirectVulkan
-            gpu: lavapipe
-        case: ${{ fromJSON(needs.trace-cases.outputs.android) }}
+      matrix: ${{ fromJSON(needs.trace-cases.outputs.android) }}
     steps:
       - name: Set Swap Space
         uses: pierotofy/set-swap-space@v1.0
@@ -423,9 +421,12 @@ jobs:
 
       - name: Retrace and validate
         env:
-          MOBILEGL_USE_ANGLE: ${{ matrix.backend.name == 'DirectGLES' && '1' || '0' }}
+          MOBILEGL_ESPRYT_USE_ANGLE: ${{ matrix.backend.name == 'DirectGLES' && '1' || '0' }}
           MOBILEGL_TRACE_ANGLE_VARIANT: ${{ matrix.case.name == 'minecraft-1.21.4-fabric-iris-bliss-in-world' && '90a62123d794' || 'ec889e6ea831' }}
           MOBILEGL_MAGMA_R11G11B10F_FALLBACK: ${{ matrix.backend.name == 'DirectVulkan' && '1' || '0' }}
+          MOBILEGL_MAGMA_FIX_ITERATIONRP_SUBGROUP_SCRATCH: ${{ matrix.backend.name == 'DirectVulkan' && matrix.case.name == 'minecraft-1.21.4-fabric-iris-iterationrp-in-world' && '1' || '0' }}
+          MOBILEGL_MAGMA_DERIVE_NUM_SUBGROUPS: ${{ matrix.backend.name == 'DirectVulkan' && matrix.case.name == 'minecraft-1.21.4-fabric-iris-iterationrp-in-world' && '1' || '0' }}
+          MOBILEGL_MAGMA_ITERATIONRP_FIX_BARRIER: ${{ matrix.backend.name == 'DirectVulkan' && matrix.case.name == 'minecraft-1.21.4-fabric-iris-iterationrp-in-world' && '1' || '0' }}
         run: |
           apk_file="android-retrace-apks/MobileGL-plugin-trace-release-${GITHUB_SHA}.apk"
           test -f "${apk_file}"

.github/workflows/test.yml

@@ -6,7 +6,19 @@ on:
       - dev
       - Feat/Backend-Direct-GLES
       - Feat/Backend-Direct-Vulkan
+      # TEMPORARY, remove before merging the MGPipe work into dev: the disaggregation
+      # branch runs the full lane on every push so a phase's landing is not gated on
+      # someone remembering to dispatch the workflow by hand.
+      - feat/disaggregated
   workflow_dispatch:
+    inputs:
+      baseline_sha:
+        description: >-
+          The commit monolith-symbol-report compares this tree against. P1's G1 says the pull
+          build is byte-identical to feat/disaggregated@087685d1, and that is what the default
+          names. The trigger set is unchanged: this job runs on workflow_dispatch only.
+        required: false
+        default: "087685d1"
 
 jobs:
   build-linux:
@@ -265,13 +277,26 @@ jobs:
         # crash stack without burning a CI round on an in-workflow debugger.
         env:
           MOBILEGL_ITEST_REQUIRE_GPU: "1"
+          MOBILEGL_MAGMA_FIX_ITERATIONRP_SUBGROUP_SCRATCH: "1"
+          MOBILEGL_MAGMA_DERIVE_NUM_SUBGROUPS: "1"
+          MOBILEGL_MAGMA_ITERATIONRP_FIX_BARRIER: "1"
         run: |
           ulimit -c unlimited
           sudo sysctl -w kernel.core_pattern='/tmp/core.%e.%p'
+          # Second, filtered pass: with the range-invalidating map flush disabled,
+          # the buffer scenarios run on the upload ring's staged-copy tier - which
+          # the default pass never reaches (the map tier absorbs every flush on
+          # Mesa), so without this the Mali fallback tier would have zero CI
+          # coverage. The flag is NOT baked into the ctest ENVIRONMENT properties,
+          # so an inline env reaches the test processes (unlike the ICD pin above).
           if [ "${{ secrets.ACTIONS_STEP_DEBUG }}" = "true" ]; then
             ctest -V -L integration-gpu --no-tests=error
+            MOBILEGL_ESPRYT_DISABLE_INVALIDATE_FLUSH=1 ctest -V -L integration-gpu \
+              -R 'Buffer|Readback|Atomic|Ssbo|Arena' --no-tests=error
           else
             ctest --output-on-failure -L integration-gpu --no-tests=error
+            MOBILEGL_ESPRYT_DISABLE_INVALIDATE_FLUSH=1 ctest --output-on-failure -L integration-gpu \
+              -R 'Buffer|Readback|Atomic|Ssbo|Arena' --no-tests=error
           fi
 
       - name: Upload core dumps
@@ -282,6 +307,399 @@ jobs:
           path: /tmp/core.*
           if-no-files-found: ignore
 
+  # THE THIRD CI MODE (ARCHITECTURE.md 13.2-(2)): the same library, built with the PipeInputs
+  # comparator compiled in, running the integration suite and a trace subset with two state models
+  # in one address space. It is a second build rather than a flag on the first because
+  # MOBILEGL_PIPE_VERIFY is a compile-time option - the snapshot, the entry compare and the
+  # compare-at-read hook do not exist in the shipped library, and are never meant to.
+  build-linux-verify:
+    runs-on: ubuntu-latest
+    timeout-minutes: 120
+    permissions:
+      actions: write
+      contents: read
+    env:
+      BUILD_DIR: build-verify
+      CCACHE_BASEDIR: ${{ github.workspace }}
+      CCACHE_COMPRESS: "true"
+      CCACHE_DIR: ${{ github.workspace }}/.ccache
+      CCACHE_MAXSIZE: 4G
+      CCACHE_NOHASHDIR: "true"
+
+    steps:
+      - name: Set Swap Space
+        uses: pierotofy/set-swap-space@v1.0
+        with:
+          swap-size-gb: 32
+
+      - name: Checkout repo
+        uses: actions/checkout@v6
+        with:
+          submodules: recursive
+
+      - name: Get CMake
+        uses: lukka/get-cmake@v4.3.3
+
+      - name: Restore ccache
+        uses: actions/cache/restore@v5
+        with:
+          path: .ccache
+          key: ${{ runner.os }}-test-${{ github.job }}-ccache-v1
+          restore-keys: |
+            ${{ runner.os }}-test-${{ github.job }}-ccache-
+
+      - name: Prepare Vulkan SDK
+        uses: humbletim/setup-vulkan-sdk@v1.2.1
+        with:
+          vulkan-query-version: 1.4.304.1
+          vulkan-components: Vulkan-Headers, Vulkan-Loader
+          vulkan-use-cache: true
+
+      - name: Update glslang external sources
+        working-directory: 3rdparty/glslang
+        run: python update_glslang_sources.py
+
+      - name: Install build dependencies
+        run: |
+          sudo apt-get update
+          sudo apt-get install -y ccache clang-20 clang++-20 lld-20 libc++-20-dev libc++abi-20-dev libvulkan-dev libegl1-mesa-dev libgles2-mesa-dev libgl1-mesa-dri mesa-vulkan-drivers ninja-build
+
+      - name: Show installed toolchain
+        run: |
+          ccache --version
+          clang-20 --version
+          clang++-20 --version
+          ld.lld-20 --version || ld.lld --version || true
+          dpkg -l 'libc++*' 'libegl*' 'libgles*' 'mesa*' 'vulkan*' || true
+
+      - name: Configure CMake
+        # Release/INFO like the shipped build on purpose. The poison arms in this configuration
+        # through MOBILEGL_PIPE_VERIFY (PipeInputs.h derives MOBILEGL_PIPE_POISON from it), so this
+        # job needs neither a Debug log level nor MOBILEGL_BUILD_DISAGGREGATED - and a Debug build
+        # would compare a different library from the one the other lanes measure. (build-linux
+        # switches to Debug under ACTIONS_STEP_DEBUG; this job deliberately does not - a Debug
+        # build flips CXX_VISIBILITY_PRESET and arms MOBILEGL_PIPE_POISON through a second, unrelated
+        # arm of its #if, so the debug switch would change what the lane is measuring.)
+        run: |
+          cmake -S . -B "${BUILD_DIR}" -G Ninja \
+            -DCMAKE_C_COMPILER=clang-20 \
+            -DCMAKE_CXX_COMPILER=clang++-20 \
+            -DCMAKE_C_COMPILER_LAUNCHER=ccache \
+            -DCMAKE_CXX_COMPILER_LAUNCHER=ccache \
+            -DCMAKE_BUILD_TYPE=Release \
+            -DMOBILEGL_LOG_ACTIVE_LEVEL=MOBILEGL_LOG_LEVEL_INFO \
+            -DMOBILEGL_BUILD_TEST=ON \
+            -DMOBILEGL_BUILD_BENCHMARK=OFF \
+            -DMOBILEGL_BUILD_INTEGRATION_TEST=ON \
+            -DMOBILEGL_ITEST_VK_ICD=/usr/share/vulkan/icd.d/lvp_icd.json \
+            -DMOBILEGL_BUILD_TRACE_REPLAY=OFF \
+            -DMOBILEGL_PIPE_VERIFY=ON \
+            -DCMAKE_POLICY_VERSION_MINIMUM=3.5
+
+      - name: Build
+        run: cmake --build "${BUILD_DIR}" --parallel "$(nproc)"
+
+      # The lane is worthless if the option silently did not take, and that is a one-character
+      # mistake away at all times (a typo'd -D is not an error in CMake). Two checks, both cheap:
+      # the comparator's entry point must be in the library, and the fill entry point with it.
+      #
+      # `nm` and NOT `nm -D`. The library is built CXX_VISIBILITY_PRESET hidden in every non-Debug
+      # configuration (CMakeLists.txt:600-604) and the MGPipe entry points are plain namespace
+      # functions with no export attribute, so not one of them appears in the DYNAMIC table: on a
+      # perfectly healthy verify build `nm -D --defined-only ... | grep -c MGPipe` answers 0 out of
+      # ~11900 exported symbols, and a gate spelled that way is red forever for a reason that has
+      # nothing to do with what it claims to test. The static symbol table has them as local `t`
+      # entries, this artifact is never stripped, and `No MG_Remote in the pull build` below already
+      # uses this spelling. The symbol count guards the remaining hole: a stripped library would
+      # make both greps fail for a third, silent reason.
+      - name: The verify library really carries the comparator
+        run: |
+          test -f "${BUILD_DIR}/libMobileGL.so"
+          defined=$(nm --defined-only "${BUILD_DIR}/libMobileGL.so" | wc -l)
+          if [ "${defined}" -lt 1000 ]; then
+            echo "::error::nm --defined-only sees only ${defined} symbols in ${BUILD_DIR}/libMobileGL.so - it looks stripped, so the two checks below could not have failed honestly"
+            exit 1
+          fi
+          for entry in MGPipeVerifyInputs MGPipeFillForVerb; do
+            if ! nm --defined-only "${BUILD_DIR}/libMobileGL.so" | grep -q "${entry}"; then
+              echo "::error::libMobileGL.so defines no ${entry}: -DMOBILEGL_PIPE_VERIFY=ON did not take, and every lane that consumes this artifact would run the comparator-free library and pass having compared nothing"
+              exit 1
+            fi
+          done
+          echo "libMobileGL.so defines MGPipeVerifyInputs and MGPipeFillForVerb (${defined} defined symbols)"
+
+      - name: Show ccache stats
+        if: always()
+        run: ccache --show-stats
+
+      - name: Release superseded ccache entry
+        if: github.ref_name == github.event.repository.default_branch
+        env:
+          GH_TOKEN: ${{ github.token }}
+          CACHE_KEY: ${{ runner.os }}-test-${{ github.job }}-ccache-v1
+        run: gh cache delete "${CACHE_KEY}" || true
+
+      - name: Save ccache
+        if: github.ref_name == github.event.repository.default_branch
+        continue-on-error: true
+        uses: actions/cache/save@v5
+        with:
+          path: .ccache
+          key: ${{ runner.os }}-test-${{ github.job }}-ccache-v1
+
+      - name: Package Linux verify runtime
+        run: |
+          mkdir -p ci-artifacts
+          mapfile -t SHARED_LIBS < <(find "${BUILD_DIR}" -type f \( -name '*.so' -o -name '*.so.*' \) -print | sort)
+          tar \
+            --exclude='*/CMakeFiles' \
+            --exclude='*.o' \
+            --exclude='*.a' \
+            --exclude='*.ninja*' \
+            --exclude='build.ninja' \
+            --exclude='cmake_install.cmake' \
+            -czf ci-artifacts/mobilegl-linux-runtime-verify.tgz \
+            "${BUILD_DIR}/CTestTestfile.cmake" \
+            "${BUILD_DIR}/MobileGL/MG_Test" \
+            "${BUILD_DIR}/MobileGL/MG_IntegrationTest" \
+            "${SHARED_LIBS[@]}"
+
+      - name: Upload Linux verify runtime
+        uses: actions/upload-artifact@v7
+        with:
+          name: mobilegl-linux-runtime-verify
+          path: ci-artifacts/mobilegl-linux-runtime-verify.tgz
+          if-no-files-found: error
+
+  # The verify lane itself, plus the two negative controls that keep it falsifiable. The controls
+  # are ALWAYS-ON steps, not a manual exercise: a gate that can only be shown to work by someone
+  # remembering to break it on purpose is a gate that has already stopped working.
+  integration-verify:
+    runs-on: ubuntu-latest
+    timeout-minutes: 180
+    needs: build-linux-verify
+
+    steps:
+      - name: Checkout repo
+        uses: actions/checkout@v6
+
+      - name: Get CMake
+        uses: lukka/get-cmake@v4.3.3
+
+      - name: Install runtime dependencies
+        run: |
+          sudo apt-get update
+          sudo apt-get install -y libvulkan1 libegl1 libegl-mesa0 libgles2 libgl1-mesa-dri mesa-vulkan-drivers
+
+      - name: Download Linux verify runtime
+        uses: actions/download-artifact@v8
+        with:
+          name: mobilegl-linux-runtime-verify
+          path: .
+
+      - name: Unpack Linux verify runtime
+        run: |
+          tar -xzf mobilegl-linux-runtime-verify.tgz
+          test -f build-verify/libMobileGL.so
+
+      - name: Normalize CTest command paths
+        run: |
+          python - <<'PY'
+          from pathlib import Path
+          import re
+
+          for path in Path('build-verify').rglob('CTestTestfile.cmake'):
+              text = path.read_text()
+              text = re.sub(r'"[^"]*/cmake-[^"]*/bin/cmake"', '"cmake"', text)
+              path.write_text(text)
+          PY
+
+      - name: Integration scenarios under MOBILEGL_PIPE_VERIFY
+        working-directory: build-verify
+        # --no-tests=error is half the gate: the verify entries only exist when the library was
+        # configured with -DMOBILEGL_PIPE_VERIFY=ON, so a build that lost the option matches no
+        # tests and reds here instead of reporting a green run of nothing. The other half is
+        # PipeVerifyArmingScenario.Armed, which fails when the library never printed its arming
+        # line - the failure mode a bare `MOBILEGL_PIPE_VERIFY=1` cannot detect by itself.
+        #
+        # SCOPE, stated so nobody reads more into a green than is there: this is every integration
+        # ENTRY under the comparator, not every integration CONFIGURATION. The `integration` job
+        # runs a second, filtered pass with MOBILEGL_ESPRYT_DISABLE_INVALIDATE_FLUSH=1 for the
+        # upload ring's staged-copy tier; that pass is 186 entries here and, at the 5-10x the
+        # comparator costs, is not affordable inside this job's budget. The tier is covered by
+        # `integration`, unverified, and P2 can take it once the comparator's cost is known.
+        env:
+          MOBILEGL_ITEST_REQUIRE_GPU: "1"
+          MOBILEGL_MAGMA_FIX_ITERATIONRP_SUBGROUP_SCRATCH: "1"
+          MOBILEGL_MAGMA_DERIVE_NUM_SUBGROUPS: "1"
+          MOBILEGL_MAGMA_ITERATIONRP_FIX_BARRIER: "1"
+        run: |
+          ulimit -c unlimited
+          sudo sysctl -w kernel.core_pattern='/tmp/core.%e.%p'
+          if [ "${{ secrets.ACTIONS_STEP_DEBUG }}" = "true" ]; then
+            ctest -V -L integration-verify --no-tests=error
+          else
+            ctest --output-on-failure -L integration-verify --no-tests=error
+          fi
+
+      # The arming lanes' logs, and ONLY those. Each lane shares one MOBILEGL_LOG_FILE_PATH and the
+      # library opens it fopen(path, "w"), so after an ambient lane of 400-odd processes the file
+      # holds the LAST one - grepping it would say nothing about the other 405 and would red a
+      # healthy lane whenever the last entry happened not to issue a verb (which is what the
+      # PoisonOmissionScenario parent, the last ambient entry, does by construction: it forks,
+      # execve()s and reads files). The DirectGLES.VerifyArming. / DirectVulkan.VerifyArming.
+      # entries are one process each on a log path nothing else writes, so this grep means exactly
+      # what it says.
+      #
+      # What it proves: arming is a property of (this library, this environment), and these two
+      # processes ran the same library with the same MOBILEGL_PIPE_VERIFY=1 as their ~400 ambient
+      # siblings. It is not, and cannot be, a per-process census - the shared log cannot support one.
+      # It catches the case ctest cannot: an arming entry that SKIPPED still reports green.
+      - name: The verify lanes armed the comparator
+        working-directory: build-verify
+        run: |
+          shopt -s nullglob
+          logs=(MobileGL/MG_IntegrationTest/pipe-verify-arming-*.log)
+          if [ ${#logs[@]} -lt 2 ]; then
+            echo "::error::found ${#logs[@]} pipe-verify-arming-*.log (expected one per backend). The VerifyArming. entries did not run, so nothing in this job establishes that the comparator was ever armed."
+            exit 1
+          fi
+          for log in "${logs[@]}"; do
+            if ! grep -q "MGPipe: verify armed" "${log}"; then
+              echo "::error::${log} carries no arming line: that lane's process ran the whole scenario without the comparator, so every green entry beside it is green for no reason"
+              exit 1
+            fi
+          done
+          echo "arming line present in all ${#logs[@]} arming-lane log(s)"
+
+      # NEGATIVE CONTROL A (gate G4). The knob perturbs one field in the snapshot arm before the
+      # entry compare, so a working comparator must abort the run. This step passes when ctest
+      # FAILS - `if ctest ...; then error` - which is the only shape that can catch a comparator
+      # that silently compares nothing.
+      #
+      # The knob reaches the test process through the JOB environment: no ctest ENVIRONMENT
+      # property on the ambient Verify. entries names it (MG_IntegrationTest/CMakeLists.txt says
+      # so out loud), and a property entry would otherwise override this and the control would
+      # prove nothing. Same precedent as MOBILEGL_ESPRYT_DISABLE_INVALIDATE_FLUSH in `integration`.
+      - name: Negative control A - a corrupted snapshot field must turn the lane red
+        working-directory: build-verify
+        env:
+          MOBILEGL_ITEST_REQUIRE_GPU: "1"
+          MOBILEGL_PIPE_VERIFY_CORRUPT: GetRenderStateParameters
+        run: |
+          FILTER='DirectGLES\.Verify\..*ClearThenReadPixels'
+          # An empty selection would ALSO make ctest exit non-zero (--no-tests=error), and this
+          # step reads non-zero as "the control worked" - so the selection is counted first. A
+          # control that passes because it ran nothing is worse than no control.
+          matched=$(ctest -N -L integration-verify -R "${FILTER}" | grep -cE '^ *Test *#[0-9]+:')
+          if [ "${matched}" -lt 1 ]; then
+            echo "::error::negative control A selected ${matched} tests; its filter no longer matches anything"
+            exit 1
+          fi
+          if ctest --output-on-failure -L integration-verify -R "${FILTER}" --no-tests=error; then
+            echo "::error::MOBILEGL_PIPE_VERIFY_CORRUPT=GetRenderStateParameters left ${matched} verify entries GREEN. The comparator is not comparing, so every green entry above is green for no reason."
+            exit 1
+          fi
+          echo "the corrupted field turned ${matched} selected entries red, as it must"
+
+      # NEGATIVE CONTROL B (gate G5). The omission skips the STAMP of one field for one verb while
+      # still copying its value - indistinguishable from a fill row nobody wrote - so the poison
+      # must abort the glGenerateMipmap. Again: this step passes when ctest fails.
+      #
+      # The entry it targets is PoisonOmissionScenario.WithoutOmissionCompletes, which is green in
+      # the ambient lane above and is the ONLY integration entry in the tree that calls
+      # glGenerateMipmap at all. It deliberately does not skip itself when the knob is set, exactly
+      # so that this control has something to turn red.
+      - name: Negative control B - an omitted fill point must turn the lane red on that verb
+        working-directory: build-verify
+        env:
+          MOBILEGL_ITEST_REQUIRE_GPU: "1"
+          MOBILEGL_PIPE_POISON_OMIT: GenerateMipmap:GetActiveTextureUnit
+        run: |
+          FILTER='DirectGLES\.Verify\.PoisonOmissionScenario\.WithoutOmissionCompletes'
+          matched=$(ctest -N -L integration-verify -R "${FILTER}" | grep -cE '^ *Test *#[0-9]+:')
+          if [ "${matched}" -lt 1 ]; then
+            echo "::error::negative control B selected ${matched} tests; its filter no longer matches anything"
+            exit 1
+          fi
+          if ctest --output-on-failure -L integration-verify -R "${FILTER}" --no-tests=error; then
+            echo "::error::MOBILEGL_PIPE_POISON_OMIT=GenerateMipmap:GetActiveTextureUnit left the verify lane GREEN. The per-verb poison is not armed, so a forgotten fill row would ship silently."
+            exit 1
+          fi
+          echo "the omitted fill point turned the lane red, as it must"
+
+      - name: Upload verify lane logs
+        if: always()
+        uses: actions/upload-artifact@v7
+        with:
+          name: integration-verify-logs
+          path: build-verify/MobileGL/MG_IntegrationTest/pipe-*.log*
+          if-no-files-found: warn
+
+      - name: Upload core dumps
+        if: failure()
+        uses: actions/upload-artifact@v7
+        with:
+          name: integration-verify-core-dumps
+          path: /tmp/core.*
+          if-no-files-found: ignore
+
+  # MobileGL/MG_Remote/Protocol/generated/protocol_generated.h is COMMITTED, and
+  # flatc is deliberately absent from the default build graph (a codegen step in
+  # the graph is how the earlier branch ended up cross-compiling an arm64 flatc
+  # and trying to run it on the host). This job is what keeps the committed
+  # header honest: build the pinned flatc, regenerate, and fail on any diff.
+  # It needs no MobileGL build, so it does not depend on build-linux.
+  flatc-check:
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Checkout repo
+        uses: actions/checkout@v6
+
+      - name: Get CMake
+        uses: lukka/get-cmake@v4.3.3
+
+      - name: Check out the FlatBuffers submodule only
+        # Just this one: the schema check has nothing to do with glslang,
+        # SPIRV-Cross or the trace fixtures.
+        run: git submodule update --init 3rdparty/flatbuffers
+
+      - name: Regenerate protocol_generated.h
+        run: python3 scripts/gen_protocol.py --build-dir "${{ runner.temp }}/flatc-build"
+
+      - name: Fail if the committed header is stale
+        run: git diff --exit-code -- MobileGL/MG_Remote/Protocol/generated/protocol_generated.h
+
+  # P0.5 interface-purity gate A (ARCHITECTURE.md:501): the two extracted headers' include closure,
+  # asserted on `-H` output because `nm --undefined-only` is blind to "included but not called" -
+  # a header whose types are never named leaves no symbol behind, and "included at all" is exactly
+  # the coupling P1 and P7 have to sever. Needs a preprocessor and three header submodules, no
+  # CMake configure and no glslang sources, so like pipe-gates it does not depend on build-linux.
+  # The script's own --self-test is always on: a negative control that stopped tripping fails the
+  # job, because a gate that cannot go red is not a gate (ROADMAP.md:7).
+  include-graph-check:
+    name: Include-closure purity gate
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Checkout repo
+        uses: actions/checkout@v6
+
+      - name: Check out the three header submodules the closure needs
+        # ska/flat_hash_map.hpp, xxhash.h and vulkan/vulkan.h are the only submodule headers
+        # Includes.h reaches; glslang and spirv_cross are vendored under include/.
+        run: git submodule update --init include/ska 3rdparty/xxHash 3rdparty/Vulkan-Headers
+
+      - name: Install clang and the X11 headers vulkan.h pulls on Linux
+        # Includes.h defines VK_USE_PLATFORM_XLIB_KHR before <vulkan/vulkan.h>, which then
+        # includes <X11/Xlib.h>; without libx11-dev every clang-mode probe dies in the
+        # preprocessor and the gate reports 5 problems that have nothing to do with purity.
+        run: sudo apt-get update && sudo apt-get install -y clang-20 libx11-dev
+
+      - name: Include-closure assertions and negative control
+        run: python3 scripts/check_include_closure.py --mode both --compiler clang++-20 --self-test --require-all
+
   benchmark:
     runs-on: ubuntu-latest
     needs: build-linux
@@ -491,14 +909,21 @@ jobs:
       - benchmark
       - integration
     outputs:
+      matrix: ${{ steps.trace-cases.outputs.matrix }}
       names: ${{ steps.trace-cases.outputs.names }}
+      verify-matrix: ${{ steps.trace-cases.outputs.verify-matrix }}
     steps:
       - name: Checkout repo
         uses: actions/checkout@v6
 
       - name: Load trace cases
         id: trace-cases
-        run: echo "names=$(python3 tools/trace_replay/trace_cases.py --ci --format names)" >> "$GITHUB_OUTPUT"
+        run: |
+          echo "matrix=$(python3 tools/trace_replay/trace_cases.py --ci --format github-test-matrix)" >> "$GITHUB_OUTPUT"
+          echo "names=$(python3 tools/trace_replay/trace_cases.py --ci --format names)" >> "$GITHUB_OUTPUT"
+          # The subset the verify build retraces ("verify": true in trace_cases.json). It is a
+          # SUBSET of the matrix above, so retrace-verify needs no fixtures of its own.
+          echo "verify-matrix=$(python3 tools/trace_replay/trace_cases.py --ci --format github-verify-matrix)" >> "$GITHUB_OUTPUT"
 
   trace-fixtures:
     name: trace fixture (${{ matrix.case }})
@@ -577,11 +1002,7 @@ jobs:
     strategy:
       fail-fast: false
       max-parallel: 4
-      matrix:
-        backend:
-          - DirectGLES
-          - DirectVulkan
-        case: ${{ fromJSON(needs.trace-cases.outputs.names) }}
+      matrix: ${{ fromJSON(needs.trace-cases.outputs.matrix) }}
 
     steps:
       - name: Set Swap Space
@@ -640,6 +1061,12 @@ jobs:
           if [ '${{ matrix.backend }}' = 'DirectVulkan' ]; then
             export MOBILEGL_MAGMA_R11G11B10F_FALLBACK=1
           fi
+          if [ '${{ matrix.backend }}' = 'DirectVulkan' ] \
+             && [ '${{ matrix.case }}' = 'minecraft-1.21.4-fabric-iris-iterationrp-in-world' ]; then
+            export MOBILEGL_MAGMA_FIX_ITERATIONRP_SUBGROUP_SCRATCH=1
+            export MOBILEGL_MAGMA_DERIVE_NUM_SUBGROUPS=1
+            export MOBILEGL_MAGMA_ITERATIONRP_FIX_BARRIER=1
+          fi
           # The blended depth-write quirk auto-enables only on Qualcomm, which no CI
           # runner has, so force it on for the OIT case it exists to fix. ForceOn
           # bypasses only the vendor gate, so this exercises the real strip on
@@ -710,10 +1137,295 @@ jobs:
           archive: false
           if-no-files-found: error
 
+  # The trace half of the third CI mode. Same replay, same goldens, but the library underneath is
+  # the verify build and MOBILEGL_PIPE_VERIFY=1 is in the environment, so every backend read of
+  # frontend state is checked against a snapshot taken at the verb boundary. Eight cases rather
+  # than the full lane's 40 (tools/trace_replay/trace_cases.json, "verify": true): the comparator
+  # is budgeted at 5-10x, and the full sweep is a phase-exit / workflow_dispatch run.
+  retrace-verify:
+    name: retrace verify (${{ matrix.backend }}, ${{ matrix.case }})
+    runs-on: ubuntu-latest
+    timeout-minutes: 240
+    needs:
+      - build-linux-verify
+      - build-retrace
+      - trace-cases
+      - trace-fixtures
+    if: ${{ always() && needs.build-linux-verify.result == 'success' && needs.build-retrace.result == 'success' && needs.trace-cases.result == 'success' }}
+    strategy:
+      fail-fast: false
+      max-parallel: 4
+      matrix: ${{ fromJSON(needs.trace-cases.outputs.verify-matrix) }}
+
+    steps:
+      - name: Set Swap Space
+        uses: pierotofy/set-swap-space@v1.0
+        with:
+          swap-size-gb: 16
+
+      - name: Checkout repo
+        uses: actions/checkout@v6
+
+      - name: Download trace fixture
+        uses: actions/download-artifact@v8
+        with:
+          name: trace-fixture-${{ matrix.case }}
+          path: trace-fixture-download
+
+      - name: Install trace fixture
+        run: |
+          mkdir -p tools/trace_replay/fixtures
+          find trace-fixture-download -type f -exec cp {} tools/trace_replay/fixtures/ \;
+
+      - name: Get CMake
+        uses: lukka/get-cmake@v4.3.3
+
+      - name: Install runtime dependencies
+        run: |
+          sudo apt-get update
+          sudo apt-get install -y libvulkan1 libegl1-mesa-dev libgles2-mesa-dev libgl1-mesa-dri mesa-vulkan-drivers
+          test -e /usr/lib/x86_64-linux-gnu/libEGL.so
+          test -e /usr/lib/x86_64-linux-gnu/libGLESv2.so
+
+      - name: Download Linux verify runtime
+        uses: actions/download-artifact@v8
+        with:
+          name: mobilegl-linux-runtime-verify
+          path: .
+
+      - name: Download trace replay
+        uses: actions/download-artifact@v8
+        with:
+          name: mobilegl-trace-replay
+          path: .
+
+      - name: Unpack the VERIFY runtime as the library under test
+        # build-retrace's CTestTestfile.cmake has the absolute path
+        # <workspace>/build-linux/libMobileGL.so frozen into every case, so the swap happens here
+        # rather than through a variable: the verify .so is put where that path points. The nm
+        # check is what makes the swap falsifiable - a run against the ordinary library would
+        # carry no comparator, ignore MOBILEGL_PIPE_VERIFY entirely, and match its golden.
+        #
+        # `nm`, not `nm -D`, for the reason spelled out in build-linux-verify: everything MGPipe is
+        # hidden-visibility in a Release build and the dynamic table has none of it.
+        run: |
+          tar -xzf mobilegl-linux-runtime-verify.tgz
+          tar -xzf mobilegl-trace-replay.tgz
+          test -f build-verify/libMobileGL.so
+          test -f build-retrace/tools/trace_replay/mobilegl_trace_replay
+          mkdir -p build-linux
+          cp build-verify/libMobileGL.so build-linux/libMobileGL.so
+          if ! nm --defined-only build-linux/libMobileGL.so | grep -q MGPipeVerifyInputs; then
+            echo "::error::the library unpacked at build-linux/libMobileGL.so defines no MGPipeVerifyInputs, so this retrace would replay against a comparator-free build and pass on its golden having verified nothing"
+            exit 1
+          fi
+          echo "the library at build-linux/libMobileGL.so is the verify build"
+
+      - name: Retrace and validate under MOBILEGL_PIPE_VERIFY
+        working-directory: build-retrace/tools/trace_replay
+        # run_trace_case.cmake turns MOBILEGL_PIPE_VERIFY into three assertions of its own (the
+        # arming line, no Fatal{PipeVerifyDiffer, no Fatal{UnmigratedPipeInput), so a case that
+        # somehow ran the wrong library reds here instead of passing on its golden.
+        # --timeout 10800: the 1800s cases run 5-10x slower with both comparator arms live, which
+        # is well past ctest's 1500s default.
+        run: |
+          ulimit -c unlimited
+          sudo sysctl -w kernel.core_pattern='/tmp/core.%e.%p'
+          export MOBILEGL_PIPE_VERIFY=1
+          if [ '${{ matrix.backend }}' = 'DirectVulkan' ]; then
+            export MOBILEGL_MAGMA_R11G11B10F_FALLBACK=1
+          fi
+          if [ '${{ matrix.backend }}' = 'DirectVulkan' ] \
+             && [ '${{ matrix.case }}' = 'improved-transparency-minecraft-26.3' ]; then
+            export MOBILEGL_MAGMA_DISABLE_BLENDED_DEPTH_WRITE=1
+          fi
+          ctest -V --no-tests=error --timeout 10800 \
+            -R '^MobileGLTraceReplay\.${{ matrix.case }}\.${{ matrix.backend }}$'
+
+      # The retrace lane's own always-on negative control, on one case so it costs one short trace:
+      # with a snapshot field corrupted, the SAME replay must fail. Without it, "40 traces, zero
+      # divergences" would be a statement about a comparator nobody watched.
+      #
+      # The rerun replays into the SAME case directory, so the verified run's images are put aside
+      # first and restored before the verdict: "Upload actual image" below runs `if: always()` and
+      # would otherwise ship the deliberately corrupted run's output under the name of the good one.
+      # The restore happens whichever way the control goes, which is why the ctest exit status is
+      # captured rather than tested inline.
+      - name: Negative control - a corrupted snapshot field must red this retrace
+        if: ${{ matrix.case == 'OpenRA' && matrix.backend == 'DirectGLES' }}
+        working-directory: build-retrace/tools/trace_replay
+        run: |
+          export MOBILEGL_PIPE_VERIFY=1
+          export MOBILEGL_PIPE_VERIFY_CORRUPT=GetRenderStateParameters
+          GOOD_OUTPUT="${RUNNER_TEMP}/openra-verified-output"
+          rm -rf "${GOOD_OUTPUT}"
+          if [ -d OpenRA ]; then
+            cp -a OpenRA "${GOOD_OUTPUT}"
+          fi
+          set +e
+          ctest -V --no-tests=error --timeout 10800 \
+            -R '^MobileGLTraceReplay\.OpenRA\.DirectGLES$'
+          control_rc=$?
+          set -e
+          if [ -d "${GOOD_OUTPUT}" ]; then
+            rm -rf OpenRA
+            mv "${GOOD_OUTPUT}" OpenRA
+            echo "restored the verified run's OpenRA output over the corrupted rerun's"
+          fi
+          if [ "${control_rc}" -eq 0 ]; then
+            echo "::error::MOBILEGL_PIPE_VERIFY_CORRUPT=GetRenderStateParameters left the OpenRA retrace GREEN, so the comparator is not comparing and the whole verify retrace lane proves nothing."
+            exit 1
+          fi
+          echo "the corrupted field turned the retrace red, as it must (ctest exit ${control_rc})"
+
+      - name: Upload core dumps
+        if: failure()
+        uses: actions/upload-artifact@v7
+        with:
+          name: retrace-verify-core-dumps-${{ matrix.backend }}-${{ matrix.case }}
+          path: /tmp/core.*
+          if-no-files-found: ignore
+
+      - name: Upload actual image
+        if: always()
+        uses: actions/upload-artifact@v7
+        with:
+          name: retrace-verify-result-${{ matrix.backend }}-${{ matrix.case }}
+          path: |
+            build-retrace/tools/trace_replay/${{ matrix.case }}/actual-images/**
+            build-retrace/tools/trace_replay/${{ matrix.case }}/${{ matrix.backend }}/output/**
+          if-no-files-found: warn
+
+  # G1's own job: the pull build must be the tree before P1, symbol for symbol and byte for byte.
+  # workflow_dispatch only - it builds the library twice from scratch, and its answer is about a
+  # BASELINE rather than about this push, so a per-push run would be measuring the wrong pair.
+  monolith-symbol-report:
+    name: monolith symbol report
+    runs-on: ubuntu-latest
+    timeout-minutes: 180
+    if: ${{ github.event_name == 'workflow_dispatch' }}
+    env:
+      CCACHE_BASEDIR: ${{ github.workspace }}
+      CCACHE_COMPRESS: "true"
+      CCACHE_DIR: ${{ github.workspace }}/.ccache
+      CCACHE_MAXSIZE: 4G
+      CCACHE_NOHASHDIR: "true"
+
+    steps:
+      - name: Set Swap Space
+        uses: pierotofy/set-swap-space@v1.0
+        with:
+          swap-size-gb: 32
+
+      - name: Checkout repo
+        uses: actions/checkout@v6
+        with:
+          submodules: recursive
+          fetch-depth: 0
+
+      - name: Get CMake
+        uses: lukka/get-cmake@v4.3.3
+
+      - name: Restore ccache
+        uses: actions/cache/restore@v5
+        with:
+          path: .ccache
+          key: ${{ runner.os }}-test-${{ github.job }}-ccache-v1
+          restore-keys: |
+            ${{ runner.os }}-test-${{ github.job }}-ccache-
+
+      - name: Prepare Vulkan SDK
+        uses: humbletim/setup-vulkan-sdk@v1.2.1
+        with:
+          vulkan-query-version: 1.4.304.1
+          vulkan-components: Vulkan-Headers, Vulkan-Loader
+          vulkan-use-cache: true
+
+      - name: Install build dependencies
+        run: |
+          sudo apt-get update
+          sudo apt-get install -y ccache clang-20 clang++-20 lld-20 libc++-20-dev libc++abi-20-dev libvulkan-dev libegl1-mesa-dev libgles2-mesa-dev libgl1-mesa-dri mesa-vulkan-drivers ninja-build binutils
+
+      # Both sides with IDENTICAL flags, LTO off, the same compiler and the same standard library:
+      # symbol_report.py's guard rails (scripts/symbol_report.py) say a mismatched pair "adds"
+      # thousands of symbols and the comparison then means nothing. The library alone - no tests,
+      # no benchmark, no integration test, no trace replay - because those targets do not ship.
+      - name: Build the baseline library (${{ inputs.baseline_sha }})
+        run: |
+          git worktree add ../baseline "${{ inputs.baseline_sha }}"
+          cd ../baseline
+          git submodule update --init --recursive
+          (cd 3rdparty/glslang && python update_glslang_sources.py)
+          cmake -S . -B build-sym-base -G Ninja \
+            -DCMAKE_C_COMPILER=clang-20 -DCMAKE_CXX_COMPILER=clang++-20 \
+            -DCMAKE_C_COMPILER_LAUNCHER=ccache -DCMAKE_CXX_COMPILER_LAUNCHER=ccache \
+            -DCMAKE_BUILD_TYPE=Release \
+            -DMOBILEGL_LOG_ACTIVE_LEVEL=MOBILEGL_LOG_LEVEL_INFO \
+            -DMOBILEGL_BUILD_TEST=OFF -DMOBILEGL_BUILD_BENCHMARK=OFF \
+            -DMOBILEGL_BUILD_INTEGRATION_TEST=OFF -DMOBILEGL_BUILD_TRACE_REPLAY=OFF \
+            -DMOBILEGL_BUILD_DISAGGREGATED=OFF \
+            -DMOBILEGL_PIPE_PUSH=OFF -DMOBILEGL_PIPE_VERIFY=OFF \
+            -DMOBILEGL_ENABLE_LTO=OFF \
+            -DCMAKE_POLICY_VERSION_MINIMUM=3.5
+          cmake --build build-sym-base --parallel "$(nproc)"
+          cp build-sym-base/libMobileGL.so "${GITHUB_WORKSPACE}/libMobileGL-baseline.so"
+
+      - name: Build the head library
+        run: |
+          (cd 3rdparty/glslang && python update_glslang_sources.py)
+          cmake -S . -B build-sym-head -G Ninja \
+            -DCMAKE_C_COMPILER=clang-20 -DCMAKE_CXX_COMPILER=clang++-20 \
+            -DCMAKE_C_COMPILER_LAUNCHER=ccache -DCMAKE_CXX_COMPILER_LAUNCHER=ccache \
+            -DCMAKE_BUILD_TYPE=Release \
+            -DMOBILEGL_LOG_ACTIVE_LEVEL=MOBILEGL_LOG_LEVEL_INFO \
+            -DMOBILEGL_BUILD_TEST=OFF -DMOBILEGL_BUILD_BENCHMARK=OFF \
+            -DMOBILEGL_BUILD_INTEGRATION_TEST=OFF -DMOBILEGL_BUILD_TRACE_REPLAY=OFF \
+            -DMOBILEGL_BUILD_DISAGGREGATED=OFF \
+            -DMOBILEGL_PIPE_PUSH=OFF -DMOBILEGL_PIPE_VERIFY=OFF \
+            -DMOBILEGL_ENABLE_LTO=OFF \
+            -DCMAKE_POLICY_VERSION_MINIMUM=3.5
+          cmake --build build-sym-head --parallel "$(nproc)"
+
+      # The monolith must not have grown a remote half. ARCHITECTURE.md:506: MG_Remote lives behind
+      # MOBILEGL_BUILD_DISAGGREGATED and nothing of it may reach a shipped pull build.
+      - name: No MG_Remote in the pull build
+        run: |
+          if nm --defined-only build-sym-head/libMobileGL.so | grep -q MG_Remote; then
+            echo "::error::the pull build defines MG_Remote symbols; the disaggregated half leaked into the monolith"
+            nm --defined-only build-sym-head/libMobileGL.so | grep MG_Remote | head -20
+            exit 1
+          fi
+          echo "no MG_Remote symbols in the pull build"
+
+      - name: Symbol report (G1)
+        run: |
+          python3 scripts/symbol_report.py \
+            --before libMobileGL-baseline.so \
+            --after build-sym-head/libMobileGL.so \
+            --threshold 0 \
+            --fail-on-symbol-set-change \
+            --fail-on-added-bytes 0 \
+            --markdown symbol-report.md \
+            --json symbol-report.json
+
+      - name: Upload the symbol report
+        if: always()
+        uses: actions/upload-artifact@v7
+        with:
+          name: monolith-symbol-report
+          path: |
+            symbol-report.md
+            symbol-report.json
+          if-no-files-found: error
+
   remove-artifact-clutter:
     name: remove artifact clutter
     runs-on: ubuntu-latest
-    needs: retrace-summary
+    # (d) retrace-verify too: this job deletes the trace-fixture-* artifacts, and the verify
+    # retraces download the same ones.
+    needs:
+      - retrace-summary
+      - retrace-verify
     if: always()
     permissions:
       actions: write
@@ -722,14 +1434,19 @@ jobs:
         env:
           GH_TOKEN: ${{ github.token }}
         run: |
+          # Both retrace lanes, not just the pull one: `retrace verify (backend, case)` downloads
+          # the same trace-fixture-<case> artifact, and a failed verify retrace is exactly when
+          # someone needs that fixture to reproduce locally. The two prefixes are stripped in
+          # order, longest first, because "retrace (" is not a prefix of "retrace verify (".
           declare -A failed_cases=()
           while IFS= read -r job_name; do
-            case_name="${job_name#retrace (*, }"
+            case_name="${job_name#retrace verify (*, }"
+            case_name="${case_name#retrace (*, }"
             case_name="${case_name%)}"
             failed_cases["${case_name}"]=1
           done < <(
             gh api --paginate "repos/${GITHUB_REPOSITORY}/actions/runs/${GITHUB_RUN_ID}/jobs?per_page=100" \
-              --jq '.jobs[] | select(.name | startswith("retrace (")) | select(.conclusion == "failure" or .conclusion == "cancelled" or .conclusion == "timed_out" or .conclusion == "action_required") | .name'
+              --jq '.jobs[] | select((.name | startswith("retrace (")) or (.name | startswith("retrace verify ("))) | select(.conclusion == "failure" or .conclusion == "cancelled" or .conclusion == "timed_out" or .conclusion == "action_required") | .name'
           )
 
           if ((${#failed_cases[@]})); then
@@ -760,3 +1477,69 @@ jobs:
           )
 
           echo "Deleted ${deleted} intermediate Linux artifact(s); retained ${retained} failed-retrace fixture(s)."
+
+  pipe-gates:
+    name: MGPipe generators and hygiene gates
+    runs-on: ubuntu-latest
+    # Deliberately independent of build-linux: these are source-level gates, they take
+    # seconds, and a broken build must not hide a drifted interface.
+
+    steps:
+      - name: Checkout repo
+        uses: actions/checkout@v6
+
+      # The seven generators all read MG_Pipe/*.def, so regenerating and diffing is what
+      # keeps the two interface tables, the wire records, the verify comparators, the
+      # PipeInputs field ids, the read-inventory coverage and the render-state member list
+      # from drifting apart from the catalogue. The generated files are committed
+      # deliberately: the build must not depend on python.
+      - name: Regenerate the MGPipe interface (G1-G7)
+        run: |
+          python3 scripts/gen_pipe.py
+          git diff --exit-code -- MobileGL/MG_Pipe/generated
+
+      # The generators' own negative controls: canned inputs that MUST trip each structural check
+      # (a field list that does not cover its struct's members, a verb set that is not the function
+      # table's). Regenerating and diffing above cannot see a check that silently stopped
+      # checking - a broken gate and a clean tree produce the same green.
+      - name: The MGPipe generators' checks can still fail
+        run: python3 scripts/gen_pipe.py --self-test
+
+      # The same question for the symbol tool the P1 gate is written in terms of.
+      - name: The symbol report's buckets and gates can still fail
+        run: python3 scripts/symbol_report.py --self-test
+
+      # Per-draw fprintf/printf instrumentation has repeatedly been committed by accident,
+      # once inside a mutex critical section. Nothing under these two trees prints to a
+      # stdio stream today - MGLOG_D compiles out in INFO builds and is the only channel
+      # they are allowed to use - so this gate starts with no exceptions, and any addition
+      # to it needs a reason in the pull request rather than a quiet whitelist entry. The
+      # alternation names every stdio spelling, not just the two that were committed:
+      # fprintf to either stream, printf, puts, and the iostream pair.
+      - name: No stdio instrumentation in MG_Backend or MG_State
+        run: |
+          if grep -rnE 'fprintf[[:space:]]*\((stderr|stdout)|(^|[^[:alnum:]_>.])printf[[:space:]]*\(|(^|[^[:alnum:]_>.:])puts[[:space:]]*\(|std::(cout|cerr)' \
+              MobileGL/MG_Backend MobileGL/MG_State; then
+            echo "::error::stdio instrumentation found; use MGLOG_D (compiled out in INFO builds)"
+            exit 1
+          fi
+          echo "no fprintf(stderr/stdout / printf( / puts( / std::cout|cerr under MobileGL/MG_Backend or MobileGL/MG_State"
+
+      # Informational: the frontend mutation surface an MGPipe aggregate generation has to
+      # cover. It becomes a gate in P2, when the mapping file exists to diff against
+      # (ROADMAP.md:18 puts the first mapping round in P2, not P1).
+      - name: MGPipe dirty-surface report
+        run: python3 scripts/gen_pipe_dirty_surface.py --summary
+
+      # Warning only for now: the disaggregation documents are still being written, and a
+      # lint that fails a rewrite in progress teaches people to ignore it. It becomes
+      # --strict when the documents settle.
+      - name: Documentation citation lint
+        run: |
+          shopt -s nullglob
+          documents=(docs/Disaggregated/*.md)
+          if [ ${#documents[@]} -eq 0 ]; then
+            echo "no disaggregation documents to check"
+            exit 0
+          fi
+          python3 scripts/check_doc_citations.py "${documents[@]}" || true

.gitignore

@@ -27,3 +27,4 @@ MobileGL/MG*/cmake-build*
 tools/trace_replay/work/
 __pycache__/
 *.py[cod]
+/.gradle

.gitmodules

@@ -34,3 +34,6 @@
 [submodule "include/ska"]
 	path = include/ska
 	url = https://github.com/MobileGL-Dev/flat_hash_map.git
+[submodule "3rdparty/flatbuffers"]
+	path = 3rdparty/flatbuffers
+	url = https://github.com/google/flatbuffers.git

CMakeLists.txt

@@ -14,6 +14,19 @@ option(MOBILEGL_ENABLE_TRACY      "Enable tracy for profiling"
 option(MOBILEGL_BUILD_TRACE_REPLAY "Build desktop apitrace replay runner"            OFF)
 option(MOBILEGL_TRACE_ANGLE_VARIANTS "Enable signed trace-APK ANGLE variant loading" OFF)
 option(MOBILEGL_IOS               "Build MobileGL for iOS instead of macOS when APPLE is set" OFF)
+# The disaggregated (two-process) shape. OFF is the shipping default and OFF
+# must stay byte-comparable to a tree without MG_Remote at all: nothing under
+# MobileGL/MG_Remote/ is compiled, no include path is added, and no library is
+# linked, so `nm --defined-only libMobileGL.so | grep -i MG_Remote` is empty.
+# That emptiness is one of the two byte-level equalities the plan's validation
+# gates keep (section 10.3).
+option(MOBILEGL_BUILD_DISAGGREGATED "Build the MG_Remote transport layer (two-process shape)" OFF)
+option(MOBILEGL_BUILD_SERVER_SPIKE "Build the P0 spike-A MobileGLServer delivery-chain executable (Android only)" OFF)
+# The PipeInputs strangler (ARCHITECTURE.md 9.2). OFF is the pull build and must stay
+# byte-identical to a tree without either option: MGB_CTX is the live GLContext, no
+# MGPipe/PipeInputs source is compiled, every MGP_FILL is ((void)0).
+option(MOBILEGL_PIPE_PUSH "Backends read frontend state through the MGPipe PipeInputs block instead of MG_State::pGLContext (ARCHITECTURE.md 9.2 phase A)" OFF)
+option(MOBILEGL_PIPE_VERIFY "Compile SnapshotFromGLContext() and the G4 per-verb shadow comparator; implies MOBILEGL_PIPE_PUSH; never shipped" OFF)
 set(MOBILEGL_LOG_ACTIVE_LEVEL "MOBILEGL_LOG_LEVEL_INFO" CACHE STRING "MobileGL active log level macro")
 set(MOBILEGL_VULKAN_LIBRARY "" CACHE FILEPATH "Vulkan loader/MoltenVK library to link for iOS builds")
 
@@ -182,6 +195,7 @@ set(ENABLE_SPVREMAPPER         OFF CACHE BOOL "Enable SPVRemapper" FORCE)
 set(ENABLE_OPT                 ON  CACHE BOOL "Enable SPIRV-Tools opt usage in glslang" FORCE)
 set(BUILD_EXTERNAL             ON  CACHE BOOL "Build external deps in External/" FORCE)
 set(ENABLE_GLSLANG_INSTALL     OFF CACHE BOOL "Install glslang targets" FORCE)
+set(SPIRV_SKIP_EXECUTABLES     ON  CACHE BOOL "Skip building SPIRV-Tools executables" FORCE)
 
 set(SPIRV_CROSS_C_API          ON  CACHE BOOL "Enable C API" FORCE)
 set(SPIRV_CROSS_ENABLE_GLSL    ON  CACHE BOOL "Enable GLSL backend" FORCE)
@@ -237,6 +251,8 @@ set(SOURCE_FILES
 
     MobileGL/MG_Util/Metrics/BufferMetrics.cpp
 
+    MobileGL/MG_Util/Metrics/PipeStats.cpp
+
     MobileGL/MG_Util/Converters/GLToStr/GLEnumConverter.cpp
     MobileGL/MG_Util/Converters/EGLToStr/EGLEnumConverter.cpp
     MobileGL/MG_Util/Converters/MGToStr/DataTypeConverter.cpp
@@ -269,6 +285,7 @@ set(SOURCE_FILES
     MobileGL/MG_Util/ShaderTranspiler/ShaderCompiler.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpvcSession.cpp
     MobileGL/MG_Util/ShaderTranspiler/ShaderSourceProcessor.cpp
+    MobileGL/MG_Util/ShaderTranspiler/TranslationCache.cpp
     MobileGL/MG_Util/ShaderTranspiler/glslang/TMglGlslIoResolver.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/FlattenInterfaceStructPass.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/EliminateFloatEqualsZeroPass.cpp
@@ -277,26 +294,43 @@ set(SOURCE_FILES
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/DecomposeWorkgroupVec3Pass.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/DecoratePositionInvariantPass.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/DemoteFloat64Pass.cpp
+    MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/FlattenFloat64StorageBlockPass.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/LowerDrawParametersPass.cpp
+    MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/LowerViewportIndexPass.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/PackDoubleVertexInputsPass.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/FlattenXfbInterfaceBlocksPass.cpp
+    MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/UniquifyIoBlockNamesPass.cpp
+    MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/StripIoBlockLocationsPass.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/SplitArrayVertexInputsPass.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/RebaseInstanceIndexPass.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/ZeroBaseVertexPass.cpp
+    MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/DeriveNumSubgroupsPass.cpp
+    MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/FixIterationRPBarrierPass.cpp
+    MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/FixIterationRPSubgroupScratchPass.cpp
+    MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/EmulateSubgroupsPass.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/NormalizeRectCoordinatesPass.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/Lower1DArrayImagesPass.cpp
+    MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/Lower1DSampledImagesPass.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/BakeImageFormatsPass.cpp
+    MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/WidenImageFormatsPass.cpp
+    MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/ClampMultisampleFetchPass.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/PrivateToEntryLocalPass.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/StripUniformLocationsPass.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/StripUboMemberRelaxedPrecisionPass.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/StripNoPerspectivePass.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/EmulateNoPerspectivePass.cpp
     MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/LegalizeFragmentOutputIndexPass.cpp
+    MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/LegalizeResourceArrayIndexPass.cpp
+    MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/FlattenAtomicCounterBlockPass.cpp
+    MobileGL/MG_Util/ShaderTranspiler/SpirvPasses/DemotePointSizePass.cpp
 
     MobileGL/MG_Util/BackendLoaders/OpenGL/Loader.cpp
     MobileGL/MG_Util/BackendLoaders/Vulkan/Loader.cpp
 
+    MobileGL/MG_Util/SelfTest/DriverBugProbes.cpp
     MobileGL/MG_Util/SelfTest/DriverPost.cpp
+    MobileGL/MG_Util/SelfTest/DriverPostIterationRPWitness.cpp
+    MobileGL/MG_Util/SelfTest/PrimitivesGeneratedNoXfbProbe.cpp
 
     MobileGL/MG_Util/Texture/PixelStoreProcessor.cpp
     MobileGL/MG_Util/Texture/TextureFormatProcessor.cpp
@@ -319,6 +353,7 @@ set(SOURCE_FILES
     MobileGL/MG_Impl/GLImpl/Program/ProgramInterface.cpp
     MobileGL/MG_Impl/GLImpl/Program/GL_ProgramPipeline.cpp
     MobileGL/MG_Impl/GLImpl/Texture/GL_Texture.cpp
+    MobileGL/MG_Impl/GLImpl/Debug/GL_Debug.cpp
     MobileGL/MG_Impl/GLImpl/Texture/Validators.cpp
     MobileGL/MG_Impl/GLImpl/Texture/ProxyTexture.cpp
     MobileGL/MG_Impl/GLImpl/VertexArray/GL_VertexArray.cpp
@@ -376,10 +411,12 @@ set(SOURCE_FILES
     MobileGL/MG_State/GLState/TextureState/TextureObject2DCube.cpp
     MobileGL/MG_State/GLState/TextureState/TextureObject3D.cpp
     MobileGL/MG_State/GLState/TextureState/TextureObjectBuffer.cpp
+    MobileGL/MG_State/GLState/TextureState/TextureObjectView.cpp
     MobileGL/MG_State/GLState/TextureState/TextureUnit.cpp
     MobileGL/MG_State/GLState/TextureState/TextureState.cpp
     MobileGL/MG_State/GLState/ProgramState/ProgramObject.cpp
     MobileGL/MG_State/GLState/ProgramState/ProgramLinkTask.cpp
+    MobileGL/MG_State/GLState/ProgramState/ProgramTranslationCache.cpp
     MobileGL/MG_State/GLState/ProgramState/ProgramSpirvTask.cpp
     MobileGL/MG_State/GLState/ProgramState/ShaderCompileTask.cpp
     MobileGL/MG_State/GLState/ProgramState/ShaderObject.cpp
@@ -395,6 +432,65 @@ set(SOURCE_FILES
     MobileGL/MG_State/GLState/RenderbufferState/RenderbufferState.cpp
 )
 
+# ---------------------------------------------------------------------------
+# MG_Remote (disaggregated transport). Everything below is gated: with the
+# option OFF not one file here is compiled and no include path is added.
+# ---------------------------------------------------------------------------
+
+# FlatBuffers is a submodule and its runtime is header-only. Guard both ways:
+# a checkout without the submodule must configure and build, just without the
+# disaggregated shape, rather than fail with a missing-header error a hundred
+# lines later. Note this only checks for the RUNTIME headers - flatc is never
+# built here (see scripts/gen_protocol.py).
+if (MOBILEGL_BUILD_DISAGGREGATED AND
+    NOT EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/3rdparty/flatbuffers/include/flatbuffers/flatbuffers.h")
+    message(WARNING
+        "MOBILEGL_BUILD_DISAGGREGATED=ON but 3rdparty/flatbuffers/include is missing. "
+        "Run `git submodule update --init 3rdparty/flatbuffers`. Building without the "
+        "disaggregated shape for this configure; the cached ON takes effect once the "
+        "submodule is present.")
+    # A NORMAL variable, deliberately not `CACHE BOOL ... FORCE`: forcing OFF into the cache
+    # made the plain re-configure after `git submodule update` stay OFF with no message at
+    # all. Shadowing the cache entry for this configure only keeps the operator's ON where it
+    # was, so the next configure - with the submodule there - honours it.
+    set(MOBILEGL_BUILD_DISAGGREGATED OFF)
+endif()
+
+# MOBILEGL_PIPE_VERIFY implies MOBILEGL_PIPE_PUSH: the comparator compares the pushed block
+# against a snapshot, so there has to be a pushed block. A normal variable, not a forced
+# cache write, for the same reason as the disaggregated fallback above.
+if (MOBILEGL_PIPE_VERIFY AND NOT MOBILEGL_PIPE_PUSH)
+    message(STATUS "MobileGL: MOBILEGL_PIPE_VERIFY=ON forces MOBILEGL_PIPE_PUSH ON for this configure")
+    set(MOBILEGL_PIPE_PUSH ON)
+endif()
+
+if (MOBILEGL_PIPE_PUSH)
+    message(STATUS "MobileGL: PipeInputs push ON, appending the MGPipe fill sources")
+    list(APPEND SOURCE_FILES
+        MobileGL/MG_Backend/MGPipe/PipeInputs.cpp
+        MobileGL/MG_Impl/Pipe/PipeFill.cpp
+    )
+endif()
+
+if (MOBILEGL_BUILD_DISAGGREGATED)
+    message(STATUS "MobileGL: disaggregated transport ON, appending MG_Remote sources")
+    list(APPEND SOURCE_FILES
+        MobileGL/MG_Remote/Transport/Ring.cpp
+        MobileGL/MG_Remote/Transport/Doorbell.cpp
+        MobileGL/MG_Remote/Transport/ShmSegment.cpp
+        # Both platform halves are listed unconditionally and each is empty on
+        # the other OS, so neither can rot behind an `if (WIN32)` nobody
+        # configures.
+        MobileGL/MG_Remote/Transport/ShmSegmentPosix.cpp
+        MobileGL/MG_Remote/Transport/ShmSegmentWin32.cpp
+        MobileGL/MG_Remote/Transport/FdPassing.cpp
+        MobileGL/MG_Remote/Transport/InProcessTransport.cpp
+        # Keeps MG_Util/Debug/Log.h - and through it the GL frontend's
+        # umbrella header - out of the header-only wire code (WireLog.h).
+        MobileGL/MG_Remote/Transport/WireLog.cpp
+    )
+endif()
+
 if (APPLE AND NOT MOBILEGL_IOS)
     list(APPEND SOURCE_FILES
         MobileGL/MG_Impl/CGLImpl/CGLImpl.cpp
@@ -445,11 +541,26 @@ set(MOBILEGL_COMPILE_DEF
         -DASIO_NO_DEPRECATED
 )
 
+if (MOBILEGL_BUILD_DISAGGREGATED)
+    list(APPEND MOBILEGL_COMPILE_DEF -DMOBILEGL_BUILD_DISAGGREGATED=1)
+endif()
+
+if (MOBILEGL_PIPE_PUSH)
+    list(APPEND MOBILEGL_COMPILE_DEF -DMOBILEGL_PIPE_PUSH=1)
+endif()
+if (MOBILEGL_PIPE_VERIFY)
+    list(APPEND MOBILEGL_COMPILE_DEF -DMOBILEGL_PIPE_VERIFY=1)
+endif()
+
 message(STATUS "MOBILEGL_COMPILE_DEF=${MOBILEGL_COMPILE_DEF}")
 
 set(MOBILEGL_INCLUDE_DIR
         ${CMAKE_SOURCE_DIR}/include
         ${CMAKE_SOURCE_DIR}/MobileGL
+        # The MGPipe boundary headers. They are reachable as <MG_Pipe/MGPipe.h> through the
+        # line above too; this entry lets the client, the backends and MG_Remote spell them
+        # as <MGPipe.h> once MG_Pipe stops being a leaf of the frontend tree.
+        ${CMAKE_SOURCE_DIR}/MobileGL/MG_Pipe
         ${spirv-tools_SOURCE_DIR}
         ${spirv-tools_SOURCE_DIR}/include
         ${spirv-tools_BINARY_DIR}
@@ -457,9 +568,16 @@ set(MOBILEGL_INCLUDE_DIR
         # Header-only submodule: no add_subdirectory, no link target. Only
         # MG_Util/Async/ShaderCompilePool.cpp includes it, and it stays behind that file's
         # pimpl so no consumer target needs this path.
-        ${CMAKE_SOURCE_DIR}/3rdparty/asio/asio/include
+        ${CMAKE_SOURCE_DIR}/3rdparty/asio/include
 )
 
+if (MOBILEGL_BUILD_DISAGGREGATED)
+    # Header-only runtime: an include path, no add_subdirectory, no link
+    # target, and above all no flatc in the build graph. protocol_generated.h
+    # is committed and regenerated by scripts/gen_protocol.py.
+    list(APPEND MOBILEGL_INCLUDE_DIR ${CMAKE_SOURCE_DIR}/3rdparty/flatbuffers/include)
+endif()
+
 add_library(${CMAKE_PROJECT_NAME} SHARED
     ${SOURCE_FILES}
 )
@@ -669,3 +787,50 @@ if (NOT ANDROID)
         add_subdirectory(tools/trace_replay)
     endif()
 endif()
+
+# The integration binary is also useful as a standalone adb-shell executable.
+# Android cannot use the desktop-only MobileGL_s target, so its CMake module
+# links libMobileGL.so and creates an AImageReader-backed window instead.
+if (ANDROID AND MOBILEGL_BUILD_INTEGRATION_TEST)
+    add_subdirectory(MobileGL/MG_IntegrationTest)
+endif()
+
+# ---------------------------------------------------------------------------
+# P0 spike A: the Android delivery chain for a second native executable.
+#
+# The disaggregated design needs a server process on Android (PLAN-B.md §8.1,
+# inheriting PLAN.md §11.1-§11.6). An APK's only exec-able install location is
+# lib/<abi>/, and the packager only puts a file there if it is named lib*.so -
+# so a second executable has to be built with an .so name and exec'd out of
+# getApplicationInfo().nativeLibraryDir. This target is the stub that proves the
+# chain end to end: it is packaged like a library, exec'd from the app's own
+# untrusted_app process, and writes a marker the parent reads back.
+#
+# Off by default and ANDROID-only, so no shipping configuration builds it. The
+# trace flavour of the plugin APK turns it on (android-plugin/build.gradle).
+# ---------------------------------------------------------------------------
+if (ANDROID AND MOBILEGL_BUILD_SERVER_SPIKE)
+    add_executable(MobileGLServer
+        ${CMAKE_CURRENT_SOURCE_DIR}/tools/spikes/server_stub/main.cpp)
+
+    # An executable that is named like a shared library still has to be a real
+    # PIE executable: Android has refused non-PIE executables since API 21, and
+    # the name alone does not change what the loader demands of the file.
+    set_target_properties(MobileGLServer PROPERTIES
+        PREFIX "lib"
+        SUFFIX ".so"
+        OUTPUT_NAME "MobileGLServer"
+        POSITION_INDEPENDENT_CODE ON)
+    target_compile_options(MobileGLServer PRIVATE -fPIE)
+    target_link_options(MobileGLServer PRIVATE -pie)
+
+    # AGP packages what the external native build drops into the per-ABI output
+    # directory, and it selects by the .so extension. CMake puts executables in
+    # CMAKE_RUNTIME_OUTPUT_DIRECTORY, which is not the directory AGP hands to
+    # CMAKE_LIBRARY_OUTPUT_DIRECTORY, so point this target's runtime output at
+    # the library directory when the generator gave us one.
+    if (CMAKE_LIBRARY_OUTPUT_DIRECTORY)
+        set_target_properties(MobileGLServer PROPERTIES
+            RUNTIME_OUTPUT_DIRECTORY "${CMAKE_LIBRARY_OUTPUT_DIRECTORY}")
+    endif()
+endif()

MobileGL/Config.h

@@ -66,22 +66,60 @@ namespace MobileGL::MG_Config {
     //   - DISPLAY: X11 session variable, not MobileGL configuration.
     //   - MOBILEGL_LOG_FILE_PATH: log-file init runs before MG_ConfigLoader::Init
     //     (see MG_Util/Debug/Log.cpp).
-    //   - MOBILEGL_VALIDATE_SPIRV: test suites like SpirvPassTest exercise
-    //     ShaderCompiler without ever running MobileGL::Initialize(), and every
-    //     Initialize() re-runs MG_ConfigLoader::Init, which would clobber a
-    //     programmatic override stored here (see ShaderCompiler.cpp,
-    //     SpirvValidationEnabled).
     struct FeaturesTable {
         // MOBILEGL_DISABLE_TIMERQUERY: do not advertise or use GPU timer queries.
         Bool DisableTimerQuery = false;
-        // MOBILEGL_USE_ANGLE: load ANGLE EGL/GLES libraries.
-        Bool UseAngle = false;
+        // MOBILEGL_ESPRYT_ENABLE_TEXTURE_VIEW: advertise GL_ARB_texture_view on DirectGLES when
+        // the host ES driver has EXT/OES_texture_view. Off by default: the host extension is
+        // present on Adreno 830 and the functional half of KHR-GL4{2,3}.texture_view still fails
+        // there, because the view's ES internalformat is normalized independently of the storage
+        // it aliases (see BackendObject_DirectGLES::BuildAdvertisedExtensions). The flag exists
+        // so that work can be done without editing the gate.
+        Bool EsprytEnableTextureView = false;
+        // MOBILEGL_ENABLE_SPIRV_VALIDATION: validate generated and transformed SPIR-V.
+        // Disabled by default because validation is a diagnostics-only cost.
+        Bool EnableSpirvValidation = false;
+        // MOBILEGL_ESPRYT_USE_ANGLE: load ANGLE EGL/GLES libraries.
+        Bool EsprytUseAngle = false;
 #if defined(MOBILEGL_TRACE_ANGLE_VARIANTS)
         // MOBILEGL_TRACE_ANGLE_VARIANT: signed trace-APK ANGLE build short hash.
         String TraceAngleVariant;
 #endif
-        // MOBILEGL_DISABLE_SUBGROUP: force-disable Vulkan shader subgroup support.
-        Bool DisableSubgroup = false;
+        // MOBILEGL_MAGMA_DISABLE_SUBGROUP: force-disable Vulkan shader subgroup support,
+        // including the opt-in emulated compute path below.
+        Bool MagmaDisableSubgroup = false;
+        // MOBILEGL_MAGMA_EMULATE_SUBGROUP: implement GL_KHR_shader_subgroup's compute
+        // stage on a 32-lane VIRTUAL subgroup lowered to workgroup-shared memory
+        // (ShaderTranspiler::EmulateSubgroupsPass). Strictly a last resort: it only ever
+        // engages when this flag is set AND the device has no native subgroup support at
+        // all - a device with real subgroup operations always uses them natively,
+        // whatever their width (the known iterationRP defect is patched by
+        // MagmaFixIterationRPSubgroupScratch below instead). Off by default.
+        Bool MagmaEmulateSubgroup = false;
+        // MOBILEGL_MAGMA_FIX_ITERATIONRP_SUBGROUP_SCRATCH: patch iterationRP's own bug - the
+        // pack declares `shared vec2 prefixSumCache[32]` for a 512-invocation exposure
+        // reduction and indexes it by gl_SubgroupID, so any device with sub-16-lane
+        // subgroups (8-lane lavapipe -> 64 subgroups) writes shared memory out of
+        // bounds. The pass grows that one array to what the device's topology needs and
+        // touches nothing else; it only rewrites modules positively matching the pack's
+        // reduction fingerprint (ShaderTranspiler::FixIterationRPSubgroupScratchPass),
+        // so every other shader passes through byte-identical - as does iterationRP
+        // itself on >= 16-lane devices. Auto is ON; ForceOff replays the pack's bug
+        // verbatim.
+        QuirkOverride MagmaFixIterationRPSubgroupScratch = QuirkOverride::Auto;
+        // MOBILEGL_MAGMA_ITERATIONRP_FIX_BARRIER: repair Program 203's missing workgroup
+        // rendezvous between its two reductions over prefixSumCache. Off by default and
+        // fingerprint-gated by FixIterationRPBarrierPass when enabled.
+        Bool MagmaIterationRPFixBarrier = false;
+        // MOBILEGL_MAGMA_DERIVE_NUM_SUBGROUPS: replace compute gl_NumSubgroups loads with
+        // ceil(workgroup invocations / gl_SubgroupSize) on the NATIVE subgroup path
+        // (ShaderTranspiler::DeriveNumSubgroupsPass). Auto is ON: GL requires
+        // gl_SubgroupID < gl_NumSubgroups, Adreno's builtin reports 1 while the same
+        // dispatch emits IDs 0..7, and the derived value is the one Vulkan guarantees
+        // whenever the pipeline can request REQUIRE_FULL_SUBGROUPS (which the renderer
+        // does whenever local_size_x is a multiple of the native width). ForceOff returns
+        // to the raw driver builtin.
+        QuirkOverride MagmaDeriveNumSubgroups = QuirkOverride::Auto;
         // MOBILEGL_ADVERTISE_FP64: add GL_ARB_gpu_shader_fp64 to the advertised extension
         // string. `double` in a shader always WORKS - it is narrowed to 32 bits before any
         // module reaches a backend (ShaderTranspiler::DemoteFloat64Pass) - but the extension
@@ -94,16 +132,39 @@ namespace MobileGL::MG_Config {
         Bool MagmaR11G11B10FFallback = false;
         // MOBILEGL_MAGMA_FRAMESINFLIGHT: requested Magma frames in flight, defaulting to 3.
         Uint32 MagmaFramesInFlight = 3;
-        // MOBILEGL_AVOID_SAMPLER_MIPMAP_MIN_FILTER: avoid mipmap min filters in samplers,
+        // MOBILEGL_ESPRYT_AVOID_SAMPLER_MIPMAP_MIN_FILTER: avoid mipmap min filters in samplers,
         // resolves certain rendering bugs on ANGLE + llvmpipe.
-        Bool AvoidSamplerMipmapMinFilter = false;
-        // MOBILEGL_AVOID_EXPLICIT_LOD_BIAS: leave an already-explicit LOD argument alone when
+        Bool EsprytAvoidSamplerMipmapMinFilter = false;
+        // MOBILEGL_ESPRYT_AVOID_EXPLICIT_LOD_BIAS: leave an already-explicit LOD argument alone when
         // emulating GL_TEXTURE_LOD_BIAS, instead of adding the bias uniform to it. Injecting
         // the uniform turns a compile-time-constant LOD into a runtime expression, which
         // sends ANGLE + llvmpipe down a mip-selection path that dereferences a NULL
         // descriptor and kills the process. Deviates from spec (Vulkan adds the bias to
         // OpImageSampleExplicitLod), so it is an avoidance for that stack only.
-        Bool AvoidExplicitLodBias = false;
+        Bool EsprytAvoidExplicitLodBias = false;
+        // MOBILEGL_ESPRYT_UNLOCATED_IO_BLOCKS: emit a tessellation/geometry program's
+        // inter-stage interface blocks WITHOUT their layout(location=) qualifier, letting ES
+        // match them by block name and member sequence instead. The Mali ES driver delivers
+        // nothing at all through a located block once a tessellation or geometry stage is in
+        // the pipeline; the driver POST measures that and turns this on by itself, so Auto is
+        // the right setting everywhere. ForceOn exists so the emulation can be exercised on a
+        // healthy driver - which is what the integration lane does, since llvmpipe and
+        // lavapipe carry a located block correctly and would otherwise never run this code -
+        // and ForceOff is the negative control. See StripIoBlockLocationsPass.
+        QuirkOverride EsprytUnlocatedIoBlocks = QuirkOverride::Auto;
+        // MOBILEGL_POINT_SIZE_DEMOTION: demote gl_PointSize out of tessellation/geometry
+        // stages into an ordinary varying (ShaderCompiler::
+        // DemoteTessellationGeometryPointSizeForProgram) instead of declining such programs
+        // on a device that advertises neither EXT/OES_tessellation_point_size /
+        // geometry_point_size (DirectGLES) nor shaderTessellationAndGeometryPointSize
+        // (DirectVulkan). Auto arms it exactly where the detection says the capability is
+        // absent, which is the right setting everywhere. ForceOn exists so the demotion can
+        // be exercised on a healthy driver - llvmpipe and lavapipe host the built-in
+        // natively and would otherwise never run this code, which is what the pinned
+        // integration lane uses - and ForceOff restores the plain declines (escape hatch /
+        // negative control). Cross-backend by design: the demotion runs in the shared
+        // phase-B chain, so one switch covers both. See DemotePointSizePass.
+        QuirkOverride PointSizeDemotion = QuirkOverride::Auto;
         // MOBILEGL_COHERENT_AS_FLUSH: app-compat for engines (e.g. Flywheel) that write
         // GPU-read data through persistent GL_MAP_FLUSH_EXPLICIT_BIT maps they never
         // flush. Persistent FLUSH_EXPLICIT map requests are rewritten to coherent
@@ -113,10 +174,38 @@ namespace MobileGL::MG_Config {
         Bool CoherentAsFlush = false;
         // MOBILEGL_TRACE_SKIP_AUTODESTROY: skip teardown in the ELF destructor (Init.cpp).
         Bool TraceSkipAutodestroy = false;
-        // MOBILEGL_DISABLE_UBO_RING: force the DirectGLES global-UBO upload back to the
+        // MOBILEGL_ESPRYT_DISABLE_UBO_RING: force the DirectGLES global-UBO upload back to the
         // per-draw glBufferSubData path instead of the persistent-mapped ring allocator
         // (negative control / driver-bug escape hatch).
-        Bool DisableUboRing = false;
+        Bool EsprytDisableUboRing = false;
+        // MOBILEGL_ESPRYT_DISABLE_UNPACK_RING: force DirectGLES texture uploads back to
+        // glTexSubImage from the client pointer instead of staging them through the
+        // persistent-mapped unpack-PBO ring (negative control / driver-bug escape
+        // hatch).
+        Bool EsprytDisableUnpackRing = false;
+        // MOBILEGL_ESPRYT_DISABLE_UPLOAD_RING: force DirectGLES app buffer updates
+        // (glBufferSubData / map flushes) back to the immediate driver upload instead
+        // of queueing them for the staged-copy flush through the persistent-mapped
+        // upload ring (negative control / driver-bug escape hatch; the immediate
+        // upload stalls on drivers that resolve the WAR hazard on the CPU, e.g. Mali).
+        Bool EsprytDisableUploadRing = false;
+        // MOBILEGL_ESPRYT_DISABLE_INVALIDATE_FLUSH: skip the glMapBufferRange(WRITE |
+        // INVALIDATE_RANGE) tier of the DirectGLES pending-range flush and go straight
+        // to the upload ring's staged glCopyBufferSubData (negative control / escape
+        // hatch for a driver whose range-invalidating map misbehaves). The map tier is
+        // what keeps a partial write into a large in-flight buffer priced by the RANGE:
+        // on Mali both the immediate glBufferSubData and a staged copy into a busy
+        // mutable store ghost the whole destination on the CPU.
+        Bool EsprytDisableInvalidateFlush = false;
+        // MOBILEGL_DISABLE_LARGE_BUFFER_ADOPTION: keep mesh-arena-sized buffer stores
+        // (>= 16MiB) on the CPU-shadow model instead of backing them with the backend's
+        // persistently+coherently mapped storage at definition time (negative control /
+        // escape hatch). Frontend-scoped: it engages only where the active backend
+        // provides AcquirePersistentMap. With adoption on, an app SubData into a busy
+        // 128MB arena is a plain memcpy into GPU-visible memory; every driver-mediated
+        // route for the same write stalls the thread or ghost-copies the whole arena on
+        // this class of Mali driver, and the arena stops costing its size again in RAM.
+        Bool DisableLargeBufferAdoption = false;
         // MOBILEGL_ESPRYT_FORCE_DS_READBACK_EMULATION: make DirectGLES skip the native ES
         // depth/stencil reads and always go through the shader-sampling emulation. Core GL
         // ES has no depth or stencil readback, but some drivers accept it anyway (Mesa does,
@@ -130,10 +219,6 @@ namespace MobileGL::MG_Config {
         // explicitly request a core profile via EGL_CONTEXT_OPENGL_PROFILE_MASK / a >=3.1
         // version request.
         Bool RelaxedSemantics = false;
-        // MOBILEGL_QUIRK_SUBGROUP_PREFIX_SCAN: overrides the shader-source quirk that
-        // rewrites the recognized workgroup prefix-scan template on Qualcomm devices with
-        // subgroups wider than 32 lanes (see ShaderSourceProcessor's quirk registry).
-        QuirkOverride SubgroupPrefixScanQuirk = QuirkOverride::Auto;
         // MOBILEGL_MAGMA_DISABLE_BLENDED_DEPTH_WRITE: overrides the DirectVulkan quirk that
         // strips depth writes from accumulation-blended pipelines (MIN/MAX or additive
         // ONE+ONE - the multi-pass depth-equality signature) on drivers without
@@ -141,10 +226,10 @@ namespace MobileGL::MG_Config {
         // gl_FragDepth writers, and fully color-masked attachments are exempt (see
         // PipelineFactory::ShouldSuppressDepthWrite). Auto detects Qualcomm.
         QuirkOverride MagmaDisableBlendedDepthWriteQuirk = QuirkOverride::Auto;
-        // MOBILEGL_DISABLE_ROBUST_BUFFER_ACCESS: leave the Vulkan robustBufferAccess device
+        // MOBILEGL_MAGMA_DISABLE_ROBUST_BUFFER_ACCESS: leave the Vulkan robustBufferAccess device
         // feature off. It is enabled by default to match GL's defined out-of-range fetch
         // behavior; this escape hatch exists to measure or dodge its GPU cost on a device.
-        Bool DisableRobustBufferAccess = false;
+        Bool MagmaDisableRobustBufferAccess = false;
         // MOBILEGL_MAGMA_MULTIDRAW_MODE: preferred DirectVulkan multi-draw dispatch tier
         // ("ext" | "indirect" | "unroll", see MultiDrawMode). Clamped to device support;
         // unset picks the best supported tier.
@@ -176,6 +261,121 @@ namespace MobileGL::MG_Config {
         // immediately stay serial by their own construction). Off by default; never
         // advertise it.
         QuirkOverride AsyncOptimisticShaderStatus = QuirkOverride::Auto;
+        // MOBILEGL_SHADER_CACHE: the three-level, in-memory shader translation memo
+        // (MG_Util/ShaderTranspiler/TranslationCache.h). The levels follow the GL
+        // entry points - L1c memoizes one glCompileShader's PARSE VERDICT, L1 a
+        // linked program's whole front end, L2 DirectGLES's emitted ESSL. Auto is
+        // ON; ForceOff turns ALL THREE off and makes every translation run from
+        // scratch. The escape hatch exists because a wrong cache hit is a silently
+        // miscompiled shader: if a device ever renders differently with the cache
+        // on, one run with this falsy says so.
+        QuirkOverride ShaderTranslationCache = QuirkOverride::Auto;
+        // MOBILEGL_ESPRYT_FORCE_VIEWPORT_ARRAY_EMULATION: DirectGLES' gl_ViewportIndex routing
+        // emulation - the builtin becomes a flat varying, the fragment stage gets a
+        // per-pass gate, and a routed draw is REPLAYED once per distinct viewport state
+        // with the real glViewport/glScissor/glDepthRangef set for it. Auto is ON, and
+        // it is ON even where the driver advertises GL_OES_viewport_array, because that
+        // extension only ever gave the SHADER a compilable name: MobileGL has never
+        // programmed a driver's INDEXED viewport state (SyncRenderState pushes index 0
+        // and nothing else), so on an extension-capable driver every index rasterized as
+        // index 0 exactly as it did without one. ForceOff returns to that behaviour -
+        // the pre-emulation path, extension passthrough where it exists and
+        // LowerViewportIndexPass' demote-to-a-plain-global where it does not - and is
+        // the negative control the emulation is measured against.
+        QuirkOverride EsprytViewportArrayEmulation = QuirkOverride::Auto;
+        // MOBILEGL_ESPRYT_WIDEN_PACKED16_STORAGE: DirectGLES stores GL_RGB565/GL_RGB5(A1)/GL_RGBA4
+        // images as 8-bit-per-channel ES storage (GL_RGB8/GL_RGBA8) instead of the driver's
+        // native 16-bit packed formats. Auto defers to a POST driver-bug probe
+        // (SelfTest::CopyImageMirrorsPacked16FieldOrder): some Mali drivers store SOME
+        // packed16 allocations with a MIRRORED field order (allocation-scoped and
+        // shape/context dependent - the failing 30x30x12 GL_TEXTURE_2D_ARRAYs are mirrored
+        // at every level), so glCopyImageSubData - a raw texel-block move - lands R/G/B/A
+        // reversed whenever exactly one endpoint sits in a mirrored allocation
+        // (KHR-GL4x.copy_image.functional rgb5/rgb5_a1/rgba4 x every *2d_array* pair).
+        // With no 16-bit packed ES image left there is no field order to disagree about; the
+        // client word still round-trips exactly, because the canonical shadow is already
+        // UNorm8 and an n-bit field encodes to UNorm8 and back losslessly for n <= 8.
+        // ForceOn widens on any driver (the llvmpipe suites use it to exercise the widened
+        // path); ForceOff keeps the native narrow storage even where the probe fires - the
+        // negative control that replays the corruption. Costs 2x the memory of the affected
+        // formats where it engages, which is why Auto is probe-gated rather than always-on.
+        QuirkOverride EsprytWidenPacked16Storage = QuirkOverride::Auto;
+        // MOBILEGL_MAGMA_PRIMGEN_QUERY_REROUTE: DirectVulkan's GL_PRIMITIVES_GENERATED
+        // reroute for draws made while transform feedback is INACTIVE. The stream query
+        // (VK_QUERY_TYPE_TRANSFORM_FEEDBACK_STREAM_EXT primitivesNeeded) is defined to count
+        // them, but a Mali driver - and Mesa lavapipe - answers 0 unless a capture span is
+        // open, which is exactly the shape the CTS uses to measure the tessellator, so ~29
+        // tessellation tests per tree size a capture buffer from the 0 and die on the
+        // zero-length map. Auto defers to a device probe at renderer bring-up
+        // (SelfTest::RunPrimitivesGeneratedNoXfbProbe), which measures two substitutes on
+        // the same capture-less draws and arms the best proven one: the dedicated
+        // VK_EXT_primitives_generated_query (exact semantics by definition; lavapipe passes
+        // it, rasterizer discard included), else a clipping-invocations pipeline-statistics
+        // pool (see the verdict vocabulary for its rasterizer-discard split). ForceOn pins
+        // the reroute structurally wherever a pool can exist (the arming-observable lane,
+        // immune to the probe's verdict moving), and ForceOff is the negative control that
+        // replays the driver's silence.
+        QuirkOverride MagmaPrimGenQueryReroute = QuirkOverride::Auto;
+        // --- MGPipe (the disaggregation plan's explicit frontend/backend boundary) ---
+        // MOBILEGL_PIPE_PUSH: per-subsystem bitmask selecting which state the frontend
+        // PUSHES over MGPipe instead of leaving the backend to pull it out of GLContext.
+        // 0 - the default and the only shipped value until the migration lands - is "pull
+        // everything", i.e. exactly today's behaviour. One bit of it also turns OFF
+        // client-side content addressing of CSOs, which is the negative control the CSO
+        // design is measured against. Accepts decimal or 0x-prefixed hex.
+        Uint64 PipePush = 0;
+        // MOBILEGL_PIPE_VERIFY: per-draw, per-FIELD shadow comparison of the pushed state
+        // against a snapshot taken from GLContext the old way, printing the first field
+        // that differs and the draw serial. Roughly 5-10x slower and never shipped; it is
+        // the semantic gate that replaces byte identity, and it catches the dangerous
+        // direction - a dirty bit that fires too RARELY - which no purity gate can see.
+        Bool PipeVerify = false;
+#if MOBILEGL_PIPE_PUSH
+        // The three knobs of the MOBILEGL_PIPE_VERIFY build (P1 brief D2). Compiled only
+        // under MOBILEGL_PIPE_PUSH so the pull build's FeaturesTable does not change size.
+        // MOBILEGL_PIPE_VERIFY_FATAL: the first divergence aborts (default). 0 logs and
+        // counts instead, for triage and for the lane that must survive to read its own
+        // log. Tri-state parse like PipeLegacyMemos: only an explicit falsy value turns it
+        // off.
+        Bool PipeVerifyFatal = true;
+        // MOBILEGL_PIPE_VERIFY_CORRUPT: a field name from kMGPipeInputFieldNames[]; the
+        // comparator perturbs that field in the SNAPSHOT arm before the entry compare, so a
+        // green verify run goes red naming it (negative control A). Unknown name is
+        // Fatal{PipeVerifyBadKnob}.
+        String PipeVerifyCorrupt;
+        // MOBILEGL_PIPE_POISON_OMIT: <Verb>:<FieldName>; the filler skips the STAMP (not
+        // the value) of that field for that verb, an omission indistinguishable from a
+        // forgotten FillPoints.def row, so that verb's read of it is
+        // Fatal{UnmigratedPipeInput} (negative control B). Unknown name is
+        // Fatal{PipeVerifyBadKnob}.
+        String PipePoisonOmit;
+#endif
+        // MOBILEGL_PIPE_STATS: dump the boundary counters (bytes, calls, roundtrips,
+        // texture pulls, upload shapes, residual-block bytes, index mirror bytes).
+        Bool PipeStats = false;
+        // MOBILEGL_PIPE_LEGACY_MEMOS: keep the pre-handle registries and TwinLookupMemos
+        // alive so the first handle waves have a real old-versus-new arm to be compared
+        // against. ON by default for the whole migration window, deleted with the pull
+        // path itself.
+        Bool PipeLegacyMemos = true;
+        // MOBILEGL_PIPE_TEXEL_RETAIN_MB: LRU budget for texels retained against a
+        // server-initiated texture re-send. Default 0, i.e. OFF: MipmapStorage already
+        // holds a complete CPU shadow, so this cache buys latency, never correctness.
+        Uint32 PipeTexelRetainMb = 0;
+        // MOBILEGL_PIPE_INDEX_MIRROR_MB: budget for the server-side index host mirror,
+        // which is what lets primitive-restart rewriting and multi-draw flattening stay on
+        // the server without shipping index bytes per draw. Over budget it degrades to
+        // per-draw staging, counted separately in the stats.
+        Uint32 PipeIndexMirrorMb = 64;
+        // MOBILEGL_PIPE_STATS_PERIOD: frames per boundary-counter summary line. 120 is the
+        // steady-state cadence; the device retrace harness never reaches the teardown dump
+        // and a trimmed fixture (create-indirect) is shorter than 120 frames, so a run that
+        // needs its numbers at all sets this low enough to land at least one window.
+        Uint32 PipeStatsPeriod = 120;
+        // MOBILEGL_PIPE_STATS_FILE: where the boundary counters' teardown JSON dump goes.
+        // Empty (the default) means no dump; the per-120-frame summary line still goes to
+        // the log whenever PipeStats is on, so a device run needs no writable path.
+        String PipeStatsFile;
     };
     extern FeaturesTable Features;
 } // namespace MobileGL::MG_Config

MobileGL/ConfigLoader.cpp

@@ -159,36 +159,108 @@ namespace MobileGL::MG_ConfigLoader {
         return static_cast<Uint32>(parsedValue);
     }
 
+    // Same contract as QueryEnvUint32, over 64 bits and accepting an explicit 0x prefix: the
+    // one consumer is a subsystem BITMASK, and a bitmask written in decimal is unreadable.
+    // Decimal otherwise - never strtoull's base 0, whose "leading zero means octal" rule
+    // silently read MOBILEGL_PIPE_PUSH=010 as 8 - and a '-' anywhere is rejected rather than
+    // wrapped, which strtoull would otherwise do without complaint (-1 -> every bit set).
+    inline Uint64 QueryEnvUint64(const String& key, Uint64 defaultValue) {
+        auto it = acceptedEnvVariablesMap->find(key);
+        if (it == acceptedEnvVariablesMap->end()) {
+            return defaultValue;
+        }
+
+        const String& value = it->second;
+        const char* text = value.c_str();
+        int base = 10;
+        if (value.size() > 2 && text[0] == '0' && (text[1] == 'x' || text[1] == 'X')) {
+            text += 2;
+            base = 16;
+        }
+        char* parseEnd = nullptr;
+        errno = 0;
+        const bool negative = value.find('-') != String::npos;
+        const unsigned long long parsedValue = negative ? 0 : std::strtoull(text, &parseEnd, base);
+        if (negative || parseEnd == text || *parseEnd != '\0' || errno == ERANGE) {
+            MGLOG_W("Config: Ignoring invalid env variable %s='%s'; expected a non-negative integer "
+                    "(decimal, or 0x-prefixed hexadecimal), using default %llu",
+                    key.c_str(), value.c_str(), static_cast<unsigned long long>(defaultValue));
+            return defaultValue;
+        }
+
+        return static_cast<Uint64>(parsedValue);
+    }
+
     inline void InitFeatures() {
         auto& features = MG_Config::Features;
         features.DisableTimerQuery = QueryEnvFlag("MOBILEGL_DISABLE_TIMERQUERY");
-        features.UseAngle = QueryEnvFlag("MOBILEGL_USE_ANGLE");
+        features.EsprytEnableTextureView = QueryEnvFlag("MOBILEGL_ESPRYT_ENABLE_TEXTURE_VIEW");
+        features.EnableSpirvValidation = QueryEnvFlag("MOBILEGL_ENABLE_SPIRV_VALIDATION");
+        features.EsprytUseAngle = QueryEnvFlag("MOBILEGL_ESPRYT_USE_ANGLE");
 #if defined(MOBILEGL_TRACE_ANGLE_VARIANTS)
         QueryEnvVariable("MOBILEGL_TRACE_ANGLE_VARIANT", features.TraceAngleVariant, "");
 #endif
-        features.DisableSubgroup = QueryEnvFlag("MOBILEGL_DISABLE_SUBGROUP");
+        features.MagmaDisableSubgroup = QueryEnvFlag("MOBILEGL_MAGMA_DISABLE_SUBGROUP");
+        features.MagmaEmulateSubgroup = QueryEnvFlag("MOBILEGL_MAGMA_EMULATE_SUBGROUP");
+        features.MagmaFixIterationRPSubgroupScratch =
+            QueryEnvQuirkOverride("MOBILEGL_MAGMA_FIX_ITERATIONRP_SUBGROUP_SCRATCH");
+        features.MagmaIterationRPFixBarrier = QueryEnvFlag("MOBILEGL_MAGMA_ITERATIONRP_FIX_BARRIER");
+        features.MagmaDeriveNumSubgroups = QueryEnvQuirkOverride("MOBILEGL_MAGMA_DERIVE_NUM_SUBGROUPS");
         features.AdvertiseFp64 = QueryEnvFlag("MOBILEGL_ADVERTISE_FP64");
         features.MagmaR11G11B10FFallback = QueryEnvFlag("MOBILEGL_MAGMA_R11G11B10F_FALLBACK");
         features.MagmaFramesInFlight = QueryEnvUint32("MOBILEGL_MAGMA_FRAMESINFLIGHT", 3, 1, 64);
-        features.AvoidSamplerMipmapMinFilter =
-            QueryEnvFlag("MOBILEGL_AVOID_SAMPLER_MIPMAP_MIN_FILTER");
-        features.AvoidExplicitLodBias = QueryEnvFlag("MOBILEGL_AVOID_EXPLICIT_LOD_BIAS");
+        features.EsprytAvoidSamplerMipmapMinFilter =
+            QueryEnvFlag("MOBILEGL_ESPRYT_AVOID_SAMPLER_MIPMAP_MIN_FILTER");
+        features.EsprytAvoidExplicitLodBias = QueryEnvFlag("MOBILEGL_ESPRYT_AVOID_EXPLICIT_LOD_BIAS");
+        features.EsprytUnlocatedIoBlocks = QueryEnvQuirkOverride("MOBILEGL_ESPRYT_UNLOCATED_IO_BLOCKS");
+        features.PointSizeDemotion = QueryEnvQuirkOverride("MOBILEGL_POINT_SIZE_DEMOTION");
         features.CoherentAsFlush = QueryEnvFlag("MOBILEGL_COHERENT_AS_FLUSH");
         features.TraceSkipAutodestroy = QueryEnvFlag("MOBILEGL_TRACE_SKIP_AUTODESTROY");
-        features.DisableUboRing = QueryEnvFlag("MOBILEGL_DISABLE_UBO_RING");
+        features.EsprytDisableUboRing = QueryEnvFlag("MOBILEGL_ESPRYT_DISABLE_UBO_RING");
+        features.EsprytDisableUnpackRing = QueryEnvFlag("MOBILEGL_ESPRYT_DISABLE_UNPACK_RING");
+        features.EsprytDisableUploadRing = QueryEnvFlag("MOBILEGL_ESPRYT_DISABLE_UPLOAD_RING");
+        features.EsprytDisableInvalidateFlush = QueryEnvFlag("MOBILEGL_ESPRYT_DISABLE_INVALIDATE_FLUSH");
+        features.DisableLargeBufferAdoption = QueryEnvFlag("MOBILEGL_DISABLE_LARGE_BUFFER_ADOPTION");
         features.EsprytForceDepthStencilReadbackEmulation =
             QueryEnvFlag("MOBILEGL_ESPRYT_FORCE_DS_READBACK_EMULATION");
         features.RelaxedSemantics = QueryEnvFlag("MOBILEGL_RELAXED_SEMANTICS");
-        features.SubgroupPrefixScanQuirk = QueryEnvQuirkOverride("MOBILEGL_QUIRK_SUBGROUP_PREFIX_SCAN");
         features.MagmaDisableBlendedDepthWriteQuirk =
             QueryEnvQuirkOverride("MOBILEGL_MAGMA_DISABLE_BLENDED_DEPTH_WRITE");
-        features.DisableRobustBufferAccess = QueryEnvFlag("MOBILEGL_DISABLE_ROBUST_BUFFER_ACCESS");
+        features.MagmaDisableRobustBufferAccess = QueryEnvFlag("MOBILEGL_MAGMA_DISABLE_ROBUST_BUFFER_ACCESS");
         features.MagmaMultiDrawMode = QueryEnvMultiDrawMode("MOBILEGL_MAGMA_MULTIDRAW_MODE");
         features.EsprytMultiDrawMode = QueryEnvGLESMultiDrawMode("MOBILEGL_ESPRYT_MULTIDRAW_MODE");
         features.AsyncShaderCompile = QueryEnvQuirkOverride("MOBILEGL_ASYNC_SHADER_COMPILE");
         features.AsyncShaderCompileThreads = QueryEnvUint32("MOBILEGL_ASYNC_SHADER_COMPILE_THREADS", 0, 0, 64);
         features.AsyncOptimisticShaderStatus =
             QueryEnvQuirkOverride("MOBILEGL_ASYNC_OPTIMISTIC_SHADER_STATUS");
+        features.ShaderTranslationCache = QueryEnvQuirkOverride("MOBILEGL_SHADER_CACHE");
+        features.EsprytViewportArrayEmulation =
+            QueryEnvQuirkOverride("MOBILEGL_ESPRYT_FORCE_VIEWPORT_ARRAY_EMULATION");
+        features.EsprytWidenPacked16Storage =
+            QueryEnvQuirkOverride("MOBILEGL_ESPRYT_WIDEN_PACKED16_STORAGE");
+        features.MagmaPrimGenQueryReroute = QueryEnvQuirkOverride("MOBILEGL_MAGMA_PRIMGEN_QUERY_REROUTE");
+        // MGPipe. Nothing here needs adding to an allow-list: InitializeAcceptedEnvVariables
+        // accepts every MOBILEGL_ / LIBGL_ prefixed variable in the environment, so a name
+        // that starts with MOBILEGL_ is visible to these queries by construction.
+        features.PipePush = QueryEnvUint64("MOBILEGL_PIPE_PUSH", 0);
+        features.PipeVerify = QueryEnvFlag("MOBILEGL_PIPE_VERIFY");
+#if MOBILEGL_PIPE_PUSH
+        // Defaults ON: read as a tri-state so only an explicitly falsy value turns it off.
+        features.PipeVerifyFatal =
+            QueryEnvQuirkOverride("MOBILEGL_PIPE_VERIFY_FATAL") != MG_Config::QuirkOverride::ForceOff;
+        QueryEnvVariable("MOBILEGL_PIPE_VERIFY_CORRUPT", features.PipeVerifyCorrupt, "");
+        QueryEnvVariable("MOBILEGL_PIPE_POISON_OMIT", features.PipePoisonOmit, "");
+#endif
+        features.PipeStats = QueryEnvFlag("MOBILEGL_PIPE_STATS");
+        // Defaults ON, so the flag has to be read as a tri-state rather than as a plain
+        // truthy check: unset must keep the memos, and only an explicitly falsy value may
+        // drop them.
+        features.PipeLegacyMemos =
+            QueryEnvQuirkOverride("MOBILEGL_PIPE_LEGACY_MEMOS") != MG_Config::QuirkOverride::ForceOff;
+        features.PipeTexelRetainMb = QueryEnvUint32("MOBILEGL_PIPE_TEXEL_RETAIN_MB", 0, 0, 4096);
+        features.PipeIndexMirrorMb = QueryEnvUint32("MOBILEGL_PIPE_INDEX_MIRROR_MB", 64, 0, 4096);
+        features.PipeStatsPeriod = QueryEnvUint32("MOBILEGL_PIPE_STATS_PERIOD", 120, 1, 1000000);
+        QueryEnvVariable("MOBILEGL_PIPE_STATS_FILE", features.PipeStatsFile, "");
     }
 
     inline void InitBackendType() {

MobileGL/Init.cpp

@@ -15,8 +15,12 @@
 #include <MG_Impl/GLImpl/Texture/ProxyTexture.h>
 #include <MG_Impl/GLImpl/Framebuffer/GL_Framebuffer.h>
 #include <MG_Impl/GLImpl/Sync/GL_Sync.h>
+#include <MG_Impl/GLImpl/Query/GL_Query.h>
 #include <MG_Util/Async/ShaderCompilePool.h>
+#include <MG_Util/Metrics/PipeStats.h>
 #include <MG_Util/ShaderTranspiler/ShaderCompiler.h>
+#include <MG_State/GLState/ProgramState/ProgramTranslationCache.h>
+#include <MG_Util/ShaderTranspiler/TranslationCache.h>
 
 #include <atomic>
 #include <mutex>
@@ -39,6 +43,11 @@ namespace MobileGL {
             if (logLifecycle) {
                 MGLOG_I("MobileGL closing...");
             }
+            // Before any subsystem the counters name goes away, and before the last frame's
+            // numbers can be lost: emits the final summary line and, when
+            // MOBILEGL_PIPE_STATS_FILE is set, the JSON dump. A no-op when the counters are
+            // off, and idempotent.
+            MG_Util::PipeStats::Shutdown();
             // First, before anything else is torn down. In-flight compile/link jobs own
             // their own inputs and are safe against everything below EXCEPT glslang's
             // process globals and the TShader/TProgram objects hanging off pGLContext,
@@ -51,6 +60,11 @@ namespace MobileGL {
             // before a re-initialized library could pair them with the wrong
             // backend's DeleteSync).
             MG_Impl::GLImpl::DestroyAllSyncObjects();
+            // Queries die with their contexts for the same reason, and their registry
+            // is the same shape of process-global map: drain it here too, while the
+            // function table can still pair each backend handle with the backend that
+            // minted it.
+            MG_Impl::GLImpl::DestroyAllQueryObjects();
             MG_Backend::pActiveBackendObject.reset();
             MG_State::pGLContext.reset();
             MG_State::pEGLContext.reset();
@@ -66,6 +80,14 @@ namespace MobileGL {
             // built-in symbol tables the prewarm latch stands for, so leaving it set would
             // make the next Initialize() skip a prewarm it genuinely needs.
             MG_Util::ShaderTranspiler::ShaderCompiler::ResetPrewarmLatch();
+            // The two-level translation memo. Nothing in it references a glslang object -
+            // both levels hold plain bytes - so this is RSS hygiene rather than a lifetime
+            // requirement, and it is safe either side of FinalizeProcess. Stats first: an
+            // fordebug build gets one line per level saying how the run went.
+            MG_Util::ShaderTranspiler::LogShaderTranslationCacheStats();
+            MG_Util::ShaderTranspiler::ClearShaderTranslationCaches();
+            MG_State::GLState::LogProgramTranslationCacheStats();
+            MG_State::GLState::ClearProgramTranslationCache();
             MG_Backend::gBackendFunctionsTable = {};
             g_isInitialized = false;
             if (logLifecycle) {
@@ -86,6 +108,10 @@ namespace MobileGL {
         MGLOG_I("Initializing MobileGL...");
         MG_ConfigLoader::Init();
         MGLOG_I("Config loaded");
+        // Immediately after the config load and before anything can count: the MGPipe
+        // boundary counters latch their enable flag here, so every counting site in the
+        // two backends is a load of an already-settled global for the rest of the run.
+        MG_Util::PipeStats::Init();
         MG_State::Init();
         MGLOG_D("MG_State initialized");
         MG_Backend::Init();

MobileGL/MG_Backend/BackendObject.h

@@ -14,6 +14,7 @@ namespace MobileGL {
     namespace MG_State::GLState {
         class FramebufferObject;
         class ITextureObject;
+        class RenderbufferObject;
     }
 
     enum class BackendType {
@@ -24,6 +25,19 @@ namespace MobileGL {
     };
 
     namespace MG_Backend {
+        // One endpoint of a glCopyImageSubData. GL 4.6 core 18.3.2 accepts GL_RENDERBUFFER
+        // alongside the ten whole-image texture targets, and a renderbuffer name lives in a
+        // namespace of its own - so an endpoint is a sum type, not an ITextureObject. At most
+        // one of the two pointers is set; neither is set when the name named nothing, which is
+        // the INVALID_VALUE the frontend validator reports.
+        struct CopyImageEndpoint {
+            SharedPtr<MG_State::GLState::ITextureObject> Texture;
+            SharedPtr<MG_State::GLState::RenderbufferObject> Renderbuffer;
+
+            Bool IsRenderbuffer() const { return Renderbuffer != nullptr; }
+            Bool Exists() const { return Texture != nullptr || Renderbuffer != nullptr; }
+        };
+
         enum class FormatCapability : Uint64 {
             Creatable = 1ull << 0,
 
@@ -160,9 +174,9 @@ namespace MobileGL {
                                    GLsizei height, GLint border);
             void (*CopyTexSubImage2D)(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint x, GLint y,
                                       GLsizei width, GLsizei height);
-            void (*CopyImageSubData)(const SharedPtr<MG_State::GLState::ITextureObject>& srcTexture,
+            void (*CopyImageSubData)(const CopyImageEndpoint& src,
                                      GLenum srcTarget, GLint srcLevel, GLint srcX, GLint srcY, GLint srcZ,
-                                     const SharedPtr<MG_State::GLState::ITextureObject>& dstTexture,
+                                     const CopyImageEndpoint& dst,
                                      GLenum dstTarget, GLint dstLevel, GLint dstX, GLint dstY, GLint dstZ,
                                      GLsizei srcWidth, GLsizei srcHeight, GLsizei srcDepth);
             void (*GenerateMipmap)(GLenum target);
@@ -178,9 +192,23 @@ namespace MobileGL {
             void (*MemoryBarrierByRegion)(GLbitfield barriers);
             void (*BindImageTexture)(GLuint unit, GLuint texture, GLint level, GLboolean layered, GLint layer,
                                      GLenum access, GLenum format);
+            // The ONLY indexed query that is genuinely a backend one, and only for the pnames
+            // MG_Impl/GLImpl/Getter/GL_Getter.cpp does not already own. Every indexed pname that
+            // names FRONTEND state - the indexed buffer bindings, the per-unit texture/sampler
+            // bindings, the image-unit bindings, the viewport rectangles, the indexed capabilities
+            // - is answered in GL_Getter::GetIntegeri_v and never reaches this entry; the
+            // 64-bit and float/double widths are derived there from the same answer, which is why
+            // no GetInteger64i_v/GetFloati_v/GetDoublei_v table entry exists. In practice this
+            // leaves GL_MAX_COMPUTE_WORK_GROUP_COUNT / _SIZE (also asked directly by
+            // MG_Util/ShaderTranspiler/CompileEnv.cpp) plus whatever pname the frontend has no
+            // case for at all.
             void (*GetIntegeri_v)(GLenum target, GLuint index, GLint* data);
-            void (*GetInteger64i_v)(GLenum target, GLuint index, GLint64* data);
-            void (*GetProgramiv)(GLuint program, GLenum pname, GLint* params);
+            // There is deliberately NO GetProgramiv entry: glGetProgramiv describes the program
+            // the APPLICATION wrote - link status, the transform-feedback mode, the compute local
+            // size - all of which are frontend link artifacts on ProgramObject, and
+            // MG_Impl/GLImpl/Program/GL_Program.cpp answers every one of them from there. Asking a
+            // backend would mean asking about a DIFFERENT program (a SPIRV-Cross-generated ESSL
+            // one, or a SPIR-V module), in a namespace the application never sees.
             // The GL program interface (glGetProgramInterfaceiv / glGetProgramResource*) is NOT
             // a backend query: it describes the program the application wrote, in the
             // application's namespace, which neither backend program is in. It is answered
@@ -236,6 +264,14 @@ namespace MobileGL {
             // (optional; null = frontend falls back to CPU accounting).
             BackendQueryHandle (*BeginXfbPrimitivesQuery)(Bool generated);
             void (*EndXfbPrimitivesQuery)(BackendQueryHandle query);
+            // Whether GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN should be answered from the
+            // frontend's own accounting wherever that accounting is exact - a capture with no
+            // geometry stage - instead of from the query above. Set by DirectGLES, whose result
+            // is whatever the ES driver's PRIMITIVES_WRITTEN counter says: Adreno reports twice
+            // the written count for a vertex-only capture that follows a large render pass,
+            // where the desktop-exact answer is the one the frontend already computed. Defaults
+            // to false, so a backend that never sets it keeps using its GPU result.
+            Bool PrefersCpuXfbPrimitiveAccounting = false;
             // Transform feedback capture spans, for backends whose own GL/ES driver
             // performs the capture (DirectGLES). Both optional; null means the backend
             // drives capture from its draw recording instead (DirectVulkan). End is
@@ -279,6 +315,12 @@ namespace MobileGL {
 
         struct DynamicBackendParameters {
             SizeT UniformBufferOffsetAlignment = 256;
+            // GL_SHADER_STORAGE_BUFFER_OFFSET_ALIGNMENT, which is a SEPARATE limit from the
+            // uniform one and is routinely larger: Adreno 830 reports 32 for uniform buffers and
+            // 64 for storage buffers. Answering the storage query with the uniform value let an
+            // application bind a storage range at an offset the driver cannot address, which it
+            // accepted without error and then wrote somewhere else entirely.
+            SizeT ShaderStorageBufferOffsetAlignment = 256;
             // GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT. 1.0 means the backend cannot filter anisotropically,
             // which is also why the extension is not advertised in that case.
             Float MaxTextureMaxAnisotropy = 1.0f;
@@ -318,8 +360,37 @@ namespace MobileGL {
             Int MaxVertexAttribs = 16;
             Int MaxComputeShaderStorageBlocks = 8;
             Int MaxCombinedShaderStorageBlocks = 32;
+            // Per-stage GL_MAX_*_SHADER_STORAGE_BLOCKS. Zero is a legal answer for the four
+            // non-compute, non-fragment stages and these defaults are the spec minimums, not
+            // placeholders: GL 4.6 table 23.64 and ES 3.2 table 21.44 both set the minimum for
+            // vertex, tessellation control, tessellation evaluation and geometry at 0, and only
+            // fragment (8 in GL, 4 in ES) and compute are guaranteed to have any. Every real ARM
+            // GLES driver takes that allowance - a Mali-G925 reports 0 for all four - so a
+            // backend that cannot honour a graphics-stage storage block MUST report 0 here
+            // rather than a hopeful number. Advertising a non-zero count the driver will refuse
+            // does not make the block work; it only moves the failure from an honest
+            // "unsupported" at query time to a backend link error the frontend never surfaces,
+            // after which every draw with that program silently renders nothing.
+            Int MaxVertexShaderStorageBlocks = 0;
+            Int MaxTessControlShaderStorageBlocks = 0;
+            Int MaxTessEvaluationShaderStorageBlocks = 0;
+            Int MaxGeometryShaderStorageBlocks = 0;
+            Int MaxFragmentShaderStorageBlocks = 8;
             Int MaxComputeUniformBlocks = 12;
             Int MaxComputeWorkGroupInvocations = 128;
+            // GL_MAX_COMPUTE_WORK_GROUP_COUNT / GL_MAX_COMPUTE_WORK_GROUP_SIZE, one value per
+            // axis. These six, with the invocations limit above, are the only indexed limits a
+            // backend genuinely OWNS - the device answers them (glGetIntegeri_v on DirectGLES,
+            // VkPhysicalDeviceLimits::maxComputeWorkGroupCount/Size on DirectVulkan) - and so
+            // the only ones that survive the retirement of the GetIntegeri_v table entry: they
+            // cross the MGPipe boundary inside MGPCaps, by inclusion of this struct (plan B
+            // section 4.4.1). Every other indexed pname names frontend state. RAW driver
+            // answers, like the invocations limit: GL_Getter and the compile environment floor
+            // them at the shared MIN_COMPUTE_WORK_GROUP_* minimums themselves. The defaults are
+            // the GL 4.3 core minimums (table 23.60) and describe the no-backend case, as
+            // MaxClipDistances' does.
+            Int MaxComputeWorkGroupCount[3] = {65535, 65535, 65535};
+            Int MaxComputeWorkGroupSize[3] = {1024, 1024, 64};
             Int MaxShaderStorageBufferBindings = 8;
             Int MaxTextureBufferSize = 65536;
             // GL_TEXTURE_BUFFER_OFFSET_ALIGNMENT; 1 means the offset is unconstrained.
@@ -334,8 +405,45 @@ namespace MobileGL {
             Int MaxComputeImageUniforms = 8;
             Int MaxDrawBuffers = 8;
             Int MaxColorAttachments = 8;
+            // GL_MAX_CLIP_DISTANCES. Zero is a legal answer here, not a placeholder, and a
+            // backend that cannot host a clip distance MUST report it: advertising eight the
+            // backend will refuse does not make gl_ClipDistance work, it only moves the failure
+            // from an honest "unsupported" at query time to a backend shader-compile error the
+            // frontend never surfaces, after which every draw with that program silently renders
+            // nothing. DirectGLES fills it from GL_EXT_clip_cull_distance, DirectVulkan from the
+            // shaderClipDistance device feature. The DEFAULT stays at the GL 4.3 core minimum
+            // because it describes the no-backend case (standalone shader compiles, unit tests),
+            // where there is no device to be honest about and BuildTBuiltInResource still has to
+            // hand glslang a workable gl_MaxClipDistances.
             Int MaxClipDistances = 8;
+            // GL_MAX_CULL_DISTANCES and GL_MAX_COMBINED_CLIP_AND_CULL_DISTANCES, under exactly
+            // the contract stated for MaxClipDistances above: ZERO IS A LEGAL ANSWER and a
+            // backend that cannot host a cull distance MUST report it. The failure this prevents
+            // is worse than the clip one, because cull distance discards the whole primitive:
+            // glslang bounds gl_CullDistance[i] against maxCullDistances and expands
+            // gl_MaxCullDistances from it, SPIRV-Cross then emits
+            // `#extension GL_EXT_clip_cull_distance : require` into the ESSL, and a host driver
+            // without that extension rejects the program in an info log nobody surfaces. These
+            // used to be bare 8s inside BuildTBuiltInResource with no backend consulted at all.
+            // The DEFAULTS are the GL 4.5 core minimums for the same reason MaxClipDistances'
+            // is: they describe the no-backend case (standalone compiles, unit tests).
+            Int MaxCullDistances = 8;
+            Int MaxCombinedClipAndCullDistances = 8;
             Int MaxViewports = 16;
+            // GL_LAYER_PROVOKING_VERTEX / GL_VIEWPORT_INDEX_PROVOKING_VERTEX: which vertex of a
+            // primitive supplies gl_Layer and gl_ViewportIndex. GL 4.6 table 23.65 makes
+            // GL_UNDEFINED_VERTEX a legal answer for both, and it is the honest default - naming
+            // a convention is a statement about behaviour, so a backend that does not pin one
+            // must not claim it does. DirectGLES fills the layer one from the ES 3.2 query and
+            // the viewport one from GL_OES_viewport_array, and leaves UNDEFINED where the
+            // capability is absent: without the viewport array extension only viewport 0 is ever
+            // rasterized, so no convention selects anything. DirectVulkan keeps UNDEFINED for
+            // both - which vertex provokes is decided per pipeline by
+            // VulkanRenderer::SelectProvokingVertexMode out of VK_EXT_provoking_vertex,
+            // provokingVertexModePerPipeline and the topology, so no single convention is true
+            // of the backend.
+            GLenum LayerProvokingVertex = GL_UNDEFINED_VERTEX;
+            GLenum ViewportIndexProvokingVertex = GL_UNDEFINED_VERTEX;
             Int MaxViewportWidth = 16384;
             Int MaxViewportHeight = 16384;
             Float ViewportBoundsRangeMin = 0.0f;
@@ -383,13 +491,55 @@ namespace MobileGL {
                 const Uint32 bit = PerLayerFramebufferAttachmentBit(target);
                 return bit != 0 && (PerLayerFramebufferAttachmentTargets & bit) != 0;
             }
+            // Whether this backend can CONSUME a shader module that still declares 64-bit floats,
+            // i.e. whether `double` survives the transpile instead of being narrowed to `float`
+            // (ShaderTranspiler::DemoteFloat64Pass). Detected, never assumed:
+            //   * DirectVulkan sets it from VkPhysicalDeviceFeatures::shaderFloat64, the feature
+            //     VUID-VkShaderModuleCreateInfo-pCode-08740 requires before a module declaring
+            //     OpCapability Float64 may be created at all. lavapipe has it; Adreno and Mali
+            //     both report VK_FALSE, so no real mobile device does.
+            //   * DirectGLES can NEVER have it. GLSL ES has no 64-bit float type in any version
+            //     or extension, so SPIRV-Cross cannot emit one ("FP64 not supported in ES
+            //     profile") and the demotion there is mathematically mandatory, always.
+            // Defaults to false so a backend that never sets it - and the no-backend case, which
+            // is what standalone shader compiles and the unit tests run under - keeps the
+            // demotion, which is the behaviour that works everywhere.
+            Bool SupportsShaderFloat64 = false;
             // Whether glVertexAttribLFormat / glVertexArrayAttribLFormat can be honoured, i.e.
             // whether a 64-bit vertex attribute can actually reach a shader unconverted. Detected,
             // never assumed: DirectVulkan needs VkPhysicalDeviceFeatures::shaderFloat64 (the
             // attribute travels as its 32-bit word pair, so no VK_FORMAT_R64* is required, but the
             // bitcast result is Float64); DirectGLES can never have it, ESSL having no fp64 type at
             // all. Defaults to false so a backend that never sets it gets the conservative answer.
+            //
+            // INDEPENDENT of SupportsShaderFloat64, and it has to be: this flag decides a VkFormat
+            // from the VAO ATTRIBUTE alone, which does not know what type the shader declared, and
+            // glVertexAttribFormat(GL_DOUBLE) feeding a plain `in vec4` is both legal and common
+            // (KHR-GL43.vertex_attrib_binding.basic-input-case4/5, advanced-bindingUpdate). A
+            // backend with native fp64 that still cannot FETCH 64 bits keeps this false and relies
+            // on the per-MODULE rule in ShaderCompiler::SanitizeAndOptimizeBinary instead: a vertex
+            // module that declares a 64-bit float INPUT is demoted whole, so the two shader-side
+            // halves (PackDoubleVertexInputsPass and VertexInputStateFactory::ToVkVertexFormat)
+            // still see one consistent world.
             Bool SupportsFloat64VertexAttributes = false;
+            // Whether a TESSELLATION stage of this backend may access gl_PointSize - i.e.
+            // whether a module declaring OpCapability TessellationPointSize can reach the
+            // driver at all. DirectVulkan sets both this and the geometry twin from the one
+            // shaderTessellationAndGeometryPointSize feature; DirectGLES sets them
+            // independently from the EXT/OES_tessellation_point_size /
+            // geometry_point_size extension pairs (PointSizeTier), which really do come
+            // separately. When absent, ProgramSpirvTask demotes the built-in to an ordinary
+            // varying program-wide (ShaderCompiler::
+            // DemoteTessellationGeometryPointSizeForProgram); MOBILEGL_POINT_SIZE_DEMOTION
+            // overrides the detection in either direction at backend init.
+            //
+            // Defaults TRUE, deliberately against the house "assume absent" rule: false
+            // ARMS a rewrite, so the conservative no-backend answer (standalone compiles,
+            // unit tests) is the one that leaves modules untouched. A backend that never
+            // sets it gets standard modules and, at worst, the old honest declines.
+            Bool SupportsTessellationPointSize = true;
+            // The geometry-stage twin (OpCapability GeometryPointSize).
+            Bool SupportsGeometryPointSize = true;
             SizeT MaxShaderStorageBlockSize = 128 * 1024 * 1024;
             Uint32 SubgroupSize = 0;
             Uint32 SubgroupSupportedStages = 0;

MobileGL/MG_Backend/DirectGLES/BackendObject_DirectGLES.cpp

@@ -8,6 +8,7 @@
 
 #include "BackendObject_DirectGLES.h"
 #include "MG_Backend/BackendObject.h"
+#include "MG_Backend/BackendObjects.h"
 #include <MG_Backend/DirectGLES/DirectGLES.h>
 #include <MG_Backend/DirectGLES/Managers.h>
 #include <MG_Backend/DirectGLES/Utils.h>
@@ -212,7 +213,10 @@ namespace MobileGL::MG_Backend::DirectGLES {
             if (options & PixelFormatNormalizeOptionBit::NoThreeChannelRenderTarget) {
                 reasons.push_back("no colour-renderable three-channel format on OpenGL ES");
             }
-            if (options & PixelFormatNormalizeOptionBit::NoSnorm16RenderTarget) {
+            // A format is either 8- or 16-bit signed normalized, so at most one of the two ever
+            // survives GetApplicablePixelFormatNormalizeOptions and the reason is not duplicated.
+            if ((options & PixelFormatNormalizeOptionBit::NoSnorm16RenderTarget) ||
+                (options & PixelFormatNormalizeOptionBit::NoSnorm8RenderTarget)) {
                 reasons.push_back("EXT_render_snorm not supported");
             }
 
@@ -303,6 +307,23 @@ namespace MobileGL::MG_Backend::DirectGLES {
             return capabilities.MaxColorTextureSamples;
         }
 
+        // The RENDERBUFFER twin, and it is a different set of pnames on purpose.
+        // GL_MAX_{COLOR,DEPTH}_TEXTURE_SAMPLES bound multisample TEXTURES; a renderbuffer is
+        // bounded by GL_MAX_SAMPLES (GL 4.6 core 9.2.4), with GL_MAX_INTEGER_SAMPLES for the
+        // integer formats. Using the texture ceilings here - which is what the renderbuffer probe
+        // did - is not merely untidy: the two texture pnames are ES 3.1 state, so on an ES 3.0
+        // context the loader's rejected-probe clamp leaves them at 1 (see the multisample clamps
+        // in the GLES loader) and the walk below would never run past one sample, recording {1}
+        // for EVERY colour format while GL_MAX_SAMPLES - ES 3.0 core, so genuinely answered -
+        // reports 4. Once the frontend validates against this list, that would reject every
+        // multisample renderbuffer on such a context.
+        Int GetGLESRenderbufferFormatMaxSamples(const MG_External::GLESCapabilities& capabilities,
+                                                GLenum imageFormat) {
+            const Bool isInteger = imageFormat == GL_RED_INTEGER || imageFormat == GL_RG_INTEGER ||
+                                   imageFormat == GL_RGB_INTEGER || imageFormat == GL_RGBA_INTEGER;
+            return isInteger ? capabilities.MaxIntegerSamples : capabilities.MaxSamples;
+        }
+
         Bool ProbeFramebufferCompletenessForTexture(const MG_External::GLESFunctionsTable& gl, TextureTarget target,
                                                     GLuint texture, TextureInternalFormat format) {
             GLuint framebuffer = 0;
@@ -406,9 +427,12 @@ namespace MobileGL::MG_Backend::DirectGLES {
             return complete;
         }
 
+        // `samples` only reaches the multisample targets; every other target ignores it. The
+        // descending sample walk (ProbeTextureSampleCounts) reuses this whole routine rather than
+        // repeating the gen/bind/completeness/delete dance.
         Bool ProbeTexture(const MG_External::GLESFunctionsTable& gl, TextureTarget target, GLenum internalFormat,
                           GLenum imageFormat, GLenum imageType, TextureInternalFormat logicalFormat,
-                          Bool* outRenderable) {
+                          Bool* outRenderable, Int samples = 1) {
             if (!IsGLESProbeTextureTarget(target) || !gl.glGenTextures || !gl.glBindTexture || !gl.glDeleteTextures) {
                 return false;
             }
@@ -428,10 +452,11 @@ namespace MobileGL::MG_Backend::DirectGLES {
 
             const Bool isMultisample = IsGLESProbeMultisampleTarget(target);
             if (isMultisample) {
+                const auto probeSamples = static_cast<GLsizei>(std::max(samples, 1));
                 if (target == TextureTarget::Texture2DMultisample && gl.glTexStorage2DMultisample) {
-                    gl.glTexStorage2DMultisample(glTarget, 1, internalFormat, 1, 1, GL_TRUE);
+                    gl.glTexStorage2DMultisample(glTarget, probeSamples, internalFormat, 1, 1, GL_TRUE);
                 } else if (target == TextureTarget::Texture2DMultisampleArray && gl.glTexStorage3DMultisample) {
-                    gl.glTexStorage3DMultisample(glTarget, 1, internalFormat, 1, 1, 1, GL_TRUE);
+                    gl.glTexStorage3DMultisample(glTarget, probeSamples, internalFormat, 1, 1, 1, GL_TRUE);
                 } else {
                     gl.glBindTexture(glTarget, static_cast<GLuint>(previousBinding));
                     gl.glDeleteTextures(1, &texture);
@@ -527,6 +552,29 @@ namespace MobileGL::MG_Backend::DirectGLES {
             return sampleCounts;
         }
 
+        // The multisample TEXTURE twin of ProbeRenderbufferSampleCounts. It used to be a
+        // hardcoded {1}, which made glGetInternalformativ(GL_SAMPLES) claim a one-sample maximum
+        // for every format on the multisample targets even where glTexImage2DMultisample happily
+        // accepts four - GL 4.6 core 8.8 makes that query the definition of the maximum, so the
+        // two answers cannot both be right. Completeness is required at every count, exactly as
+        // the renderbuffer walk requires it; the caller only reaches here once the one-sample
+        // probe has already succeeded, so 1 terminates the list without being re-probed.
+        Vector<Int> ProbeTextureSampleCounts(const MG_External::GLESFunctionsTable& gl, TextureTarget target,
+                                             GLenum internalFormat, GLenum imageFormat, GLenum imageType,
+                                             TextureInternalFormat logicalFormat, Int maxSamples) {
+            Vector<Int> sampleCounts;
+            for (Int samples = std::max(maxSamples, 1); samples > 1; samples >>= 1) {
+                Bool renderable = false;
+                const Bool created = ProbeTexture(gl, target, internalFormat, imageFormat, imageType, logicalFormat,
+                                                  &renderable, samples);
+                if (created && renderable) {
+                    sampleCounts.push_back(samples);
+                }
+            }
+            sampleCounts.push_back(1);
+            return sampleCounts;
+        }
+
         void PopulateFormatCapabilitiesImpl(const MG_External::GLESFunctionsTable& gl,
                                             const MG_External::GLESCapabilities& capabilities,
                                             FormatCapabilityCache& cache) {
@@ -627,7 +675,11 @@ namespace MobileGL::MG_Backend::DirectGLES {
                             AddFullFormatCaps(cache, targetIndex, formatIndex,
                                               BuildTextureCapsFromProbe(logicalFormat, target, nativeRenderable));
                             if (IsGLESProbeMultisampleTarget(target)) {
-                                cache.SampleCounts[targetIndex][formatIndex] = {1};
+                                const Int maxSamples =
+                                    GetGLESFormatMaxSamples(capabilities, logicalFormat, nativeInfo.ImageFormat);
+                                cache.SampleCounts[targetIndex][formatIndex] = ProbeTextureSampleCounts(
+                                    gl, probeTarget, nativeInfo.InternalFormat, nativeInfo.ImageFormat,
+                                    nativeInfo.ImageType, logicalFormat, maxSamples);
                             }
                         }
                         shouldProbeFallback = !nativeCreated || !nativeRenderable;
@@ -645,7 +697,11 @@ namespace MobileGL::MG_Backend::DirectGLES {
                                 LogGLESFormatCaveat(logicalFormat, targetIndex, fallbackInfo);
                             }
                             if (IsGLESProbeMultisampleTarget(target)) {
-                                cache.SampleCounts[targetIndex][formatIndex] = {1};
+                                const Int maxSamples =
+                                    GetGLESFormatMaxSamples(capabilities, logicalFormat, fallbackInfo.ImageFormat);
+                                cache.SampleCounts[targetIndex][formatIndex] = ProbeTextureSampleCounts(
+                                    gl, probeTarget, fallbackInfo.InternalFormat, fallbackInfo.ImageFormat,
+                                    fallbackInfo.ImageType, logicalFormat, maxSamples);
                             }
                         }
                     }
@@ -678,7 +734,7 @@ namespace MobileGL::MG_Backend::DirectGLES {
                         AddFullFormatCaps(cache, renderbufferTargetIndex, formatIndex,
                                           GetRenderbufferFeatureCaps(logicalFormat));
                         const Int maxSamples =
-                            GetGLESFormatMaxSamples(capabilities, logicalFormat, nativeInfo.ImageFormat);
+                            GetGLESRenderbufferFormatMaxSamples(capabilities, nativeInfo.ImageFormat);
                         cache.SampleCounts[renderbufferTargetIndex][formatIndex] =
                             ProbeRenderbufferSampleCounts(gl, nativeInfo.InternalFormat, logicalFormat, maxSamples);
                     } else {
@@ -692,7 +748,7 @@ namespace MobileGL::MG_Backend::DirectGLES {
                         LogGLESFormatCaveat(logicalFormat, renderbufferTargetIndex, renderbufferFallbackInfo);
                     }
                     const Int maxSamples =
-                        GetGLESFormatMaxSamples(capabilities, logicalFormat, renderbufferFallbackInfo.ImageFormat);
+                        GetGLESRenderbufferFormatMaxSamples(capabilities, renderbufferFallbackInfo.ImageFormat);
                     cache.SampleCounts[renderbufferTargetIndex][formatIndex] = ProbeRenderbufferSampleCounts(
                         gl, renderbufferFallbackInfo.InternalFormat, logicalFormat, maxSamples);
                 }
@@ -710,11 +766,11 @@ namespace MobileGL::MG_Backend::DirectGLES {
                 .ExtraVendor = Nullopt,              // Extra vendor
                 .RendererGLInfo =
                     {
-                        .TargetGLVersion = {4, 0, 0},   // GL target version
+                        .TargetGLVersion = {4, 6, 0},   // GL target version
                         .TargetGLSLVersion = {4, 6, 0}, // Target Shading Language Version
-                        // Baseline advertisement (no timer queries / anisotropy yet); reconciled
-                        // once the ES capabilities exist, see UpdateAdvertisedCapabilityExtensions.
-                        .Extensions = BuildAdvertisedExtensions(false, false),
+                        // Baseline advertisement (no runtime capabilities yet); reconciled once
+                        // the ES capabilities exist, see UpdateAdvertisedCapabilityExtensions.
+                        .Extensions = BuildAdvertisedExtensions(false, false, false, false, false, false),
                         .IsCompatibilityProfile = false // Is Compatibility Profile
                     },
                 .StaticBackendCapability = {.AllowVSOnlyPrograms = false} // Backend Capability
@@ -734,9 +790,12 @@ namespace MobileGL::MG_Backend::DirectGLES {
         // thread can only observe the extension string after the
         // advertisement for its context has settled; rebuilding the whole
         // list keeps the re-run after a context recreation idempotent.
-        void UpdateAdvertisedCapabilityExtensions(Bool anisotropicFilteringSupported) {
-            MutableRendererInfo().RendererGLInfo.Extensions =
-                BuildAdvertisedExtensions(AreTimerQueriesSupported(), anisotropicFilteringSupported);
+        void UpdateAdvertisedCapabilityExtensions(const MG_External::GLESCapabilities& capabilities) {
+            MutableRendererInfo().RendererGLInfo.Extensions = BuildAdvertisedExtensions(
+                AreTimerQueriesSupported(), capabilities.SupportsTextureFilterAnisotropy,
+                capabilities.SupportsDrawIndirect,
+                capabilities.SupportsDrawIndirect && capabilities.SupportsBaseInstance,
+                capabilities.SupportsTextureView, capabilities.SupportsTextureCubeMapArray);
         }
     } // namespace
 
@@ -745,6 +804,29 @@ namespace MobileGL::MG_Backend::DirectGLES {
         PopulateFormatCapabilitiesImpl(gl, capabilities, cache);
     }
 
+    Int ClampSamplesToBackendSupport(SizeT targetIndex, TextureInternalFormat logicalFormat, GLenum imageFormat,
+                                     Int samples) {
+        if (samples <= 1) {
+            return samples;
+        }
+
+        Int maxSamples = 0;
+        const SizeT formatIndex = static_cast<SizeT>(logicalFormat);
+        if (pActiveBackendObject && targetIndex < kFormatCapabilityTargetCount &&
+            formatIndex < kFormatCapabilityFormatCount) {
+            // Descending, so the head is the largest count this device actually allocated.
+            const Vector<Int>& probedCounts =
+                pActiveBackendObject->GetFormatCapabilities().SampleCounts[targetIndex][formatIndex];
+            if (!probedCounts.empty()) {
+                maxSamples = probedCounts.front();
+            }
+        }
+        if (maxSamples <= 0) {
+            maxSamples = GetGLESFormatMaxSamples(g_GLESCapabilities, logicalFormat, imageFormat);
+        }
+        return std::min(samples, std::max(maxSamples, 1));
+    }
+
     BackendObject_DirectGLES::~BackendObject_DirectGLES() {
         DestroyEGLContext();
     }
@@ -779,11 +861,11 @@ namespace MobileGL::MG_Backend::DirectGLES {
             return false;
         }
         DirectGLES::SetGLESCapabilities(m_GLESCapabilities);
-        // Now that g_GLESCapabilities knows about GL_EXT_disjoint_timer_query and
-        // GL_EXT_texture_filter_anisotropic, reconcile the advertisement (see the comment on
-        // UpdateAdvertisedCapabilityExtensions for why it cannot happen when the extension
-        // list is first built).
-        UpdateAdvertisedCapabilityExtensions(m_GLESCapabilities.SupportsTextureFilterAnisotropy);
+        // Now that g_GLESCapabilities knows the host extensions, entry points, and ES version,
+        // reconcile every runtime-gated advertisement (see the comment on
+        // UpdateAdvertisedCapabilityExtensions for why this cannot happen when the list is first
+        // built).
+        UpdateAdvertisedCapabilityExtensions(m_GLESCapabilities);
         UpdateDynamicBackendParameters();
         PopulateFormatCapabilities(m_GLESFunctions, m_GLESCapabilities, MutableFormatCapabilities());
         PrintFormatCapabilities(GetFormatCapabilities());
@@ -924,11 +1006,20 @@ namespace MobileGL::MG_Backend::DirectGLES {
         return MutableRendererInfo();
     }
 
-    Vector<GLExtension> BuildAdvertisedExtensions(Bool timerQueriesSupported, Bool anisotropicFilteringSupported) {
+    Vector<GLExtension> BuildAdvertisedExtensions(Bool timerQueriesSupported, Bool anisotropicFilteringSupported,
+                                                  Bool drawIndirectSupported,
+                                                  Bool nonZeroIndirectBaseInstanceSupported,
+                                                  Bool textureViewSupported, Bool cubeMapArraySupported) {
         Vector<GLExtension> extensions = {
-            V_OpenGL30, V_OpenGL31, V_OpenGL32, V_OpenGL33, V_OpenGL40, E_GL_ARB_draw_buffers_blend,
+            // The version tokens have to reach the version the backend actually claims:
+            // TargetGLVersion is {4,6,0}, and a list that stopped at OpenGL40 told an
+            // application feature-detecting off these tokens the opposite of what
+            // GL_MAJOR_VERSION / GL_MINOR_VERSION told it.
+            V_OpenGL30, V_OpenGL31, V_OpenGL32, V_OpenGL33, V_OpenGL40, V_OpenGL41, V_OpenGL42, V_OpenGL43,
+            V_OpenGL44, V_OpenGL45, V_OpenGL46,
+            E_GL_ARB_draw_buffers_blend,
             E_GL_ARB_compute_shader, E_GL_ARB_shader_storage_buffer_object, E_GL_ARB_shader_image_load_store,
-            E_GL_ARB_program_interface_query, E_GL_ARB_framebuffer_object, E_GL_EXT_framebuffer_object,
+            E_GL_ARB_clear_buffer_object, E_GL_ARB_program_interface_query, E_GL_ARB_framebuffer_object, E_GL_EXT_framebuffer_object,
             E_GL_ARB_depth_texture, E_GL_ARB_buffer_storage, E_GL_ARB_texture_storage,
             E_GL_ARB_texture_storage_multisample, E_GL_ARB_clear_texture, E_GL_ARB_direct_state_access,
             E_GL_ARB_multi_draw_indirect, E_GL_ARB_indirect_parameters, E_GL_ARB_shader_draw_parameters,
@@ -951,10 +1042,94 @@ namespace MobileGL::MG_Backend::DirectGLES {
             // has had the same texture parameter since ES 3.1, which every device MobileGL
             // runs on provides.
             E_GL_ARB_stencil_texturing,
+            // Core since 3.2 and implemented here on both backends - glDrawElementsBaseVertex,
+            // glDrawRangeElementsBaseVertex, glDrawElementsInstancedBaseVertex and
+            // glMultiDrawElementsBaseVertex all reach real per-draw vertex rebasing. The string
+            // was simply never emitted, which left KHR-GL4*.draw_elements_base_vertex_tests
+            // NotSupported on a feature that works.
+            E_GL_ARB_draw_elements_base_vertex,
+            // The whole sync-object family is real and core since 3.2: glFenceSync, glIsSync,
+            // glDeleteSync, glClientWaitSync, glWaitSync and glGetSynciv all live in GLImpl over a
+            // backend fence (a host GLsync here, a VkFence on DirectVulkan), and glGetInteger64v
+            // answers GL_MAX_SERVER_WAIT_TIMEOUT. The string matters for the same reason
+            // ARB_uniform_buffer_object's does: LWJGL builds GLCapabilities from the extension
+            // list, and a caller that finds GL_ARB_sync missing never resolves the entry points -
+            // then calls through null if it uses fences anyway. Nothing in the CTS gates on this
+            // string, so it is advertised on the strength of the implementation, not a test unlock.
+            E_GL_ARB_sync,
+            // Atomic counters, core since 4.2. glGetActiveAtomicCounterBufferiv and the whole
+            // GL_ATOMIC_COUNTER_BUFFER_* query family are real in GLImpl, and SyncAtomicCounterBuffers
+            // re-issues the counter buffer as an SSBO binding in the range reserved at the top of
+            // the ES driver's shader-storage points, so a counter dispatch reads and writes the
+            // buffer the application bound. DirectVulkan reaches the same place through its own
+            // descriptor resolution, so the string is symmetric.
+            E_GL_ARB_shader_atomic_counters,
+            // glVertexAttribDivisor, core since 3.3 and real on both backends. Applications
+            // (Better Clouds' GLCompat among them) accept the extension string as an
+            // ALTERNATIVE to a 3.3 context when deciding whether instanced rendering is
+            // available, so withholding it makes MobileGL look less capable than it is.
+            E_GL_ARB_instanced_arrays,
+            // The whole of KHR_debug lives in GLImpl - the message log, the group stack and the
+            // object-label table are MobileGL's own state, not the host driver's - so it is as
+            // available here as it is on DirectVulkan, which has advertised it all along.
+            E_GL_KHR_debug,
+            // Core GL 3.0-4.3 plumbing that has been real here for as long as the backend has
+            // existed, and that was simply never named. None of these unlocks a single CTS case -
+            // the conformance suite reaches all of them through the version - so they are
+            // advertised for the OTHER consumer of this list: LWJGL builds GLCapabilities from the
+            // string set, and an application that gates its ENTRY POINTS on the string rather than
+            // on the version never resolves them and then calls through null. Each is backed by
+            // the entry points named beside it.
+            //
+            // glBindVertexArray / glGenVertexArrays / glDeleteVertexArrays / glIsVertexArray.
+            E_GL_ARB_vertex_array_object,
+            // The 14 glSamplerParameter* / glGetSamplerParameter* entry points, including the
+            // integer-valued Iiv/Iuiv forms.
+            E_GL_ARB_sampler_objects,
+            // glMapBufferRange + glFlushMappedBufferRange, which ARB_buffer_storage's persistent
+            // maps are already built on top of.
+            E_GL_ARB_map_buffer_range,
+            // glCopyBufferSubData plus the GL_COPY_READ_BUFFER / GL_COPY_WRITE_BUFFER targets.
+            E_GL_ARB_copy_buffer,
+            // glCopyImageSubData, wired to a real backend hook on both backends.
+            E_GL_ARB_copy_image,
+            // GL_TEXTURE_SWIZZLE_{R,G,B,A,RGBA}, which this backend syncs through to the ES
+            // driver's identical parameters.
+            E_GL_ARB_texture_swizzle,
+            // GL_INT_2_10_10_10_REV / GL_UNSIGNED_INT_2_10_10_10_REV on glVertexAttribPointer plus
+            // the eight glVertexAttribP* entry points.
+            E_GL_ARB_vertex_type_2_10_10_10_rev,
+            // The R/RG internal formats. Named separately from the float ones because an
+            // application may check either.
+            E_GL_ARB_texture_rg,
+            // GL_DEPTH_COMPONENT32F and GL_DEPTH32F_STENCIL8.
+            E_GL_ARB_depth_buffer_float,
+            // The floating-point colour formats. Unlike the rest of this block this string DOES
+            // gate CTS cases - KHR-GL4*.internalformat.texture2d.*{16f,32f} is keyed on it with no
+            // core-version fallback, so eight cases per version list were NotSupported on formats
+            // the backend has always had.
+            E_GL_ARB_texture_float,
+            // glViewportArrayv / glViewportIndexedf{,v} / glScissorArrayv / glScissorIndexed{,v} /
+            // glDepthRangeArrayv / glDepthRangeIndexed / glGetFloati_v / glGetDoublei_v, over the
+            // 16 viewports GL_MAX_VIEWPORTS reports and the per-viewport routing emulation.
+            E_GL_ARB_viewport_array,
             // Advertised with GL_NUM_PROGRAM_BINARY_FORMATS = 0, which the
             // extension explicitly permits. It is also the only thing that
             // exposes glProgramParameteri before GL 4.1.
             E_GL_ARB_get_program_binary};
+        // Minecraft 26.3 checks this prerequisite before it even considers
+        // GL_ARB_multi_draw_indirect. ES 3.1 supplies both single-draw entry points; the loader
+        // folds the version and pointer checks into SupportsDrawIndirect.
+        if (drawIndirectSupported) {
+            extensions.push_back(E_GL_ARB_draw_indirect);
+        }
+        // ARB_base_instance also defines the last word of an indirect command. Direct calls are
+        // emulated on every Espryt device, but without host GL_EXT_base_instance a native indirect
+        // draw cannot shift divisor attributes by a GPU-authored non-zero value, so do not promise
+        // that incomplete case.
+        if (drawIndirectSupported && nonZeroIndirectBaseInstanceSupported) {
+            extensions.push_back(E_GL_ARB_base_instance);
+        }
         // GL_KHR_parallel_shader_compile is MobileGL's own capability, not the host ES
         // driver's: the compiler threads are MobileGL's, and glCompileShader/glLinkProgram
         // are serviced entirely inside the frontend. Whether the device driver advertises
@@ -986,6 +1161,47 @@ namespace MobileGL::MG_Backend::DirectGLES {
         if (timerQueriesSupported && !MG_Config::Features.DisableTimerQuery) {
             extensions.push_back(E_GL_ARB_timer_query);
         }
+        // Cube map arrays are core from GL 4.0 and from ES 3.2, but on a pre-ES-3.2 driver without
+        // EXT/OES_texture_cube_map_array there is nothing underneath: the texture gets no storage
+        // and a samplerCubeArray shader does not even compile, which is exactly what the POST
+        // reports. So the string follows the host capability rather than the version.
+        //
+        // Named for the application's benefit rather than the suite's: measured on Adreno 830,
+        // KHR-GL43.texture_gather.plain-gather-*-cube-array already passed without the string, so
+        // this unlocks no conformance case. It is advertised because the feature is real and
+        // because an application that feature-detects cube map arrays off the string (rather than
+        // off the 4.0 version) would otherwise decline a path this backend serves.
+        if (cubeMapArraySupported) {
+            extensions.push_back(E_GL_ARB_texture_cube_map_array);
+        }
+        // Only advertised when the host ES driver has EXT/OES_texture_view. ES has no core
+        // texture views at any version and no honest emulation exists: a view is a SECOND NAME
+        // over the SAME storage, so that writes through either are visible through the other and
+        // the two carry independent per-texture parameters at the same time - which is exactly
+        // what applications use it for (Better Clouds samples one D24S8 through its own name with
+        // DEPTH_STENCIL_TEXTURE_MODE = STENCIL_INDEX and through a view with DEPTH_COMPONENT, in
+        // a single shading pass). A copy-based fallback satisfies neither half, and fails
+        // silently; withholding the string and answering glTextureView with INVALID_OPERATION is
+        // the only behaviour that cannot be mistaken for success.
+        //
+        // The host extension is necessary and NOT sufficient, which is why this second gate
+        // exists. Adreno 830 has EXT_texture_view, and on it the whole functional half of
+        // KHR-GL4{2,3}.texture_view fails: base_and_max_levels, reference_counting and
+        // view_sampling Fail and view_classes crashes, while only the two pure-API cases
+        // (errors, gettexparameter - neither of which touches the host view) pass. The cause is
+        // known and is MobileGL's, not the driver's: SyncTextureViewToBackend normalizes the
+        // VIEW's ES internalformat independently of the storage it aliases, so whenever the two
+        // land on different renderability carriers the host rejects the pair, the error is
+        // swallowed, and the view is left as a storage-less name that samples as zeros.
+        // DirectVulkan builds the view as a second VkImageView over one VkImage and has no such
+        // seam - it passes 5 of the 7 cases on the same device - so the string stays there.
+        //
+        // Until that reconciliation exists, advertising here would be the same lie the comment
+        // above refuses to tell, just with an extra prerequisite met. Set
+        // MOBILEGL_ESPRYT_ENABLE_TEXTURE_VIEW=1 to re-enable it for that work.
+        if (textureViewSupported && MG_Config::Features.EsprytEnableTextureView) {
+            extensions.push_back(E_GL_ARB_texture_view);
+        }
         // Only advertised when the host ES driver actually filters anisotropically: the sampler
         // state is accepted regardless, but forwarding it would be a no-op without the extension,
         // and an app that trusts the string (LWJGL builds GLCapabilities from it) would silently
@@ -1039,8 +1255,6 @@ namespace MobileGL::MG_Backend::DirectGLES {
             funcsTable.GL.MemoryBarrierByRegion = MemoryBarrierByRegion;
             funcsTable.GL.BindImageTexture = BindImageTexture;
             funcsTable.GL.GetIntegeri_v = GetIntegeri_v;
-            funcsTable.GL.GetInteger64i_v = GetInteger64i_v;
-            funcsTable.GL.GetProgramiv = GetProgramiv;
             funcsTable.GL.ShaderStorageBlockBinding = ShaderStorageBlockBinding;
             funcsTable.GL.Clear = Clear;
             funcsTable.GL.ClearBufferfi = ClearBufferfi;
@@ -1090,6 +1304,12 @@ namespace MobileGL::MG_Backend::DirectGLES {
             // geometry shader's amplification.
             funcsTable.GL.BeginXfbPrimitivesQuery = BeginXfbPrimitivesQuery;
             funcsTable.GL.EndXfbPrimitivesQuery = EndXfbPrimitivesQuery;
+            // ...but where it CAN see the whole capture - no geometry stage - the frontend's
+            // own count is the desktop-exact one and the ES driver's is only as good as the
+            // vendor made it (Adreno doubles PRIMITIVES_WRITTEN for a vertex-only capture that
+            // follows a large render pass). The query above stays installed: it is still what
+            // answers an amplifying span, and PRIMITIVES_GENERATED always.
+            funcsTable.GL.PrefersCpuXfbPrimitiveAccounting = true;
             funcsTable.GL.IsQueryResultAvailable = IsQueryResultAvailable;
             funcsTable.GL.GetQueryResult64 = GetQueryResult64;
             funcsTable.GL.DeleteBackendQuery = DeleteBackendQuery;
@@ -1119,6 +1339,8 @@ namespace MobileGL::MG_Backend::DirectGLES {
 
     void BackendObject_DirectGLES::UpdateDynamicBackendParameters() {
         m_dynamicParameters.UniformBufferOffsetAlignment = m_GLESCapabilities.UniformBufferOffsetAlignment;
+        m_dynamicParameters.ShaderStorageBufferOffsetAlignment =
+            m_GLESCapabilities.ShaderStorageBufferOffsetAlignment;
         m_dynamicParameters.MaxTextureMaxAnisotropy = m_GLESCapabilities.MaxTextureMaxAnisotropy;
         m_dynamicParameters.AliasedLineWidthRangeMin = m_GLESCapabilities.AliasedLineWidthRangeMin;
         m_dynamicParameters.AliasedLineWidthRangeMax = m_GLESCapabilities.AliasedLineWidthRangeMax;
@@ -1169,9 +1391,38 @@ namespace MobileGL::MG_Backend::DirectGLES {
                      static_cast<Int>(MG_State::GLState::VertexArrayObject::MAX_VERTEX_ATTRIBS));
         m_dynamicParameters.MaxComputeShaderStorageBlocks = m_GLESCapabilities.MaxComputeShaderStorageBlocks;
         m_dynamicParameters.MaxCombinedShaderStorageBlocks = m_GLESCapabilities.MaxCombinedShaderStorageBlocks;
+        // Per-stage storage-block counts, forwarded from the host driver rather than invented.
+        // A stage the driver cannot serve reports 0, which is a legal answer everywhere these
+        // limits appear (GL 4.6 table 23.64, ES 3.2 table 21.44 - the minimum is 0 for every
+        // graphics stage except fragment) and is the only answer that lets an application take
+        // its own fallback instead of building a program the driver will refuse to link. The
+        // stage limit cannot exceed the combined limit or the number of binding points there
+        // are to bind buffers to, so clamp to both.
+        const auto clampStageStorageBlocks = [this](Int stageLimit) {
+            return std::min({std::max(stageLimit, 0), std::max(m_dynamicParameters.MaxCombinedShaderStorageBlocks, 0),
+                             std::max(m_dynamicParameters.MaxShaderStorageBufferBindings, 0)});
+        };
+        m_dynamicParameters.MaxShaderStorageBufferBindings = m_GLESCapabilities.MaxShaderStorageBufferBindings;
+        m_dynamicParameters.MaxVertexShaderStorageBlocks =
+            clampStageStorageBlocks(m_GLESCapabilities.MaxVertexShaderStorageBlocks);
+        m_dynamicParameters.MaxTessControlShaderStorageBlocks =
+            clampStageStorageBlocks(m_GLESCapabilities.MaxTessControlShaderStorageBlocks);
+        m_dynamicParameters.MaxTessEvaluationShaderStorageBlocks =
+            clampStageStorageBlocks(m_GLESCapabilities.MaxTessEvaluationShaderStorageBlocks);
+        m_dynamicParameters.MaxGeometryShaderStorageBlocks =
+            clampStageStorageBlocks(m_GLESCapabilities.MaxGeometryShaderStorageBlocks);
+        m_dynamicParameters.MaxFragmentShaderStorageBlocks =
+            clampStageStorageBlocks(m_GLESCapabilities.MaxFragmentShaderStorageBlocks);
         m_dynamicParameters.MaxComputeUniformBlocks = m_GLESCapabilities.MaxComputeUniformBlocks;
         m_dynamicParameters.MaxComputeWorkGroupInvocations = m_GLESCapabilities.MaxComputeWorkGroupInvocations;
-        m_dynamicParameters.MaxShaderStorageBufferBindings = m_GLESCapabilities.MaxShaderStorageBufferBindings;
+        // The six per-axis compute limits: the driver's raw glGetIntegeri_v answers, the same
+        // numbers GLFunctionsTable::GetIntegeri_v forwards live. Carried here so that MGPCaps has
+        // them once the table entry retires (plan B section 4.4.1); GL_Getter floors them.
+        for (SizeT axis = 0; axis < 3; ++axis) {
+            m_dynamicParameters.MaxComputeWorkGroupCount[axis] = m_GLESCapabilities.MaxComputeWorkGroupCount[axis];
+            m_dynamicParameters.MaxComputeWorkGroupSize[axis] = m_GLESCapabilities.MaxComputeWorkGroupSize[axis];
+        }
+        // (MaxShaderStorageBufferBindings is assigned above, before the per-stage clamp reads it.)
         // This is the number glGetIntegerv(GL_MAX_TEXTURE_BUFFER_SIZE) hands the application, and
         // on a host without buffer textures it is knowingly a floor MobileGL cannot honour rather
         // than a driver answer (m_GLESCapabilities.MaxTextureBufferSizeIsDriverReported says
@@ -1224,14 +1475,61 @@ namespace MobileGL::MG_Backend::DirectGLES {
                     DynParams::PerLayerFramebufferAttachmentBit(TextureTarget::TextureCubeMapArray);
             }
         }
-        // Not a driver question and never will be: OpenGL ES has no double-precision vertex format
-        // and ESSL has no fp64 type to consume one with, so a 64-bit vertex attribute has nowhere to
-        // land on this backend regardless of what the driver underneath happens to support.
+        // Not a driver question and never will be: GLSL ES has no 64-bit float type in ANY version
+        // or extension, so SPIRV-Cross cannot emit one ("FP64 not supported in ES profile") and a
+        // module that still declared Float64 would never reach the driver at all. The demotion is
+        // mathematically mandatory here, on every device, forever - which is why this stays false
+        // regardless of what the driver underneath happens to support.
+        m_dynamicParameters.SupportsShaderFloat64 = false;
+        // Follows the line above, and must: OpenGL ES has no double-precision vertex format and no
+        // fp64 type to consume one with, so a 64-bit vertex attribute has nowhere to land here.
         m_dynamicParameters.SupportsFloat64VertexAttributes = false;
+        // Whether a tessellation / geometry stage's ESSL may name gl_PointSize at all: the two
+        // extension pairs the loader probed, independently, because they really do come
+        // separately. False arms the shared phase-B demotion
+        // (ShaderCompiler::DemoteTessellationGeometryPointSizeForProgram), whose ESSL then
+        // never names the built-in in those stages and needs no extension.
+        // MOBILEGL_POINT_SIZE_DEMOTION=1 pretends both are absent so the demotion can be
+        // exercised on a healthy driver (the pinned integration lane); =0 restores the
+        // detected answer's declines.
+        m_dynamicParameters.SupportsTessellationPointSize =
+            m_GLESCapabilities.TessellationPointSizeSupport !=
+            MG_External::GLESCapabilities::PointSizeTier::None;
+        m_dynamicParameters.SupportsGeometryPointSize =
+            m_GLESCapabilities.GeometryPointSizeSupport !=
+            MG_External::GLESCapabilities::PointSizeTier::None;
+        switch (MG_Config::Features.PointSizeDemotion) {
+        case MG_Config::QuirkOverride::ForceOn:
+            MGLOG_I("DirectGLES: MOBILEGL_POINT_SIZE_DEMOTION=1 - treating tessellation/geometry "
+                    "gl_PointSize as unhosted so the demotion runs on this driver");
+            m_dynamicParameters.SupportsTessellationPointSize = false;
+            m_dynamicParameters.SupportsGeometryPointSize = false;
+            break;
+        case MG_Config::QuirkOverride::ForceOff:
+            MGLOG_I("DirectGLES: MOBILEGL_POINT_SIZE_DEMOTION=0 - keeping the built-in and the "
+                    "plain declines regardless of the driver's extensions");
+            m_dynamicParameters.SupportsTessellationPointSize = true;
+            m_dynamicParameters.SupportsGeometryPointSize = true;
+            break;
+        case MG_Config::QuirkOverride::Auto:
+            break;
+        }
         m_dynamicParameters.MaxDrawBuffers = m_GLESCapabilities.MaxDrawBuffers;
         m_dynamicParameters.MaxColorAttachments = m_GLESCapabilities.MaxColorAttachments;
         m_dynamicParameters.MaxClipDistances = m_GLESCapabilities.MaxClipDistances;
+        // The loader already gated both on GL_EXT_clip_cull_distance and left 0 without it, which
+        // is the answer that keeps glslang from accepting a gl_CullDistance the ESSL compiler
+        // would reject.
+        m_dynamicParameters.MaxCullDistances = m_GLESCapabilities.MaxCullDistances;
+        m_dynamicParameters.MaxCombinedClipAndCullDistances = m_GLESCapabilities.MaxCombinedClipAndCullDistances;
         m_dynamicParameters.MaxViewports = m_GLESCapabilities.MaxViewports;
+        // Whatever the driver said about which vertex supplies gl_Layer, and GL_UNDEFINED_VERTEX
+        // for gl_ViewportIndex on every driver without GL_OES_viewport_array - which is both test
+        // devices. That is not a shortfall being hidden: without the extension only viewport 0 is
+        // ever rasterized, so no vertex "selects" a viewport index and naming a convention would
+        // describe behaviour this backend does not implement.
+        m_dynamicParameters.LayerProvokingVertex = m_GLESCapabilities.LayerProvokingVertex;
+        m_dynamicParameters.ViewportIndexProvokingVertex = m_GLESCapabilities.ViewportIndexProvokingVertex;
         m_dynamicParameters.MaxViewportWidth = m_GLESCapabilities.MaxViewportWidth;
         m_dynamicParameters.MaxViewportHeight = m_GLESCapabilities.MaxViewportHeight;
         m_dynamicParameters.ViewportBoundsRangeMin = m_GLESCapabilities.ViewportBoundsRangeMin;

MobileGL/MG_Backend/DirectGLES/BackendObject_DirectGLES.h

@@ -18,6 +18,16 @@ namespace MobileGL::MG_Backend::DirectGLES {
                                     const MG_External::GLESCapabilities& capabilities,
                                     FormatCapabilityCache& cache);
 
+    // Clamps a requested sample count down to what the ES driver can really deliver for this
+    // format on this format-capability target: the probed per-format list when there is one, the
+    // driver's per-class GL_MAX_*_SAMPLES otherwise. The frontend deliberately validates against
+    // the count MobileGL advertises instead (GL_Getter's GetAdvertisedMaxSamples), which on a
+    // driver reporting GL_MAX_INTEGER_SAMPLES 1 is higher than the driver accepts, so every ES
+    // allocation call has to come through here. The shadow state keeps the requested count, so
+    // GL_TEXTURE_SAMPLES and framebuffer completeness still answer what the application asked for.
+    Int ClampSamplesToBackendSupport(SizeT targetIndex, TextureInternalFormat logicalFormat, GLenum imageFormat,
+                                     Int samples);
+
     class BackendObject_DirectGLES : public BackendObject {
     public:
         ~BackendObject_DirectGLES() override;
@@ -67,9 +77,13 @@ namespace MobileGL::MG_Backend::DirectGLES {
     const RendererInfo& GetRendererIdentity();
 
     // The full OpenGL extension list Espryt advertises (glGetString(GL_EXTENSIONS))
-    // for a device whose timer queries / anisotropic filtering are (or are not) usable.
+    // for a device whose timer queries / anisotropic filtering / native indirect draws /
+    // non-zero indirect baseInstance semantics / EXT-OES texture views are (or are not) usable.
     // The MOBILEGL_DISABLE_TIMERQUERY escape hatch is applied inside.
-    Vector<GLExtension> BuildAdvertisedExtensions(Bool timerQueriesSupported, Bool anisotropicFilteringSupported);
+    Vector<GLExtension> BuildAdvertisedExtensions(Bool timerQueriesSupported, Bool anisotropicFilteringSupported,
+                                                  Bool drawIndirectSupported,
+                                                  Bool nonZeroIndirectBaseInstanceSupported,
+                                                  Bool textureViewSupported, Bool cubeMapArraySupported);
 
     // Format: <OpenGL ES Renderer>, OpenGL ES <Major>.<Minor> — the exact string an
     // initialized backend returns from GetBackendAPIVersionString (and that ends up

MobileGL/MG_Backend/DirectGLES/DirectGLES.h

@@ -76,9 +76,9 @@ namespace MobileGL::MG_Backend::DirectGLES {
                         GLsizei height, GLint border);
     void CopyTexSubImage2D(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint x, GLint y, GLsizei width,
                            GLsizei height);
-    void CopyImageSubData(const SharedPtr<MG_State::GLState::ITextureObject>& srcTexture,
+    void CopyImageSubData(const CopyImageEndpoint& src,
                           GLenum srcTarget, GLint srcLevel, GLint srcX, GLint srcY, GLint srcZ,
-                          const SharedPtr<MG_State::GLState::ITextureObject>& dstTexture,
+                          const CopyImageEndpoint& dst,
                           GLenum dstTarget, GLint dstLevel, GLint dstX, GLint dstY, GLint dstZ,
                           GLsizei srcWidth, GLsizei srcHeight, GLsizei srcDepth);
     void GenerateMipmap(GLenum target);
@@ -92,8 +92,6 @@ namespace MobileGL::MG_Backend::DirectGLES {
     void BindImageTexture(GLuint unit, GLuint texture, GLint level, GLboolean layered, GLint layer, GLenum access,
                           GLenum format);
     void GetIntegeri_v(GLenum target, GLuint index, GLint* data);
-    void GetInteger64i_v(GLenum target, GLuint index, GLint64* data);
-    void GetProgramiv(GLuint program, GLenum pname, GLint* params);
     void ShaderStorageBlockBinding(GLuint program, const GLchar* storageBlockName, GLuint storageBlockBinding);
     Bool InitWindowSurface(NativeWindowType window);
     Bool InitPbufferSurface(EGLint width, EGLint height);

MobileGL/MG_Backend/DirectGLES/Managers.h

@@ -21,6 +21,29 @@ namespace MobileGL::MG_Backend::DirectGLES {
     String EmulateBaseInstanceInVertexShader(String source, GLenum shaderType);
     String PromoteDrawParameterGlobalsToUniforms(String source, GLenum shaderType);
 
+    // The ESSL half of the gl_ViewportIndex routing emulation, in the order a program's stages
+    // meet it. Both are pure String -> String rewrites over what SPIRV-Cross emitted once
+    // LowerViewportIndexPass has demoted the builtin to the plain global `mg_ViewportIndex`.
+    //
+    // The producing stage's global becomes an ordinary flat varying; true when there was one to
+    // promote, which is also the answer to "does this program route viewports at all".
+    Bool PromoteViewportIndexGlobalToVarying(String& source);
+    // The fragment stage grows a matching flat input, the mg_ViewportPassMask uniform the draw
+    // path writes, and a wrapper entry point that discards every fragment whose primitive routed
+    // to an index the current replay pass is not drawing. False when the stage has no entry point
+    // to wrap, which leaves the program renderable but unrouted.
+    Bool InjectViewportIndexPassGate(String& source);
+
+    // Whether a vertex shader may declare a storage block at all, given what the host driver
+    // reports for GL_MAX_VERTEX_SHADER_STORAGE_BLOCKS. Pure, and separated from the capability
+    // global purely so the decision can be tested without one.
+    //
+    // The indirect half of the gl_BaseInstance lowering in PromoteDrawParameterGlobalsToUniforms
+    // is the only thing that needs this, and it needs exactly one block. A driver reporting 0 is
+    // conformant - the minimum is 0 in GL 4.6 table 23.64 and ES 3.2 table 21.44 - and ARM's
+    // GLES driver does report 0, so this is a live path, not a defensive one.
+    Bool VertexStageStorageBlockUsable(Int maxVertexShaderStorageBlocks);
+
     // True once the process has entered exit(): past that point the EGL library and
     // the driver may already be unloaded, so a backend twin's destructor must not
     // call into g_GLESFuncs (the observed crash is a jump through an unmapped driver
@@ -79,9 +102,98 @@ namespace MobileGL::MG_Backend::DirectGLES {
     // Brings the whole draw-relevant frontend state onto the native ES context and binds
     // the program; every GL draw entry point calls it exactly once before issuing draws.
     void PrepareForDraw(DrawSyncFlags syncBits);
-    // GLES core supports only GL_PRIMITIVE_RESTART_FIXED_INDEX. Throws when the app enabled
-    // the arbitrary GL_PRIMITIVE_RESTART with a non-fixed index for this index type.
-    void CheckPrimitiveRestartSupported(GLenum indexType);
+    // What an indexed draw has to do about primitive restart before it can be issued.
+    //
+    // Desktop GL restarts on an application-chosen index (glPrimitiveRestartIndex under
+    // GL_PRIMITIVE_RESTART); GLES core restarts only on the all-ones value of the index type
+    // (GL_PRIMITIVE_RESTART_FIXED_INDEX), which the render-state push enables for BOTH caps.
+    // That leaves three cases, and the difference between the last two is not cosmetic - one
+    // adds restarts, the other has to take away restarts the driver would otherwise make.
+    enum class RestartSubstitutionKind : Uint8 {
+        // Nothing to do: restart is off, the fixed-index cap is on, or the application's
+        // restart index already IS the type's all-ones value. The overwhelmingly common answer.
+        None,
+        // The application's index is representable in this index type and differs from the
+        // all-ones value: the index DATA has to be rewritten so the driver restarts where the
+        // application asked.
+        RewriteIndices,
+        // The application's index cannot be held by this index type at all. GL 4.6 core 10.3.6
+        // compares the fetched index, zero-extended, against the full 32-bit
+        // PRIMITIVE_RESTART_INDEX, so no index can match and the draw restarts NOWHERE - but the
+        // render-state push has already enabled the driver's fixed-index restart, so the
+        // all-ones value has to be un-restarted for the duration of the draw.
+        SuppressRestart,
+    };
+    RestartSubstitutionKind ResolveRestartSubstitution(GLenum indexType);
+
+    // Turns the driver's fixed-index restart off for one draw and back on afterwards, for the
+    // SuppressRestart case above. Separate from the substitution below because the multi-draw
+    // tiers need it on its own: they rewrite the index stream themselves and only ever need the
+    // cap half. Inert for every other kind, and it never touches the render-state shadow - it
+    // puts the driver back exactly where SyncRenderState left it.
+    class ScopedSuppressedPrimitiveRestart {
+    public:
+        explicit ScopedSuppressedPrimitiveRestart(RestartSubstitutionKind kind);
+        ~ScopedSuppressedPrimitiveRestart();
+        ScopedSuppressedPrimitiveRestart(const ScopedSuppressedPrimitiveRestart&) = delete;
+        ScopedSuppressedPrimitiveRestart& operator=(const ScopedSuppressedPrimitiveRestart&) = delete;
+
+    private:
+        Bool m_suppressed = false;
+    };
+
+    // Swaps in a scratch element array buffer holding a copy of the index data in which the
+    // application's restart index has been replaced by the value GLES restarts on. Inert
+    // (and free) unless ResolveRestartSubstitution asks for it. The swap lives for the
+    // object's lifetime, so it covers every pass of a viewport-routed draw, and the previous
+    // GL_ELEMENT_ARRAY_BUFFER name is restored on destruction - which matters beyond tidiness,
+    // because the VAO twin memoises that it already synced that binding.
+    //
+    // The copy may be WIDER than the source (see IndexType): when the source already contains
+    // the type's all-ones value as an ordinary vertex index, that value cannot double as the
+    // restart sentinel, and widening is the only way to keep both meanings. Callers must
+    // therefore take the index type from this object, not from their own argument.
+    class ScopedRestartIndexSubstitution {
+    public:
+        // count/indices describe the draw's index range when the CPU knows it. Pass
+        // count == 0 for an indirect draw, whose count lives in GPU memory: the whole bound
+        // element array buffer is rewritten instead, so every element keeps its position and
+        // a GPU-resident firstIndex - an ELEMENT index, so it survives widening too - still
+        // addresses the index it named.
+        ScopedRestartIndexSubstitution(GLenum indexType, GLsizei count, const void* indices);
+        ~ScopedRestartIndexSubstitution();
+        ScopedRestartIndexSubstitution(const ScopedRestartIndexSubstitution&) = delete;
+        ScopedRestartIndexSubstitution& operator=(const ScopedRestartIndexSubstitution&) = delete;
+
+        // False only when a substitution was needed and could not be made. The draw must
+        // then be skipped: issuing it would let the driver silently drop every restart and
+        // weld the primitives on either side together, which is worse than drawing nothing.
+        Bool DrawIsValid() const { return m_valid; }
+        // The element-array offset (or client pointer) the draw must use. Identical to what
+        // was passed in unless a substitution was made.
+        const void* Indices() const { return m_indices; }
+        // The index type the draw must be issued with. Identical to the constructor's unless
+        // the copy had to be widened to keep an all-ones vertex index distinguishable from the
+        // restart sentinel.
+        GLenum IndexType() const { return m_indexType; }
+
+    private:
+        // Declared before m_capOverride so it is initialised first (members initialise in
+        // declaration order): the whole decision is made once, and both the cap override and the
+        // constructor body read the same answer.
+        RestartSubstitutionKind m_kind = RestartSubstitutionKind::None;
+        ScopedSuppressedPrimitiveRestart m_capOverride;
+        const void* m_indices = nullptr;
+        GLenum m_indexType = 0;
+        Uint m_previousBinding = 0;
+        Bool m_substituted = false;
+        Bool m_valid = true;
+    };
+
+    // Drops the scratch element array buffer the substitution above stages through. Like
+    // MultiDrawImpl's scratch names it is abandoned rather than deleted: the name belongs to
+    // the dead ES context, and deleting it would target whatever its successor handed out.
+    void OnRestartSubstitutionContextDestroyed();
     // Feed the current program's gl_BaseInstance / gl_DrawID / gl_BaseVertex emulation
     // uniforms. All are no-ops when the program does not read the corresponding builtin.
     void SetCurrentBaseInstance(Uint32 baseInstance);
@@ -103,6 +215,58 @@ namespace MobileGL::MG_Backend::DirectGLES {
     // link.
     Bool CurrentProgramMayNeedPerSubDrawBuiltins(Bool batchCarriesBaseVertices);
 
+    // ---- gl_ViewportIndex routing emulation, draw half ---------------------------------------
+    //
+    // GLES has ONE viewport, ONE scissor rectangle and ONE depth range; GL 4.1 has sixteen of
+    // each, selected per primitive by gl_ViewportIndex. There is no ES entry point to program the
+    // other fifteen with (GL_OES_viewport_array exists but Adreno 830 does not have it, verified
+    // three ways), so the only way to rasterize a primitive against index i's rectangle is to
+    // make index i's rectangle THE viewport for the duration of a draw - which means issuing the
+    // draw once per distinct viewport state and letting the fragment stage throw away the
+    // primitives that belong to the other indices (the gate Managers.cpp injects).
+    //
+    // Indices whose whole state tuple (viewport rectangle, scissor rectangle, scissor-test enable,
+    // depth range) is identical share ONE pass, so the overwhelmingly common case - every index
+    // still holding what glViewport/glScissor/glDepthRange broadcast to all sixteen - collapses
+    // to a single pass with an all-ones gate mask, i.e. one draw and no behaviour change at all.
+    //
+    // Whether emulation runs. Off only under MOBILEGL_ESPRYT_FORCE_VIEWPORT_ARRAY_EMULATION falsy, which
+    // restores the pre-emulation path as a negative control.
+    Bool ViewportArrayEmulationEnabled();
+    // Whether ANY program built in this process has come out with a viewport gate. Sticky once
+    // true; it exists so that BeginViewportRoutingPasses - which runs on every draw of every
+    // workload - can answer with one static load in the case that matters, which is every
+    // application that has never heard of gl_ViewportIndex.
+    extern Bool g_anyProgramRoutesViewportIndex;
+    // Number of times the current draw has to be issued. Always >= 1, and exactly 1 - with no
+    // state touched - whenever the current program does not route viewports, whenever every
+    // configured index shares one state, and whenever replaying would multiply a side effect the
+    // fragment gate cannot undo (transform feedback, rasterizer discard). Also seeds the pass
+    // mask uniform for that single-pass case, so a gated fragment shader never runs against the
+    // zero every GLSL uniform starts at - which would discard the whole draw.
+    Uint BeginViewportRoutingPasses();
+    // Push pass `pass`'s viewport / scissor / scissor-test / depth range onto the ES context and
+    // set the gate mask to the indices it serves. Only called when the count above exceeds 1.
+    void ApplyViewportRoutingPass(Uint pass);
+    // Restore the gate mask and mark the render-state shadow dirty, so the next ordinary draw
+    // re-pushes index 0's state. Takes the count so it can do nothing at all in the common case.
+    void EndViewportRoutingPasses(Uint passCount);
+
+    // Issue one draw, replayed once per viewport-routing pass. Every application-visible draw
+    // entry point wraps its native glDraw* call in this; the internal blit and clear helpers
+    // deliberately do not, because they bind their own programs, which never route.
+    template <typename IssueDraw>
+    inline void ForEachViewportRoutingPass(IssueDraw&& issue) {
+        const Uint passCount = BeginViewportRoutingPasses();
+        for (Uint pass = 0; pass < passCount; ++pass) {
+            if (passCount > 1) {
+                ApplyViewportRoutingPass(pass);
+            }
+            issue();
+        }
+        EndViewportRoutingPasses(passCount);
+    }
+
     template <typename StateObject, typename BackendObject>
     class StateBackendObjectRegistry {
     public:
@@ -129,7 +293,28 @@ namespace MobileGL::MG_Backend::DirectGLES {
             // Twin creation is the moment a driver-owned id starts needing a guarded
             // destructor; cold path, so the once-guard costs nothing per draw.
             EnsureProcessTeardownSentinel();
+            // Sweep BEFORE the entry reference below exists: the map is open-addressed and an
+            // erase relocates the rest of the probe cluster, so collecting once that reference
+            // is taken would invalidate it. The sweep is therefore owed from an earlier call
+            // rather than triggered by this one.
+            if (m_creationTick >= kCreationGCInterval) {
+                m_creationTick = 0;
+                CollectGarbage();
+            }
+            const SizeT entryCountBeforeInsert = m_entries.size();
             auto& entry = m_entries[stateObj.get()];
+            if (m_entries.size() != entryCountBeforeInsert) {
+                // A key the registry has never held. Nothing tells the backend that a texture or
+                // renderbuffer was DELETED - the twin, and the driver storage it owns, lives
+                // until a collection - and CollectGarbageIfNeeded is ticked only from the
+                // per-draw sync paths, which a CTS-shaped workload runs about ten times per
+                // case. 1024 of those ticks then span ~100 cases, so ~100 cases' worth of dead
+                // (and, for this suite, gigabyte-sized) objects stay allocated at once. Object
+                // CHURN rather than draw count is what makes the sweep urgent, so a twin the
+                // registry has never seen ticks it too - and it does so on the path that is
+                // about to allocate, which is exactly when the memory is needed.
+                ++m_creationTick;
+            }
             if (entry.stateRef.expired()) {
                 // The previous owner of this address is gone and the allocator handed it
                 // to a new object: its twin describes ids the new state object never made.
@@ -203,8 +388,12 @@ namespace MobileGL::MG_Backend::DirectGLES {
 
     private:
         static constexpr Uint32 kGCInterval = 1024;
+        // Creations are far rarer than draws, so this counts in a much smaller unit than
+        // kGCInterval does.
+        static constexpr Uint32 kCreationGCInterval = 64;
         BackendMap m_entries;
         Uint32 m_gcTick = 0;
+        Uint32 m_creationTick = 0;
         Bool m_isCollecting = false;
     };
 
@@ -272,6 +461,16 @@ namespace MobileGL::MG_Backend::DirectGLES {
             // the owning thread replaying them: guard both fields with pendingMutex.
             Bool pendingRespecify = false;
             VecRange1D pendingRanges;
+            // App bytes for an ADOPTED store, awaiting their GPU-ordered landing (ring
+            // stage + glCopyBufferSubData at the next sync; see
+            // BufferBackendOps::ResidentSubData). The frontend keeps such writes out of
+            // the coherent mapping - an in-place host write tears the in-flight frames
+            // still reading the old bytes. Guarded by pendingMutex like pendingRanges.
+            struct PendingResidentWrite {
+                SizeT offset = 0;
+                Vector<Uint8> bytes;
+            };
+            Vector<PendingResidentWrite> pendingResidentWrites;
             std::mutex pendingMutex;
             // Buffer-mutation epoch (see CurrentBufferMutationEpoch) at which this
             // resource last probed IsBufferDrawClean == true, 0 = never (epochs start
@@ -346,13 +545,37 @@ namespace MobileGL::MG_Backend::DirectGLES {
         // client-attribute staging buffers): scrub every buffer-binding shadow that
         // could false-skip when the name is recycled.
         void NoteBufferIdDeleted(Uint id);
+        // Bumped whenever a live GLESBufferResource's driver id is retired and re-minted
+        // while its frontend buffer stays alive (persistent-map adoption, immutable-store
+        // retire). The VAO twins' baked glVertexAttribPointer / element-array bindings
+        // key on FRONTEND versions, which a backend-side re-mint does not move - without
+        // this generation the driver VAO would keep fetching through the deleted id (or
+        // its retained store) forever. Compared and stamped by
+        // BackendVertexArrayObject::SyncToBackend.
+        extern Uint64 g_bufferBackendIdGeneration;
         // Redundant-bind cache for INDEXED buffer bindings (glBindBufferBase/Range on
-        // GL_UNIFORM_BUFFER / GL_SHADER_STORAGE_BUFFER): skips the GL call when the
-        // (id, range) already at that index matches, like the array-buffer/texture/
-        // sampler caches already do. Invalidated on MakeCurrent (context may reset).
+        // GL_UNIFORM_BUFFER / GL_SHADER_STORAGE_BUFFER / GL_TRANSFORM_FEEDBACK_BUFFER):
+        // skips the GL call when the (id, range) already at that index matches, like the
+        // array-buffer/texture/sampler caches already do. Invalidated on MakeCurrent
+        // (context may reset).
+        // Binds the transform feedback capture points [0, bufferCount) from the frontend
+        // state, and touches nothing else - in particular it never binds a zero the
+        // application did not ask for. See the definition for why that matters on Mali.
+        void SyncTransformFeedbackBindingPoints(SizeT bufferCount);
         void BindBufferBaseCached(GLenum glTarget, Uint index, Uint id);
         void BindBufferRangeCached(GLenum glTarget, Uint index, Uint id, GLintptr offset, GLsizeiptr size);
         void InvalidateIndexedBufferBindingCache();
+        // The transform feedback capture points are per-transform-feedback-OBJECT state, so
+        // every glBindTransformFeedback swaps all of them under the shadow above. XfbImpl
+        // calls this on each bind/delete.
+        void InvalidateTransformFeedbackBindingShadows();
+        // Re-issues the GL_ATOMIC_COUNTER_BUFFER binding points a program's shaders declare as
+        // GL_SHADER_STORAGE_BUFFER bindings at the reserved slots the transpiled ESSL was built
+        // against (BackendProgramObjectImpl::GetAtomicCounterBindings /
+        // GetAtomicCounterEsslBindingTop). ES has no counter-buffer target at all, so without
+        // this the shader reads a storage block nobody ever bound a buffer to and the buffer the
+        // application bound never reaches the driver.
+        void SyncAtomicCounterBuffers(const Vector<Int>& glBindings, Int esslBindingTop);
         // Buffer-storage pool maintenance. TrimBufferPool evicts over-budget entries
         // (called once per frame from Present); ClearBufferPool drops all pooled ids
         // without glDeleteBuffers (called when the ES context is going away).
@@ -395,6 +618,57 @@ namespace MobileGL::MG_Backend::DirectGLES {
         // Present()-time upkeep: records the frame's high-water mark for reclamation
         // and deletes grown-away ring stores once the GPU is done with them.
         void UboRingOnPresent();
+
+        // --- Texture unpack-PBO ring ----------------------------------------------
+        // The same persistent-mapped bump allocator, staging TEXTURE UPLOADS. A
+        // glTexSubImage from client memory hands the driver a pointer it must read
+        // before the call returns, so the copy has to be ordered against whatever GPU
+        // work still reads the destination texture: Mali resolves that by BLOCKING the
+        // calling thread (osup_sync_object_wait) instead of ghosting, and Minecraft
+        // re-uploads animated atlas sprites and the lightmap every tick into textures
+        // the in-flight frame is still sampling. Staging the bytes into a
+        // GPU-visible unpack PBO and passing an OFFSET instead lets the driver queue
+        // the copy in the command stream with no CPU wait at all.
+        //
+        // Same reclamation contract as the UBO ring: no ring bytes are recycled before
+        // the frame that referenced them completed on the GPU, so a staged block stays
+        // intact for as long as the queued transfer can still be reading it. The store
+        // therefore settles at roughly (bytes staged per frame) x (frames in flight),
+        // which is what to watch if this ring ever shows up in an RSS regression: it
+        // grows on demand from 4 MiB and is capped, not unbounded.
+        //
+        // False when the feature is disabled (MOBILEGL_ESPRYT_DISABLE_UNPACK_RING),
+        // EXT_buffer_storage / fences are missing, the ES context is not current, or
+        // ring creation already failed under this context. Callers then upload from
+        // the client pointer exactly as before.
+        Bool UnpackRingAvailable();
+        // Bump-allocate `size` bytes aligned to 64 (a PBO-sourced glTexSubImage only
+        // owes the driver the pixel type's own alignment). Grows the ring when the
+        // in-flight span would be overrun; false when the request exceeds the ring's
+        // size cap or storage (re)creation fails.
+        Bool UnpackRingAllocate(SizeT size, SizeT& outOffset);
+        void* UnpackRingMappedPtr();
+        Uint UnpackRingBufferId();
+        // Largest single staging request the ring can ever satisfy.
+        SizeT UnpackRingMaxBytes();
+        void UnpackRingOnPresent();
+
+        // --- Buffer upload ring ---------------------------------------------------
+        // The same persistent-mapped bump allocator, staging APP BUFFER UPDATES
+        // (glBufferSubData / non-persistent map flushes) whose destination store may
+        // still be referenced by in-flight GPU work. Mali resolves that WAR hazard by
+        // BLOCKING the calling glBufferSubData (osup_sync_object_wait) until every
+        // referencing job retires - Minecraft 26.3 rewrites its chunk-section and
+        // dynamic-transform UBOs and streams chunk meshes with per-frame SubData, and
+        // each such call serialized against the whole GPU queue (~1 fps while chunks
+        // stream in, and again on every camera pan). App SubData ranges are queued on
+        // the resource instead (the frontend shadow already holds the bytes) and
+        // draw-time sync drains them: bytes staged into this ring, then one
+        // glCopyBufferSubData per merged range - the copy is ordered on the GPU
+        // timeline, so the hazard costs no CPU wait. Reclamation contract identical
+        // to the other two rings. MOBILEGL_ESPRYT_DISABLE_UPLOAD_RING restores the
+        // historical immediate-upload path (negative control / escape hatch).
+        void UploadRingOnPresent();
     } // namespace BufferImpl
 
     namespace VertexArrayImpl {
@@ -459,12 +733,51 @@ namespace MobileGL::MG_Backend::DirectGLES {
             PendingAttribValueMask& GetPendingAttribValueMaskMemo() { return m_pendingAttribValueMask; }
 
         private:
+            // Narrows one enabled GL_DOUBLE array into a tightly packed float32 stream held in
+            // this VAO's own scratch buffer and declares the attribute against it. ES has no
+            // 64-bit vertex format, but the source bytes are ordinary IEEE-754 doubles and every
+            // fp64 value in every shader is already narrowed to 32 bits (DemoteFloat64Pass), so
+            // narrowing the ARRAY is the coherent completion of that decision rather than
+            // dropping it. Returns false when the stream cannot be built, in which case the
+            // caller must DISABLE the array - leaving a 64-bit array enabled with no pointer is
+            // what the Adreno driver turns into a SIGSEGV at the next draw.
+            Bool SyncFloat64AttributeAsFloat32(Uint attribIndex, const MG_State::GLState::VertexAttribute& attrib,
+                                               Uint32 fetchBaseInstance);
+
+            // What the converted float32 stream in m_convertedAttributeBufferIds[i] was built
+            // from. A hit skips the CPU conversion and the re-upload; the buffer's change serial
+            // is part of the key, so a glBufferSubData into the source invalidates it.
+            struct ConvertedFloat64Stream {
+                Bool valid = false;
+                Uint64 sourceLifetimeId = 0;
+                Uint64 sourceChangeSerial = 0;
+                SizeT sourceOffset = 0;
+                SizeT sourceStride = 0;
+                SizeT componentCount = 0;
+                SizeT elementCount = 0;
+            };
+
             ResolvedDrawBuffers m_resolvedDrawBuffers;
             PendingAttribValueMask m_pendingAttribValueMask;
             Uint m_backendVAOId = 0;
             Array<Uint, MG_State::GLState::VertexArrayObject::MAX_VERTEX_ATTRIBS> m_clientAttributeBufferIds;
+            // Scratch stores for the buffer-backed GL_DOUBLE narrowing. Deliberately separate
+            // from m_clientAttributeBufferIds: that one holds the per-draw upload of a
+            // CLIENT-MEMORY array, and an attribute index can carry both shapes over its life.
+            Array<Uint, MG_State::GLState::VertexArrayObject::MAX_VERTEX_ATTRIBS> m_convertedAttributeBufferIds;
+            Array<ConvertedFloat64Stream, MG_State::GLState::VertexArrayObject::MAX_VERTEX_ATTRIBS>
+                m_convertedAttributeStreams;
+            // True while at least one attribute of this VAO is fed by a converted stream. Such a
+            // stream is derived from buffer CONTENT, which no VAO version covers, so the config
+            // version early-out in SyncToBackend must not be trusted while it is set.
+            Bool m_hasConvertedFloat64Attribute = false;
             Bool m_isInitialized = false;
             Uint16 m_syncedIndexBufferVersion = 0;
+            // Identity of the buffer the version above was stamped against. Raw and never
+            // dereferenced: the slot version is a wrapping Uint16 (see the ResolvedDrawBuffers
+            // IBO memo and the packed_pixels postmortem at BindCurrentFBO), so the version
+            // alone would read a wrapped-back count with a different buffer bound as clean.
+            const MG_State::GLState::BufferObject* m_syncedIndexBufferObject = nullptr;
             // Aggregate gate over the per-attribute walk below: the frontend bumps its config
             // version on every per-attribute version bump (the three Bump*Version functions are
             // its only writers), so an unchanged config version proves every per-attribute
@@ -480,6 +793,11 @@ namespace MobileGL::MG_Backend::DirectGLES {
             // Kept here because it describes what was last EMITTED, which is what the next sync
             // has to correct.
             Uint32 m_syncedFetchBaseInstance = 0;
+            // BufferImpl::g_bufferBackendIdGeneration as of this twin's last emit. A
+            // mismatch means some live buffer's driver id was re-minted since; the ids
+            // baked into the driver VAO's attribute/element bindings may be dead even
+            // though every frontend version matches, so the next sync re-emits them all.
+            Uint64 m_syncedBufferIdGeneration = 0;
         };
 
         extern StateBackendObjectRegistry<MG_State::GLState::VertexArrayObject, BackendVertexArrayObject>
@@ -586,9 +904,21 @@ namespace MobileGL::MG_Backend::DirectGLES {
         // Returns `data` untouched when no widening applies. Pure CPU and context-free so a unit
         // test can exercise the exact packing the driver is handed; `widenedData` is the caller's
         // scratch buffer and has to outlive the returned pointer.
+        // `alphaOneCodeOverride`, when non-zero, replaces the value written into the synthetic
+        // alpha channel: an image carrier that holds a NORMALIZED format's channel CODES has to
+        // pad alpha with that channel's saturated CODE (65535, 32767, 3), which neither of the
+        // transfer type's own "ones" is.
         const void* PrepareChannelWidenedUpload(Uint componentCount, const IntVec3& texelSize, const void* data,
                                                 SizeT byteSize, GLenum uploadType, Vector<Uint8>& widenedData,
-                                                Bool integerData = false);
+                                                Bool integerData = false, Uint32 alphaOneCodeOverride = 0u);
+
+        // Splits a GL_UNSIGNED_INT_2_10_10_10_REV shadow (rgb10_a2, rgb10_a2ui) into the four
+        // GL_UNSIGNED_SHORT channel CODES its GL_RGBA16UI image carrier is uploaded as: red in
+        // bits 0-9, green 10-19, blue 20-29, alpha 30-31. Pure CPU and context-free so a unit test
+        // can pin the exact fields; `widenedData` is the caller's scratch and has to outlive the
+        // returned pointer.
+        const void* PreparePackedIntWidenedUpload(const IntVec3& texelSize, const void* data, SizeT byteSize,
+                                                  Vector<Uint8>& widenedData);
 
         struct StateTextureBasicInfo { // Used for tracking texture state changes
             TextureInternalFormat internalFormat = TextureInternalFormat::Unknown;
@@ -621,12 +951,38 @@ namespace MobileGL::MG_Backend::DirectGLES {
             BackendTextureObject(const BackendTextureObject&) = delete;
             BackendTextureObject& operator=(const BackendTextureObject&) = delete;
             void SyncMipmapsToBackend(const SharedPtr<MG_State::GLState::ITextureObject>& stateTextureObject);
+            // The storage half of the sync for a texture created by glTextureView. Instead of
+            // allocating storage and replaying uploads, it makes this object's ES name BE a view
+            // of the storage texture's ES name (EXT/OES_texture_view), which is what gives the
+            // two names one image and independent per-texture parameters at the same time. The
+            // parameter and sampler halves are unchanged and run on this name as on any other.
+            void SyncTextureViewToBackend(const SharedPtr<MG_State::GLState::ITextureObject>& stateTextureObject);
+            void StampViewSyncKeys(const SharedPtr<MG_State::GLState::ITextureObject>& stateTextureObject);
+            // The storage half of the sync for a texture created by glTextureView. Instead of
+            // allocating storage and replaying uploads, it makes this object's ES name BE a view
+            // of the storage texture's ES name (EXT/OES_texture_view), which is what gives the
+            // two names one image and independent per-texture parameters at the same time. The
+            // parameter and sampler halves are unchanged and run on this name as on any other.
             void SyncBuiltinSamplerToBackend(const SharedPtr<MG_State::GLState::ITextureObject>& stateTextureObject);
             void SyncTextureParamsToBackend(const SharedPtr<MG_State::GLState::ITextureObject>& stateTextureObject);
-            void RequireImageBindableStorage();
+            // Marks the texture as one whose ES storage has to be image-bindable, which for a
+            // non-core image format means re-minting it in the widening's carrier. Takes the state
+            // object because the levels already uploaded have to be marked dirty again: the
+            // re-mint allocates fresh storage and only replays what the shadow still calls dirty.
+            void RequireImageBindableStorage(
+                const SharedPtr<MG_State::GLState::ITextureObject>& stateTextureObject);
+            // Whether this texture's ES storage was minted in an image carrier rather than in the
+            // frontend format's own layout - the readback has to ask, because for a NORMALIZED
+            // carrier the storage is an integer texture holding codes and glGetTexImage still owes
+            // the application floats.
+            Bool RequiresImageBindableStorage() const { return m_imageBindableStorageRequired; }
             void Bind(GLenum target, Uint unit = TempTextureUnit);
             Uint GetBackendTextureId() const;
 
+            // The id to hand glBindImageTexture for a SPLIT buffer image, or 0 when this texture
+            // takes no split. See m_bufferImageSplitViewId.
+            Uint GetBufferImageSplitViewId() const { return m_bufferImageSplitViewId; }
+
             // Aggregate first-level clean gate for the per-draw trio
             // SyncTextureParamsToBackend + SyncBuiltinSamplerToBackend +
             // SyncMipmapsToBackend: EXACTLY the conjunction of their own early-outs
@@ -639,6 +995,10 @@ namespace MobileGL::MG_Backend::DirectGLES {
             // `contextId`/`samplingGeneration` are the frontend context's current
             // values, hoisted by the caller so a per-draw list walk reads them once
             // instead of per texture. `t` must be the live frontend texture.
+            // True while a driver-side re-mint has left the parameter caches describing a texture
+            // that no longer exists; SyncTextureObjectToBackend re-pushes them in the same sync.
+            Bool NeedsParameterResync() const { return m_forceTextureParamsResync || m_forceSamplerResync; }
+
             Bool IsDrawSyncClean(const MG_State::GLState::ITextureObject* t, Uint64 contextId,
                                  Uint64 samplingGeneration) const {
                 if (!m_isInitialized || m_syncedShapeContextId == 0 || m_syncedShapeContextId != contextId ||
@@ -663,6 +1023,36 @@ namespace MobileGL::MG_Backend::DirectGLES {
             void RecreateBackendTexture();
 
             Uint m_backendTextureId = 0;
+            // A SECOND buffer-texture name over the SAME buffer object, viewed in the split's
+            // single-channel base format, used only as the glBindImageTexture target.
+            //
+            // The split needs the view to say r32f where the application said rg32f, but a buffer
+            // texture that is image-bound may ALSO be read through a samplerBuffer - and the
+            // sampler side is not subscript-rewritten, so re-describing the application's own
+            // texture broke it: texelFetch(s, i) returned component 2i of the base view instead of
+            // texel i's pair. That is exactly and only
+            // KHR-GL42/43.shader_image_load_store.advanced-sync-imageAccess, which image-stores
+            // into a GL_RG32F buffer texture and then reads the same texture through both an
+            // imageBuffer and a samplerBuffer in one shader, comparing the two.
+            //
+            // Two names over one buffer cost nothing and alias exactly: a buffer texture owns no
+            // storage, so both views are the application's bytes, and the split's whole premise is
+            // that the two describe the same memory. The application's own name therefore keeps
+            // the format it asked for - rg32f IS a legal SAMPLED buffer-texture format in ES 3.2,
+            // it is only the IMAGE binding ES cannot spell - and the private name below carries
+            // the split the shader was rewritten against. 0 when this texture takes no split.
+            Uint m_bufferImageSplitViewId = 0;
+            // For a texture created by glTextureView: the ES name of the storage texture this
+            // one was last made a view OF. EXT_texture_view may be called only once per name, so
+            // a storage texture that got re-minted underneath (RecreateBackendTexture) has to be
+            // detected here and answered with a fresh name for the view as well - otherwise the
+            // view would keep aliasing storage that no longer exists.
+            Uint m_viewSourceBackendTextureId = 0;
+            // For a texture created by glTextureView: the ES name of the storage texture this
+            // one was last made a view OF. EXT_texture_view may be called only once per name, so
+            // a storage texture that got re-minted underneath (RecreateBackendTexture) has to be
+            // detected here and answered with a fresh name for the view as well - otherwise the
+            // view would keep aliasing storage that no longer exists.
             // ES context generation the id was created under; a dtor running after
             // that context died must not delete a foreign (recycled) name.
             Uint m_contextGeneration = 0;
@@ -696,7 +1086,14 @@ namespace MobileGL::MG_Backend::DirectGLES {
             Uint16 m_syncedShapeParamsVersion = 0;
             SamplerParameters m_cacheSamplerParameters;
             UintVec2 m_cacheLodRange = {0, 1000};
+            // All three representations plus the form, because none of them alone identifies the
+            // border colour the driver texture is holding: two integer borders can share one float
+            // (anything differing above 2^24), and a Float -> Int transition can leave every number
+            // unchanged while still needing a different driver entry point.
             FloatVec4 m_cacheBorderColor = {0.0f, 0.0f, 0.0f, 0.0f};
+            IntVec4 m_cacheBorderColorI = {0, 0, 0, 0};
+            UintVec4 m_cacheBorderColorUI = {0, 0, 0, 0};
+            BorderColorForm m_cacheBorderColorForm = BorderColorForm::Float;
             Vec4<TextureSwizzleParam> m_cacheSwizzleParams = {TextureSwizzleParam::Red, TextureSwizzleParam::Green,
                                                               TextureSwizzleParam::Blue, TextureSwizzleParam::Alpha};
             // GL_DEPTH_STENCIL_TEXTURE_MODE. GL_DEPTH_COMPONENT is the GL and ES default, so a
@@ -711,6 +1108,15 @@ namespace MobileGL::MG_Backend::DirectGLES {
             // parameter already pushed onto it: the params-version early-out has to be overridden
             // once, or an unchanged version would skip the re-push forever.
             Bool m_forceTextureParamsResync = false;
+            // The same problem for the FILTER state, which lives in m_cacheSamplerParameters and
+            // is gated on the frontend sampler's version rather than on the params version. A
+            // re-mint leaves that cache describing values the new driver texture never received,
+            // and an unchanged sampler version would then skip re-pushing them forever. This
+            // matters more than mis-filtering: ES makes a texture INCOMPLETE when its filters do
+            // not suit its level set (any integer texture with a non-NEAREST filter, or a
+            // single-level texture with a mipmapping filter), and an incomplete texture samples
+            // (0, 0, 0, 1) rather than its contents.
+            Bool m_forceSamplerResync = false;
         };
 
         void ActivateTextureUnit(Uint unit);
@@ -799,6 +1205,11 @@ namespace MobileGL::MG_Backend::DirectGLES {
 
             using FramebufferObject = MG_State::GLState::FramebufferObject;
             FramebufferObject::FramebufferAttachmentVersionArray m_syncedFrontendAttachmentVersions = {0};
+            // g_attachmentBackendIdGeneration as of this twin's last attachment walk. A
+            // mismatch means some backend texture id was re-minted since, and any of this
+            // twin's attachment points may still hold the dead id even though the frontend
+            // attachment versions match - so the walk re-attaches everything first.
+            Uint64 m_syncedBackendIdGeneration = 0;
         };
 
         extern StateBackendObjectRegistry<MG_State::GLState::FramebufferObject, BackendFramebufferObject>
@@ -888,6 +1299,19 @@ namespace MobileGL::MG_Backend::DirectGLES {
         extern Array<MG_State::GLState::FramebufferObject*, SizeT(FramebufferTarget::FramebufferTargetCount)>
             g_fboSyncedObjects;
 
+        // Bumped whenever a live backend texture's driver id is re-minted while its
+        // frontend texture may still be attached to application FBOs
+        // (BackendTextureObject::RecreateBackendTexture - e.g. a respecify of a texture
+        // whose backend storage went immutable). The FBO twins' attachment memos key on
+        // FRONTEND attachment versions, which a backend-side re-mint does not move, so
+        // the driver FBO would keep the deleted texture name attached forever. The
+        // SyncCurrentFBO gate compares this generation (below) to re-enter the sync,
+        // and each twin re-arms its per-attachment memo on a mismatch (SyncToBackend).
+        extern Uint64 g_attachmentBackendIdGeneration;
+        // What g_attachmentBackendIdGeneration was when SyncCurrentFBO last stamped each
+        // target; part of the synced tuple above.
+        extern Array<Uint64, SizeT(FramebufferTarget::FramebufferTargetCount)> g_fboSyncedBackendIdGenerations;
+
         // Driver-level READ/DRAW framebuffer-binding shadow. Every backend
         // glBindFramebuffer routes through BindFramebufferId so scoped helpers can
         // save/restore the current binding without a glGetIntegerv round-trip (that
@@ -991,23 +1415,51 @@ namespace MobileGL::MG_Backend::DirectGLES {
     // Image uniforms take their unit from the layout(binding=N) qualifier baked into
     // the transpiled ESSL; unlike samplers they must not (and in ES cannot) be
     // assigned through glUniform1i.
+    //
+    // ALL THIRTY-THREE of them, in the one contiguous block ARB_shader_image_load_store allocated
+    // (GL_IMAGE_1D 0x904C through GL_UNSIGNED_INT_IMAGE_2D_MULTISAMPLE_ARRAY 0x906C). The list
+    // used to hold only the fifteen whose TARGET exists in ES, which read as a reasonable
+    // shortcut and was two bugs: an image uniform this says "no" to is one
+    // CollectImageFormatBakeInputs never walks, so its non-core format is neither baked nor
+    // widened and SPIRV-Cross throws for the whole stage ("Attempting to use image format not
+    // supported in ES profile"), and it is also one SyncToBackend then treats as a SAMPLER and
+    // assigns with glUniform1i, which ES makes an INVALID_OPERATION. A GL_TEXTURE_CUBE_MAP_ARRAY
+    // image - which ES 3.2 has in core, so it is not even an emulated target - hit both.
     inline Bool IsImageUniformType(GLenum type) {
         switch (type) {
+        case 0x904C: /*GL_IMAGE_1D*/
         case 0x904D: /*GL_IMAGE_2D*/
         case 0x904E: /*GL_IMAGE_3D*/
+        case 0x904F: /*GL_IMAGE_2D_RECT*/
         case 0x9050: /*GL_IMAGE_CUBE*/
         case 0x9051: /*GL_IMAGE_BUFFER*/
+        case 0x9052: /*GL_IMAGE_1D_ARRAY*/
         case 0x9053: /*GL_IMAGE_2D_ARRAY*/
+        case 0x9054: /*GL_IMAGE_CUBE_MAP_ARRAY*/
+        case 0x9055: /*GL_IMAGE_2D_MULTISAMPLE*/
+        case 0x9056: /*GL_IMAGE_2D_MULTISAMPLE_ARRAY*/
+        case 0x9057: /*GL_INT_IMAGE_1D*/
         case 0x9058: /*GL_INT_IMAGE_2D*/
         case 0x9059: /*GL_INT_IMAGE_3D*/
+        case 0x905A: /*GL_INT_IMAGE_2D_RECT*/
         case 0x905B: /*GL_INT_IMAGE_CUBE*/
         case 0x905C: /*GL_INT_IMAGE_BUFFER*/
+        case 0x905D: /*GL_INT_IMAGE_1D_ARRAY*/
         case 0x905E: /*GL_INT_IMAGE_2D_ARRAY*/
+        case 0x905F: /*GL_INT_IMAGE_CUBE_MAP_ARRAY*/
+        case 0x9060: /*GL_INT_IMAGE_2D_MULTISAMPLE*/
+        case 0x9061: /*GL_INT_IMAGE_2D_MULTISAMPLE_ARRAY*/
+        case 0x9062: /*GL_UNSIGNED_INT_IMAGE_1D*/
         case 0x9063: /*GL_UNSIGNED_INT_IMAGE_2D*/
         case 0x9064: /*GL_UNSIGNED_INT_IMAGE_3D*/
+        case 0x9065: /*GL_UNSIGNED_INT_IMAGE_2D_RECT*/
         case 0x9066: /*GL_UNSIGNED_INT_IMAGE_CUBE*/
         case 0x9067: /*GL_UNSIGNED_INT_IMAGE_BUFFER*/
+        case 0x9068: /*GL_UNSIGNED_INT_IMAGE_1D_ARRAY*/
         case 0x9069: /*GL_UNSIGNED_INT_IMAGE_2D_ARRAY*/
+        case 0x906A: /*GL_UNSIGNED_INT_IMAGE_CUBE_MAP_ARRAY*/
+        case 0x906B: /*GL_UNSIGNED_INT_IMAGE_2D_MULTISAMPLE*/
+        case 0x906C: /*GL_UNSIGNED_INT_IMAGE_2D_MULTISAMPLE_ARRAY*/
             return true;
         default:
             return false;
@@ -1015,6 +1467,9 @@ namespace MobileGL::MG_Backend::DirectGLES {
     }
 
     namespace PrgramImpl {
+        // Defined further down, next to CollectImageFormatBakeInputs; only referenced here.
+        struct ImageFormatBakeInputs;
+
         class BackendProgramObjectImpl {
         public:
             // Per-link cache of a sampler-style uniform's backend location: built once in
@@ -1074,7 +1529,7 @@ namespace MobileGL::MG_Backend::DirectGLES {
             BackendProgramObjectImpl();
             ~BackendProgramObjectImpl();
             void SyncToBackend(const SharedPtr<MG_State::GLState::ProgramObject>& stateProgramObject);
-            void Use() const;
+            void Use();
             void SetBaseInstance(Uint32 baseInstance) const;
             void SetBaseInstanceWordIndex(Int32 wordIndex) const;
             void SetDrawID(Uint32 drawId) const;
@@ -1085,6 +1540,14 @@ namespace MobileGL::MG_Backend::DirectGLES {
             // Same for gl_BaseVertex: only a program that reads it pays for the per-draw
             // uniform write, and only such a program needs the reset after one.
             Bool ReadsBaseVertex() const { return m_baseVertexUniformLocation >= 0; }
+            // Which viewport indices the next draw's fragments may keep, one bit each. Written
+            // once per replay pass; see ForEachViewportRoutingPass.
+            void SetViewportPassMask(Uint32 indexMask) const;
+            // True when this build injected the fragment-stage viewport gate, i.e. when a
+            // pre-rasterization stage routes by gl_ViewportIndex AND the fragment stage can act
+            // on it. The uniform is the honest test for both halves: it exists only where the
+            // gate was injected, and the gate is injected only where a stage routes.
+            Bool RoutesViewportIndex() const { return m_viewportPassMaskUniformLocation >= 0; }
             Int GetIndirectParamsBinding() const { return m_indirectParamsBinding; }
             Uint GetBackendProgramId() const { return m_backendProgramId; }
             // False when the last SyncToBackend could not produce a usable program (a
@@ -1100,6 +1563,33 @@ namespace MobileGL::MG_Backend::DirectGLES {
             // qualifier, so the overrides are baked into the source). A mismatch means the
             // program is stale exactly like the clamp masks above.
             Uint64 GetShaderStorageBlockBindingSignature() const { return m_shaderStorageBlockBindingSignature; }
+            // GL atomic-counter binding points the transpiled stages declare (sorted, unique),
+            // and the top of the reserved shader-storage range their counter blocks were
+            // transpiled against - the slot for GL binding N is `top - N`. Empty for every
+            // program that uses no atomic counter, which is what keeps the per-draw cost of the
+            // counter sync at one empty-vector test.
+            const Vector<Int>& GetAtomicCounterBindings() const { return m_atomicCounterGlBindings; }
+            Int GetAtomicCounterEsslBindingTop() const { return m_atomicCounterEsslBindingTop; }
+            // GL_PATCH_VERTICES the synthesized pass-through tessellation control stage was built
+            // for, or -1 when this program needed no such stage. Another of the same shape as the
+            // signatures above: the value is compiled INTO the synthesized stage as
+            // `layout(vertices = N) out`, so a program built for one patch size is stale for
+            // another and the draw path has to say so. -1 compares equal to itself for every
+            // program that has a control stage of its own, i.e. for all but a handful.
+            Int GetPassthroughTessControlPatchVertices() const {
+                return m_passthroughTessControlPatchVertices;
+            }
+            // GL_PATCH_DEFAULT_{OUTER,INNER}_LEVEL the same synthesized stage was built with, for
+            // the same reason: ES has neither the state nor an entry point to forward it to, so
+            // glPatchParameterfv's values are compiled in as literals and a program built with one
+            // set is stale for another. Meaningless (and never read) when the patch-vertices field
+            // above is -1, which is the gate the draw path tests first.
+            const FloatVec4& GetPassthroughTessControlOuterLevel() const {
+                return m_passthroughTessControlOuterLevel;
+            }
+            const FloatVec2& GetPassthroughTessControlInnerLevel() const {
+                return m_passthroughTessControlInnerLevel;
+            }
 
             Bool HasGlobalUboBlock() const { return m_globalUboBackendBlockIndex >= 0; }
             const Vector<Int>& GetUniformBlockBackendIndices() const { return m_uniformBlockBackendIndices; }
@@ -1148,6 +1638,34 @@ namespace MobileGL::MG_Backend::DirectGLES {
         private:
             void CacheResourceLocations(const SharedPtr<MG_State::GLState::ProgramObject>& stateProgramObject);
 
+            // Builds, compiles and attaches the pass-through tessellation control stage GL 4.6
+            // core 11.2.2 describes, for a program that has an evaluation stage and none of its
+            // own - which ES 3.2 rejects outright. Called from SyncToBackend after every real
+            // stage has been attached and before the link; see the definition for why it cannot
+            // regress a program that works today.
+            void AttachPassthroughTessControlStage(
+                const MG_State::GLState::ProgramObject& stateProgramObject, Int tessEvalShaderIndex,
+                const Vector<Vector<unsigned int>>& shaderSpirvs, const String& vertexStageEssl,
+                const String& tessEvalStageEssl);
+
+            // One stage's SPIR-V through the DirectGLES pass chain and SPIRV-Cross, producing
+            // the raw emitted ESSL and the interface blocks this stage's XFB flattening
+            // rewrote. This is the segment the L2 shader-translation memo keys on, so every
+            // input it reads must appear in EsslTranslationKeyInputs - see the definition's
+            // header comment in Managers.cpp and MG_Util/ShaderTranspiler/TranslationCache.h.
+            // False means SPIRV-Cross refused the module; `outError` then carries its message.
+            Bool TranspileSpirvToEssl(const Vector<unsigned int>& spirvCode, GLenum glShaderType,
+                                      const std::set<String>& xfbCaptureBlockNames,
+                                      const ImageFormatBakeInputs& imageFormatBake,
+                                      const UnorderedMap<String, Int>& storageBlockBindingOverrides,
+                                      const std::map<String, String>& inputBlockRenames,
+                                      const std::map<String, String>& outputBlockRenames,
+                                      Bool stripInputBlockLocations, Bool stripOutputBlockLocations,
+                                      Int atomicCounterEsslBindingTop, Bool enableSpirvValidation,
+                                      String& outSource,
+                                      std::set<String>& outFlattenedXfbBlockNames,
+                                      Vector<Int>& outAtomicCounterGlBindings, String& outError) const;
+
             Uint m_backendProgramId = 0;
             // GL name of the frontend program this was last synced from; diagnostics only, so
             // an unusable backend program can be traced back to the glCreateProgram id the app
@@ -1158,6 +1676,7 @@ namespace MobileGL::MG_Backend::DirectGLES {
             Int m_drawIdUniformLocation = -1;
             Int m_baseVertexUniformLocation = -1;
             Int m_baseInstanceWordIndexUniformLocation = -1;
+            Int m_viewportPassMaskUniformLocation = -1;
             Int m_indirectParamsBinding = -1;
             Uint32 m_snormFallbackClampOutputMask = 0;
             Uint32 m_unormFallbackClampOutputMask = 0;
@@ -1166,8 +1685,23 @@ namespace MobileGL::MG_Backend::DirectGLES {
             Uint m_fragColorBroadcastCount = 1;
             // 0 is the signature of an empty override set, i.e. what almost every program has.
             Uint64 m_shaderStorageBlockBindingSignature = 0;
+            Vector<Int> m_atomicCounterGlBindings;
+            Int m_atomicCounterEsslBindingTop = -1;
+            // -1 for every program that has a tessellation control stage of its own (or none at
+            // all); otherwise the GL_PATCH_VERTICES the synthesized pass-through stage was built
+            // with. See GetPassthroughTessControlPatchVertices.
+            Int m_passthroughTessControlPatchVertices = -1;
+            // The default tessellation levels baked into that same stage. Only meaningful while
+            // the field above is not -1.
+            FloatVec4 m_passthroughTessControlOuterLevel = FloatVec4(1.0f, 1.0f, 1.0f, 1.0f);
+            FloatVec2 m_passthroughTessControlInnerLevel = FloatVec2(1.0f, 1.0f);
             Bool m_isInitialized = false;
             Bool m_backendProgramUsable = false;
+            // Set by SyncToBackend every time it relinks the driver program, cleared by the
+            // next Use(). Use() dedupes on a GL program NAME, and a relink replaces the
+            // executable behind that name without changing it - see the note at the
+            // glLinkProgram in SyncToBackend for what the driver runs otherwise.
+            Bool m_rebindAfterRelink = false;
 
             Int m_globalUboBackendBlockIndex = -1;
             Int m_globalUboBackendBlockSize = 0;
@@ -1257,6 +1791,14 @@ namespace MobileGL::MG_Backend::DirectGLES {
             // Some format in play - declared or baked - is outside the GLSL ES core image
             // format set, so the emitted ESSL needs the GL_NV_image_formats directive.
             Bool needsExtendedImageFormats = false;
+            // Some DECLARED format in play is one WidenImageFormatsForEssl will re-declare in a
+            // core carrier. Answered from the uniform reflection rather than from a module parse
+            // on purpose: the widening is armed on every driver, so a per-stage BuildModule to
+            // find out would land on every stage of every program - which is the cost
+            // SpirvGateFeatures exists to avoid. Program-wide, so it can over-arm a stage that
+            // declares no image; the pass then finds nothing, reports no change, and the caller
+            // keeps the module it already had.
+            Bool declaresWidenableImageFormat = false;
         };
         ImageFormatBakeInputs CollectImageFormatBakeInputs(
             const MG_State::GLState::ProgramObject& stateProgramObject);

MobileGL/MG_Backend/DirectGLES/MultiDraw.cpp

@@ -9,6 +9,8 @@
 #include "MultiDraw.h"
 #include "Managers.h"
 #include <MG_State/GLState/Core.h>
+#include <MG_Pipe/PipeInputsSwitch.h>
+#include <MG_Util/Metrics/PipeStats.h>
 #include <cstring>
 #include <limits>
 
@@ -29,21 +31,26 @@ namespace MobileGL::MG_Backend::DirectGLES::MultiDrawImpl {
             }
         }
 
-        // The all-ones value of an index type, which is what GL restarts on once
-        // primitive restart is in play. CheckPrimitiveRestartSupported has already
-        // rejected the arbitrary-index form of GL_PRIMITIVE_RESTART, so an enabled
-        // restart always restarts here and nowhere else.
+        // The index value this batch restarts on, compared at 32 bits against the zero-extended
+        // source index. Normally the all-ones value of the source type, which is what
+        // GL_PRIMITIVE_RESTART_FIXED_INDEX and GLES both restart on; with desktop
+        // GL_PRIMITIVE_RESTART it is instead whatever glPrimitiveRestartIndex named. The rebased
+        // tier turns whichever it is into 0xFFFFFFFF in its widened stream, which is what the
+        // driver restarts on.
+        //
+        // No truncation, deliberately, and the same rule ResolveRestartSubstitution applies: a
+        // restart index the source type cannot hold simply matches nothing, so returning it
+        // verbatim is already "this batch restarts nowhere".
         Uint32 RestartSentinelFor(GLenum type) {
-            switch (type) {
-            case GL_UNSIGNED_BYTE: return 0xFFu;
-            case GL_UNSIGNED_SHORT: return 0xFFFFu;
-            default: return 0xFFFFFFFFu;
+            if (ResolveRestartSubstitution(type) != RestartSubstitutionKind::None) {
+                return MGB_CTX->GetPrimitiveRestartIndex();
             }
+            return MG_Util::FixedRestartIndexForGLType(type);
         }
 
         Bool RestartActive() {
-            return MG_State::pGLContext->IsCapabilityEnabled(CapabilityInput::PrimitiveRestart) ||
-                   MG_State::pGLContext->IsCapabilityEnabled(CapabilityInput::PrimitiveRestartFixedIndex);
+            return MGB_CTX->IsCapabilityEnabled(CapabilityInput::PrimitiveRestart) ||
+                   MGB_CTX->IsCapabilityEnabled(CapabilityInput::PrimitiveRestartFixedIndex);
         }
 
         // Vertices per primitive for the modes whose sub-draws may be concatenated into a
@@ -78,7 +85,7 @@ namespace MobileGL::MG_Backend::DirectGLES::MultiDrawImpl {
 
         Uint BoundDrawIndirectBufferId() {
             const auto& indirect =
-                MG_State::pGLContext->GetBufferBindingSlot(BufferTarget::DrawIndirect).GetBoundObject();
+                MGB_CTX->GetBufferBindingSlot(BufferTarget::DrawIndirect).GetBoundObject();
             if (!indirect) return 0;
             const auto* resource = BufferImpl::EnsureBufferResource(indirect);
             return resource ? resource->id : 0;
@@ -86,7 +93,7 @@ namespace MobileGL::MG_Backend::DirectGLES::MultiDrawImpl {
 
         const SharedPtr<MG_State::GLState::BufferObject>& BoundIndexBuffer() {
             static const SharedPtr<MG_State::GLState::BufferObject> none;
-            const auto& vao = MG_State::pGLContext->GetBoundVertexArray();
+            const auto& vao = MGB_CTX->GetBoundVertexArray();
             if (!vao) return none;
             return vao->GetIndexBufferBindingSlot().GetBoundObject();
         }
@@ -151,7 +158,10 @@ namespace MobileGL::MG_Backend::DirectGLES::MultiDrawImpl {
         // are bound as storage blocks. Respecifies rather than sub-updates: glBufferData
         // orphans the previous store, so the upload never waits on a dispatch still reading
         // the old contents out of the same name.
-        Bool UploadScratch(ScratchBuffer& buffer, SizeT bytes, const void* data) {
+        // statsClass: which MGPipe byte population these bytes belong to. Counted here
+        // rather than at the four call sites so a new tier cannot forget it.
+        Bool UploadScratch(ScratchBuffer& buffer, SizeT bytes, const void* data,
+                           MG_Util::PipeStats::ByteClass statsClass) {
             if (bytes == 0) return true;
             if (!EnsureScratchName(buffer)) return false;
             BufferImpl::BindBufferId(BufferImpl::TempBufferTarget, buffer.id);
@@ -164,6 +174,9 @@ namespace MobileGL::MG_Backend::DirectGLES::MultiDrawImpl {
             buffer.cursor = 0;
             if (data) {
                 g_GLESFuncs.glBufferSubData(BufferImpl::TempBufferTarget, 0, static_cast<GLsizeiptr>(bytes), data);
+                if (MG_Util::PipeStats::Enabled()) {
+                    MG_Util::PipeStats::AddBytes(statsClass, static_cast<Uint64>(bytes));
+                }
             }
             return true;
         }
@@ -178,7 +191,8 @@ namespace MobileGL::MG_Backend::DirectGLES::MultiDrawImpl {
         constexpr SizeT kRingAlignment = 16; // >= 4, so both command and uint32-index offsets stay legal
         constexpr SizeT kMinRingBytes = 1u << 16;
 
-        Bool UploadScratchRing(ScratchBuffer& buffer, SizeT bytes, const void* data, SizeT& outOffset) {
+        Bool UploadScratchRing(ScratchBuffer& buffer, SizeT bytes, const void* data,
+                               MG_Util::PipeStats::ByteClass statsClass, SizeT& outOffset) {
             outOffset = 0;
             if (bytes == 0) return true;
             if (!EnsureScratchName(buffer)) return false;
@@ -202,6 +216,9 @@ namespace MobileGL::MG_Backend::DirectGLES::MultiDrawImpl {
             if (data) {
                 g_GLESFuncs.glBufferSubData(BufferImpl::TempBufferTarget, static_cast<GLintptr>(outOffset),
                                             static_cast<GLsizeiptr>(bytes), data);
+                if (MG_Util::PipeStats::Enabled()) {
+                    MG_Util::PipeStats::AddBytes(statsClass, static_cast<Uint64>(bytes));
+                }
             }
             buffer.cursor += aligned;
             return true;
@@ -275,10 +292,20 @@ namespace MobileGL::MG_Backend::DirectGLES::MultiDrawImpl {
         // its remaining feasibility checks inside its implementation, where the data it
         // has to walk is already in hand.
         GLESMultiDrawMode ResolveTierForBatch(Bool programReadsDrawID, Bool perSubDrawBaseVertex,
-                                              Bool hasIndexBuffer) {
+                                              Bool hasIndexBuffer, Bool arbitraryRestart) {
             ResolveTierOnce();
             GLESMultiDrawMode tier = g_resolvedTier;
 
+            // Desktop GL_PRIMITIVE_RESTART restarts on an application-chosen index; the driver
+            // only ever restarts on the all-ones value. Every tier but the rebased one hands
+            // the application's own index data to the driver, which would then see no restarts
+            // at all and weld the primitives together. The rebased tier is the one that
+            // REWRITES the stream, and RestartSentinelFor already tells it which value to
+            // translate, so it is the only tier this batch can take.
+            if (arbitraryRestart) {
+                return GLESMultiDrawMode::DrawElements;
+            }
+
             // Batched tiers issue one driver entry for the whole batch, so the emulated
             // gl_DrawID uniform can only hold one value across every sub-draw. A program
             // that reads gl_DrawID gets an unrolled tier, which feeds each sub-draw its
@@ -402,7 +429,8 @@ namespace MobileGL::MG_Backend::DirectGLES::MultiDrawImpl {
 
             const SizeT commandBytes = g_commandStaging.size() * sizeof(DrawElementsIndirectCommand);
             SizeT commandBase = 0;
-            if (!UploadScratchRing(g_indirectCommands, commandBytes, g_commandStaging.data(), commandBase)) {
+            if (!UploadScratchRing(g_indirectCommands, commandBytes, g_commandStaging.data(),
+                                   MG_Util::PipeStats::ByteClass::StageIndirectCmd, commandBase)) {
                 return false;
             }
 
@@ -414,14 +442,18 @@ namespace MobileGL::MG_Backend::DirectGLES::MultiDrawImpl {
             const Uint previousIndirectBinding = BoundDrawIndirectBufferId();
             BufferImpl::BindBufferId(GL_DRAW_INDIRECT_BUFFER, g_indirectCommands.id);
             if (batched) {
-                g_GLESFuncs.glMultiDrawElementsIndirectEXT(mode, type, reinterpret_cast<const void*>(commandBase),
-                                                           drawcount, 0);
+                ForEachViewportRoutingPass([&] {
+                    g_GLESFuncs.glMultiDrawElementsIndirectEXT(mode, type, reinterpret_cast<const void*>(commandBase),
+                                                               drawcount, 0);
+                });
             } else {
                 for (GLsizei i = 0; i < drawcount; ++i) {
                     if (feedDrawID) SetCurrentDrawID(static_cast<Uint32>(i));
                     if (feedBaseVertex) SetCurrentBaseVertex(basevertex ? basevertex[i] : 0);
                     const SizeT commandOffset = commandBase + static_cast<SizeT>(i) * sizeof(DrawElementsIndirectCommand);
-                    g_GLESFuncs.glDrawElementsIndirect(mode, type, reinterpret_cast<const void*>(commandOffset));
+                    ForEachViewportRoutingPass([&] {
+                        g_GLESFuncs.glDrawElementsIndirect(mode, type, reinterpret_cast<const void*>(commandOffset));
+                    });
                 }
                 if (feedDrawID) SetCurrentDrawID(0);
                 if (feedBaseVertex) SetCurrentBaseVertex(0);
@@ -442,8 +474,10 @@ namespace MobileGL::MG_Backend::DirectGLES::MultiDrawImpl {
                 if (count[i] <= 0) continue;
                 if (feedDrawID) SetCurrentDrawID(static_cast<Uint32>(i));
                 if (feedBaseVertex) SetCurrentBaseVertex(basevertex ? basevertex[i] : 0);
-                g_GLESFuncs.glDrawElementsBaseVertex(mode, count[i], type, indices[i],
-                                                     basevertex ? basevertex[i] : 0);
+                ForEachViewportRoutingPass([&] {
+                    g_GLESFuncs.glDrawElementsBaseVertex(mode, count[i], type, indices[i],
+                                                         basevertex ? basevertex[i] : 0);
+                });
             }
             if (feedDrawID) SetCurrentDrawID(0);
             if (feedBaseVertex) SetCurrentBaseVertex(0);
@@ -482,6 +516,16 @@ namespace MobileGL::MG_Backend::DirectGLES::MultiDrawImpl {
 
             const Bool restartActive = RestartActive();
             const Uint32 restartSentinel = RestartSentinelFor(type);
+            // Widening to GL_UNSIGNED_INT gives a UBYTE/USHORT source a sentinel it can never
+            // spell, so those batches are lossless. A UINT source that already uses 0xFFFFFFFF as
+            // a real vertex index while restarting on a different one is the one shape 32 bits
+            // cannot express - the same corner the single-draw substitution reports.
+            if (restartActive && indexSize == 4 && restartSentinel != 0xFFFFFFFFu) {
+                MGLOG_E_ONCE("GL_PRIMITIVE_RESTART with restart index %u over GL_UNSIGNED_INT multi-draw indices: "
+                             "any index that is already 0xFFFFFFFF will restart too, because the rewritten stream "
+                             "has no wider sentinel to move to.",
+                             restartSentinel);
+            }
             g_indexStaging.resize(total);
             SizeT cursor = 0;
             for (GLsizei i = 0; i < drawcount; ++i) {
@@ -501,7 +545,8 @@ namespace MobileGL::MG_Backend::DirectGLES::MultiDrawImpl {
             }
 
             SizeT indexBase = 0;
-            if (!UploadScratchRing(g_rebasedIndices, total * sizeof(Uint32), g_indexStaging.data(), indexBase)) {
+            if (!UploadScratchRing(g_rebasedIndices, total * sizeof(Uint32), g_indexStaging.data(),
+                                   MG_Util::PipeStats::ByteClass::StageIndexClient, indexBase)) {
                 return false;
             }
 
@@ -515,8 +560,10 @@ namespace MobileGL::MG_Backend::DirectGLES::MultiDrawImpl {
                 // driver sees none - but gl_BaseVertex still has to report the value the
                 // application passed for this sub-draw.
                 if (feedBaseVertex) SetCurrentBaseVertex(basevertex ? basevertex[i] : 0);
-                g_GLESFuncs.glDrawElements(mode, count[i], GL_UNSIGNED_INT,
-                                           reinterpret_cast<const void*>(indexBase + cursor * sizeof(Uint32)));
+                ForEachViewportRoutingPass([&] {
+                    g_GLESFuncs.glDrawElements(mode, count[i], GL_UNSIGNED_INT,
+                                               reinterpret_cast<const void*>(indexBase + cursor * sizeof(Uint32)));
+                });
                 cursor += static_cast<SizeT>(count[i]);
             }
             if (feedDrawID) SetCurrentDrawID(0);
@@ -704,10 +751,16 @@ void main() {
             if (total == 0) return; // nothing to draw; the ordinary tiers no-op just as well
 
             if (!EnsureComputeProgram()) return;
-            if (!UploadScratch(g_drawInfo, g_drawInfoStaging.size() * sizeof(Uint32), g_drawInfoStaging.data())) {
+            if (!UploadScratch(g_drawInfo, g_drawInfoStaging.size() * sizeof(Uint32), g_drawInfoStaging.data(),
+                               MG_Util::PipeStats::ByteClass::StageIndirectCmd)) {
+                return;
+            }
+            // data == nullptr: pure respecify, the compute pass writes the contents, so no
+            // host bytes cross here and nothing is counted.
+            if (!UploadScratch(g_flattenedIndices, total * sizeof(Uint32), nullptr,
+                               MG_Util::PipeStats::ByteClass::StageIndexClient)) {
                 return;
             }
-            if (!UploadScratch(g_flattenedIndices, total * sizeof(Uint32), nullptr)) return;
 
             BufferImpl::BindBufferBaseCached(GL_SHADER_STORAGE_BUFFER, 0, sourceResource->id);
             BufferImpl::BindBufferBaseCached(GL_SHADER_STORAGE_BUFFER, 1, g_drawInfo.id);
@@ -844,8 +897,14 @@ void main() {
     void DrawElementsBatch(GLenum mode, const GLsizei* count, GLenum type, const GLvoid* const* indices,
                            GLsizei drawcount, const GLint* basevertex) {
         if (drawcount <= 0 || !count || !indices) return;
-        // State-independent and possibly throwing, so it runs before any GL work.
-        CheckPrimitiveRestartSupported(type);
+        // Read before any GL work, because it decides the tier below: a desktop restart index
+        // the driver does not know about can only be honoured by the tier that rewrites the
+        // index stream (see ResolveTierForBatch). A restart index this index type cannot hold
+        // needs no rewrite at all - nothing can match it - but it does need the driver's own
+        // fixed-index restart held off for the batch, which is what the scope below does.
+        const RestartSubstitutionKind restartKind = ResolveRestartSubstitution(type);
+        const Bool arbitraryRestart = restartKind == RestartSubstitutionKind::RewriteIndices;
+        const ScopedSuppressedPrimitiveRestart restartCapOverride(restartKind);
 
         const Bool hasIndexBuffer = BoundIndexBuffer() != nullptr;
 
@@ -870,7 +929,9 @@ void main() {
         if (flattened.indexCount != 0) {
             const Uint previousIndexBinding = BoundIndexBufferId();
             BufferImpl::BindBufferId(GL_ELEMENT_ARRAY_BUFFER, flattened.bufferId);
-            g_GLESFuncs.glDrawElements(mode, static_cast<GLsizei>(flattened.indexCount), GL_UNSIGNED_INT, nullptr);
+            ForEachViewportRoutingPass([&] {
+                g_GLESFuncs.glDrawElements(mode, static_cast<GLsizei>(flattened.indexCount), GL_UNSIGNED_INT, nullptr);
+            });
             BufferImpl::BindBufferId(GL_ELEMENT_ARRAY_BUFFER, previousIndexBinding);
             return;
         }
@@ -879,7 +940,8 @@ void main() {
         // the tier choice and the per-sub-draw feeds use those, not the guess above.
         const Bool feedDrawID = CurrentProgramReadsDrawID();
         const Bool feedBaseVertex = basevertex != nullptr && CurrentProgramReadsBaseVertex();
-        const GLESMultiDrawMode tier = ResolveTierForBatch(feedDrawID, feedBaseVertex, hasIndexBuffer);
+        const GLESMultiDrawMode tier =
+            ResolveTierForBatch(feedDrawID, feedBaseVertex, hasIndexBuffer, arbitraryRestart);
 
         Bool drawn = false;
         switch (tier) {
@@ -911,8 +973,10 @@ void main() {
         // Every tier above may decline a batch whose shape it cannot express. The two
         // below are the floor: a base-vertex replay where the driver has one, and the
         // rewritten index stream where it does not. Both are safe for any batch these
-        // entry points can receive.
-        if (!drawn) {
+        // entry points can receive - except that the base-vertex replay hands the
+        // application's own indices to the driver, which cannot restart on a desktop
+        // restart index, so that batch has only the rewriting floor.
+        if (!drawn && !arbitraryRestart) {
             drawn = RunBaseVertexLoop(mode, count, type, indices, drawcount, basevertex, feedDrawID, feedBaseVertex);
         }
         if (!drawn) {

MobileGL/MG_Backend/DirectGLES/Utils.cpp

@@ -11,9 +11,13 @@
 #include "Managers.h"
 #include "MG_Backend/BackendObjects.h"
 #include "MG_Util/Converters/GLToMG/FramebufferEnumConverter.h"
+#include "MG_Util/SelfTest/DriverBugProbes.h"
 #include "MG_Util/Texture/TextureFormatProcessor.h"
+#include "MG_Util/ShaderTranspiler/ShaderCompiler.h"
+#include <Config.h>
 
 #include <MG_State/GLState/Core.h>
+#include <MG_Pipe/PipeInputsSwitch.h>
 #include <MG_Util/BackendLoaders/OpenGL/Loader.h>
 #include <MG_Util/Converters/GLToStr/GLEnumConverter.h>
 #include <MG_Util/Converters/MGToGL/TextureEnumConverter.h>
@@ -25,6 +29,7 @@
 #include <cmath>
 #include <cctype>
 #include <cstring>
+#include <format>
 #include <regex>
 
 namespace MobileGL::MG_Backend::DirectGLES {
@@ -123,6 +128,14 @@ namespace MobileGL::MG_Backend::DirectGLES {
                     requestedInternalFormat,
                     TextureImpl::GetRenderTargetNormalizeOptions(g_GLESCapabilities, targetIndex));
             }
+            // Outside the caveat branch on purpose: the driver CAN create the native narrow
+            // storage - the capability probes say so - it just cannot be trusted as a raw-copy
+            // endpoint. Texture and renderbuffer targets both come through here, which is what
+            // keeps a renderbuffer -> texture copy of these formats same-ES-format when the
+            // widening engages.
+            if (TextureImpl::UsesWidenedPacked16NormStorage(internalFormat)) {
+                options |= PixelFormatNormalizeOptionBit::WidenPacked16Norm;
+            }
             NormalizePixelFormat(requestedInternalFormat, options, outInternalFormat, outFormat, outType);
         }
     } // namespace
@@ -171,9 +184,45 @@ namespace MobileGL::MG_Backend::DirectGLES {
             if (!capabilities.SupportsRenderSnorm || !capabilities.SupportsNorm16Texture) {
                 options |= PixelFormatNormalizeOptionBit::NoSnorm16RenderTarget;
             }
+            // 8-bit signed-normalized storage is core ES, so only the rendering half is in
+            // question here; the 16-bit bit above additionally needs EXT_texture_norm16 for the
+            // encoding to exist at all.
+            if (!capabilities.SupportsRenderSnorm) {
+                options |= PixelFormatNormalizeOptionBit::NoSnorm8RenderTarget;
+            }
             return options;
         }
 
+        Bool UsesWidenedPacked16NormStorage(TextureInternalFormat internalFormat) {
+            switch (internalFormat) {
+            // TextureInternalFormat::RGB5 is both GL_RGB5 and GL_RGB565 - the GL-to-MG
+            // converter folds the two spellings onto one logical format.
+            case TextureInternalFormat::RGB5:
+            case TextureInternalFormat::RGB5A1:
+            case TextureInternalFormat::RGBA4:
+                break;
+            default:
+                return false;
+            }
+            switch (MG_Config::Features.EsprytWidenPacked16Storage) {
+            case MG_Config::QuirkOverride::ForceOn:
+                return true;
+            case MG_Config::QuirkOverride::ForceOff:
+                return false;
+            case MG_Config::QuirkOverride::Auto:
+                break;
+            }
+            // Behind the backend gate on purpose: the memoized probe latches its first answer
+            // for the whole process, and before the backend is up the GL function table may
+            // not be resolved yet - a probe run then would latch "cannot tell" as "clean"
+            // forever. Once the backend exists, the first narrow-format image this process
+            // creates runs the probe on a live context.
+            if (pActiveBackendObject == nullptr) {
+                return false;
+            }
+            return MG_Util::SelfTest::CopyImageMirrorsPacked16FieldOrder(g_GLESFuncs);
+        }
+
         void GenerateTextureFormatInfo(TextureInternalFormat internalFormat, GLenum* outInternalFormat,
                                        GLenum* outFormat, GLenum* outType, TextureTarget target) {
 #ifdef TRACY_ENABLE
@@ -227,6 +276,105 @@ namespace MobileGL::MG_Backend::DirectGLES {
         Bool BackendRenderbufferFormatAddsAlpha(TextureInternalFormat internalFormat) {
             return BackendFormatAddsAlpha(internalFormat, GetRenderbufferFormatCapabilityTargetIndex());
         }
+
+        ImageBindableStorageWidening GetImageBindableStorageWidening(TextureInternalFormat internalFormat) {
+            const GLenum requested = MG_Util::ConvertTextureInternalFormatToGLEnum(internalFormat);
+            const auto carrier = static_cast<GLenum>(
+                MG_Util::ShaderTranspiler::ShaderCompiler::WidenedCoreEsslImageFormat(requested));
+            if (carrier == 0) {
+                return {};
+            }
+            // EXACTLY the arming WidenImageFormatsForEssl uses, and it has to be: the shader, the
+            // storage and the bind must all widen or none of them may, or the shader addresses a
+            // texel size the storage does not have (which every driver tested accepts silently,
+            // reading and writing out of bounds).
+            //
+            // A driver WITH GL_NV_image_formats can spell the narrow format - but only for the
+            // formats SPIRV-Cross will actually print. It throws for its is_desktop_only_format
+            // set instead of emitting a token, and the throw loses the stage whatever the driver
+            // would have accepted: on Mesa, which advertises the extension, `layout(r8ui)
+            // uimage2D` still lost its whole program until the widening ran for it too.
+            if (g_GLESCapabilities.SupportsExtendedImageFormats &&
+                MG_Util::ShaderTranspiler::ShaderCompiler::SpirvCrossCanPrintEsslImageFormat(requested)) {
+                return {};
+            }
+            ImageBindableStorageWidening widening;
+            widening.InternalFormat = carrier;
+            widening.SourceChannels =
+                MG_Util::ShaderTranspiler::ShaderCompiler::ImageFormatChannelCount(requested);
+            switch (carrier) {
+            case GL_RGBA32UI:
+            case GL_RGBA16UI:
+            case GL_RGBA8UI:
+            case GL_RGBA32I:
+            case GL_RGBA16I:
+            case GL_RGBA8I:
+                widening.IntegerData = true;
+                break;
+            default:
+                widening.IntegerData = false;
+                break;
+            }
+            // The carrier is a core ES format in every case, so it needs no fallback options of
+            // its own; this call is only here to spell the transfer pair that describes it.
+            MG_Util::TextureFormatProcessor::NormalizePixelFormat(carrier, Flags<PixelFormatNormalizeOptionBit>{},
+                                                                  nullptr, &widening.Format, &widening.Type);
+            // The two carriers that are not channel widenings, whose transfer pair has to say so.
+            // Every other entry keeps the frontend format's own component type - a GL_RG16F shadow
+            // is halves and so is its GL_RGBA16F carrier, so padding the channels is the whole
+            // conversion. These two shadows are a PACKED 32-bit word per texel
+            // (TextureFormatProcessor::NormalizePixelFormat), and no ES driver accepts either
+            // packed type for the carrier's level, so the transfer names the carrier's own layout
+            // and PrepareImageWidenedUpload splits the word into it.
+            switch (internalFormat) {
+            case TextureInternalFormat::R11FG11FB10F:
+                // GL_UNSIGNED_INT_10F_11F_11F_REV -> GL_RGBA / GL_FLOAT, legal for GL_RGBA16F.
+                widening.Format = GL_RGBA;
+                widening.Type = GL_FLOAT;
+                widening.SourceEncoding = ImageWidenSourceEncoding::PackedFloat11f11f10f;
+                break;
+            case TextureInternalFormat::RGB10A2UI:
+            case TextureInternalFormat::RGB10A2:
+                // GL_UNSIGNED_INT_2_10_10_10_REV -> the GL_RGBA_INTEGER / GL_UNSIGNED_SHORT the
+                // GL_RGBA16UI carrier already asked for above; only the split is new. The two
+                // formats share it: rgb10_a2's channel codes are the same fields rgb10_a2ui's are,
+                // and what the shader divides them by is not the transfer's business.
+                widening.SourceEncoding = ImageWidenSourceEncoding::PackedInt2101010Rev;
+                break;
+            default:
+                break;
+            }
+            // The seven normalized formats whose carrier holds CODES rather than values. Both
+            // halves of the transfer need to know: a missing alpha is padded with the saturated
+            // code rather than the integer 1, and glGetTexImage has to divide the codes back out.
+            bool signedNormalized = false;
+            Uint32 channelMax[4] = {0u, 0u, 0u, 0u};
+            if (MG_Util::ShaderTranspiler::ShaderCompiler::NormalizedImageCarrierCodes(requested, channelMax,
+                                                                                       signedNormalized)) {
+                for (SizeT channel = 0; channel < 4; ++channel) {
+                    widening.ChannelMax[channel] = channelMax[channel];
+                }
+                widening.SignedNormalized = signedNormalized;
+            }
+            return widening;
+        }
+
+        GLenum GetImageBindableBufferSplitFormat(TextureInternalFormat internalFormat) {
+            const GLenum requested = MG_Util::ConvertTextureInternalFormatToGLEnum(internalFormat);
+            const auto base = static_cast<GLenum>(
+                MG_Util::ShaderTranspiler::ShaderCompiler::SplitCoreEsslBufferImageFormat(requested));
+            if (base == 0) {
+                return GL_UNKNOWN_MGL;
+            }
+            // EXACTLY the arming WidenImageFormatsForEssl uses, for the reason the widening's is:
+            // the shader, the glTexBuffer view and the glBindImageTexture argument must all split
+            // or none of them may, or the shader subscripts a view the buffer is not described as.
+            if (g_GLESCapabilities.SupportsExtendedImageFormats &&
+                MG_Util::ShaderTranspiler::ShaderCompiler::SpirvCrossCanPrintEsslImageFormat(requested)) {
+                return GL_UNKNOWN_MGL;
+            }
+            return base;
+        }
     } // namespace TextureImpl
     namespace PrgramImpl {
         String ProcessOutColorLocations(const String& glslCode) {
@@ -569,6 +717,84 @@ namespace MobileGL::MG_Backend::DirectGLES {
             return glslCode;
         }
 
+        String RequestViewportArrayExtension(String glslCode, Bool needed) {
+#ifdef TRACY_ENABLE
+            ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
+#endif
+            // gl_ViewportIndex is desktop GL 4.1 core and is in ESSL only under
+            // GL_OES_viewport_array. SPIRV-Cross prints the identifier as-is and requests no
+            // extension for it - three lines away from the BuiltInLayer case, which DOES ask for
+            // one on ES - so an untouched decompile reaches the driver naming a builtin its core
+            // language has never heard of. The stage then fails to compile, the program is marked
+            // unusable and every draw made with it renders nothing while raising no GL error.
+            //
+            // Same `needed` contract as RequestExtendedImageFormats, and the same hard rule:
+            // `#extension` on a name the driver does not advertise is itself a compile error
+            // (ARM's compiler is strict about it), so this must never be emitted speculatively.
+            // A driver without the extension does not come through here at all - its module took
+            // the LowerViewportIndexPass fallback and the emitted source no longer names the
+            // builtin.
+            static constexpr const char* kDirective = "#extension GL_OES_viewport_array : require\n";
+            static constexpr const char* kExtName = "GL_OES_viewport_array";
+            if (!needed || glslCode.find(kExtName) != String::npos) {
+                return glslCode;
+            }
+            // Right after the #version line, for the reason spelled out above: it is the only
+            // position that must stay first, and ForceSupporterOutput's scan for the LAST
+            // #extension directive still finds whichever one that ends up being.
+            const SizeT versionPos = glslCode.find("#version");
+            if (versionPos == String::npos) {
+                return kDirective + glslCode;
+            }
+            const SizeT lineEnd = glslCode.find('\n', versionPos);
+            if (lineEnd == String::npos) {
+                return glslCode + "\n" + kDirective;
+            }
+            glslCode.insert(lineEnd + 1, kDirective);
+            return glslCode;
+        }
+
+        const char* PointSizeExtensionName(MG_External::GLESCapabilities::PointSizeTier tier, Bool tessellation) {
+            using Tier = MG_External::GLESCapabilities::PointSizeTier;
+            switch (tier) {
+                case Tier::ExtensionEXT:
+                    return tessellation ? "GL_EXT_tessellation_point_size" : "GL_EXT_geometry_point_size";
+                case Tier::ExtensionOES:
+                    return tessellation ? "GL_OES_tessellation_point_size" : "GL_OES_geometry_point_size";
+                default:
+                    return nullptr;
+            }
+        }
+
+        String RequestPointSizeExtension(String glslCode, const char* extensionName) {
+#ifdef TRACY_ENABLE
+            ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
+#endif
+            // The gl_ViewportIndex story, one built-in over: ESSL 320 makes the tessellation and
+            // geometry STAGES core but leaves gl_PointSize out of their gl_PerVertex entirely,
+            // and SPIRV-Cross - which only ever sees a SPIR-V BuiltIn PointSize decoration -
+            // prints the identifier with no directive behind it. Same hard rule as the two
+            // neighbours: never emitted speculatively, because `#extension` on a name the driver
+            // does not advertise is a compile error of its own.
+            if (extensionName == nullptr || glslCode.find(extensionName) != String::npos) {
+                return glslCode;
+            }
+            const String directive = String("#extension ") + extensionName + " : require\n";
+            // Right after the #version line, the one position that must stay first;
+            // ForceSupporterOutput's scan for the LAST #extension directive still finds
+            // whichever one that ends up being.
+            const SizeT versionPos = glslCode.find("#version");
+            if (versionPos == String::npos) {
+                return directive + glslCode;
+            }
+            const SizeT lineEnd = glslCode.find('\n', versionPos);
+            if (lineEnd == String::npos) {
+                return glslCode + "\n" + directive;
+            }
+            glslCode.insert(lineEnd + 1, directive);
+            return glslCode;
+        }
+
         String BakeImageFormatQualifiers(String glslCode,
                                          const UnorderedMap<String, String>& esslFormatByUniformName) {
 #ifdef TRACY_ENABLE
@@ -657,6 +883,86 @@ namespace MobileGL::MG_Backend::DirectGLES {
             return result;
         }
 
+        std::optional<String> ExtractPerVertexBlockMembers(const String& essl, const Bool input) {
+#ifdef TRACY_ENABLE
+            ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
+#endif
+            // Deliberately a scan for the DECLARATION rather than a regex over the whole text:
+            // "gl_PerVertex" also appears inside the block's own body in some emissions, and the
+            // direction keyword has to be the one immediately preceding the name for the match to
+            // mean what this needs it to mean.
+            const auto isIdentifierChar = [](char c) {
+                return std::isalnum(static_cast<unsigned char>(c)) != 0 || c == '_';
+            };
+            const String keyword = input ? String("in") : String("out");
+            SizeT pos = 0;
+            while ((pos = essl.find("gl_PerVertex", pos)) != String::npos) {
+                // Walk back over whitespace to the direction keyword.
+                SizeT before = pos;
+                while (before > 0 && std::isspace(static_cast<unsigned char>(essl[before - 1]))) --before;
+                const Bool matches = before >= keyword.size() &&
+                                     essl.compare(before - keyword.size(), keyword.size(), keyword) == 0 &&
+                                     (before == keyword.size() ||
+                                      !isIdentifierChar(essl[before - keyword.size() - 1]));
+                if (!matches) {
+                    pos += 1;
+                    continue;
+                }
+                const SizeT open = essl.find('{', pos);
+                if (open == String::npos) return std::nullopt;
+                const SizeT close = essl.find('}', open);
+                if (close == String::npos) return std::nullopt;
+                return essl.substr(open + 1, close - open - 1);
+            }
+            return std::nullopt;
+        }
+
+        String BuildPassthroughTessControlEssl(const Uint esslVersion, const Uint patchVertices,
+                                               const String& inPerVertexMembers,
+                                               const String& outPerVertexMembers,
+                                               const FloatVec4& defaultOuterLevel,
+                                               const FloatVec2& defaultInnerLevel) {
+#ifdef TRACY_ENABLE
+            ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
+#endif
+            // Tessellation is core in ES 3.2 and reachable in 3.1 only through
+            // GL_EXT_tessellation_shader. The caller has already established that the driver runs
+            // the evaluation stage at all, so the only question here is which spelling to use.
+            const Bool core = esslVersion >= 320;
+            String source = "#version " + std::to_string(core ? 320u : 310u) + " es\n";
+            if (!core) {
+                source += "#extension GL_EXT_tessellation_shader : require\n";
+            }
+            source += "precision highp float;\n";
+            source += "precision highp int;\n";
+            source += "layout(vertices = " + std::to_string(patchVertices) + ") out;\n";
+            // Mirrored, never invented. An empty member list means the neighbouring stage did not
+            // redeclare the block either, and the driver's own built-in declaration is then what
+            // both sides agree on - redeclaring here would be the thing that broke the match.
+            if (!inPerVertexMembers.empty()) {
+                source += "in gl_PerVertex {" + inPerVertexMembers + "} gl_in[gl_MaxPatchVertices];\n";
+            }
+            if (!outPerVertexMembers.empty()) {
+                source += "out gl_PerVertex {" + outPerVertexMembers + "} gl_out[];\n";
+            }
+            source += "void main() {\n";
+            // Only gl_Position is forwarded. That is the whole of what the pass-through owes the
+            // evaluation stage: a program whose evaluation stage reads anything else per-vertex
+            // was declined before this was ever called (ModuleReadsLocatedInput), and gl_PointSize
+            // from a tessellation stage is a separate capability on both targets.
+            source += "    gl_out[gl_InvocationID].gl_Position = gl_in[gl_InvocationID].gl_Position;\n";
+            for (Uint i = 0; i < 4; ++i) {
+                source += "    gl_TessLevelOuter[" + std::to_string(i) +
+                          "] = " + MG_Util::ShaderTranspiler::TessellationLevelLiteral(defaultOuterLevel[i]) + ";\n";
+            }
+            for (Uint i = 0; i < 2; ++i) {
+                source += "    gl_TessLevelInner[" + std::to_string(i) +
+                          "] = " + MG_Util::ShaderTranspiler::TessellationLevelLiteral(defaultInnerLevel[i]) + ";\n";
+            }
+            source += "}\n";
+            return source;
+        }
+
         namespace {
             Bool IsImagePassIdentifierChar(char c) {
                 return std::isalnum(static_cast<unsigned char>(c)) || c == '_';
@@ -768,6 +1074,8 @@ namespace MobileGL::MG_Backend::DirectGLES {
 
             struct ImageUniformDecl {
                 String name;
+                String aliasName;   // the repair-tagged name the rewritten declaration takes; empty
+                                    // for a declaration this pass leaves alone
                 String writeName;   // the writeonly half's name, when split
                 String layout;      // raw contents of layout(...)
                 String qualifiers;  // memory/precision qualifiers, normalized, no trailing space
@@ -775,19 +1083,35 @@ namespace MobileGL::MG_Backend::DirectGLES {
                 String arraySuffix; // "" or "[7]"
                 SizeT declStart = 0;
                 SizeT declLength = 0;
+                SizeT nameStart = 0;   // the name token alone, for a rename that edits nothing else
+                SizeT nameLength = 0;
                 SizeT referenceCount = 0; // uses this pass recognized and accounted for
                 Bool loaded = false;
                 Bool stored = false;
                 Bool unknownUse = false;
                 Bool split = false;
+                // SPIRV-Cross already tagged this one readonly or writeonly, so it needs no
+                // qualifier repair - only the rename that keeps two stages from merging it.
+                Bool preTaggedReadonly = false;
+                Bool preTaggedWriteonly = false;
             };
 
             // A rebuilt declaration. Keeps SPIRV-Cross's own word order (`uniform readonly
             // highp image2D`) so the image-rebinding regex in Managers.cpp still matches what
             // comes out of here, whichever order the two passes end up running in.
+            //
+            // `forceCoherent` is for the SPLIT pair only. GLSL guarantees that a write through
+            // one image variable is visible to a read through a DIFFERENT one only when both are
+            // declared coherent, and the split turns a same-variable read-after-write - which
+            // desktop GLSL orders by construction, so the source almost never says `coherent` -
+            // into exactly that cross-variable shape. Without it the driver may serve the load
+            // from a cache that never saw the store through the writeonly half.
             String BuildImageDeclaration(const ImageUniformDecl& decl, const char* memoryQualifier,
-                                         const String& variableName) {
+                                         const String& variableName, Bool forceCoherent = false) {
                 String out = "layout(" + decl.layout + ") uniform ";
+                if (forceCoherent && !ContainsIdentifier(decl.qualifiers, "coherent")) {
+                    out += "coherent ";
+                }
                 out += memoryQualifier;
                 out += ' ';
                 if (!decl.qualifiers.empty()) {
@@ -802,11 +1126,11 @@ namespace MobileGL::MG_Backend::DirectGLES {
                 return out;
             }
 
-            // A name for the writeonly half that no identifier in the shader (and no other
-            // half already minted) can collide with.
-            String MakeImageWriteAliasName(const String& name, const String& source,
-                                           const Vector<String>& taken) {
-                String candidate = String(IMAGE_WRITE_ALIAS_PREFIX) + name;
+            // A name for a rewritten declaration that no identifier in the shader (and no other
+            // alias already minted for this stage) can collide with.
+            String MakeImageAliasName(const String& prefix, const String& name, const String& source,
+                                      const Vector<String>& taken) {
+                String candidate = prefix + name;
                 // "__" anywhere in an identifier is reserved (GLSL ES 3.20 3.7), which a name
                 // that already starts with '_' would otherwise produce.
                 for (SizeT doubled = candidate.find("__"); doubled != String::npos;
@@ -829,12 +1153,265 @@ namespace MobileGL::MG_Backend::DirectGLES {
                 SizeT length;
                 String text;
             };
+
+            // The offset just past the `;` that terminates the call whose argument list opens at
+            // `openParen`, or npos when what follows is not a plain statement. Parentheses alone
+            // are counted: every other bracket a GLSL argument list can contain is balanced
+            // inside them, and imageStore returns void, so a well-formed call site is always
+            // `imageStore(...);` and anything else is a shape this pass declines to edit.
+            SizeT FindEndOfCallStatement(const String& code, SizeT openParen) {
+                Int depth = 0;
+                SizeT scan = openParen;
+                for (; scan < code.size(); ++scan) {
+                    if (code[scan] == '(') {
+                        ++depth;
+                    } else if (code[scan] == ')' && --depth == 0) {
+                        break;
+                    }
+                }
+                if (scan >= code.size()) return String::npos;
+                const SizeT after = code.find_first_not_of(" \t\r\n", scan + 1);
+                if (after == String::npos || code[after] != ';') return String::npos;
+                return after + 1;
+            }
         } // namespace
 
-        String SplitReadWriteImageUniforms(const String& glslCode) {
+        namespace {
+            // The digits of an array extent or of an element subscript, or -1 for "not a plain
+            // decimal literal".
+            //
+            // One trailing `u`/`U` is PART of the literal rather than grounds for rejection.
+            // SPIRV-Cross prints an index in the type SPIR-V gave it, and
+            // LegalizeResourceArrayIndexPass mints its per-element constants in the type of the
+            // index it replaced (ConstantLikeIndex reads that index's own type_id), so an image
+            // array reached through anything unsigned - `for (uint i = 0u; i < 4u; ++i)`, or any
+            // expression on gl_LocalInvocationIndex, which is uint by definition - arrives here
+            // spelled `g_image[0u]`. Reading that as "not a literal" declined the array and left
+            // it on one layout(binding = N), which hands its elements the consecutive units
+            // N, N+1, ... - exactly the silently-wrong-units defect the split exists to remove.
+            Int ParseNonNegativeIntLiteral(const String& text) {
+                if (text.empty()) return -1;
+                SizeT digitCount = text.size();
+                if (text[digitCount - 1] == 'u' || text[digitCount - 1] == 'U') --digitCount;
+                if (digitCount == 0) return -1;
+                Int value = 0;
+                for (SizeT i = 0; i < digitCount; ++i) {
+                    const char c = text[i];
+                    if (c < '0' || c > '9') return -1;
+                    value = value * 10 + (c - '0');
+                    if (value > 4096) return -1; // no image array is anywhere near this
+                }
+                return value;
+            }
+        } // namespace
+
+        String RemapImageArrayElementUnits(const String& glslCode, const Vector<ImageArrayUnitPlan>& plans,
+                                           Vector<String>* outDeclined) {
 #ifdef TRACY_ENABLE
             ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
+            if (outDeclined != nullptr) outDeclined->clear();
+            if (plans.empty() || glslCode.find("image") == String::npos) return glslCode;
+
+            // Same declaration shape as the split pass reads, with the array extent captured.
+            static const std::regex imageDeclRegex(
+                R"(layout\s*\(([^)]*)\)\s*uniform\s+)"
+                R"(((?:(?:readonly|writeonly|coherent|volatile|restrict|highp|mediump|lowp)\s+)*))"
+                R"(([iu]?image[A-Za-z0-9_]*)\s+([A-Za-z_][A-Za-z0-9_]*)\s*(?:\[\s*([0-9]*)\s*\])?\s*;)");
+            static const std::regex bindingValueRegex(R"(binding\s*=\s*\d+)");
+
+            struct StageImageDecl {
+                String name;
+                String layout;
+                String qualifiers;
+                String type;
+                Int elementCount = 1;
+                SizeT declStart = 0;
+                SizeT declLength = 0;
+            };
+            // Every image declaration in the stage; the plans are program-wide and name arrays
+            // this stage may not declare at all.
+            Vector<StageImageDecl> decls;
+            for (std::sregex_iterator it(glslCode.begin(), glslCode.end(), imageDeclRegex), last; it != last; ++it) {
+                const std::smatch& match = *it;
+                StageImageDecl decl;
+                decl.layout = match[1].str();
+                decl.qualifiers = NormalizeDeclarationSpacing(match[2].str());
+                decl.type = match[3].str();
+                decl.name = match[4].str();
+                decl.elementCount = match[5].matched ? ParseNonNegativeIntLiteral(match[5].str()) : 1;
+                decl.declStart = static_cast<SizeT>(match.position(0));
+                decl.declLength = match[0].str().size();
+                decls.push_back(Move(decl));
+            }
+
+            Vector<ImageSourceEdit> edits;
+            Vector<String> takenNames;
+            for (const ImageArrayUnitPlan& plan : plans) {
+                const auto decline = [&](const char* why) {
+                    if (outDeclined != nullptr) outDeclined->push_back(plan.name + ": " + why);
+                };
+                if (plan.units.size() < 2) continue;
+
+                const StageImageDecl* decl = nullptr;
+                for (const auto& candidate : decls) {
+                    if (candidate.name == plan.name) {
+                        decl = &candidate;
+                        break;
+                    }
+                }
+                if (decl == nullptr) {
+                    // Absent from this stage entirely is the normal outcome - the reflection is
+                    // program-wide and this pass runs per stage. Named but not RECOGNIZED is not:
+                    // it means the declaration is spelled in some shape the regex above does not
+                    // read, and staying quiet about that is how the wrong units got shipped.
+                    if (ContainsIdentifier(glslCode, plan.name)) {
+                        decline("the stage names it but declares it in a shape this pass cannot read");
+                    }
+                    continue;
+                }
+                if (decl->elementCount < 0 || static_cast<SizeT>(decl->elementCount) != plan.units.size()) {
+                    decline("the emitted array extent disagrees with the reflected element count");
+                    continue;
+                }
+
+                Bool consecutive = true;
+                Bool everyElementHasAUnit = true;
+                for (SizeT element = 0; element < plan.units.size(); ++element) {
+                    const Int unit = plan.units[element];
+                    if (unit < 0) {
+                        everyElementHasAUnit = false;
+                        break;
+                    }
+                    if (unit != plan.units[0] + static_cast<Int>(element)) consecutive = false;
+                }
+                if (!everyElementHasAUnit) {
+                    decline("an element has no image unit");
+                    continue;
+                }
+                // Already exactly what ESSL would do on its own. The caller filters these out;
+                // repeating the test here keeps the pass correct on its own terms.
+                if (consecutive) continue;
+
+                // Every use has to be `name[<literal>]`. The literal is what the split turns
+                // into a name, and by the time this runs there is always one:
+                // LegalizeResourceArrayIndexingForEssl has already folded or lowered every
+                // dynamic image-array subscript in the module, because ESSL forbids one
+                // outright ("image arrays indexed with non-constant expressions are forbidden
+                // in GLSL ES"). A subscript that is still an expression here is therefore a
+                // stage that was never going to compile, and guessing which element it meant
+                // would only change which unit it addressed wrongly.
+                struct ElementUse {
+                    SizeT start;   // the first character of the name
+                    SizeT length;  // through the closing ']'
+                    SizeT element;
+                };
+                Vector<ElementUse> uses;
+                const char* refusal = nullptr;
+                for (SizeT pos = glslCode.find(plan.name); pos != String::npos;
+                     pos = glslCode.find(plan.name, pos + 1)) {
+                    if (pos > 0 && IsImagePassIdentifierChar(glslCode[pos - 1])) continue;
+                    const SizeT after = pos + plan.name.size();
+                    if (after < glslCode.size() && IsImagePassIdentifierChar(glslCode[after])) continue;
+                    if (pos >= decl->declStart && pos < decl->declStart + decl->declLength) {
+                        continue; // the declaration's own name
+                    }
+                    const SizeT open = glslCode.find_first_not_of(" \t\r\n", after);
+                    if (open == String::npos || glslCode[open] != '[') {
+                        refusal = "it is reached by something other than a subscript, so there is no "
+                                  "element index to rewrite";
+                        break;
+                    }
+                    Int depth = 0;
+                    SizeT scan = open;
+                    for (; scan < glslCode.size(); ++scan) {
+                        if (glslCode[scan] == '[') {
+                            ++depth;
+                        } else if (glslCode[scan] == ']' && --depth == 0) {
+                            break;
+                        }
+                    }
+                    if (scan >= glslCode.size() || open + 1 >= scan) {
+                        refusal = "it is reached by something other than a subscript, so there is no "
+                                  "element index to rewrite";
+                        break;
+                    }
+                    const Int element = ParseNonNegativeIntLiteral(
+                        NormalizeDeclarationSpacing(glslCode.substr(open + 1, scan - open - 1)));
+                    if (element < 0 || element >= decl->elementCount) {
+                        refusal = "its subscript is not a literal element index, so which unit the "
+                                  "access reaches cannot be decided here";
+                        break;
+                    }
+                    uses.push_back({pos, scan + 1 - pos, static_cast<SizeT>(element)});
+                }
+                if (refusal != nullptr) {
+                    decline(refusal);
+                    continue;
+                }
+
+                // One SCALAR declaration per element, each carrying its own binding. ESSL nails
+                // an ARRAY's elements to consecutive units and offers no way to move them, so
+                // the only spelling that reaches an arbitrary set of units is one declaration
+                // per unit - and with every subscript a literal, every use has exactly one of
+                // them to be rewritten to.
+                //
+                // It costs precisely the image uniforms the application declared, which is why
+                // there is no budget test here: an array of four elements becomes four scalars
+                // however far apart their units are.
+                const SizeT elementCount = plan.units.size();
+                Vector<String> elementNames;
+                String replacement;
+                for (SizeT element = 0; element < elementCount; ++element) {
+                    const String elementName =
+                        MakeImageAliasName(IMAGE_ARRAY_ELEMENT_PREFIX,
+                                           plan.name + "_" + std::to_string(element), glslCode, takenNames);
+                    takenNames.push_back(elementName);
+                    elementNames.push_back(elementName);
+
+                    String layout = decl->layout;
+                    const String bindingText = "binding = " + std::to_string(plan.units[element]);
+                    if (std::regex_search(layout, bindingValueRegex)) {
+                        layout = std::regex_replace(layout, bindingValueRegex, bindingText);
+                    } else {
+                        layout = bindingText + (layout.empty() ? String() : ", " + layout);
+                    }
+                    if (element != 0) replacement += '\n';
+                    replacement += "layout(" + layout + ") uniform ";
+                    if (!decl->qualifiers.empty()) {
+                        replacement += decl->qualifiers;
+                        replacement += ' ';
+                    }
+                    replacement += decl->type + " " + elementName + ";";
+                }
+                edits.push_back({decl->declStart, decl->declLength, Move(replacement)});
+
+                // `name[k]` -> the scalar declared for element k, subscript and all.
+                for (const ElementUse& use : uses) {
+                    edits.push_back({use.start, use.length, elementNames[use.element]});
+                }
+            }
+            if (edits.empty()) return glslCode;
+
+            // Back to front, so an earlier edit's offsets stay valid. No two edits overlap: each
+            // one covers either a whole declaration or a whole `name[k]`, the declaration's own
+            // name is skipped when the uses are collected, and one occurrence of a name yields at
+            // most one edit.
+            std::sort(edits.begin(), edits.end(),
+                      [](const ImageSourceEdit& a, const ImageSourceEdit& b) { return a.start > b.start; });
+            String result = glslCode;
+            for (const ImageSourceEdit& edit : edits) {
+                result.replace(edit.start, edit.length, edit.text);
+            }
+            return result;
+        }
+
+        String SplitReadWriteImageUniforms(const String& glslCode, Uint* outSplitCount) {
+#ifdef TRACY_ENABLE
+            ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
+#endif
+            // Written before any early return, so the caller never reads a stale count.
+            if (outSplitCount != nullptr) *outSplitCount = 0;
             if (glslCode.find("image") == String::npos) {
                 return glslCode;
             }
@@ -853,10 +1430,12 @@ namespace MobileGL::MG_Backend::DirectGLES {
             for (std::sregex_iterator it(glslCode.begin(), glslCode.end(), imageDeclRegex), last; it != last; ++it) {
                 const std::smatch& match = *it;
                 const String qualifiers = match[2].str();
-                // Already legal: SPIRV-Cross decided one way, leave it alone.
-                if (ContainsIdentifier(qualifiers, "readonly") || ContainsIdentifier(qualifiers, "writeonly")) {
-                    continue;
-                }
+                const Bool hasReadonly = ContainsIdentifier(qualifiers, "readonly");
+                const Bool hasWriteonly = ContainsIdentifier(qualifiers, "writeonly");
+                // Carrying BOTH is a spelling no per-stage access analysis produces (SPIRV-Cross
+                // clears one decoration or the other as soon as it sees a load or a store), so it
+                // came from the application and is identical in every stage. Nothing to do.
+                if (hasReadonly && hasWriteonly) continue;
 
                 Bool hasFormat = false;
                 Bool exemptFormat = false;
@@ -865,10 +1444,12 @@ namespace MobileGL::MG_Backend::DirectGLES {
                     hasFormat = true;
                     exemptFormat = IsMemoryQualifierExemptImageFormat(token);
                 }
-                // No format qualifier at all is a different (and, in ES, unconditionally
-                // illegal) shape that GL_EXT_shader_image_load_formatted would be needed for;
-                // SPIRV-Cross refuses to emit it for an ES target, so nothing to do here.
-                if (!hasFormat || exemptFormat) continue;
+                // A declaration carrying neither qualifier is illegal ES unless its format is
+                // r32f/r32i/r32ui, and no format qualifier at all is a shape SPIRV-Cross refuses
+                // to emit for an ES target. Either way there is no repair to make - and no rename
+                // to make either, because a declaration with no access qualifier is spelled the
+                // same in every stage.
+                if (!hasReadonly && !hasWriteonly && (!hasFormat || exemptFormat)) continue;
 
                 ImageUniformDecl decl;
                 decl.layout = match[1].str();
@@ -878,6 +1459,10 @@ namespace MobileGL::MG_Backend::DirectGLES {
                 decl.arraySuffix = NormalizeDeclarationSpacing(match[5].str());
                 decl.declStart = static_cast<SizeT>(match.position(0));
                 decl.declLength = match[0].str().size();
+                decl.nameStart = static_cast<SizeT>(match.position(4));
+                decl.nameLength = match[4].str().size();
+                decl.preTaggedReadonly = hasReadonly;
+                decl.preTaggedWriteonly = hasWriteonly;
                 decls.push_back(Move(decl));
             }
             if (decls.empty()) {
@@ -892,12 +1477,17 @@ namespace MobileGL::MG_Backend::DirectGLES {
             };
 
             // Walk every `image*(` call and attribute its first argument to a declaration.
-            struct StoreSite {
+            // EVERY recognized use is recorded, not only the stores: a declaration this pass
+            // renames has to take all of its uses with it, and the "every occurrence was one I
+            // saw" check below is what makes the recorded set provably the complete set.
+            struct ImageUseSite {
                 SizeT declIndex;
                 SizeT start;
                 SizeT length;
+                SizeT callOpen; // the '(' of the call this argument belongs to
+                Bool stores;    // an imageStore, i.e. the use a split redirects to the write half
             };
-            Vector<StoreSite> storeSites;
+            Vector<ImageUseSite> useSites;
             for (SizeT pos = glslCode.find("image"); pos != String::npos; pos = glslCode.find("image", pos + 1)) {
                 if (pos > 0 && IsImagePassIdentifierChar(glslCode[pos - 1])) continue; // uimage2D, myimageFoo
                 SizeT tokenEnd = pos;
@@ -942,12 +1532,16 @@ namespace MobileGL::MG_Backend::DirectGLES {
                 switch (ClassifyImageBuiltin(builtin)) {
                 case ImageBuiltinAccess::Load:
                     decl.loaded = true;
+                    useSites.push_back({declIndex, argStart, argEnd - argStart, openParen, false});
                     break;
                 case ImageBuiltinAccess::Store:
                     decl.stored = true;
-                    storeSites.push_back({declIndex, argStart, argEnd - argStart});
+                    useSites.push_back({declIndex, argStart, argEnd - argStart, openParen, true});
                     break;
                 case ImageBuiltinAccess::None:
+                    // imageSize/imageSamples touch nothing, but they still NAME the variable, so
+                    // a rename has to reach them.
+                    useSites.push_back({declIndex, argStart, argEnd - argStart, openParen, false});
                     break;
                 default:
                     decl.unknownUse = true;
@@ -964,30 +1558,122 @@ namespace MobileGL::MG_Backend::DirectGLES {
             }
 
             Vector<ImageSourceEdit> edits;
-            Vector<String> takenAliases;
+            Vector<String> takenNames;
             for (auto& decl : decls) {
                 if (decl.unknownUse) continue; // leave it exactly as it was; no guessing
+                // EVERY declaration this pass rewrites is also RENAMED, under the prefix of the
+                // repair it is about to receive - the qualifier below is a decision about ONE
+                // STAGE's accesses, and GLSL requires a uniform declared in two stages to be
+                // declared IDENTICALLY (GLSL 4.3 4.3.9 / GLSL ES 3.20 4.3.9). A shader that
+                // stores to an image in the vertex stage and loads it in the fragment stage gets
+                // `writeonly` on one and `readonly` on the other, and on Adreno the linker merges
+                // the two same-named declarations and SILENTLY DISCARDS the vertex-stage stores:
+                // no GL error, no link log, LINK_STATUS = 1, and the image still holding its
+                // initial contents afterwards
+                // (KHR-GL4x.shader_image_load_store.advanced-memory-dependentInvocation, and any
+                // shader pack that writes an image in one stage to read it in another).
+                //
+                // Keyed on the REPAIR and not on the stage, which is what makes the rename
+                // exactly as wide as the problem. Two stages that use the image the same way
+                // reach the same prefix and emit byte-identical declarations, so they keep ONE
+                // shared uniform and there is nothing mismatched to merge; two that use it
+                // differently reach different prefixes and cannot be merged at all. Tagging by
+                // stage instead also broke the merge - but it broke it for the agreeing stages
+                // too, turning one image uniform into one PER STAGE that names it, and Adreno
+                // allocates image locations per distinct uniform: the five stages of
+                // KHR-GL43.shading_language_420pack.binding_images_texture_type_* went from 6
+                // image uniforms to 30 and the link failed outright with "Error: Image Image
+                // location or component exceeds max allowed." on an Adreno 830, where Mali and
+                // Mesa both accept the same text.
+                //
+                // Nothing downstream reads these names: the two passes that key on the GL uniform
+                // name (RebindImageUniformsToFrontendUnits, BakeImageFormatQualifiers) both run
+                // BEFORE this one, RemoveLayoutBinding recognises an image declaration by its TYPE
+                // token, and CacheResourceLocations skips image uniforms outright because ES image
+                // units come only from layout(binding=N). The declarations this pass LEAVES ALONE -
+                // already readonly/writeonly in the source, or r32f/r32i/r32ui, which need no
+                // qualifier - keep their names, and they are exactly the ones that already match
+                // across stages.
+                if (decl.preTaggedReadonly || decl.preTaggedWriteonly) {
+                    // No repair: SPIRV-Cross already emitted a legal qualifier. But it derived
+                    // that qualifier from THIS STAGE's accesses, so a uniform stored in one stage
+                    // and loaded in another arrives here `writeonly` in one and `readonly` in the
+                    // other under ONE name - precisely the same-name/mismatched-qualifier pair
+                    // Adreno merges while silently discarding the writing stage's stores
+                    // (advanced-memory-dependentInvocation; a raw-ES probe reproduces it with no
+                    // MobileGL in the process, and renaming either half fixes it). Keyed on the
+                    // qualifier for the same reason the repair below is: two stages that agree
+                    // spell the same alias and stay merged, so no shader gains an image uniform.
+                    const char* preTagPrefix =
+                        decl.preTaggedReadonly ? IMAGE_READONLY_ALIAS_PREFIX : IMAGE_WRITEONLY_ALIAS_PREFIX;
+                    decl.aliasName = MakeImageAliasName(preTagPrefix, decl.name, glslCode, takenNames);
+                    takenNames.push_back(decl.aliasName);
+                    // The name token alone: the qualifiers are already right, and re-emitting the
+                    // whole declaration would only risk changing them.
+                    edits.push_back({decl.nameStart, decl.nameLength, decl.aliasName});
+                    continue;
+                }
+
+                const char* aliasPrefix = decl.loaded && decl.stored ? IMAGE_SPLIT_READ_ALIAS_PREFIX
+                                          : decl.stored             ? IMAGE_WRITEONLY_ALIAS_PREFIX
+                                                                    : IMAGE_READONLY_ALIAS_PREFIX;
+                decl.aliasName = MakeImageAliasName(aliasPrefix, decl.name, glslCode, takenNames);
+                takenNames.push_back(decl.aliasName);
                 if (decl.loaded && decl.stored) {
-                    decl.writeName = MakeImageWriteAliasName(decl.name, glslCode, takenAliases);
-                    takenAliases.push_back(decl.writeName);
+                    // Minted from the ALREADY access-tagged name, so the write half of a split
+                    // can never collide with the single declaration another stage's repair mints
+                    // for the same image.
+                    decl.writeName =
+                        MakeImageAliasName(IMAGE_WRITE_ALIAS_PREFIX, decl.aliasName, glslCode, takenNames);
+                    takenNames.push_back(decl.writeName);
                     decl.split = true;
+                    if (outSplitCount != nullptr) ++*outSplitCount;
+                    // Both halves carry `coherent`; see BuildImageDeclaration. The
+                    // single-declaration cases below stay as they were - nothing aliases them, so
+                    // there is no visibility to restore and no reason to pay for the cache
+                    // behaviour.
                     edits.push_back({decl.declStart, decl.declLength,
-                                     BuildImageDeclaration(decl, "readonly", decl.name) + "\n" +
-                                         BuildImageDeclaration(decl, "writeonly", decl.writeName)});
+                                     BuildImageDeclaration(decl, "readonly", decl.aliasName,
+                                                           /*forceCoherent=*/true) +
+                                         "\n" +
+                                         BuildImageDeclaration(decl, "writeonly", decl.writeName,
+                                                               /*forceCoherent=*/true)});
                 } else if (decl.stored) {
                     edits.push_back({decl.declStart, decl.declLength,
-                                     BuildImageDeclaration(decl, "writeonly", decl.name)});
+                                     BuildImageDeclaration(decl, "writeonly", decl.aliasName)});
                 } else {
                     // Loaded only, or only ever handed to imageSize (or unused): readonly is
                     // the qualifier that keeps every one of those legal.
                     edits.push_back({decl.declStart, decl.declLength,
-                                     BuildImageDeclaration(decl, "readonly", decl.name)});
+                                     BuildImageDeclaration(decl, "readonly", decl.aliasName)});
                 }
             }
-            for (const StoreSite& site : storeSites) {
+            for (const ImageUseSite& site : useSites) {
                 const ImageUniformDecl& decl = decls[site.declIndex];
-                if (!decl.split) continue;
-                edits.push_back({site.start, site.length, decl.writeName});
+                // Empty exactly when the declaration was poisoned above and left untouched; its
+                // uses must keep naming the variable that is still called that.
+                if (decl.aliasName.empty()) continue;
+                edits.push_back(
+                    {site.start, site.length, decl.split && site.stores ? decl.writeName : decl.aliasName});
+                if (!decl.split || !site.stores) continue;
+                // ...and an explicit barrier behind it. `coherent` on both halves is what makes
+                // the store VISIBLE to a load through the other variable, but it says nothing
+                // about ORDER within one invocation - and the whole reason a declaration is split
+                // is that the shader both stores and loads through it, which on the ES side is now
+                // a write to one variable followed by a read of another the compiler has no reason
+                // to believe alias. Adreno duly serves the load from before the store
+                // (KHR-GL4x.shader_image_load_store.advanced-memory-order's store/load/compare
+                // loop reads back the previous iteration's value). memoryBarrierImage() is the
+                // GLSL primitive for exactly that ordering, is core GLSL ES 3.10 in every stage,
+                // and is not an execution barrier, so it is legal in non-uniform control flow too.
+                //
+                // Confined to the split pair: a single-declaration repair has nothing aliasing it
+                // and must not pay for this, and a shader that never got split never sees it at
+                // all.
+                const SizeT statementEnd = FindEndOfCallStatement(glslCode, site.callOpen);
+                if (statementEnd != String::npos) {
+                    edits.push_back({statementEnd, 0, " memoryBarrierImage();"});
+                }
             }
             if (edits.empty()) {
                 return glslCode;
@@ -1609,11 +2295,11 @@ namespace MobileGL::MG_Backend::DirectGLES {
         static Bool StoreClientRows(SizeT dstPixelBytes, SizeT swapGroupSize, GLsizei width, GLsizei sliceHeight,
                                     GLsizei sliceCount, void* pixels, Bool applyPackImageParams, FillRow&& fillRow) {
             const auto& pixelPackBufferObject =
-                MG_State::pGLContext->GetBufferBindingSlot(BufferTarget::PixelPack).GetBoundObject();
+                MGB_CTX->GetBufferBindingSlot(BufferTarget::PixelPack).GetBoundObject();
 
             // Destination layout is computed from the client-side PACK parameters; only the actual pixel
             // rows are written so skip regions of the destination stay untouched.
-            const auto packParams = MG_State::pGLContext->GetPixelStoreParameters(false);
+            const auto packParams = MGB_CTX->GetPixelStoreParameters(false);
             const SizeT rowPixels = static_cast<SizeT>(packParams.RowLength > 0 ? packParams.RowLength : width);
             const SizeT dstRowStride = AlignReadbackRow(rowPixels * dstPixelBytes, packParams.Alignment);
             const SizeT imageRows =

MobileGL/MG_Backend/DirectGLES/Utils.h

@@ -46,6 +46,15 @@ namespace MobileGL::MG_Backend::DirectGLES {
         Flags<PixelFormatNormalizeOptionBit> GetRenderTargetNormalizeOptions(
             const MG_External::GLESCapabilities& capabilities, SizeT targetIndex);
 
+        // Whether this format's ES storage is widened to 8-bit-per-channel because the
+        // driver stores some packed16 allocations with a mirrored field order
+        // (PixelFormatNormalizeOptionBit::WidenPacked16Norm). True only for
+        // GL_RGB565/GL_RGB5(_A1)/GL_RGBA4, and only where the POST probe measured the
+        // divergence (or MOBILEGL_ESPRYT_WIDEN_PACKED16_STORAGE forces it). The transfer paths
+        // consult it too: the packed-norm re-upload leg must stand down when the ES storage
+        // is no longer 16-bit packed.
+        Bool UsesWidenedPacked16NormStorage(TextureInternalFormat internalFormat);
+
         void GenerateTextureFormatInfo(TextureInternalFormat internalFormat, GLenum* outInternalFormat,
                                        GLenum* outFormat, GLenum* outType,
                                        TextureTarget target = TextureTarget::Unknown);
@@ -60,6 +69,115 @@ namespace MobileGL::MG_Backend::DirectGLES {
         Bool BackendTextureFormatAddsAlpha(TextureInternalFormat internalFormat, TextureTarget target);
         Bool BackendRenderbufferFormatAddsAlpha(TextureInternalFormat internalFormat);
         Bool ShouldUseCaveatRenderbufferFormat(TextureInternalFormat internalFormat);
+
+        // The CHANNEL WIDENING an image-bindable texture's ES storage takes, so that a format
+        // GLSL ES cannot spell as an image is carried by one it can.
+        //
+        // GL has forty image formats, GLSL ES core has thirteen, and no test device advertises
+        // GL_NV_image_formats - so a shader declaring one of the other twenty-six has no legal
+        // ESSL at all and glBindImageTexture rejects the narrow format outright for most of them
+        // (GL_INVALID_VALUE for nineteen of twenty-six on Adreno, twenty-five on both Malis).
+        // Seventeen have a core format of the SAME per-channel width and component type,
+        // differing only in channel count, and in one of those the emulation is EXACT: GL already
+        // defines an imageLoad from a narrower format as (r, 0, 0, 1) and an imageStore as
+        // dropping the components the format does not have, so the carrier's surplus channels
+        // hold values GL has already named. WidenImageFormatsPass pins them in the shader; this
+        // is the storage half, and DirectGLES::TextureImpl::SyncImageTextureBinding the bind
+        // half. All three ask WidenedCoreEsslImageFormat, so they cannot pick different carriers.
+        //
+        // Reports nothing (InternalFormat == GL_UNKNOWN_MGL) for a format that is core already,
+        // for the nine with no exact carrier (r11f_g11f_b10f, rgb10_a2, rgb10_a2ui, rgba16, rg16,
+        // r16, rgba16_snorm, rg16_snorm, r16_snorm - those keep the honest "no GLSL ES spelling"
+        // diagnostic rather than a silent approximation), and on a driver that HAS
+        // GL_NV_image_formats, where the shader keeps the declared format and no widening may
+        // happen behind it.
+        //
+        // The widened triple REPLACES what GenerateTextureFormatInfo chose, including any
+        // renderability substitution: an image that cannot be image-bound is useless whatever its
+        // attachment behaviour, so the image constraint wins. In practice that only bites
+        // RG8_SNORM/R8_SNORM on a driver without EXT_render_snorm, where the storage stays
+        // signed-normalized instead of becoming the half float that fallback would have picked -
+        // so an image-bound texture in one of those two formats is no longer attachable, and
+        // glGetTexImage on it falls through to the CPU shadow, which a shader-side imageStore
+        // does not update. Accepted deliberately: before the widening, an image binding in either
+        // format was refused outright by every driver tested and the stage that declared it never
+        // compiled at all, so nothing that works today is being given up.
+        //
+        // KNOWN GAP, for the same "all three layers move together" reason: a widened texture that
+        // is ALSO an FBO colour attachment gains one to three writable channels, and a draw into
+        // it can leave values in channels GL says are 0 and 1. Sampling and imageLoad are covered
+        // (the swizzle composition in SyncTextureParamsToBackend and the shader-side mask), but a
+        // glReadPixels/glGetTexImage that asks for more channels than the frontend format has
+        // would see them. Closing it needs the per-draw-buffer colour mask the three-channel
+        // widening already carries (FramebufferImpl::g_alphaWidenedDrawBufferMask) generalized
+        // from "alpha" to a channel count, which is its own change.
+        // How the FRONTEND's CPU shadow for a widened format is laid out relative to the carrier's
+        // transfer, i.e. what the upload has to do to it. Almost every entry is `Components`: the
+        // shadow already holds SourceChannels components of exactly the carrier's own type, so
+        // padding it out to four is the whole conversion. The packed entries do not - their shadow
+        // is ONE 32-bit word per texel - and reading such a word as components of the carrier's
+        // type takes twelve or sixteen bytes out of four and shears the level.
+        enum class ImageWidenSourceEncoding : Uint8 {
+            Components = 0,
+            // r11f_g11f_b10f: GL_UNSIGNED_INT_10F_11F_11F_REV -> four GL_FLOATs of an rgba16f.
+            PackedFloat11f11f10f,
+            // rgb10_a2 and rgb10_a2ui: GL_UNSIGNED_INT_2_10_10_10_REV -> four GL_UNSIGNED_SHORT
+            // channel CODES of an rgba16ui. The same split serves both: the two formats differ
+            // only in what the codes MEAN, which is the shader's business and not the transfer's.
+            PackedInt2101010Rev,
+        };
+
+        struct ImageBindableStorageWidening {
+            GLenum InternalFormat = GL_UNKNOWN_MGL;
+            GLenum Format = GL_UNKNOWN_MGL;
+            GLenum Type = GL_UNKNOWN_MGL;
+            // Channels the FRONTEND format has, i.e. how many of the carrier's four the client
+            // data fills. The rest are uploaded as 0, and the fourth as the format's implied 1.
+            Uint SourceChannels = 0;
+            // Whether that implied 1 is the integer one or a saturated normalized field - the
+            // transfer type cannot tell the two apart (GL_UNSIGNED_BYTE serves both RG8 and
+            // RG8UI), so the carrier decides.
+            Bool IntegerData = false;
+            // What the upload has to do to the frontend shadow before it describes the level to
+            // the driver (PrepareImageWidenedUpload).
+            ImageWidenSourceEncoding SourceEncoding = ImageWidenSourceEncoding::Components;
+            // Non-zero when the carrier holds this format's channels as the INTEGER CODES of a
+            // NORMALIZED value - the seven 16-bit and 10-bit normalized formats, which core ESSL
+            // has no image format of any width for and which a float carrier would requantise.
+            // Each entry is the largest code that channel can hold, i.e. the denominator of GL 4.6
+            // 2.3.5; SignedNormalized picks which of the two conversions it is the denominator of.
+            //
+            // Two things depend on it, both because the ES storage no longer shares the frontend
+            // format's component class: the upload pads a missing alpha with ChannelMax[3] instead
+            // of the transfer type's own "one" (through a uint carrier the saturated field IS the
+            // one), and glGetTexImage divides the codes back out into the floats the application
+            // is still owed.
+            Uint ChannelMax[4] = {0u, 0u, 0u, 0u};
+            Bool SignedNormalized = false;
+
+            Bool CarriesNormalizedCodes() const { return ChannelMax[0] != 0u; }
+            explicit operator Bool() const { return InternalFormat != GL_UNKNOWN_MGL; }
+        };
+        ImageBindableStorageWidening GetImageBindableStorageWidening(TextureInternalFormat internalFormat);
+
+        // The single-channel core format an image-bindable BUFFER texture's view is SPLIT into, or
+        // GL_UNKNOWN_MGL for a format that needs no split (or has no core base).
+        //
+        // A buffer texture cannot be widened: its texels are the application's buffer object, at
+        // the size and layout the application gave it, and it is usually also a vertex, index or
+        // storage buffer whose bytes are not ours to restride. But an rg32f view of N texels and
+        // an r32f view of 2N texels describe exactly the SAME bytes, so the split changes only
+        // how the shader subscripts them - component j of texel i is texel 2i + j of the base
+        // view - which WidenImageFormatsPass rewrites every access to do. The same rule as the
+        // widening decides WHETHER: a driver that can spell rg32f for an imageBuffer needs
+        // nothing.
+        //
+        // KNOWN GAP, and the reason this is not applied to a texture that is merely sampled: a
+        // buffer texture that is BOTH image-bound and read through a samplerBuffer would have its
+        // sampled view split too, and the sampler side is not rewritten. Accepted for the same
+        // reason the storage widening's gaps are - on a driver where the split applies at all
+        // there is no legal ESSL for the image declaration, so such a program did not compile.
+        GLenum GetImageBindableBufferSplitFormat(TextureInternalFormat internalFormat);
     } // namespace TextureImpl
 
     namespace FramebufferImpl {} // namespace FramebufferImpl
@@ -154,6 +272,32 @@ namespace MobileGL::MG_Backend::DirectGLES {
         // extension - requesting an unadvertised extension is itself a compile error, so this is
         // never emitted speculatively. A no-op when not needed or already present.
         String RequestExtendedImageFormats(String glslCode, Bool needed);
+        // Adds `#extension GL_OES_viewport_array : require` when the emitted ESSL names
+        // gl_ViewportIndex. SPIRV-Cross prints that identifier and asks for nothing (unlike
+        // gl_Layer, which it backs with GL_NV_viewport_array2 on ES) and ESSL has no core
+        // spelling for it at any version, so the request has to be made here or the stage does
+        // not compile - which loses the whole program, not just the multi-viewport routing.
+        // `needed` is the caller's answer for the same reason as above: only it knows whether the
+        // driver advertises the extension, and requesting an unadvertised one is itself a compile
+        // error, so this is never emitted speculatively. A no-op when not needed or already
+        // present.
+        String RequestViewportArrayExtension(String glslCode, Bool needed);
+        // Adds `#extension <extensionName> : require` when a TESSELLATION or GEOMETRY stage's
+        // emitted ESSL names gl_PointSize. Desktop GL has that built-in in gl_PerVertex for every
+        // vertex-processing stage; ESSL does NOT have it in those two at any version - not even
+        // 320, where the stages themselves are core - until EXT/OES_tessellation_point_size resp.
+        // EXT/OES_geometry_point_size is requested. SPIRV-Cross prints the identifier bare and
+        // asks for nothing, exactly as it does for gl_ViewportIndex, so without this the stage
+        // fails to compile with "`gl_PointSize' undeclared" and the WHOLE program is replaced by
+        // program 0 - the draw renders nothing and any transform-feedback capture it was carrying
+        // is rejected outright. `extensionName` is the caller's answer, nullptr when the driver
+        // advertises neither spelling, because requesting an unadvertised extension is itself a
+        // compile error. A no-op when nullptr or already present.
+        String RequestPointSizeExtension(String glslCode, const char* extensionName);
+        // The extension name RequestPointSizeExtension should be given for `tier`, or nullptr for
+        // PointSizeTier::None. `tessellation` picks the tessellation spellings over the geometry
+        // ones; the two extensions are separate and neither implies the other.
+        const char* PointSizeExtensionName(MG_External::GLESCapabilities::PointSizeTier tier, Bool tessellation);
         // Writes a format layout qualifier into the image declarations named in
         // `esslFormatByUniformName` that still have none. The completion half of the image-format
         // bake, and ONLY that: the SPIR-V pass (BakeImageFormatsPass) is what normally puts the
@@ -168,9 +312,115 @@ namespace MobileGL::MG_Backend::DirectGLES {
         // stops being safe to edit by hand.
         String BakeImageFormatQualifiers(String glslCode, const UnorderedMap<String, String>& esslFormatByUniformName);
         String RemoveLayoutBinding(const String& glslCode);
+        // Prefix of the per-element scalar declarations RemapImageArrayElementUnits splits an
+        // image array into; the suffix is the array's own name and the element's index.
+        constexpr const char* IMAGE_ARRAY_ELEMENT_PREFIX = "mg_imageElem_";
+        // One image ARRAY whose elements the application pointed at units that are not
+        // consecutive-from-element-zero.
+        struct ImageArrayUnitPlan {
+            String name;       // the array's name, exactly as the emitted ESSL declares it
+            Vector<Int> units; // the frontend image unit element k has to reach
+        };
+        // Desktop GL lets an application give each element of an image array an ARBITRARY unit
+        // (glUniform1i per element). ES has no such call at all - "ES image units come
+        // exclusively from the layout(binding=N) qualifier" - and one declaration carries one
+        // binding, so ESSL nails an array's elements to the CONSECUTIVE units N, N+1, N+2, ...
+        // MobileGL used to stamp element [0]'s unit as the binding and let the rest fall where
+        // they fell: KHR-GL4x.shader_image_load_store.advanced-sso-simple assigns 0,2,4,6 and
+        // 1,3,5,7, so its two programs actually addressed 0,1,2,3 and 1,2,3,4 - one layer got the
+        // wrong value and three were never written, with no GL error and no link log. The same
+        // defect for SAMPLER arrays was fixed API-side (SubscriptUniformNameForElement); an image
+        // array has no API side to fix, because ES makes glUniform1i on an image uniform an
+        // INVALID_OPERATION.
+        //
+        // Repaired by SPLITTING the array into one SCALAR image uniform per element, each with
+        // its own layout(binding = N), and rewriting `name[k]` to the scalar declared for
+        // element k. One declaration carries one binding, so one declaration per unit is the
+        // only spelling that reaches an arbitrary set of them.
+        //
+        // That rewrite needs every k in the emitted text to be a LITERAL, and it is:
+        // LegalizeResourceArrayIndexingForEssl has already folded or lowered every dynamic
+        // image-array subscript in the module, because ESSL forbids one outright ("image arrays
+        // indexed with non-constant expressions are forbidden in GLSL ES", Mesa 26.1.4 at
+        // ES 3.2, on a raw GLES probe with no MobileGL in the loop). The earlier shape here -
+        // widening the array to cover the whole span of units and routing each subscript through
+        // a `const highp int` offset table - was written before that pass covered images, and
+        // the table lookup was itself one of the non-constant expressions the same probe refuses.
+        // The split also costs exactly the image uniforms the application declared, where the
+        // widening cost the whole SPAN (seven for the four elements of
+        // KHR-GL42.shader_image_load_store.advanced-sso-simple), so there is no budget for it to
+        // fail to fit in.
+        //
+        // Declines - leaving the array exactly as it was, and naming it in `outDeclined` for the
+        // caller to report - when the emitted extent disagrees with the reflection, when the
+        // array is reached by anything other than a subscript, or when a subscript is not a
+        // literal element index. Silence was the whole defect here, so a decline must be audible.
+        //
+        // Must run AFTER RebindImageUniformsToFrontendUnits and BakeImageFormatQualifiers (both
+        // key on the GL uniform name and on a binding already being stamped) and BEFORE
+        // SplitReadWriteImageUniforms (so each element that is both read and written is split
+        // with its own binding already on it) and RemoveLayoutBinding (which is what preserves
+        // image bindings). Like them, it is downstream of the L2 shader-translation memo, so the
+        // per-program units it reads need no entry in BuildEsslTranslationKey.
+        String RemapImageArrayElementUnits(const String& glslCode, const Vector<ImageArrayUnitPlan>& plans,
+                                           Vector<String>* outDeclined = nullptr);
+        // The member list of a `gl_PerVertex { ... }` redeclaration in already-emitted ESSL -
+        // the text between the braces, verbatim - or nullopt when the shader does not redeclare
+        // the block in that direction. `input` selects the `in gl_PerVertex` form over the
+        // `out` one.
+        //
+        // Exists so BuildPassthroughTessControlEssl can MIRROR the stages it has to sit between
+        // rather than guess at them. Whether SPIRV-Cross redeclares the built-in block, and with
+        // which members, depends on what the application's shader touched; a synthesized stage
+        // that redeclares a different shape than its neighbours is an ES link error against a
+        // program that has no other problem.
+        std::optional<String> ExtractPerVertexBlockMembers(const String& essl, Bool input);
+        // The pass-through tessellation control stage GL 4.6 core 11.2.2 describes: "the input
+        // patch is passed through unmodified", the output patch has PATCH_VERTICES vertices, and
+        // the levels come from the PATCH_DEFAULT_OUTER_LEVEL / PATCH_DEFAULT_INNER_LEVEL state.
+        //
+        // Desktop GL makes the control stage OPTIONAL. OpenGL ES 3.2 does not: it has no
+        // PATCH_DEFAULT_*_LEVEL state at all (only glPatchParameteri, for PATCH_VERTICES) and
+        // rejects a program that has an evaluation stage without a control stage - with an EMPTY
+        // info log, verified on an Adreno 830 with no MobileGL in the process. MobileGL's own
+        // frontend link succeeds, so the program reports GL_LINK_STATUS = TRUE, program 0 is
+        // bound in its place, and every draw silently renders nothing.
+        //
+        // `inPerVertexMembers` / `outPerVertexMembers` are the member lists to redeclare gl_in
+        // and gl_out with - normally taken from the neighbouring stages' own emitted ESSL via
+        // ExtractPerVertexBlockMembers, and empty to leave the driver's built-in declaration
+        // alone, which is what matching a neighbour that did not redeclare requires.
+        //
+        // All four outer levels and both inner levels are written unconditionally: writing a
+        // level the evaluation stage's domain does not use is legal and ignored, and it saves
+        // this from having to know the domain. They are the GL_PATCH_DEFAULT_OUTER_LEVEL /
+        // GL_PATCH_DEFAULT_INNER_LEVEL state, baked in as literals - ES has no such state and no
+        // glPatchParameterfv to forward to, so compiling them in is the only way to honour them.
+        // That makes them part of what a built program is stale against, exactly as PATCH_VERTICES
+        // is: see the staleness clause in DirectGLES.cpp's SyncCurrentProgram, which compares both.
+        //
+        // The same stage, for the same reason, that DirectVulkan synthesizes in
+        // ProgramFactory::BuildPassthroughTessControlSource - Vulkan likewise requires both
+        // tessellation stages. Kept as two generators rather than one because the two targets
+        // disagree on everything but the algorithm: desktop GLSL 450 against ESSL, a fixed
+        // gl_PerVertex shape that Vulkan matches structurally against a mirrored one, and a
+        // VkShaderModule against a driver shader object.
+        String BuildPassthroughTessControlEssl(Uint esslVersion, Uint patchVertices,
+                                               const String& inPerVertexMembers,
+                                               const String& outPerVertexMembers,
+                                               const FloatVec4& defaultOuterLevel,
+                                               const FloatVec2& defaultInnerLevel);
         // Prefix of the writeonly half a read+write image uniform is split into (see
-        // SplitReadWriteImageUniforms); the suffix is the image's own name.
+        // SplitReadWriteImageUniforms); the suffix is the image's own (already access-tagged) name.
         constexpr const char* IMAGE_WRITE_ALIAS_PREFIX = "mg_imageWrite_";
+        // The three names SplitReadWriteImageUniforms renames a rewritten image declaration
+        // under, one per REPAIR it can apply. Which one a stage picks is decided by that stage's
+        // own accesses, so two stages that use an image the same way arrive at the SAME name and
+        // two that use it differently arrive at different ones - which is exactly the property
+        // the rename exists for, at no cost to the stages that agree. Exposed for the tests.
+        constexpr const char* IMAGE_READONLY_ALIAS_PREFIX = "mg_imageRo_";
+        constexpr const char* IMAGE_WRITEONLY_ALIAS_PREFIX = "mg_imageWo_";
+        constexpr const char* IMAGE_SPLIT_READ_ALIAS_PREFIX = "mg_imageRw_";
         // ESSL refuses an image variable that carries a format qualifier other than r32f /
         // r32i / r32ui unless it also carries `readonly` or `writeonly` (GLSL ES 3.10 4.9 /
         // 3.20 4.10; glslang enforces it verbatim in ParseHelper.cpp's layoutObjectCheck).
@@ -182,15 +432,74 @@ namespace MobileGL::MG_Backend::DirectGLES {
         // bare declaration, so the frontend raises no error and the illegal ESSL only shows
         // up as a device compile failure - and then as a silently no-op draw.
         //
-        // Restores a legal declaration:
-        //  * loaded only            -> add `readonly`
-        //  * stored only            -> add `writeonly`
+        // Restores a legal declaration, and RENAMES it after the repair it applied while doing so:
+        //  * loaded only            -> add `readonly`, rename under IMAGE_READONLY_ALIAS_PREFIX
+        //  * stored only            -> add `writeonly`, rename under IMAGE_WRITEONLY_ALIAS_PREFIX
         //  * both                   -> emit TWO declarations on the same binding and of the
-        //                              same type, `readonly <name>` and `writeonly
-        //                              <IMAGE_WRITE_ALIAS_PREFIX><name>`, and point every
-        //                              imageStore at the second one. Several image variables
-        //                              may share an image unit as long as they have the same
-        //                              type and format, which is exactly what the pair is.
+        //                              same type, `coherent readonly
+        //                              <IMAGE_SPLIT_READ_ALIAS_PREFIX><name>` and `coherent
+        //                              writeonly <IMAGE_WRITE_ALIAS_PREFIX><that name>`, point
+        //                              every imageStore at the second one, and follow each of
+        //                              those stores with `memoryBarrierImage();`. Several image
+        //                              variables may share an image unit as long as they have
+        //                              the same type and format, which is exactly what the pair
+        //                              is.
+        //
+        // The rename is the other half of the repair and applies to all three cases. The qualifier
+        // chosen above is a decision about ONE STAGE's accesses, and GLSL requires a uniform
+        // declared in two stages to be declared identically - so a shader that stores an image from
+        // the vertex stage and loads it from the fragment stage came out of here `writeonly` in one
+        // and `readonly` in the other. Adreno merges the two same-named declarations and silently
+        // drops the vertex-stage STORES: no GL error, no link log, LINK_STATUS = 1, and the image
+        // still reads back its initial contents
+        // (KHR-GL4x.shader_image_load_store.advanced-memory-dependentInvocation; a raw-ES probe
+        // isolated the trigger to the same-name/mismatched-qualifier pair, and only when both
+        // carry `coherent`). Renaming leaves no cross-stage variable to merge.
+        //
+        // The name is keyed on the REPAIR, not on the stage, and that distinction is the whole
+        // point: two stages that use an image the same way emit byte-identical declarations, so
+        // letting them keep one shared name costs nothing and merging them is correct, while two
+        // stages that use it differently land on different prefixes and cannot be merged at all.
+        // A per-STAGE tag also satisfied the first requirement but violated the second: it made
+        // the SAME image a distinct uniform in every stage that named it, and Adreno allocates
+        // image LOCATIONS per distinct uniform. KHR-GL43.shading_language_420pack.
+        // binding_images_texture_type_* declares three read+write images in each of its five
+        // stages; merged that is 6 image uniforms, per-stage-tagged it is 30, and the Adreno 830
+        // linker answered "Error: Image Image location or component exceeds max allowed. Error:
+        // Linking failed." - which, the frontend having already published LINK_STATUS = TRUE from
+        // glslang's link, surfaced only as every draw silently doing nothing and the images
+        // reading back zero. Mali and Mesa link the same text, so nothing but a device gate
+        // catches this.
+        //
+        // A declaration SPIRV-Cross already tagged `readonly` or `writeonly` needs no qualifier
+        // repair, but it is NOT stage-independent: that tag is derived from the accesses of the
+        // stage being emitted, so an image stored in the vertex stage and loaded in the fragment
+        // stage arrives here as `coherent writeonly g_image` and `coherent readonly g_image` -
+        // one name, two spellings, which is exactly the pair Adreno merges. Those declarations
+        // are therefore renamed too, keyed on the qualifier they already carry (readonly ->
+        // IMAGE_READONLY_ALIAS_PREFIX, writeonly -> IMAGE_WRITEONLY_ALIAS_PREFIX) and with
+        // nothing but the identifier changed. Stages that agree still reach the same alias and
+        // stay merged, so this costs no shader an extra image uniform.
+        //
+        // The declarations this pass still leaves untouched keep their names: one carrying BOTH
+        // readonly and writeonly (a spelling no access analysis produces, so it came from the
+        // application and is identical everywhere), and one carrying NEITHER, which is legal only
+        // for the r32f/r32i/r32ui formats and is likewise spelled the same in every stage.
+        //
+        // The `coherent` on both halves of the pair is load-bearing, not decoration: GLSL only
+        // guarantees a write through one image variable is visible to a read through a DIFFERENT
+        // one when both are coherent, and the split is what makes a same-variable
+        // read-after-write cross-variable. The single-declaration repairs above do not get it -
+        // nothing aliases them.
+        //
+        // The barrier is the other half of the same problem, and coherent alone did not cover it:
+        // visibility is not ORDER. Within one invocation the ES compiler sees a write to one
+        // variable and a read of another it has no reason to believe alias, and is free to serve
+        // the read from before the write - which is what advanced-memory-order's store/load/
+        // compare loop measured on Adreno. memoryBarrierImage() orders exactly those two, is core
+        // GLSL ES 3.10 in every stage, and is not an execution barrier, so it is legal in
+        // non-uniform control flow. It costs something in a shader that stores to a read+write
+        // image in a loop, which is why it is confined to the split pair.
         //
         // Budget note: the split DOUBLES the image-uniform count of the stage it fires in, so
         // a driver advertising a tight GL_MAX_{FRAGMENT,VERTEX,...}_IMAGE_UNIFORMS can turn a
@@ -200,8 +509,14 @@ namespace MobileGL::MG_Backend::DirectGLES {
         //
         // Runs on the transpiled ESSL, so it must see the bindings the frontend units were
         // already rewritten to and must run before those bindings are stripped - see the call
-        // site in Managers.cpp.
-        String SplitReadWriteImageUniforms(const String& glslCode);
+        // site in Managers.cpp. Its output is a function of the emitted text alone - it needs no
+        // stage and no per-program state - so it adds nothing to BuildEsslTranslationKey either.
+        //
+        // `outSplitCount`, when given, receives the number of declarations that were actually
+        // doubled - i.e. exactly how many image uniforms this stage gained over what the
+        // application declared. Zero for every shader but a handful, and the only number the
+        // budget note above can be reported with.
+        String SplitReadWriteImageUniforms(const String& glslCode, Uint* outSplitCount = nullptr);
         // Prefix of the per-sampler float uniform that carries GL_TEXTURE_LOD_BIAS into
         // the shader (see EmulateTextureLodBias); the suffix is the sampler's own name.
         constexpr const char* LOD_BIAS_UNIFORM_PREFIX = "mg_lodBias_";
@@ -217,7 +532,7 @@ namespace MobileGL::MG_Backend::DirectGLES {
         // avoidExplicitLodBias leaves lookups that already carry an explicit LOD untouched,
         // so their constant level stays constant; only the implicit-LOD forms take the bias.
         // Off by default and only ever set on ANGLE + llvmpipe, where injecting the uniform
-        // into a constant LOD crashes the driver (MOBILEGL_AVOID_EXPLICIT_LOD_BIAS).
+        // into a constant LOD crashes the driver (MOBILEGL_ESPRYT_AVOID_EXPLICIT_LOD_BIAS).
         String EmulateTextureLodBias(const String& glslCode, Bool avoidExplicitLodBias = false);
     } // namespace PrgramImpl
 

MobileGL/MG_Backend/DirectVulkan/BackendObject_DirectVulkan.cpp

@@ -9,7 +9,10 @@
 #include "BackendObject_DirectVulkan.h"
 #include "MG_Backend/BackendObject.h"
 #include "DirectVulkan.h"
+#include "SubgroupSupportPolicy.h"
 #include "MG_State/GLState/FramebufferState/FramebufferObject.h"
+#include "MG_State/GLState/Core.h"
+#include <MG_Pipe/PipeInputsSwitch.h>
 #include "MG_State/GLState/TextureState/TextureState.h"
 #include "MG_Util/Classifiers/TextureEnumClassifier.h"
 #include "MG_Util/Converters/MGToGL/TextureEnumConverter.h"
@@ -383,6 +386,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         }
         UpdateDynamicBackendParameters();
         UpdateAdvertisedExtensions();
+        if (MGB_CTX_LIVE) {
+            MGB_CTX->InvalidateCompileEnv();
+        }
         PopulateFormatCapabilities(physicalDevice.handle, vkGetPhysicalDeviceFormatProperties, m_vulkanCaps,
                                    MutableFormatCapabilities());
         PrintFormatCapabilities(GetFormatCapabilities());
@@ -495,22 +501,31 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             .RendererName = "Magma",
             .BackendName = "Direct (Vulkan)",
             .ExtraVendor = Nullopt,
-            .RendererGLInfo = {.TargetGLVersion = {4, 0, 0},
+            .RendererGLInfo = {.TargetGLVersion = {4, 6, 0},
                                .TargetGLSLVersion = {4, 6, 0},
-                               // Baseline advertisement (no shader subgroup, no timer queries); a
-                               // live backend reconciles its copy in UpdateAdvertisedExtensions.
-                               .Extensions = BuildAdvertisedExtensions(false, false, false),
+                               // Baseline advertisement (no runtime-gated capabilities); a live
+                               // backend reconciles its copy in UpdateAdvertisedExtensions.
+                               .Extensions = BuildAdvertisedExtensions(false, false, false, false, false),
                                .IsCompatibilityProfile = false},
             .StaticBackendCapability = {.AllowVSOnlyPrograms = false}};
         return rendererInfo;
     }
 
     Vector<GLExtension> BuildAdvertisedExtensions(Bool shaderSubgroupSupported, Bool timerQueriesSupported,
-                                                  Bool anisotropicFilteringSupported) {
+                                                  Bool anisotropicFilteringSupported,
+                                                  Bool nonZeroIndirectBaseInstanceSupported,
+                                                  Bool cubeMapArraySupported) {
         Vector<GLExtension> extensions = {
-            V_OpenGL30, V_OpenGL31, V_OpenGL32, V_OpenGL33, V_OpenGL40, E_GL_ARB_draw_buffers_blend,
+            // The version tokens have to reach the version the backend actually claims:
+            // TargetGLVersion is {4,6,0}, and a list that stopped at OpenGL40 told an
+            // application feature-detecting off these tokens the opposite of what
+            // GL_MAJOR_VERSION / GL_MINOR_VERSION told it.
+            V_OpenGL30, V_OpenGL31, V_OpenGL32, V_OpenGL33, V_OpenGL40, V_OpenGL41, V_OpenGL42, V_OpenGL43,
+            V_OpenGL44, V_OpenGL45, V_OpenGL46,
+            E_GL_ARB_draw_buffers_blend,
             E_GL_ARB_compute_shader, E_GL_ARB_shader_storage_buffer_object, E_GL_ARB_shader_image_load_store,
-            E_GL_ARB_program_interface_query, E_GL_ARB_framebuffer_object, E_GL_ARB_multi_draw_indirect,
+            E_GL_ARB_clear_buffer_object, E_GL_ARB_program_interface_query, E_GL_ARB_framebuffer_object, E_GL_ARB_draw_indirect,
+            E_GL_ARB_multi_draw_indirect,
             E_GL_ARB_indirect_parameters, E_GL_EXT_framebuffer_object, E_GL_ARB_depth_texture, E_GL_ARB_buffer_storage,
             E_GL_ARB_texture_storage, E_GL_ARB_texture_storage_multisample, E_GL_ARB_texture_multisample,
             E_GL_ARB_clear_texture, E_GL_ARB_direct_state_access, E_GL_ARB_shader_draw_parameters,
@@ -526,11 +541,91 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             // Sampling the stencil aspect through DEPTH_STENCIL_TEXTURE_MODE. Core from 4.3,
             // so on a 4.0 context the string is the only way to reach it.
             E_GL_ARB_stencil_texturing,
+            // Unconditional, unlike DirectGLES: a GL texture view is a second set of VkImageViews
+            // over the same VkImage with a sub-range and possibly a reinterpreted VkFormat, which
+            // is core Vulkan on every device MobileGL runs on. Format-reinterpreting views need
+            // VK_IMAGE_CREATE_MUTABLE_FORMAT_BIT on the image, which SyncTextureResource sets for
+            // every immutable-storage texture (see the comment there).
+            E_GL_ARB_texture_view,
+            // Core since 3.2 and implemented here on both backends - glDrawElementsBaseVertex,
+            // glDrawRangeElementsBaseVertex, glDrawElementsInstancedBaseVertex and
+            // glMultiDrawElementsBaseVertex all reach real per-draw vertex rebasing. The string
+            // was simply never emitted, which left KHR-GL4*.draw_elements_base_vertex_tests
+            // NotSupported on a feature that works.
+            E_GL_ARB_draw_elements_base_vertex,
+            // The whole sync-object family is real and core since 3.2: glFenceSync, glIsSync,
+            // glDeleteSync, glClientWaitSync, glWaitSync and glGetSynciv all live in GLImpl over a
+            // backend fence (a VkFence here, an EGLSync/GLsync on DirectGLES), and glGetInteger64v
+            // answers GL_MAX_SERVER_WAIT_TIMEOUT. The string matters for the same reason
+            // ARB_uniform_buffer_object's does: LWJGL builds GLCapabilities from the extension
+            // list, and a caller that finds GL_ARB_sync missing never resolves the entry points -
+            // then calls through null if it uses fences anyway. Nothing in the CTS gates on this
+            // string, so it is advertised on the strength of the implementation, not a test unlock.
+            E_GL_ARB_sync,
+            // Atomic counters, core since 4.2. glGetActiveAtomicCounterBufferiv and the whole
+            // GL_ATOMIC_COUNTER_BUFFER_* query family are real in GLImpl, and the counter buffer
+            // now reaches the shader on BOTH backends - Magma resolves the lowered
+            // gl_AtomicCounterBlock_<N> from the atomic-counter binding points rather than the
+            // shader-storage ones (see ResolveStorageBufferDescriptor). Withheld here until that
+            // landed, because the counter silently read whatever was bound as SSBO N instead.
+            E_GL_ARB_shader_atomic_counters,
+            // glVertexAttribDivisor, core since 3.3 and real on both backends. Applications
+            // (Better Clouds' GLCompat among them) accept the extension string as an
+            // ALTERNATIVE to a 3.3 context when deciding whether instanced rendering is
+            // available, so withholding it makes MobileGL look less capable than it is.
+            E_GL_ARB_instanced_arrays,
+            // Core GL 3.0-4.3 plumbing that has been real here for as long as the backend has
+            // existed, and that was simply never named. None of these unlocks a single CTS case -
+            // the conformance suite reaches all of them through the version - so they are
+            // advertised for the OTHER consumer of this list: LWJGL builds GLCapabilities from the
+            // string set, and an application that gates its ENTRY POINTS on the string rather than
+            // on the version never resolves them and then calls through null. Each is backed by
+            // the entry points named beside it. Kept identical to the DirectGLES block so the two
+            // backends do not disagree about what MobileGL is.
+            //
+            // glBindVertexArray / glGenVertexArrays / glDeleteVertexArrays / glIsVertexArray.
+            E_GL_ARB_vertex_array_object,
+            // The 14 glSamplerParameter* / glGetSamplerParameter* entry points, including the
+            // integer-valued Iiv/Iuiv forms.
+            E_GL_ARB_sampler_objects,
+            // glMapBufferRange + glFlushMappedBufferRange, which ARB_buffer_storage's persistent
+            // maps are already built on top of.
+            E_GL_ARB_map_buffer_range,
+            // glCopyBufferSubData plus the GL_COPY_READ_BUFFER / GL_COPY_WRITE_BUFFER targets.
+            E_GL_ARB_copy_buffer,
+            // glCopyImageSubData, wired to a real backend hook on both backends.
+            E_GL_ARB_copy_image,
+            // GL_TEXTURE_SWIZZLE_{R,G,B,A,RGBA}, which map onto a VkImageView's component swizzle.
+            E_GL_ARB_texture_swizzle,
+            // GL_INT_2_10_10_10_REV / GL_UNSIGNED_INT_2_10_10_10_REV on glVertexAttribPointer plus
+            // the eight glVertexAttribP* entry points.
+            E_GL_ARB_vertex_type_2_10_10_10_rev,
+            // The R/RG internal formats. Named separately from the float ones because an
+            // application may check either.
+            E_GL_ARB_texture_rg,
+            // GL_DEPTH_COMPONENT32F and GL_DEPTH32F_STENCIL8.
+            E_GL_ARB_depth_buffer_float,
+            // The floating-point colour formats. Unlike the rest of this block this string DOES
+            // gate CTS cases - KHR-GL4*.internalformat.texture2d.*{16f,32f} is keyed on it with no
+            // core-version fallback, so eight cases per version list were NotSupported on formats
+            // the backend has always had.
+            E_GL_ARB_texture_float,
+            // glViewportArrayv / glViewportIndexedf{,v} / glScissorArrayv / glScissorIndexed{,v} /
+            // glDepthRangeArrayv / glDepthRangeIndexed / glGetFloati_v / glGetDoublei_v, over the
+            // 16 viewports GL_MAX_VIEWPORTS reports.
+            E_GL_ARB_viewport_array,
             // Advertised with GL_NUM_PROGRAM_BINARY_FORMATS = 0, which the
             // extension explicitly permits. It is also the only thing that
             // exposes glProgramParameteri before GL 4.1.
             E_GL_ARB_get_program_binary};
-        if (shaderSubgroupSupported && !MG_Config::Features.DisableSubgroup) {
+        // Vulkan's drawIndirectFirstInstance feature is optional. Direct base-instance calls work
+        // without it, but ARB_base_instance also promises non-zero firstInstance in GPU indirect
+        // commands; the renderer supplies true only when that word is legal and gl_InstanceID can
+        // be rebased to OpenGL's zero-based semantics.
+        if (nonZeroIndirectBaseInstanceSupported) {
+            extensions.push_back(E_GL_ARB_base_instance);
+        }
+        if (shaderSubgroupSupported && !MG_Config::Features.MagmaDisableSubgroup) {
             extensions.push_back(E_GL_KHR_shader_subgroup);
         }
         // GL_KHR_parallel_shader_compile is MobileGL's own capability, not the Vulkan
@@ -548,12 +643,13 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         if (MG_Util::Async::AsyncShaderCompileEnabled()) {
             extensions.push_back(E_GL_KHR_parallel_shader_compile);
         }
-        // GL_ARB_gpu_shader_fp64 is opt-in (MOBILEGL_ADVERTISE_FP64). Every `double` in a
-        // shader compiles and runs already - it is narrowed to 32 bits before the module
-        // reaches this backend - so an application that simply uses doubles needs nothing
-        // advertised. What the extension additionally promises is 64-bit PRECISION, which no
-        // mobile GPU has and the narrowing cannot fake, so advertising it by default would
-        // make an application that checks the string take a path MobileGL cannot honour.
+        // GL_ARB_gpu_shader_fp64 is opt-in (MOBILEGL_ADVERTISE_FP64), and stays opt-in even on a
+        // device that HAS shaderFloat64. Every `double` in a shader compiles and runs either way
+        // - narrowed to 32 bits where the device has no 64-bit floats, kept whole where it does -
+        // so an application that simply uses doubles needs nothing advertised. What the extension
+        // additionally promises is the whole GL_ARB_gpu_shader_fp64 SURFACE (glUniform*d
+        // conformance, the fp64 built-ins, the state queries), and turning the string on is a
+        // decision about all of it rather than about the shader path alone.
         if (MG_Config::Features.AdvertiseFp64) {
             extensions.push_back(E_GL_ARB_gpu_shader_fp64);
         }
@@ -570,6 +666,18 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             extensions.push_back(E_GL_EXT_texture_filter_anisotropic);
             extensions.push_back(E_GL_ARB_texture_filter_anisotropic);
         }
+        // A cube map array is a 6n-layer VkImage viewed as VK_IMAGE_VIEW_TYPE_CUBE_ARRAY, and that
+        // view type cannot be created without the imageCubeArray device feature - so the string
+        // follows the feature, not the version, exactly as the per-layer attachment bit does.
+        //
+        // Named for the application's benefit rather than the suite's: measured on Adreno 830,
+        // KHR-GL43.texture_gather.plain-gather-*-cube-array already passed without the string, so
+        // this unlocks no conformance case. It is advertised because the feature is real and
+        // because an application that feature-detects cube map arrays off the string (rather than
+        // off the 4.0 version) would otherwise decline a path this backend serves.
+        if (cubeMapArraySupported) {
+            extensions.push_back(E_GL_ARB_texture_cube_map_array);
+        }
         return extensions;
     }
 
@@ -633,8 +741,6 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             funcsTable.GL.MemoryBarrierByRegion = MemoryBarrierByRegion;
             funcsTable.GL.BindImageTexture = BindImageTexture;
             funcsTable.GL.GetIntegeri_v = GetIntegeri_v;
-            funcsTable.GL.GetInteger64i_v = GetInteger64i_v;
-            funcsTable.GL.GetProgramiv = GetProgramiv;
             funcsTable.GL.ShaderStorageBlockBinding = ShaderStorageBlockBinding;
             funcsTable.GL.FenceSync = FenceSync;
             funcsTable.GL.ClientWaitSync = ClientWaitSync;
@@ -678,6 +784,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         m_vulkanCaps = capabilities;
         UpdateDynamicBackendParameters();
         UpdateAdvertisedExtensions();
+        if (MGB_CTX_LIVE) {
+            MGB_CTX->InvalidateCompileEnv();
+        }
         MutableFormatCapabilities().Clear();
     }
 
@@ -688,9 +797,17 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // real device timestamp support. ApplyVulkanCapabilitiesForTesting may
         // run without a renderer; no timer query is advertised then. Rebuilding
         // the whole list keeps re-runs idempotent.
+        // The opt-in emulated compute path (SubgroupSupportPolicy.h) carries the
+        // extension by itself on devices with no native subgroup support at all; a
+        // device with native subgroups always advertises - and uses - those.
+        const Bool subgroupSupportAdvertised =
+            m_vulkanCaps.SupportsShaderSubgroup ||
+            ShouldEmulateSubgroups(m_vulkanCaps.SupportsShaderSubgroup);
         m_rendererInfo.RendererGLInfo.Extensions = BuildAdvertisedExtensions(
-            m_vulkanCaps.SupportsShaderSubgroup, pVulkanRenderer && pVulkanRenderer->IsTimerQuerySupported(),
-            pVulkanRenderer && pVulkanRenderer->IsSamplerAnisotropySupported());
+            subgroupSupportAdvertised, pVulkanRenderer && pVulkanRenderer->IsTimerQuerySupported(),
+            pVulkanRenderer && pVulkanRenderer->IsSamplerAnisotropySupported(),
+            pVulkanRenderer && pVulkanRenderer->IsNonZeroIndirectBaseInstanceSupported(),
+            m_vulkanCaps.SupportsImageCubeArray);
     }
 
     void BackendObject_DirectVulkan::UpdateDynamicBackendParameters() {
@@ -738,6 +855,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
         static constexpr SizeT kMaxAdvertisedShaderStorageBlockSize = 512ull * 1024ull * 1024ull;
         m_dynamicParameters.UniformBufferOffsetAlignment = m_vulkanCaps.UniformBufferOffsetAlignment;
+        m_dynamicParameters.ShaderStorageBufferOffsetAlignment = m_vulkanCaps.ShaderStorageBufferOffsetAlignment;
         m_dynamicParameters.AliasedLineWidthRangeMin = m_vulkanCaps.AliasedLineWidthRangeMin;
         m_dynamicParameters.AliasedLineWidthRangeMax = m_vulkanCaps.AliasedLineWidthRangeMax;
         // Without the samplerAnisotropy feature the limit is unusable, so report 1.0 (no anisotropy)
@@ -820,9 +938,50 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             clampLimit("GL_MAX_COMPUTE_UNIFORM_BLOCKS", m_vulkanCaps.MaxComputeUniformBlocks,
                        kMaxAdvertisedBufferBlocks);
         m_dynamicParameters.MaxComputeWorkGroupInvocations = m_vulkanCaps.MaxComputeWorkGroupInvocations;
+        // The six per-axis compute limits, from the same VkPhysicalDeviceLimits fields
+        // GLFunctionsTable::GetIntegeri_v (DirectVulkan.cpp) reads live. Carried here so that
+        // MGPCaps has them once the table entry retires (plan B section 4.4.1); GL_Getter floors
+        // them. Not clamped: unlike the block counts these are not amounts an application
+        // allocates, and the frontend already raises them to the GL minimum.
+        for (SizeT axis = 0; axis < 3; ++axis) {
+            m_dynamicParameters.MaxComputeWorkGroupCount[axis] = m_vulkanCaps.MaxComputeWorkGroupCount[axis];
+            m_dynamicParameters.MaxComputeWorkGroupSize[axis] = m_vulkanCaps.MaxComputeWorkGroupSize[axis];
+        }
         m_dynamicParameters.MaxShaderStorageBufferBindings =
             clampLimit("GL_MAX_SHADER_STORAGE_BUFFER_BINDINGS", m_vulkanCaps.MaxShaderStorageBufferBindings,
                        kMaxAdvertisedBufferBlocks);
+        // Per-stage GL_MAX_*_SHADER_STORAGE_BLOCKS. Vulkan has one descriptor limit for every
+        // stage (maxPerStageDescriptorStorageBuffers, which is what MaxComputeShaderStorageBlocks
+        // carries), so the stage limits differ only by whether the stage can have blocks at all.
+        //
+        // Deliberately NOT gated on vertexPipelineStoresAndAtomics, unlike the per-stage image
+        // uniforms below. That gate reads as the obvious one and is wrong here in practice: a
+        // Mali-G925-Immortalis reports vertexPipelineStoresAndAtomics=false (supported AND
+        // enabled) and yet runs all 433 KHR-GL43.constant_expressions.*_tess_* cases correctly
+        // through this backend - those write their result through a storage block declared in a
+        // tessellation stage. Gating would report 0 and turn 433 passing cases into
+        // "unsupported", removing function that demonstrably works.
+        //
+        // The asymmetry with DirectGLES is real and is the point. There, 0 prevents a program
+        // the driver refuses outright at link time; the honest limit converts a silent
+        // wrong-render into a capability an application can route around. Here there is no such
+        // failure to prevent, so the limit stays at what the device can address. If a Vulkan
+        // device is ever found that genuinely rejects such a pipeline, the gate belongs at
+        // pipeline creation where the rejection is observable, not on a feature bit this driver
+        // reports inaccurately.
+        {
+            const Int maxPerStageStorageBlocks =
+                std::min(std::max(m_dynamicParameters.MaxComputeShaderStorageBlocks, 0),
+                         std::min(std::max(m_dynamicParameters.MaxCombinedShaderStorageBlocks, 0),
+                                  std::max(m_dynamicParameters.MaxShaderStorageBufferBindings, 0)));
+            m_dynamicParameters.MaxVertexShaderStorageBlocks = maxPerStageStorageBlocks;
+            m_dynamicParameters.MaxTessControlShaderStorageBlocks = maxPerStageStorageBlocks;
+            m_dynamicParameters.MaxTessEvaluationShaderStorageBlocks = maxPerStageStorageBlocks;
+            // The one hard capability in the set: no geometry stage means no blocks in it.
+            m_dynamicParameters.MaxGeometryShaderStorageBlocks =
+                m_vulkanCaps.SupportsGeometryShader ? maxPerStageStorageBlocks : 0;
+            m_dynamicParameters.MaxFragmentShaderStorageBlocks = maxPerStageStorageBlocks;
+        }
         m_dynamicParameters.MaxTextureBufferSize = clampLimit(
             "GL_MAX_TEXTURE_BUFFER_SIZE", m_vulkanCaps.MaxTextureBufferSize, kMaxAdvertisedTextureBufferSize);
         m_dynamicParameters.TextureBufferOffsetAlignment = m_vulkanCaps.TextureBufferOffsetAlignment;
@@ -849,8 +1008,35 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         const Int maxSupportedDrawBuffers = static_cast<Int>(MG_State::GLState::FramebufferObject::MAX_DRAW_BUFFERS);
         m_dynamicParameters.MaxDrawBuffers = std::min(m_vulkanCaps.MaxDrawBuffers, maxSupportedDrawBuffers);
         m_dynamicParameters.MaxColorAttachments = std::min(m_vulkanCaps.MaxColorAttachments, maxSupportedDrawBuffers);
-        m_dynamicParameters.MaxClipDistances = m_vulkanCaps.MaxClipDistances;
+        // Same shape as the image-uniform limits three lines above: maxClipDistances is reported
+        // by every device, but declaring ClipDistance in a module needs the shaderClipDistance
+        // FEATURE, which VulkanRenderer enables exactly where the physical device has it. Without
+        // it the limit describes a capacity no shader may use, so report none.
+        m_dynamicParameters.MaxClipDistances =
+            m_vulkanCaps.SupportsShaderClipDistance ? std::max(m_vulkanCaps.MaxClipDistances, 0) : 0;
+        // The cull pair, gated on its own feature. shaderCullDistance is separate from
+        // shaderClipDistance and VulkanRenderer enables it independently, so it gets its own
+        // gate rather than riding on the clip one.
+        m_dynamicParameters.MaxCullDistances =
+            m_vulkanCaps.SupportsShaderCullDistance ? std::max(m_vulkanCaps.MaxCullDistances, 0) : 0;
+        // GL 4.6 core 11.1.3.10: the combined limit is at least as large as either half. A device
+        // with only one of the two features must not report a combined capacity that implies the
+        // other, so the gate is "either feature" and the value never drops below what is enabled.
+        m_dynamicParameters.MaxCombinedClipAndCullDistances =
+            (m_vulkanCaps.SupportsShaderClipDistance || m_vulkanCaps.SupportsShaderCullDistance)
+                ? std::max({m_vulkanCaps.MaxCombinedClipAndCullDistances, m_dynamicParameters.MaxClipDistances,
+                            m_dynamicParameters.MaxCullDistances})
+                : 0;
         m_dynamicParameters.MaxViewports = m_vulkanCaps.MaxViewports;
+        // Assigned explicitly rather than left to the struct's defaults, like every other
+        // parameter here, so a second fill cannot inherit a stale value. GL_UNDEFINED_VERTEX is
+        // the truthful answer for DirectVulkan and a legal one (GL 4.6 table 23.65): which vertex
+        // provokes is chosen per pipeline by VulkanRenderer::SelectProvokingVertexMode out of
+        // VK_EXT_provoking_vertex, provokingVertexModePerPipeline and the topology, so there is no
+        // one convention to name. Vulkan's own default is FIRST, which is the opposite of the
+        // GL_LAST_VERTEX_CONVENTION this used to claim unconditionally.
+        m_dynamicParameters.LayerProvokingVertex = GL_UNDEFINED_VERTEX;
+        m_dynamicParameters.ViewportIndexProvokingVertex = GL_UNDEFINED_VERTEX;
         m_dynamicParameters.MaxViewportWidth = m_vulkanCaps.MaxViewportWidth;
         m_dynamicParameters.MaxViewportHeight = m_vulkanCaps.MaxViewportHeight;
         m_dynamicParameters.ViewportBoundsRangeMin = m_vulkanCaps.ViewportBoundsRangeMin;
@@ -895,26 +1081,59 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                     DynParams::PerLayerFramebufferAttachmentBit(TextureTarget::TextureCubeMapArray);
             }
         }
-        // Never, on any device, and no longer for the reason it used to be. It used to track
-        // shaderFloat64 because a `dvec3` input needed the Float64 capability to exist in the
-        // module at all; a 64-bit vertex FETCH was already impossible (VK_FORMAT_R64*_SFLOAT is
-        // optional and lavapipe reports zero bufferFeatures for all four), so the attribute
-        // arrived as its 32-bit word pair and PackDoubleVertexInputsPass bitcast it back.
+        // The device feature the whole fp64 story hangs off. With it, a module keeps its
+        // OpCapability Float64 and real doubles reach the driver; without it the transpile
+        // narrows every 64-bit float to 32 (ShaderTranspiler::DemoteFloat64Pass), because
+        // VUID-VkShaderModuleCreateInfo-pCode-08740 forbids the capability outright and no
+        // pipeline could be built from such a module. lavapipe reports it; Adreno and Mali both
+        // report VK_FALSE, so on every real mobile device this is false and the demotion runs
+        // exactly as it always has.
+        m_dynamicParameters.SupportsShaderFloat64 = m_vulkanCaps.SupportsShaderFloat64;
+        // shaderTessellationAndGeometryPointSize, both stage families from the one feature.
+        // False arms the shared phase-B point-size demotion, whose modules then carry no
+        // TessellationPointSize/GeometryPointSize capability and build without the feature.
+        // MOBILEGL_POINT_SIZE_DEMOTION=1 pretends it is absent so the demotion can be
+        // exercised on a healthy driver (lavapipe advertises the feature); =0 restores the
+        // detected answer's declines.
+        {
+            Bool supportsStagePointSize = m_vulkanCaps.SupportsTessellationAndGeometryPointSize;
+            switch (MG_Config::Features.PointSizeDemotion) {
+            case MG_Config::QuirkOverride::ForceOn:
+                MGLOG_I("DirectVulkan: MOBILEGL_POINT_SIZE_DEMOTION=1 - treating tessellation/geometry "
+                        "gl_PointSize as unhosted so the demotion runs on this driver");
+                supportsStagePointSize = false;
+                break;
+            case MG_Config::QuirkOverride::ForceOff:
+                MGLOG_I("DirectVulkan: MOBILEGL_POINT_SIZE_DEMOTION=0 - keeping the built-in and the "
+                        "plain declines regardless of the device feature");
+                supportsStagePointSize = true;
+                break;
+            case MG_Config::QuirkOverride::Auto:
+                break;
+            }
+            m_dynamicParameters.SupportsTessellationPointSize = supportsStagePointSize;
+            m_dynamicParameters.SupportsGeometryPointSize = supportsStagePointSize;
+        }
+        // Never, on any device, and DELIBERATELY NOT COUPLED to the line above even though it
+        // once tracked the same feature. It used to, because a `dvec` input needed Float64 to
+        // exist in the module at all; a 64-bit vertex FETCH was already impossible
+        // (VK_FORMAT_R64*_SFLOAT is optional and lavapipe reports zero bufferFeatures for all
+        // four), so the attribute arrived as its 32-bit word pair and PackDoubleVertexInputsPass
+        // bitcast it back.
         //
-        // The shader half of that is gone: every 64-bit float is narrowed before any module
-        // reaches a backend (ShaderTranspiler::DemoteFloat64Pass), so there is no `double` input
-        // left to bitcast INTO, and feeding a UINT-formatted attribute to what is now a `float`
-        // input would be silent garbage. Reconstructing the value would mean decoding the
-        // IEEE-754 double bit pattern in the shader - software fp64, which is precisely what the
-        // demotion exists to avoid - and on Espryt it would additionally need the ES driver to
-        // fetch 2N uint components where the application declared N doubles, which a dvec3 or
-        // dvec4 cannot even express within one attribute location.
+        // Re-coupling it does not work, and the reason is worth recording because it is not
+        // obvious: this flag decides the VkFormat from the VAO ATTRIBUTE alone, and the attribute
+        // does not know what the shader declared. glVertexAttribFormat(GL_DOUBLE) against a plain
+        // `in vec4` is not only legal but the common case
+        // (KHR-GL43.vertex_attrib_binding.basic-input-case4 does exactly that, and case5 adds
+        // normalized=GL_TRUE), and advanced-bindingUpdate feeds a dvec3 the same way - GL defines
+        // all of them as "doubles in memory, converted to float". Turning the flag on turns the
+        // narrowing OFF for every one of them and the attributes come back unfetched.
         //
-        // So glVertexAttribLFormat / glVertexAttribLPointer are declined here exactly as they
-        // already were on Espryt and on every real mobile device (Adreno and Mali both report
-        // shaderFloat64 == VK_FALSE), and for the same visible reason. A `dvec3` INPUT still
-        // compiles and draws - it is a `vec3` after demotion - as long as the application feeds
-        // it with glVertexAttribPointer(GL_FLOAT) rather than 64-bit data.
+        // What keeps the two halves honest instead is a per-MODULE decision: a vertex module that
+        // declares a 64-bit float INPUT is demoted whole, even where the backend has native fp64,
+        // so `dvec` inputs are `vec` inputs on this backend exactly as they always were. See
+        // ShaderCompiler::SanitizeAndOptimizeBinary.
         m_dynamicParameters.SupportsFloat64VertexAttributes = false;
         m_dynamicParameters.MaxShaderStorageBlockSize =
             std::min(m_vulkanCaps.MaxShaderStorageBlockSize, kMaxAdvertisedShaderStorageBlockSize);
@@ -924,6 +1143,18 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             m_dynamicParameters.SubgroupSupportedFeatures =
                 mapSubgroupFeatures(m_vulkanCaps.SubgroupSupportedOperations);
             m_dynamicParameters.SubgroupQuadOperationsInAllStages = m_vulkanCaps.SubgroupQuadOperationsInAllStages;
+        } else if (ShouldEmulateSubgroups(m_vulkanCaps.SupportsShaderSubgroup)) {
+            // MOBILEGL_MAGMA_EMULATE_SUBGROUP on a device with no native subgroups: the
+            // advertised values describe the 32-lane virtual subgroup the compute
+            // lowering implements (SubgroupSupportPolicy.h / EmulateSubgroupsPass).
+            // GL requires the advertisement and the execution to agree, and on this
+            // path the emulation is what executes; only the compute stage is offered.
+            m_dynamicParameters.SubgroupSize = kEmulatedSubgroupSize;
+            m_dynamicParameters.SubgroupSupportedStages = kEmulatedSubgroupStages;
+            m_dynamicParameters.SubgroupSupportedFeatures = kEmulatedSubgroupFeatures;
+            m_dynamicParameters.SubgroupQuadOperationsInAllStages = false;
+            MGLOG_I("DirectVulkan: emulating 32-lane compute subgroups "
+                    "(MOBILEGL_MAGMA_EMULATE_SUBGROUP, no native subgroup support)");
         } else {
             m_dynamicParameters.SubgroupSize = 0;
             m_dynamicParameters.SubgroupSupportedStages = 0;

MobileGL/MG_Backend/DirectVulkan/BackendObject_DirectVulkan.h

@@ -62,19 +62,21 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     // POST screen shows.
 
     // Static identity of the Magma renderer (renderer/backend names, target GL/GLSL
-    // versions, ExtraVendor) with the baseline extension advertisement (no shader
-    // subgroup, no timer queries). A live backend copies this in its constructor and
+    // versions, ExtraVendor) with the baseline extension advertisement (no runtime-gated
+    // capabilities). A live backend copies this in its constructor and
     // reconciles the Extensions in UpdateAdvertisedExtensions once real capabilities
     // exist; callers that need the advertised list for a known capability set must
     // use BuildAdvertisedExtensions instead.
     const RendererInfo& GetRendererIdentity();
 
     // The full OpenGL extension list Magma advertises (glGetString(GL_EXTENSIONS)) for
-    // a device with the given raw capabilities. The MOBILEGL_DISABLE_SUBGROUP and
+    // a device with the given raw capabilities. The MOBILEGL_MAGMA_DISABLE_SUBGROUP and
     // MOBILEGL_DISABLE_TIMERQUERY escape hatches are applied inside, so callers pass
     // the detected device support (passing an already-gated value is harmless).
     Vector<GLExtension> BuildAdvertisedExtensions(Bool shaderSubgroupSupported, Bool timerQueriesSupported,
-                                                  Bool anisotropicFilteringSupported);
+                                                  Bool anisotropicFilteringSupported,
+                                                  Bool nonZeroIndirectBaseInstanceSupported,
+                                                  Bool cubeMapArraySupported);
 
     // Format: <GPU Name>, Vulkan <Vulkan Version>, Driver <Driver Version> — the exact
     // string an initialized backend returns from GetBackendAPIVersionString (and that

MobileGL/MG_Backend/DirectVulkan/DirectVulkan.cpp

@@ -10,9 +10,11 @@
 #include "DirectVulkanResourceState.h"
 #include "MG_Backend/BackendObjects.h"
 #include "MG_State/GLState/Core.h"
+#include <MG_Pipe/PipeInputsSwitch.h>
 #include "MG_State/GLState/ErrorState/ErrorInfo.h"
 #include "MG_Impl/GLImpl/Framebuffer/GL_Framebuffer.h"
 #include "MG_Util/Converters/GLToMG/TextureEnumConverter.h"
+#include "MG_Util/Metrics/PipeStats.h"
 #include "MG_Util/Metrics/TextureMetrics.h"
 #include "MG_Util/Miscellany/IndexGenerator.h"
 #include <atomic>
@@ -69,9 +71,14 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             // slot's ownership unambiguous.
             Uint64 programLifetimeId = 0;
             Uint32 backendStateVersion = 0;
+            // glShaderStorageBlockBinding deliberately does NOT bump the backend state
+            // version, and the pipeline composite is unnamed so the in-place patch in
+            // DirectVulkan::ShaderStorageBlockBinding can never reach its slot - the
+            // mirror replay bumps only the program's block-binding version. Without this
+            // key the composite's slot kept serving the pre-rebind block.binding.
+            Uint32 blockBindingVersion = 0;
             Vector<StorageBlockResource> storageBlocks;
             Vector<BufferVariableResource> bufferVariables;
-            GLint computeWorkGroupSize[3] = {1, 1, 1};
         };
 
         struct DrawElementsIndirectCommand {
@@ -156,18 +163,33 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             auto& cache = g_programResourceCaches[program.GetExternalIndex()];
             const Uint64 programLifetimeId = program.GetLifetimeId();
             const Uint32 backendStateVersion = program.GetBackendStateVersion();
+            const Uint32 blockBindingVersion = program.GetBlockBindingVersion();
             // The lifetime id must match too: a new program that reuses a deleted
             // program's name and happens to land on the same backendStateVersion (both
             // count from zero) would otherwise be served the dead program's reflection.
             if (cache.programLifetimeId == programLifetimeId &&
                 cache.backendStateVersion == backendStateVersion &&
                 (!cache.storageBlocks.empty() || !cache.bufferVariables.empty())) {
+                if (cache.blockBindingVersion != blockBindingVersion) {
+                    // Only the block bindings moved (glShaderStorageBlockBinding, or the
+                    // pipeline composite's mirror replay - neither touches the backend
+                    // state version): the reflection itself is unchanged, so re-apply the
+                    // overrides by name instead of re-running spirv-reflect. Overrides
+                    // only ever accumulate, so a block without one still holds its
+                    // declared binding.
+                    for (auto& block : cache.storageBlocks) {
+                        const Int rebound = program.GetShaderStorageBlockBindingOverride(block.name);
+                        if (rebound >= 0) block.binding = static_cast<Uint32>(rebound);
+                    }
+                    cache.blockBindingVersion = blockBindingVersion;
+                }
                 return cache;
             }
 
             cache = {};
             cache.programLifetimeId = programLifetimeId;
             cache.backendStateVersion = backendStateVersion;
+            cache.blockBindingVersion = blockBindingVersion;
 
             Vector<SpvReflectShaderModule> modules;
             Vector<Bool> validModules;
@@ -187,16 +209,6 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             }
 
             for (auto& module : modules) {
-                for (Uint32 entryIndex = 0; entryIndex < module.entry_point_count; ++entryIndex) {
-                    const auto& entryPoint = module.entry_points[entryIndex];
-                    if ((entryPoint.shader_stage & SPV_REFLECT_SHADER_STAGE_COMPUTE_BIT) == 0) {
-                        continue;
-                    }
-                    cache.computeWorkGroupSize[0] = static_cast<GLint>(std::max<Uint32>(entryPoint.local_size.x, 1));
-                    cache.computeWorkGroupSize[1] = static_cast<GLint>(std::max<Uint32>(entryPoint.local_size.y, 1));
-                    cache.computeWorkGroupSize[2] = static_cast<GLint>(std::max<Uint32>(entryPoint.local_size.z, 1));
-                }
-
                 uint32_t bindingCount = 0;
                 SpvReflectResult result = spvReflectEnumerateDescriptorBindings(&module, &bindingCount, nullptr);
                 if (result != SPV_REFLECT_RESULT_SUCCESS || bindingCount == 0) {
@@ -256,15 +268,15 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         }
 
         MG_State::GLState::ProgramObject* TryGetDirectVulkanProgram(GLuint program) {
-            if (!MG_State::pGLContext->ValidateProgramName(program)) {
+            if (!MGB_CTX->ValidateProgramName(program)) {
                 return nullptr;
             }
-            auto& programObject = MG_State::pGLContext->GetProgramObject(program);
+            auto& programObject = MGB_CTX->GetProgramObject(program);
             return programObject.get();
         }
 
         const Uint8* ResolveIndirectCommandBytes(const void* indirect, SizeT requiredBytes, const char* label) {
-            auto drawBuffer = MG_State::pGLContext->GetBufferBindingSlot(BufferTarget::DrawIndirect).GetBoundObject();
+            auto drawBuffer = MGB_CTX->GetBufferBindingSlot(BufferTarget::DrawIndirect).GetBoundObject();
             if (drawBuffer) {
                 drawBuffer->SyncPersistentMappedRange();
                 const SizeT commandOffset = reinterpret_cast<SizeT>(indirect);
@@ -323,64 +335,64 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
     void ClearBufferfi(GLenum buffer, GLint drawbuffer, GLfloat depth, GLint stencil) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::ClearBufferfi called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::ClearBufferfi called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::ClearBufferfi called with null GL context");
         pVulkanRenderer->ClearBufferfi(buffer, drawbuffer, depth, stencil);
     }
 
     void ClearBufferfv(GLenum buffer, GLint drawbuffer, const GLfloat* value) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::ClearBufferfv called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::ClearBufferfv called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::ClearBufferfv called with null GL context");
         pVulkanRenderer->ClearBufferfv(buffer, drawbuffer, value);
     }
 
     void ClearBufferuiv(GLenum buffer, GLint drawbuffer, const GLuint* value) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::ClearBufferuiv called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::ClearBufferuiv called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::ClearBufferuiv called with null GL context");
         pVulkanRenderer->ClearBufferuiv(buffer, drawbuffer, value);
     }
 
     void ClearBufferiv(GLenum buffer, GLint drawbuffer, const GLint* value) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::ClearBufferiv called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::ClearBufferiv called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::ClearBufferiv called with null GL context");
         pVulkanRenderer->ClearBufferiv(buffer, drawbuffer, value);
     }
 
     void ClearNamedFramebufferfv(const SharedPtr<MG_State::GLState::FramebufferObject>& framebuffer, GLenum buffer,
                                  GLint drawbuffer, const GLfloat* value) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::ClearNamedFramebufferfv called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::ClearNamedFramebufferfv called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::ClearNamedFramebufferfv called with null GL context");
         pVulkanRenderer->ClearNamedFramebufferfv(framebuffer, buffer, drawbuffer, value);
     }
 
     void ClearNamedFramebufferiv(const SharedPtr<MG_State::GLState::FramebufferObject>& framebuffer, GLenum buffer,
                                  GLint drawbuffer, const GLint* value) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::ClearNamedFramebufferiv called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::ClearNamedFramebufferiv called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::ClearNamedFramebufferiv called with null GL context");
         pVulkanRenderer->ClearNamedFramebufferiv(framebuffer, buffer, drawbuffer, value);
     }
 
     void ClearNamedFramebufferuiv(const SharedPtr<MG_State::GLState::FramebufferObject>& framebuffer, GLenum buffer,
                                   GLint drawbuffer, const GLuint* value) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::ClearNamedFramebufferuiv called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::ClearNamedFramebufferuiv called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::ClearNamedFramebufferuiv called with null GL context");
         pVulkanRenderer->ClearNamedFramebufferuiv(framebuffer, buffer, drawbuffer, value);
     }
 
     void ClearNamedFramebufferfi(const SharedPtr<MG_State::GLState::FramebufferObject>& framebuffer, GLenum buffer,
                                  GLint drawbuffer, GLfloat depth, GLint stencil) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::ClearNamedFramebufferfi called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::ClearNamedFramebufferfi called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::ClearNamedFramebufferfi called with null GL context");
         pVulkanRenderer->ClearNamedFramebufferfi(framebuffer, buffer, drawbuffer, depth, stencil);
     }
 
     void MultiDrawElementsIndirect(GLenum mode, GLenum type, const void* indirect, GLsizei drawcount, GLsizei stride) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::MultiDrawElementsIndirect called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::MultiDrawElementsIndirect called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::MultiDrawElementsIndirect called with null GL context");
         pVulkanRenderer->MultiDrawElementsIndirect(mode, type, indirect, drawcount, stride);
     }
     void MultiDrawArraysIndirect(GLenum mode, const void* indirect, GLsizei drawcount, GLsizei stride) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::MultiDrawArraysIndirect called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::MultiDrawArraysIndirect called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::MultiDrawArraysIndirect called with null GL context");
 
         if (drawcount <= 0) {
             return;
@@ -388,7 +400,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
         // With a bound GL_DRAW_INDIRECT_BUFFER the command parameters may be GPU-written
         // (e.g. by a compute shader), so consume them natively on the GPU.
-        auto drawBuffer = MG_State::pGLContext->GetBufferBindingSlot(BufferTarget::DrawIndirect).GetBoundObject();
+        auto drawBuffer = MGB_CTX->GetBufferBindingSlot(BufferTarget::DrawIndirect).GetBoundObject();
         if (drawBuffer) {
             pVulkanRenderer->MultiDrawArraysIndirect(mode, indirect, drawcount, stride);
             return;
@@ -431,13 +443,13 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     void MultiDrawElementsIndirectCount(GLenum mode, GLenum type, const void* indirect, GLintptr drawcount,
                                         GLsizei maxdrawcount, GLsizei stride) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::MultiDrawElementsIndirectCount called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::MultiDrawElementsIndirectCount called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::MultiDrawElementsIndirectCount called with null GL context");
         pVulkanRenderer->MultiDrawElementsIndirectCount(mode, type, indirect, drawcount, maxdrawcount, stride);
     }
     void MultiDrawArraysIndirectCount(GLenum mode, const void* indirect, GLintptr drawcount,
                                       GLsizei maxdrawcount, GLsizei stride) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::MultiDrawArraysIndirectCount called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::MultiDrawArraysIndirectCount called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::MultiDrawArraysIndirectCount called with null GL context");
 
         if (maxdrawcount <= 0) {
             return;
@@ -451,7 +463,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             return;
         }
 
-        auto parameterBuffer = MG_State::pGLContext->GetBufferBindingSlot(BufferTarget::Parameter).GetBoundObject();
+        auto parameterBuffer = MGB_CTX->GetBufferBindingSlot(BufferTarget::Parameter).GetBoundObject();
         if (!parameterBuffer || drawcount < 0 || static_cast<SizeT>(drawcount) + sizeof(Uint32) > parameterBuffer->GetSize()) {
             MGLOG_E_ONCE("MultiDrawArraysIndirectCount skipped: invalid GL_PARAMETER_BUFFER binding or range");
             return;
@@ -482,7 +494,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     void DrawElementsInstancedBaseVertexBaseInstance(GLenum mode, GLsizei count, GLenum type, const void* indices,
                                                      GLsizei instancecount, GLint basevertex, GLuint baseinstance) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::DrawElementsInstancedBaseVertexBaseInstance called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::DrawElementsInstancedBaseVertexBaseInstance called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::DrawElementsInstancedBaseVertexBaseInstance called with null GL context");
 
         DrawIndexedCmd payload{};
         payload.mode = mode;
@@ -509,7 +521,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     }
     void DrawElementsIndirect(GLenum mode, GLenum type, const void* indirect) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::DrawElementsIndirect called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::DrawElementsIndirect called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::DrawElementsIndirect called with null GL context");
 
         const SizeT indexSize = MG_Util::GetGLTypeSize(type);
         if (indexSize == 0) {
@@ -519,7 +531,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
         // With a bound GL_DRAW_INDIRECT_BUFFER the command parameters may be GPU-written
         // (e.g. by a compute shader), so consume them natively on the GPU.
-        auto drawBuffer = MG_State::pGLContext->GetBufferBindingSlot(BufferTarget::DrawIndirect).GetBoundObject();
+        auto drawBuffer = MGB_CTX->GetBufferBindingSlot(BufferTarget::DrawIndirect).GetBoundObject();
         if (drawBuffer) {
             pVulkanRenderer->MultiDrawElementsIndirect(mode, type, indirect, 1, 0);
             return;
@@ -553,7 +565,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     void DrawArraysInstancedBaseInstance(GLenum mode, GLint first, GLsizei count, GLsizei instancecount,
                                          GLuint baseinstance) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::DrawArraysInstancedBaseInstance called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::DrawArraysInstancedBaseInstance called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::DrawArraysInstancedBaseInstance called with null GL context");
 
         DrawCmd payload{};
         payload.mode = mode;
@@ -568,11 +580,11 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     }
     void DrawArraysIndirect(GLenum mode, const void* indirect) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::DrawArraysIndirect called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::DrawArraysIndirect called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::DrawArraysIndirect called with null GL context");
 
         // With a bound GL_DRAW_INDIRECT_BUFFER the command parameters may be GPU-written
         // (e.g. by a compute shader), so consume them natively on the GPU.
-        auto drawBuffer = MG_State::pGLContext->GetBufferBindingSlot(BufferTarget::DrawIndirect).GetBoundObject();
+        auto drawBuffer = MGB_CTX->GetBufferBindingSlot(BufferTarget::DrawIndirect).GetBoundObject();
         if (drawBuffer) {
             pVulkanRenderer->MultiDrawArraysIndirect(mode, indirect, 1, 0);
             return;
@@ -602,47 +614,47 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     void CopyTexImage2D(GLenum target, GLint level, GLenum internalformat, GLint x, GLint y, GLsizei width,
                         GLsizei height, GLint border) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::CopyTexImage2D called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::CopyTexImage2D called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::CopyTexImage2D called with null GL context");
         pVulkanRenderer->CopyTexSubImage2D(target, level, 0, 0, x, y, width, height);
     }
     void CopyTexSubImage2D(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint x, GLint y, GLsizei width,
                            GLsizei height) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::CopyTexSubImage2D called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::CopyTexSubImage2D called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::CopyTexSubImage2D called with null GL context");
         pVulkanRenderer->CopyTexSubImage2D(target, level, xoffset, yoffset, x, y, width, height);
     }
-    void CopyImageSubData(const SharedPtr<MG_State::GLState::ITextureObject>& srcTexture,
+    void CopyImageSubData(const CopyImageEndpoint& src,
                           GLenum srcTarget, GLint srcLevel, GLint srcX, GLint srcY, GLint srcZ,
-                          const SharedPtr<MG_State::GLState::ITextureObject>& dstTexture,
+                          const CopyImageEndpoint& dst,
                           GLenum dstTarget, GLint dstLevel, GLint dstX, GLint dstY, GLint dstZ,
                           GLsizei srcWidth, GLsizei srcHeight, GLsizei srcDepth) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::CopyImageSubData called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::CopyImageSubData called with null GL context");
-        pVulkanRenderer->CopyImageSubData(srcTexture, srcTarget, srcLevel, srcX, srcY, srcZ,
-                                          dstTexture, dstTarget, dstLevel, dstX, dstY, dstZ,
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::CopyImageSubData called with null GL context");
+        pVulkanRenderer->CopyImageSubData(src, srcTarget, srcLevel, srcX, srcY, srcZ,
+                                          dst, dstTarget, dstLevel, dstX, dstY, dstZ,
                                           srcWidth, srcHeight, srcDepth);
     }
     void GenerateMipmap(GLenum target) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::GenerateMipmap called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::GenerateMipmap called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::GenerateMipmap called with null GL context");
         pVulkanRenderer->GenerateMipmap(target);
     }
 
     void DispatchCompute(GLuint numGroupsX, GLuint numGroupsY, GLuint numGroupsZ) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::DispatchCompute called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::DispatchCompute called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::DispatchCompute called with null GL context");
         pVulkanRenderer->DispatchCompute(numGroupsX, numGroupsY, numGroupsZ);
     }
 
     void DispatchComputeIndirect(GLintptr indirect) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::DispatchComputeIndirect called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::DispatchComputeIndirect called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::DispatchComputeIndirect called with null GL context");
         pVulkanRenderer->DispatchComputeIndirect(indirect);
     }
 
     void MemoryBarrier(GLbitfield barriers) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::MemoryBarrier called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::MemoryBarrier called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::MemoryBarrier called with null GL context");
         pVulkanRenderer->MemoryBarrier(barriers);
     }
 
@@ -661,130 +673,40 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         (void)format;
     }
 
+    // The two compute limits are the only indexed pnames a backend genuinely owns: they come
+    // from the physical device, and MG_Impl/GLImpl/Getter/GL_Getter.cpp asks for them here so it
+    // can raise the answer to the GL required minimum. The same six numbers are carried in
+    // DynamicBackendParameters::MaxComputeWorkGroupCount/Size (filled at capability init from
+    // the same limits), which is their MGPCaps carrier once this entry retires - the
+    // AdvertisedLimitsScenario pins the two against each other. Every other indexed pname names FRONTEND
+    // state (the indexed buffer bindings, the per-unit texture/sampler bindings, the image-unit
+    // bindings, the viewport rectangles, the indexed capabilities) and is answered there before
+    // the table is consulted, so the arms this function used to carry for
+    // GL_SHADER_STORAGE_BUFFER_* and GL_IMAGE_BINDING_* were unreachable duplicates - and not
+    // even faithful ones: the frontend reports the range glBindBufferRange was ASKED for,
+    // verbatim, while these clamped it to the buffer's current storage.
     void GetIntegeri_v(GLenum target, GLuint index, GLint* data) {
         if (!data) return;
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::GetIntegeri_v called with null VulkanRenderer");
+        if (index >= 3) {
+            *data = 0;
+            return;
+        }
         switch (target) {
         case GL_MAX_COMPUTE_WORK_GROUP_COUNT:
-            if (index >= 3) {
-                *data = 0;
-                return;
-            }
             *data = static_cast<GLint>(
                 pVulkanRenderer->GetPhysicalDevice().properties.limits.maxComputeWorkGroupCount[index]);
             return;
         case GL_MAX_COMPUTE_WORK_GROUP_SIZE:
-            if (index >= 3) {
-                *data = 0;
-                return;
-            }
             *data = static_cast<GLint>(
                 pVulkanRenderer->GetPhysicalDevice().properties.limits.maxComputeWorkGroupSize[index]);
             return;
-        case GL_SHADER_STORAGE_BUFFER_BINDING: {
-            auto& point = MG_State::pGLContext->GetBufferBindingPoint(BufferTarget::ShaderStorage, index);
-            auto& obj = point.GetBoundObject();
-            *data = obj ? static_cast<GLint>(obj->GetExternalIndex()) : 0;
-            return;
-        }
-        case GL_SHADER_STORAGE_BUFFER_START: {
-            auto& point = MG_State::pGLContext->GetBufferBindingPoint(BufferTarget::ShaderStorage, index);
-            *data = static_cast<GLint>(point.GetRange().start);
-            return;
-        }
-        case GL_SHADER_STORAGE_BUFFER_SIZE: {
-            auto& point = MG_State::pGLContext->GetBufferBindingPoint(BufferTarget::ShaderStorage, index);
-            auto& obj = point.GetBoundObject();
-            if (!obj) {
-                *data = 0;
-                return;
-            }
-            const auto& range = point.GetRange();
-            const auto start = std::min(range.start, obj->GetSize());
-            const auto end = std::min(range.end, obj->GetSize());
-            *data = static_cast<GLint>(end - start);
-            return;
-        }
-        case GL_IMAGE_BINDING_NAME:
-        case GL_IMAGE_BINDING_LEVEL:
-        case GL_IMAGE_BINDING_LAYERED:
-        case GL_IMAGE_BINDING_LAYER:
-        case GL_IMAGE_BINDING_ACCESS:
-        case GL_IMAGE_BINDING_FORMAT: {
-            if (index >= MG_State::GLState::TextureState::MAX_TEXTURE_IMAGE_UNITS) {
-                *data = 0;
-                return;
-            }
-            auto& imageBinding = MG_State::pGLContext->GetImageTextureBinding(static_cast<Int>(index));
-            if (target == GL_IMAGE_BINDING_NAME) {
-                *data = imageBinding.Texture ? static_cast<GLint>(imageBinding.Texture->GetExternalIndex()) : 0;
-            } else if (target == GL_IMAGE_BINDING_LEVEL) {
-                *data = imageBinding.Level;
-            } else if (target == GL_IMAGE_BINDING_LAYERED) {
-                *data = imageBinding.Layered;
-            } else if (target == GL_IMAGE_BINDING_LAYER) {
-                *data = imageBinding.Layer;
-            } else if (target == GL_IMAGE_BINDING_ACCESS) {
-                *data = static_cast<GLint>(imageBinding.Access);
-            } else {
-                *data = static_cast<GLint>(imageBinding.Format);
-            }
-            return;
-        }
         default:
             *data = 0;
             return;
         }
     }
 
-    void GetInteger64i_v(GLenum target, GLuint index, GLint64* data) {
-        if (!data) return;
-        switch (target) {
-        case GL_SHADER_STORAGE_BUFFER_START: {
-            auto& point = MG_State::pGLContext->GetBufferBindingPoint(BufferTarget::ShaderStorage, index);
-            *data = static_cast<GLint64>(point.GetRange().start);
-            return;
-        }
-        case GL_SHADER_STORAGE_BUFFER_SIZE: {
-            auto& point = MG_State::pGLContext->GetBufferBindingPoint(BufferTarget::ShaderStorage, index);
-            auto& obj = point.GetBoundObject();
-            if (!obj) {
-                *data = 0;
-                return;
-            }
-            const auto& range = point.GetRange();
-            const auto start = std::min(range.start, obj->GetSize());
-            const auto end = std::min(range.end, obj->GetSize());
-            *data = static_cast<GLint64>(end - start);
-            return;
-        }
-        default:
-            *data = 0;
-            return;
-        }
-    }
-
-    void GetProgramiv(GLuint program, GLenum pname, GLint* params) {
-        if (!params) return;
-        auto* programObject = TryGetDirectVulkanProgram(program);
-        if (!programObject) {
-            params[0] = 0;
-            return;
-        }
-        switch (pname) {
-        case GL_COMPUTE_WORK_GROUP_SIZE: {
-            auto& cache = GetProgramResourceCache(*programObject);
-            params[0] = cache.computeWorkGroupSize[0];
-            params[1] = cache.computeWorkGroupSize[1];
-            params[2] = cache.computeWorkGroupSize[2];
-            return;
-        }
-        default:
-            params[0] = 0;
-            return;
-        }
-    }
-
     void ShaderStorageBlockBinding(GLuint program, const GLchar* storageBlockName, GLuint storageBlockBinding) {
         auto* programObject = TryGetDirectVulkanProgram(program);
         if (!programObject || storageBlockName == nullptr) return;
@@ -792,7 +714,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             ? pActiveBackendObject->GetDynamicParameters().MaxShaderStorageBufferBindings
             : 0;
         if (storageBlockBinding >= static_cast<GLuint>(maxBindings)) {
-            MG_State::pGLContext->RecordError(
+            MGB_CTX->RecordError(
                 ErrorCode::InvalidValue,
                 MakeUnique<GenericErrorInfo>("DirectVulkan", __func__, "Shader storage binding is out of range."));
             return;
@@ -817,24 +739,24 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     }
     void ReadPixels(GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, void* pixels) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::ReadPixels called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::ReadPixels called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::ReadPixels called with null GL context");
         pVulkanRenderer->ReadPixels(x, y, width, height, format, type, pixels);
     }
     void GetTexImage(GLenum target, GLint level, GLenum format, GLenum type, GLvoid* pixels) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::GetTexImage called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::GetTexImage called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::GetTexImage called with null GL context");
         pVulkanRenderer->GetTexImage(target, level, format, type, pixels);
     }
     void GetTextureImage(const SharedPtr<MG_State::GLState::ITextureObject>& texture, TextureUploadTarget uploadTarget,
                          GLint level, GLenum format, GLenum type, GLsizei bufSize, GLvoid* pixels) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::GetTextureImage called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::GetTextureImage called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::GetTextureImage called with null GL context");
         pVulkanRenderer->GetTextureImage(texture, uploadTarget, level, format, type, bufSize, pixels);
     }
 
     void Clear(GLbitfield mask) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::Clear called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::Clear called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::Clear called with null GL context");
         pVulkanRenderer->Clear(mask);
     }
 
@@ -862,7 +784,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             return false;
         }
         const Uint8* indexBytes = nullptr;
-        const auto& vao = *MG_State::pGLContext->GetBoundVertexArray();
+        const auto& vao = *MGB_CTX->GetBoundVertexArray();
         const auto& indexBufferShared = vao.GetIndexBufferBindingSlot().GetBoundObject();
         if (indexBufferShared != nullptr) {
             const SizeT offset = reinterpret_cast<SizeT>(indices);
@@ -893,7 +815,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
     void DrawArrays(GLenum mode, GLint first, GLsizei count) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::DrawArrays called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::DrawArrays called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::DrawArrays called with null GL context");
 
         if (mode == GL_LINE_LOOP) {
             if (count < 2) {
@@ -918,7 +840,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
     void DrawElements(GLenum mode, GLsizei count, GLenum type, const void* indices) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::DrawElements called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::DrawElements called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::DrawElements called with null GL context");
 
         if (mode == GL_LINE_LOOP) {
             Vector<Uint32> closedIndices;
@@ -941,7 +863,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
     void MultiDrawArrays(GLenum mode, const GLint* first, const GLsizei* count, GLsizei drawcount) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::MultiDrawArrays called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::MultiDrawArrays called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::MultiDrawArrays called with null GL context");
         if (drawcount <= 0) {
             return;
         }
@@ -986,7 +908,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // MultiDrawIndexedCmd left the client-memory shape addressing a view whose byte
         // offset is a hardcoded 0, so UploadAndBindIndexBuffer saw a null client pointer,
         // declined the whole batch and painted nothing.)
-        const auto& vao = *MG_State::pGLContext->GetBoundVertexArray();
+        const auto& vao = *MGB_CTX->GetBoundVertexArray();
         if (vao.GetIndexBufferBindingSlot().GetBoundObject() == nullptr) {
             for (GLsizei i = 0; i < drawcount; ++i) {
                 if (count[i] <= 0) {
@@ -1047,13 +969,13 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     void MultiDrawElements(GLenum mode, const GLsizei* count, GLenum type, const GLvoid* const* indices,
                            GLsizei drawcount) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::MultiDrawElements called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::MultiDrawElements called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::MultiDrawElements called with null GL context");
         MultiDrawElementsImpl(mode, count, type, indices, drawcount, nullptr);
     }
 
     void DrawElementsBaseVertex(GLenum mode, GLsizei count, GLenum type, const GLvoid* indices, GLint basevertex) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::DrawElementsBaseVertex called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::DrawElementsBaseVertex called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::DrawElementsBaseVertex called with null GL context");
         if (mode == GL_LINE_LOOP) {
             Vector<Uint32> closedIndices;
             if (BuildClosedLineLoopIndices(count, type, indices, closedIndices)) {
@@ -1077,14 +999,14 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     void MultiDrawElementsBaseVertex(GLenum mode, const GLsizei* count, GLenum type, const GLvoid* const* indices,
                                      GLsizei drawcount, const GLint* basevertex) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::MultiDrawElementsBaseVertex called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::MultiDrawElementsBaseVertex called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::MultiDrawElementsBaseVertex called with null GL context");
         MultiDrawElementsImpl(mode, count, type, indices, drawcount, basevertex);
     }
 
     void BlitFramebuffer(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1,
                          GLint dstY1, GLbitfield mask, GLenum filter) {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::BlitFramebuffer called with null VulkanRenderer");
-        MOBILEGL_ASSERT(MG_State::pGLContext, "DirectVulkan::BlitFramebuffer called with null GL context");
+        MOBILEGL_ASSERT(MGB_CTX_LIVE, "DirectVulkan::BlitFramebuffer called with null GL context");
         pVulkanRenderer->BlitFramebuffer(srcX0, srcY0, srcX1, srcY1, dstX0, dstY0, dstX1, dstY1, mask, filter);
     }
 
@@ -1185,6 +1107,12 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             SharedPtr<VkTimerQueryManager::TimestampRecord> end;
             // Kind::Occlusion - pool slots recorded between Begin/End; summed at result time.
             Vector<Uint32> occlusionSlots;
+            // Kind::XfbGenerated - reroute-pool slots for the span's XFB-INACTIVE
+            // draws, where the renderer's reroute is armed (the affected driver's
+            // stream query counts nothing without an open capture; see
+            // VulkanRenderer::BeginXfbQueryForDraw). Summed alongside the stream
+            // slots above, which keep the span's XFB-active draws.
+            Vector<Uint32> rerouteSlots;
             // Renderer generation the records were written under (see
             // g_rendererGeneration). A stale generation resolves as available
             // with a final zero result: the records' pool indices and frame
@@ -1194,11 +1122,19 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             // stale queries are always safe to delete.
             Uint64 rendererGeneration = 0;
             // Kind::XfbGenerated - the frontend's paused-draw primitive counter when the
-            // query began. VK_QUERY_TYPE_TRANSFORM_FEEDBACK_STREAM_EXT counts only what the
-            // capture saw, so a draw made while the span was paused is invisible to it -
-            // but GL_PRIMITIVES_GENERATED counts what the last vertex processing stage
-            // emitted regardless. The delta closes that gap at result time.
+            // query began. On the affected drivers VK_QUERY_TYPE_TRANSFORM_FEEDBACK_STREAM_EXT
+            // counts only what the capture saw, so a draw made while the span was paused is
+            // invisible to it - but GL_PRIMITIVES_GENERATED counts what the last vertex
+            // processing stage emitted regardless. The delta closes that gap at result time.
             Uint64 pausedPrimitiveSnapshot = 0;
+            // ...unless the GPU already counted those paused draws when the span opened -
+            // through the reroute pool (VulkanRenderer::BeginXfbQueryForDraw reroutes every
+            // draw with no open capture, paused ones included) or, where the probe measured
+            // the stream query as counting capture-less draws, through the stream slot the
+            // paused draw still takes. Adding the CPU delta on top would count them twice,
+            // and the CPU counter is the weaker source anyway: only 3 of the ~15 draw entry
+            // points write it and it answers 0 for GL_PATCHES.
+            Bool pausedPrimitivesCountedByGpu = false;
         };
     } // namespace
 
@@ -1292,13 +1228,14 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         if (query->kind == VulkanTimerQuery::Kind::XfbWritten ||
             query->kind == VulkanTimerQuery::Kind::XfbGenerated) {
             Uint64 primitives = 0;
-            if (!pVulkanRenderer->ResolveXfbQueryResult(query->occlusionSlots,
+            if (!pVulkanRenderer->ResolveXfbQueryResult(query->occlusionSlots, query->rerouteSlots,
                                                         query->kind == VulkanTimerQuery::Kind::XfbGenerated,
                                                         primitives)) {
                 return false;
             }
-            if (query->kind == VulkanTimerQuery::Kind::XfbGenerated && MG_State::pGLContext != nullptr) {
-                primitives += MG_State::pGLContext->GetTransformFeedbackPausedPrimitiveCounter() -
+            if (query->kind == VulkanTimerQuery::Kind::XfbGenerated &&
+                !query->pausedPrimitivesCountedByGpu && MGB_CTX_LIVE) {
+                primitives += MGB_CTX->GetTransformFeedbackPausedPrimitiveCounter() -
                               query->pausedPrimitiveSnapshot;
             }
             *outNanoseconds = primitives;
@@ -1345,7 +1282,10 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         query->kind = generated ? VulkanTimerQuery::Kind::XfbGenerated : VulkanTimerQuery::Kind::XfbWritten;
         query->rendererGeneration = GetRendererGeneration();
         query->pausedPrimitiveSnapshot =
-            MG_State::pGLContext ? MG_State::pGLContext->GetTransformFeedbackPausedPrimitiveCounter() : 0;
+            MGB_CTX_LIVE ? MGB_CTX->GetTransformFeedbackPausedPrimitiveCounter() : 0;
+        // Read AFTER StartXfbQueryCapture, which is where a failed reroute-pool creation
+        // disarms: the answer is then what this span will actually do for every draw.
+        query->pausedPrimitivesCountedByGpu = generated && pVulkanRenderer->ArePausedDrawsGpuCounted();
         return query;
     }
 
@@ -1356,7 +1296,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             return;
         }
         pVulkanRenderer->StopXfbQueryCapture(
-            query->kind == VulkanTimerQuery::Kind::XfbGenerated ? 1u : 0u, query->occlusionSlots);
+            query->kind == VulkanTimerQuery::Kind::XfbGenerated ? 1u : 0u, query->occlusionSlots,
+            query->rerouteSlots);
     }
 
     BackendQueryHandle BeginOcclusionQuery() {
@@ -1390,5 +1331,12 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     void Present() {
         MOBILEGL_ASSERT(pVulkanRenderer, "DirectVulkan::Present called with null VulkanRenderer");
         pVulkanRenderer->Present();
+        // THE frame boundary for the MGPipe counters, at the backend entry point rather
+        // than inside VulkanRenderer::Present: that function has an early return for the
+        // no-usable-swapchain case, and a suspended frame is still a frame the counters
+        // must close.
+        if (MG_Util::PipeStats::Enabled()) {
+            MG_Util::PipeStats::OnPresent();
+        }
     }
 } // namespace MobileGL::MG_Backend::DirectVulkan

MobileGL/MG_Backend/DirectVulkan/DirectVulkan.h

@@ -82,9 +82,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                         GLsizei height, GLint border);
     void CopyTexSubImage2D(GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint x, GLint y, GLsizei width,
                            GLsizei height);
-    void CopyImageSubData(const SharedPtr<MG_State::GLState::ITextureObject>& srcTexture,
+    void CopyImageSubData(const CopyImageEndpoint& src,
                           GLenum srcTarget, GLint srcLevel, GLint srcX, GLint srcY, GLint srcZ,
-                          const SharedPtr<MG_State::GLState::ITextureObject>& dstTexture,
+                          const CopyImageEndpoint& dst,
                           GLenum dstTarget, GLint dstLevel, GLint dstX, GLint dstY, GLint dstZ,
                           GLsizei srcWidth, GLsizei srcHeight, GLsizei srcDepth);
     void GenerateMipmap(GLenum target);
@@ -95,8 +95,6 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     void BindImageTexture(GLuint unit, GLuint texture, GLint level, GLboolean layered, GLint layer, GLenum access,
                           GLenum format);
     void GetIntegeri_v(GLenum target, GLuint index, GLint* data);
-    void GetInteger64i_v(GLenum target, GLuint index, GLint64* data);
-    void GetProgramiv(GLuint program, GLenum pname, GLint* params);
     void ShaderStorageBlockBinding(GLuint program, const GLchar* storageBlockName, GLuint storageBlockBinding);
     void ReadPixels(GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, void* pixels);
     void GetTexImage(GLenum target, GLint level, GLenum format, GLenum type, GLvoid* pixels);

MobileGL/MG_Backend/DirectVulkan/Renderer/PipelineFactory.cpp

@@ -201,11 +201,16 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         XXHASH_VERIFY(XXH64_update(m_hashState, &payload.renderPass, sizeof(payload.renderPass)));
         XXHASH_VERIFY(XXH64_update(m_hashState, &payload.colorAttachmentCount, sizeof(payload.colorAttachmentCount)));
         XXHASH_VERIFY(XXH64_update(m_hashState, &payload.rasterizationSamples, sizeof(payload.rasterizationSamples)));
+        XXHASH_VERIFY(XXH64_update(m_hashState, &payload.sampleShadingEnable, sizeof(payload.sampleShadingEnable)));
+        XXHASH_VERIFY(XXH64_update(m_hashState, &payload.minSampleShading, sizeof(payload.minSampleShading)));
+        XXHASH_VERIFY(XXH64_update(m_hashState, &payload.sampleMask, sizeof(payload.sampleMask)));
         XXHASH_VERIFY(XXH64_update(m_hashState, &payload.subpass, sizeof(payload.subpass)));
         XXHASH_VERIFY(XXH64_update(m_hashState, &payload.topology, sizeof(payload.topology)));
         XXHASH_VERIFY(
             XXH64_update(m_hashState, &payload.primitiveRestartEnable, sizeof(payload.primitiveRestartEnable)));
         XXHASH_VERIFY(XXH64_update(m_hashState, &payload.patchControlPoints, sizeof(payload.patchControlPoints)));
+        XXHASH_VERIFY(XXH64_update(m_hashState, &payload.passthroughTessControlKey,
+                                   sizeof(payload.passthroughTessControlKey)));
         XXHASH_VERIFY(XXH64_update(m_hashState, &payload.viewportCount, sizeof(payload.viewportCount)));
         XXHASH_VERIFY(XXH64_update(m_hashState, &payload.polygonMode, sizeof(payload.polygonMode)));
         XXHASH_VERIFY(XXH64_update(m_hashState, &payload.cullMode, sizeof(payload.cullMode)));
@@ -435,6 +440,14 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
         VkPipelineMultisampleStateCreateInfo ms{VK_STRUCTURE_TYPE_PIPELINE_MULTISAMPLE_STATE_CREATE_INFO};
         ms.rasterizationSamples = payload.rasterizationSamples;
+        ms.sampleShadingEnable = payload.sampleShadingEnable ? VK_TRUE : VK_FALSE;
+        // Ignored by Vulkan unless sampleShadingEnable is set, but written unconditionally so the
+        // struct's bytes match the hash the payload was keyed by.
+        ms.minSampleShading = payload.minSampleShading;
+        // GL_SAMPLE_MASK / glSampleMaski. Left at nullptr - which Vulkan reads as all-ones - until
+        // now, so glSampleMaski was a silent no-op on this backend while DirectGLES forwarded it.
+        // The pointer has to outlive the vkCreateGraphicsPipelines call, which the payload does.
+        ms.pSampleMask = payload.sampleMask;
 
         VkPipelineDepthStencilStateCreateInfo depthStencil{VK_STRUCTURE_TYPE_PIPELINE_DEPTH_STENCIL_STATE_CREATE_INFO};
         depthStencil.depthTestEnable = payload.depthTestEnable ? VK_TRUE : VK_FALSE;

MobileGL/MG_Backend/DirectVulkan/Renderer/PipelineFactory.h

@@ -37,11 +37,39 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             VkRenderPass renderPass = VK_NULL_HANDLE;
             Uint32 colorAttachmentCount = 1;
             VkSampleCountFlagBits rasterizationSamples = VK_SAMPLE_COUNT_1_BIT;
+            // glEnable(GL_SAMPLE_SHADING) + glMinSampleShading, which Vulkan bakes into the
+            // pipeline rather than exposing as dynamic state - so both are part of the pipeline's
+            // identity and both are hashed. The renderer leaves the enable false unless the
+            // device's sampleRateShading feature was enabled
+            // (VUID-VkPipelineMultisampleStateCreateInfo-sampleShadingEnable-00784).
+            Bool sampleShadingEnable = false;
+            Float minSampleShading = 0.0f;
+            // glEnable(GL_SAMPLE_MASK) + glSampleMaski, the fixed-function coverage mask, already
+            // reduced to what GL says this draw gets (VulkanRenderer::ResolveEffectiveSampleMask:
+            // all-ones unless the target is genuinely multisampled). Pipeline state like the two
+            // above - Vulkan has no dynamic sample mask before VK_EXT_extended_dynamic_state3 -
+            // so it is hashed with them, and all-ones has to keep producing the pipeline a null
+            // pSampleMask always did.
+            //
+            // TWO words, though GL only ever fills the first. GL_MAX_SAMPLE_MASK_WORDS is clamped
+            // to 1 on both backends, so glSampleMaski writes index 0 and nothing else - but the
+            // count Vulkan READS is ceil(rasterizationSamples / 32), which is 2 on a 64-sample
+            // target, and GetAdvertisedMaxSamples does not cap the driver's sample count. A
+            // single Uint32 here let such a pipeline read one word past the member (the next
+            // struct field). The second word is all-ones: full coverage for samples 32..63, which
+            // is the only honest answer when GL has no state describing them.
+            Uint32 sampleMask[2] = {0xffffffffu, 0xffffffffu};
             Uint32 subpass = 0;
             VkPrimitiveTopology topology = VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST;
             Bool primitiveRestartEnable = false;
             // GL_PATCH_VERTICES; only read for a PATCH_LIST topology.
             Uint32 patchControlPoints = 3;
+            // ProgramFactory::ComputePassthroughTessControlKey of the synthesized pass-through
+            // tessellation control stage below, or 0 when this pipeline has none. Hashed, because
+            // the levels glPatchParameterfv set are compiled INTO that module and are not a
+            // function of the program or of patchControlPoints - see the note on
+            // passthroughTessControlStage.
+            Uint64 passthroughTessControlKey = 0;
             // How many of ARB_viewport_array's viewports this pipeline rasterizes into. 1 for
             // every program that never assigns gl_ViewportIndex, which is all of them outside the
             // conformance suite - the wide shape costs a longer vkCmdSetViewport/Scissor per state
@@ -87,8 +115,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             // renderer could not build one, and CreatePipeline refuses the pipeline - the same
             // refusal it applies when `stages` itself is half-tessellated.
             //
-            // NOT hashed: it is a pure function of the program and of patchControlPoints, both
-            // of which ComputeHash already mixes in.
+            // NOT hashed directly: it is a pure function of the program, of patchControlPoints and
+            // of the default tessellation levels - the first two of which ComputeHash already
+            // mixes in, and the third of which arrives through passthroughTessControlKey above.
             VkPipelineShaderStageCreateInfo passthroughTessControlStage{};
             const VkPipelineVertexInputStateCreateInfo* vertexInputState = nullptr;
             // Diagnostic only; may be null. Read solely from the pipeline-creation failure path.

MobileGL/MG_Backend/DirectVulkan/Renderer/ProgramFactory.h

@@ -76,6 +76,40 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         using CompileOptionFlags = Flags<CompileOptionBit>;
         using HashType = Uint64;
 
+        // The gl_PerVertex members a pass-through tessellation control stage may have to carry,
+        // in the order glslang declares them - which is the order a redeclaration must use.
+        // Which of them exist is a function of the neighbouring stage's GLSL VERSION
+        // (gl_CullDistance joins the block at #version 450), so the mask is read off that
+        // stage's SPIR-V rather than assumed. See ReflectPerVertexInputMembers.
+        enum class PerVertexMemberBit : Uint32 {
+            Position = 1u << 0,
+            PointSize = 1u << 1,
+            ClipDistance = 1u << 2,
+            CullDistance = 1u << 3,
+        };
+        // What a program parsed below #version 450 carries, and the fallback when a module's
+        // block cannot be read.
+        static constexpr Uint32 kDefaultPerVertexMembers =
+            static_cast<Uint32>(PerVertexMemberBit::Position) | static_cast<Uint32>(PerVertexMemberBit::PointSize) |
+            static_cast<Uint32>(PerVertexMemberBit::ClipDistance);
+
+        struct UpdateAfterBindLimits {
+            Bool enabled = false;
+            Uint32 maxPerStageSamplers = 0;
+            Uint32 maxPerStageUniformBuffers = 0;
+            Uint32 maxPerStageStorageBuffers = 0;
+            Uint32 maxPerStageSampledImages = 0;
+            Uint32 maxPerStageStorageImages = 0;
+            Uint32 maxPerStageResources = 0;
+            Uint32 maxSetSamplers = 0;
+            Uint32 maxSetUniformBuffers = 0;
+            Uint32 maxSetUniformBuffersDynamic = 0;
+            Uint32 maxSetStorageBuffers = 0;
+            Uint32 maxSetStorageBuffersDynamic = 0;
+            Uint32 maxSetSampledImages = 0;
+            Uint32 maxSetStorageImages = 0;
+        };
+
         struct VkProgramObject {
             static constexpr Uint32 kMaxVertexInputLocations = 32;
 
@@ -88,6 +122,10 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
             // Layout data (previously in separate VkProgramLayout)
             VkDescriptorSetLayout descriptorSetLayout = VK_NULL_HANDLE;
+            // True only when this layout passed every descriptor-indexing feature and
+            // update-after-bind limit gate at reflection time. It controls both the
+            // layout/binding flags and the pool class used by UniformManager.
+            Bool usesUpdateAfterBind = false;
             VkPipelineLayout pipelineLayout = VK_NULL_HANDLE;
             Vector<DescriptorBindingKind> bindingKinds;
             // The bindings this program actually declares, ascending. bindingKinds is sized to the
@@ -164,6 +202,32 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             // tessellation stages are present or neither
             // (VUID-VkGraphicsPipelineCreateInfo-pStages-00730). So the draw path has to supply
             // the pass-through stage GL describes; see GetOrCreatePassthroughTessControlStage.
+            // True when this program was built AS a transform-feedback capture variant but its
+            // last pre-rasterization module does NOT carry the Xfb execution mode - so the
+            // renderer must decline the capture span instead of issuing
+            // vkCmdBeginTransformFeedbackEXT against it
+            // (VUID-vkCmdBeginTransformFeedbackEXT-None-04128).
+            //
+            // Two ways to get here, and neither is visible from GL state, which is all
+            // BeginXfbCaptureForDraw otherwise consults: the clip/XFB validation backstop had to
+            // rewind past the capture decoration, or XfbCaptureDecoratePass resolved none of the
+            // requested varyings and returned without changing anything (its own MGLOG_E path)
+            // while its runner still reported success. Both used to ship a non-Xfb module under
+            // an Xfb-flagged cache entry - the flag and the layout are part of the program cache
+            // key, so it was sticky for every later captured draw of the program, not a glitch.
+            Bool xfbCaptureDeclined = false;
+            // The program has a tessellation or geometry module declaring TessellationPointSize /
+            // GeometryPointSize on a device whose shaderTessellationAndGeometryPointSize feature
+            // is off, so a pipeline built from it is invalid usage
+            // (VUID-RuntimeSpirv-PointSize-06439). Its draws are refused in SetupDraw rather than
+            // handed to the driver - the same contract PipelineFactory's half-tessellated refusal
+            // implements one level up, and the counterpart of the DirectGLES arm that reports a
+            // driver with neither point-size extension by name.
+            //
+            // Sticky by construction, which is what makes ONE log line honest: the flag lives on
+            // the cache entry, so every later draw of the same program variant reads the same
+            // answer instead of re-deciding it.
+            Bool pointSizeCapabilityUnsupported = false;
             Bool needsPassthroughTessControl = false;
             // ...and the pass-through this renderer can synthesize carries gl_Position and
             // nothing else, so it is only correct when the evaluation stage's inputs are
@@ -173,6 +237,15 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             // instead (PipelineFactory::CreatePipeline refuses the pipeline and the draw is
             // skipped). See ReflectPassthroughTessControlNeed.
             Bool passthroughTessControlEmulatable = false;
+            // Which gl_PerVertex members the evaluation stage's `in gl_PerVertex gl_in[]` block
+            // actually carries, as a PerVertexMemberBit mask read off its SPIR-V. The synthesized
+            // control stage has to redeclare the SAME shape: glslang appends gl_CullDistance to
+            // that block from #version 450 upward, so a 450/460 program - and every ESSL program,
+            // which the source processor rewrites to "#version 460 core" - carries four members
+            // where a 430 program carries three. A fixed three-member pass-through fed the
+            // evaluation stage a differently-shaped block, which is the black-frame-no-error case
+            // this whole family is written around.
+            Uint32 passthroughPerVertexMembers = 0;
             // Frame-boundary counter value of the last GetOrCreateProgram hit; drives
             // cache eviction (see OnFrameBoundary). Mutable: the draw snapshot's memoised
             // entry pointer re-stamps use through a const reference (StampProgramUse).
@@ -196,6 +269,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 // a pipeline failure would be reported against the wrong SPIR-V.
                 stageSpirvDigests = std::move(other.stageSpirvDigests);
                 descriptorSetLayout = other.descriptorSetLayout;
+                usesUpdateAfterBind = other.usesUpdateAfterBind;
                 pipelineLayout = other.pipelineLayout;
                 bindingKinds = std::move(other.bindingKinds);
                 activeBindings = std::move(other.activeBindings);
@@ -224,11 +298,14 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 fragmentInputComponentCount = other.fragmentInputComponentCount;
                 fragmentReplacesDepth = other.fragmentReplacesDepth;
                 readsBaseVertexBuiltin = other.readsBaseVertexBuiltin;
+                writesViewportIndexBuiltin = other.writesViewportIndexBuiltin;
                 needsPassthroughTessControl = other.needsPassthroughTessControl;
                 passthroughTessControlEmulatable = other.passthroughTessControlEmulatable;
+                passthroughPerVertexMembers = other.passthroughPerVertexMembers;
                 lastUsedFrame = other.lastUsedFrame;
                 other.hash = 0;
                 other.descriptorSetLayout = VK_NULL_HANDLE;
+                other.usesUpdateAfterBind = false;
                 other.pipelineLayout = VK_NULL_HANDLE;
                 other.hasStorageImages = false;
                 other.declinedDescriptors = false;
@@ -240,8 +317,10 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 other.fragmentInputComponentCount = 0;
                 other.fragmentReplacesDepth = false;
                 other.readsBaseVertexBuiltin = false;
+                other.writesViewportIndexBuiltin = false;
                 other.needsPassthroughTessControl = false;
                 other.passthroughTessControlEmulatable = false;
+                other.passthroughPerVertexMembers = 0;
                 other.lastUsedFrame = 0;
             }
             VkProgramObject& operator=(VkProgramObject&& other) noexcept {
@@ -254,6 +333,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 modules = std::move(other.modules);
                 stageSpirvDigests = std::move(other.stageSpirvDigests); // travels with `modules` - see the move ctor
                 descriptorSetLayout = other.descriptorSetLayout;
+                usesUpdateAfterBind = other.usesUpdateAfterBind;
                 pipelineLayout = other.pipelineLayout;
                 bindingKinds = std::move(other.bindingKinds);
                 activeBindings = std::move(other.activeBindings);
@@ -282,11 +362,14 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 fragmentInputComponentCount = other.fragmentInputComponentCount;
                 fragmentReplacesDepth = other.fragmentReplacesDepth;
                 readsBaseVertexBuiltin = other.readsBaseVertexBuiltin;
+                writesViewportIndexBuiltin = other.writesViewportIndexBuiltin;
                 needsPassthroughTessControl = other.needsPassthroughTessControl;
                 passthroughTessControlEmulatable = other.passthroughTessControlEmulatable;
+                passthroughPerVertexMembers = other.passthroughPerVertexMembers;
                 lastUsedFrame = other.lastUsedFrame;
                 other.hash = 0;
                 other.descriptorSetLayout = VK_NULL_HANDLE;
+                other.usesUpdateAfterBind = false;
                 other.pipelineLayout = VK_NULL_HANDLE;
                 other.hasStorageImages = false;
                 other.declinedDescriptors = false;
@@ -298,8 +381,10 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 other.fragmentInputComponentCount = 0;
                 other.fragmentReplacesDepth = false;
                 other.readsBaseVertexBuiltin = false;
+                other.writesViewportIndexBuiltin = false;
                 other.needsPassthroughTessControl = false;
                 other.passthroughTessControlEmulatable = false;
+                other.passthroughPerVertexMembers = 0;
                 other.lastUsedFrame = 0;
                 return *this;
             }
@@ -343,12 +428,41 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             virtual void OnProgramEvicted(HashType programHash, VkDescriptorSetLayout descriptorSetLayout) = 0;
         };
 
-        explicit ProgramFactory(VkDevice device, const VulkanRendererConfig& config, Uint32 maxBindings = 16,
-                                Bool shaderDrawParametersEnabled = false,
-                                Bool unformattedFloatStorageImagesEnabled = false)
+        // How this factory's compute modules implement GL_KHR_shader_subgroup. Computed
+        // once at renderer initialization (SubgroupSupportPolicy.h + the device's
+        // subgroup properties) so lowering can never disagree with the advertised
+        // capabilities. Native subgroup operations always execute natively; the two
+        // repair passes patch modules AROUND them, and the emulation only replaces them
+        // on opted-in devices with no subgroup support at all.
+        struct SubgroupLoweringPolicy {
+            Bool emulateSubgroups = false;      // MOBILEGL_MAGMA_EMULATE_SUBGROUP, no-native-support devices
+            Bool fixIterationRPSubgroupScratch = false; // patch iterationRP's under-declared scratch
+            Bool fixIterationRPBarrier = false; // repair Program 203's shared-scratch race
+            Bool deriveNumSubgroups = false;    // repair the NumSubgroups builtin
+            Bool requireFullSubgroups = false;  // computeFullSubgroups enabled on the device
+            Uint32 nativeSubgroupSize = 0;
+            // Full-subgroup launches are bounded by this device limit; a dispatch whose
+            // workgroup needs more subgroups than this cannot request the flag.
+            Uint32 maxComputeWorkgroupSubgroups = 0;
+            // VkPhysicalDeviceLimits::maxComputeSharedMemorySize; bounds the scratch the
+            // emulation pass may add (0 falls back to the Vulkan minimum, 16384).
+            Uint32 maxComputeSharedMemoryBytes = 0;
+        };
+
+        explicit ProgramFactory(VkDevice device, const VulkanRendererConfig& config, Uint32 maxBindings,
+                                Bool shaderDrawParametersEnabled,
+                                Bool unformattedFloatStorageImagesEnabled,
+                                Bool tessellationAndGeometryPointSizeEnabled,
+                                Bool enableSpirvValidation,
+                                UpdateAfterBindLimits updateAfterBindLimits,
+                                SubgroupLoweringPolicy subgroupPolicy)
             : m_device(device), m_maxBindings(maxBindings), m_config(config),
               m_shaderDrawParametersEnabled(shaderDrawParametersEnabled),
-              m_unformattedFloatStorageImagesEnabled(unformattedFloatStorageImagesEnabled) {
+              m_unformattedFloatStorageImagesEnabled(unformattedFloatStorageImagesEnabled),
+              m_tessellationAndGeometryPointSizeEnabled(tessellationAndGeometryPointSizeEnabled),
+              m_enableSpirvValidation(enableSpirvValidation),
+              m_updateAfterBindLimits(updateAfterBindLimits),
+              m_subgroupPolicy(subgroupPolicy) {
             VkProgramObject::s_device = device;
         }
         // Destroys the pass-through tessellation control modules. Runs while the device is
@@ -392,6 +506,14 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         static VkShaderStageFlagBits ToVkStage(ShaderStage stage);
         static VkFormat ConvertSpirvImageFormatToVkFormat(SpvImageFormat format);
         static SamplerNumericDomain UniformTypeToSamplerNumericDomain(GLenum glType);
+        // The same question for an IMAGE uniform (`image2D`, `uimageBuffer`, ...), which the
+        // sampler form above deliberately does not answer. Kept separate rather than folded in
+        // because the two are asked in different places for different reasons: a sampler's domain
+        // decides a sampled VIEW format, an image's decides what a placeholder descriptor for an
+        // UNBOUND image unit must be (see UniformManager::AcquireUnboundTexelBufferView and
+        // GetUnboundStorageImageTexture) - a formatless `writeonly` declaration reflects no
+        // format at all, and the numeric domain is then the only thing that constrains it.
+        static SamplerNumericDomain UniformTypeToImageNumericDomain(GLenum glType);
         // True when any entry point declares the DepthReplacing execution mode, i.e. the
         // shader assigns gl_FragDepth. Exposed so the blended depth-write quirk's exemption
         // can be pinned by tests. A false negative loses the exemption, so such a shader is
@@ -421,18 +543,40 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // the caller then has no control stage to inject, and CreatePipeline refuses the
         // pipeline rather than handing the driver a half-tessellated one.
         //
-        // Keyed on the patch size because GL takes the output patch size from PATCH_VERTICES,
-        // which is draw state, not link state - the CTS case that motivated this links at the
-        // default 3 and draws at 4. The pipeline cache already re-keys on patchControlPoints,
-        // so the module a pipeline was built with is part of that pipeline's identity.
-        // Compiling is bounded by the number of distinct patch sizes a program draws with
-        // (MAX_PATCH_VERTICES = 32 in the worst case, one or two in practice) and only ever
-        // happens for the rare program that has no control stage at all.
-        VkPipelineShaderStageCreateInfo GetOrCreatePassthroughTessControlStage(Uint32 patchVertices);
+        // Keyed on the patch size, the six default tessellation levels AND the gl_PerVertex
+        // member set, because all three decide what the generator emits. The size comes from
+        // PATCH_VERTICES and the levels from PATCH_DEFAULT_OUTER_LEVEL / PATCH_DEFAULT_INNER_LEVEL
+        // - draw state rather than link state, and the CTS case that motivated this links at the
+        // default 3 and draws at 4. The member set comes from the neighbouring evaluation stage's
+        // own SPIR-V, so two programs at different GLSL versions need different modules. The
+        // pipeline cache re-keys on the same inputs, so the module a pipeline was built with is
+        // part of that pipeline's identity. Compiling is bounded by the number of distinct
+        // (size, levels, members) combinations a program draws with - one or two in practice -
+        // and only ever happens for the rare program that has no control stage at all.
+        VkPipelineShaderStageCreateInfo GetOrCreatePassthroughTessControlStage(Uint32 patchVertices,
+                                                                              const FloatVec4& defaultOuterLevel,
+                                                                              const FloatVec2& defaultInnerLevel,
+                                                                              Uint32 perVertexMembers);
 
         // Source of the module above. Exposed for tests: the generated GLSL is the whole
         // contract with the evaluation stage, so it is worth pinning independently of a device.
-        static String BuildPassthroughTessControlSource(Uint32 patchVertices);
+        static String BuildPassthroughTessControlSource(Uint32 patchVertices, const FloatVec4& defaultOuterLevel,
+                                                        const FloatVec2& defaultInnerLevel, Uint32 perVertexMembers);
+
+        // The identity of one such module: everything the generator bakes in, folded into a
+        // 64-bit key over the raw bits (so -0.0 and +0.0 key apart, which is harmless, and NaN
+        // keys to itself, which is what matters). Shared with PipelineFactory, which mixes the
+        // same value into the pipeline hash so a pipeline can never be handed a module built for
+        // different levels or a different block shape.
+        static Uint64 ComputePassthroughTessControlKey(Uint32 patchVertices, const FloatVec4& defaultOuterLevel,
+                                                       const FloatVec2& defaultInnerLevel, Uint32 perVertexMembers);
+
+        // The PerVertexMemberBit mask of the INPUT per-vertex block a module declares, read
+        // straight out of its SPIR-V (OpMemberDecorate ... BuiltIn on the struct behind the one
+        // Input variable that is an array of a Block-decorated struct). Zero when the module has
+        // no such block. Exposed for tests, which is the only way to pin the shape agreement
+        // without a device.
+        static Uint32 ReflectPerVertexInputMembers(const Vector<Uint>& spirv);
 
     private:
         struct ProgramLookupCache {
@@ -443,13 +587,17 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         };
 
         static TextureTarget UniformTypeToTextureTarget(GLenum glType);
-        void ReflectVertexInputs(const Vector<SharedPtr<MG_State::GLState::ShaderObject>>& shaders,
+        // `stages` is ALWAYS ProgramObject::GetLinkedShaderStages() - one entry per module of
+        // `spirv`, at the same index. Taking the stages rather than the shader objects is what
+        // keeps the program's live attach list, which is a longer and differently-indexed list
+        // the moment a glAttachShader lands after the link, from being passed here by mistake.
+        void ReflectVertexInputs(const Vector<ShaderStage>& stages,
                      const Vector<Vector<Uint>>& spirv,
                      VkProgramObject& entry) const;
-        void ReflectViewportIndexUsage(const Vector<SharedPtr<MG_State::GLState::ShaderObject>>& shaders,
+        void ReflectViewportIndexUsage(const Vector<ShaderStage>& stages,
                                        const Vector<Vector<Uint>>& spirv,
                                        VkProgramObject& entry) const;
-        void ReflectFragmentOutputs(const Vector<SharedPtr<MG_State::GLState::ShaderObject>>& shaders,
+        void ReflectFragmentOutputs(const Vector<ShaderStage>& stages,
                         const Vector<Vector<Uint>>& spirv,
                         VkProgramObject& entry) const;
         void ReflectLayout(const MG_State::GLState::ProgramObject& program, const Vector<Vector<Uint>>& spirv,
@@ -457,7 +605,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // Fills needsPassthroughTessControl / passthroughTessControlEmulatable off the linked
         // modules. Const and reflection-only: it decides nothing about the pipeline, it only
         // records what the evaluation stage's input interface is made of.
-        void ReflectPassthroughTessControlNeed(const Vector<SharedPtr<MG_State::GLState::ShaderObject>>& shaders,
+        void ReflectPassthroughTessControlNeed(const Vector<ShaderStage>& stages,
                                                const Vector<Vector<Uint>>& spirv,
                                                VkProgramObject& entry) const;
 
@@ -471,6 +619,19 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // True only when the logical device enabled both
         // shaderStorageImageReadWithoutFormat and shaderStorageImageWriteWithoutFormat.
         Bool m_unformattedFloatStorageImagesEnabled = false;
+        // True when the logical device enabled shaderTessellationAndGeometryPointSize. When it is
+        // FALSE a program whose tessellation or geometry module declares TessellationPointSize /
+        // GeometryPointSize is refused at build time (see VkProgramObject::
+        // pointSizeCapabilityUnsupported) instead of being handed to the driver as invalid usage.
+        Bool m_tessellationAndGeometryPointSizeEnabled = false;
+        // Startup snapshot used only by internally synthesized shader modules, which do not
+        // originate from a ProgramLinkTask.
+        Bool m_enableSpirvValidation = false;
+        // Device feature and limit gate resolved before vkCreateDevice. Keeping it in
+        // the factory lets each reflected layout choose ordinary descriptors when its
+        // own counts would exceed the update-after-bind budget.
+        UpdateAfterBindLimits m_updateAfterBindLimits{};
+        SubgroupLoweringPolicy m_subgroupPolicy{};
         // See SetDefaultFramebufferHeight. 0 means "not known yet"; the FragCoordYFlip bit is
         // never set before the swapchain exists, so no variant can be compiled against it.
         Uint32 m_defaultFramebufferHeight = 0;
@@ -480,11 +641,20 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // See GetCacheStructureEpoch(). Starts at 1 so a zero-initialized memo can never match.
         Uint64 m_cacheStructureEpoch = 1;
         IEvictionObserver* m_evictionObserver = nullptr;
-        // Pass-through tessellation control stages by input patch size. Never evicted: at most
-        // MAX_PATCH_VERTICES entries exist for the lifetime of the device, and every pipeline
-        // ever built from one keeps referencing its module. A failed build is cached as
+        // Pass-through tessellation control stages by the identity of what was compiled into
+        // them - the input patch size and the six default tessellation levels, folded into one
+        // 64-bit key by ComputePassthroughTessControlKey (the levels are float state, so the map
+        // cannot simply be keyed on the patch size any more). A failed build is cached as
         // VK_NULL_HANDLE so a broken generator costs one compile, not one per draw.
-        UnorderedMap<Uint32, VkPipelineShaderStageCreateInfo> m_passthroughTessControlStages;
+        //
+        // Hard-capped, because the key is application-controlled: glPatchParameterfv clamps
+        // nothing, so an application that recomputes a level per frame mints a new key per frame.
+        // Reaching the cap destroys every module and starts over (see the flush in
+        // GetOrCreatePassthroughTessControlStage); the cap is far above what any program that
+        // holds its levels still will ever need. The gl_PerVertex member set is in the key too
+        // and adds only a handful of values, so it does not move the cap in practice.
+        static constexpr SizeT kMaxPassthroughTessControlStages = 64;
+        UnorderedMap<Uint64, VkPipelineShaderStageCreateInfo> m_passthroughTessControlStages;
         static inline XXH64_state_t* m_hashState = XXH64_createState();
     };
 } // namespace MobileGL::MG_Backend::DirectVulkan

MobileGL/MG_Backend/DirectVulkan/Renderer/UniformManager.h

@@ -26,12 +26,26 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     public:
         struct SamplerBindingOverride {
             Uint32 binding = 0;
+            Uint32 element = 0;
             MG_State::GLState::ITextureObject* texture = nullptr;
             const MG_State::GLState::SamplerObject* sampler = nullptr;
             VkImageView imageView = VK_NULL_HANDLE;
+            VkImageLayout imageLayout = VK_IMAGE_LAYOUT_UNDEFINED;
+            Bool forceNearestFiltering = false;
         };
 
-        Bool Initialize(VkDevice device, VkBufferManager* bufferManager,
+        struct SamplerImageFeedbackBinding {
+            Uint32 samplerBinding = 0;
+            Uint32 samplerElement = 0;
+            MG_State::GLState::ITextureObject* texture = nullptr;
+            const MG_State::GLState::SamplerObject* sampler = nullptr;
+            SamplerNumericDomain numericDomain = SamplerNumericDomain::Unknown;
+        };
+
+        // `physicalDevice` is only ever asked for format properties: a placeholder descriptor for
+        // an unbound texel-buffer binding has to be built from a format the DEVICE accepts as a
+        // texel buffer, and there is no other route to that answer from here.
+        Bool Initialize(VkDevice device, VkPhysicalDevice physicalDevice, VkBufferManager* bufferManager,
                         ProgramFactory* programFactory,
                         VkDeviceSize minUniformBufferOffsetAlignment, Uint32 frameCount,
                         Uint32 maxBindings = 16, Uint32 setsPerFrame = 64,
@@ -79,6 +93,12 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         Bool CollectStorageImageTextures(const MG_State::GLState::ProgramObject& program,
                                          const ProgramFactory::VkProgramObject& programObj,
                                          Vector<MG_State::GLState::ITextureObject*>& outTextures) const;
+        Bool CollectSamplerImageFeedback(
+            const MG_State::GLState::ProgramObject& program,
+            const ProgramFactory::VkProgramObject& programObj,
+            Vector<SamplerImageFeedbackBinding>& outBindings) const;
+        static Bool SamplerOverlapsWritableImageSubresource(Int samplerBaseLevel, Int samplerMaxLevel,
+                                                             GLint imageLevel, GLenum imageAccess);
         // samplerDescriptorsUnchangedHint: the caller (SetupDraw fast path) proved that
         // every input of every combined-image-sampler resolution is unchanged since the
         // previous draw's resolve - same (texture, sampler) per binding, texture params
@@ -91,7 +111,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                                        Uint32 frameIndex,
                                        VkPipelineBindPoint bindPoint = VK_PIPELINE_BIND_POINT_GRAPHICS,
                                        const SamplerBindingOverride* samplerBindingOverride = nullptr,
-                                       Bool samplerDescriptorsUnchangedHint = false);
+                                       Bool samplerDescriptorsUnchangedHint = false,
+                                       const Vector<SamplerBindingOverride>* samplerBindingOverrides = nullptr);
 
         // Pure format-policy helper kept public for host regression tests. Formatted storage
         // images use their shader qualifier; transformed float images use glBindImageTexture's
@@ -114,6 +135,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             VkDescriptorPool handle = VK_NULL_HANDLE;
             Uint32 maxSets = 0;
             Uint32 allocatedSets = 0;
+            Bool updateAfterBind = false;
         };
 
         // A cached descriptor set together with the pool it was allocated from, so a
@@ -157,7 +179,43 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         static MG_State::GLState::ITextureObject* ResolveSamplerTextureRaw(
             const MG_State::GLState::ProgramObject& program,
             const ProgramFactory::VkProgramObject& programObj, Uint32 binding, Uint32 element);
-        SharedPtr<MG_State::GLState::ITextureObject> GetFallbackTexture(TextureTarget target) const;
+        // `numericDomain` is the sampler's class, and it matters only for the multisample arm -
+        // see GetFallbackMultisampleTexture for why the single-sampled fallback can ignore it.
+        SharedPtr<MG_State::GLState::ITextureObject> GetFallbackTexture(
+            TextureTarget target, SamplerNumericDomain numericDomain) const;
+        // The multisample arm of GetFallbackTexture. One object per (target, numeric domain) and
+        // no upload path: a multisample image cannot be written by a transfer, so its texels stay
+        // undefined - which is what GL promises for a texelFetch on an incomplete multisample
+        // texture - and it cannot carry MUTABLE_FORMAT, so its format has to match the sampler's
+        // class outright rather than being reinterpreted at view time.
+        SharedPtr<MG_State::GLState::ITextureObject> GetFallbackMultisampleTexture(
+            TextureTarget target, SamplerNumericDomain numericDomain) const;
+        // ---- placeholders for UNBOUND image-backed descriptors -------------------------
+        // GL lets a program declare `samplerBuffer`, `imageBuffer` or `image2D` and bind nothing
+        // to the unit it names: the fetch is then undefined (GL 4.6 core 8.9 for an incomplete
+        // buffer texture, 8.26 for an image unit with no texture) - undefined VALUES, not a
+        // dropped draw. Vulkan has no unwritten descriptor, so something valid has to sit in the
+        // set or the whole draw or dispatch is lost, which is what these two build. Same shape as
+        // VkBufferManager::AcquireUnboundStorageDescriptor, one level up: per FORMAT rather than
+        // one shared object, because a descriptor whose format disagrees with the shader's
+        // declaration is invalid Vulkan even when nothing ever reads it.
+        //
+        // `declaredFormat` is the format the SHADER declared (VK_FORMAT_UNDEFINED for a sampled
+        // texel buffer, which never carries one, or for a formatless `writeonly` image);
+        // `numericDomain` decides the format when there is no declaration and is the fallback
+        // class when the device cannot use the declared one as a texel buffer.
+        VkBufferView AcquireUnboundTexelBufferView(VkFormat declaredFormat, SamplerNumericDomain numericDomain,
+                                                   Bool storage);
+        // A 1x1 (x1 layer, or 6 faces for a cube) texture of `format`, shaped for `target` so the
+        // view the descriptor gets has the view type the shader's image declaration demands.
+        // Null for a target with no single-sampled placeholder shape - multisample images, whose
+        // descriptor needs a multisample view that this cannot stand in for.
+        SharedPtr<MG_State::GLState::ITextureObject> GetUnboundStorageImageTexture(TextureTarget target,
+                                                                                    VkFormat format) const;
+        // The (target, format) pair a storage-image binding's placeholder is keyed by, resolved
+        // from reflection alone. False when the binding has no placeholder shape.
+        Bool ResolveUnboundStorageImagePlaceholder(const ProgramFactory::VkProgramObject& programObj, Uint32 binding,
+                                                   TextureTarget& outTarget, VkFormat& outFormat) const;
         // `element` indexes a sampler ARRAY inside one binding; each element carries its own
         // independently assigned GL texture unit, so it selects the texture, the sampler
         // override and the fallback separately from its neighbours.
@@ -223,8 +281,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         void BindDescriptorSetDeduped(VkCommandBuffer commandBuffer, VkPipelineBindPoint bindPoint,
                                       VkPipelineLayout pipelineLayout, VkDescriptorSet descriptorSet,
                                       const Vector<Uint32>& dynamicOffsets);
-        Bool CreateDescriptorPool(Uint32 maxSets, VkDescriptorPool& outPool) const;
-        Bool GrowFrameDescriptorPool(FrameResources& frame, Uint32 frameIndex);
+        Bool CreateDescriptorPool(Uint32 maxSets, Bool updateAfterBind, VkDescriptorPool& outPool) const;
+        Bool GrowFrameDescriptorPool(FrameResources& frame, Uint32 frameIndex, Bool updateAfterBind);
         VkResult AllocateDescriptorSetsFromActivePool(
             Uint32 frameIndex, const ProgramFactory::VkProgramObject& programObj, VkDescriptorSet& outDescriptorSet);
         VkResult AcquireDescriptorSet(Uint32 frameIndex,
@@ -232,6 +290,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                                       VkDescriptorSet& outDescriptorSet);
 
         VkDevice m_device = VK_NULL_HANDLE;
+        VkPhysicalDevice m_physicalDevice = VK_NULL_HANDLE;
         VkBufferManager* m_bufferManager = nullptr;
         ProgramFactory* m_programFactory = nullptr;
         Vector<FrameResources> m_frames;
@@ -244,6 +303,18 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         VkTextureManager* m_textureManager = nullptr;
         VkSamplerManager* m_samplerManager = nullptr;
         mutable SharedPtr<MG_State::GLState::ITextureObject> m_fallbackTexture2D;
+        // Keyed by (arrayed, numeric domain); see GetFallbackMultisampleTexture. Lazily populated,
+        // never evicted - at most six tiny 1x1 images - and torn down with the manager.
+        mutable UnorderedMap<Uint32, SharedPtr<MG_State::GLState::ITextureObject>> m_fallbackMultisampleTextures;
+        // See AcquireUnboundTexelBufferView / GetUnboundStorageImageTexture. Both are lazily
+        // populated, never evicted (a program's declared formats are a fixed, tiny set) and torn
+        // down with the manager. The texel views are keyed by format AND by storage-vs-sampled
+        // because the two descriptor kinds demand different format FEATURES of the device, so one
+        // format can be usable for one and not the other. Deliberately NOT the per-frame
+        // texelBufferViews list: those are destroyed at every frame boundary, and these must
+        // outlive it or the placeholder would be rebuilt for every unbound binding every frame.
+        UnorderedMap<Uint64, VkBufferView> m_unboundTexelBufferViews;
+        mutable UnorderedMap<Uint64, SharedPtr<MG_State::GLState::ITextureObject>> m_unboundStorageImageTextures;
 
         // Per-draw scratch buffers for BindProgramUniformBuffers: reused (clear keeps
         // capacity) so the descriptor-write path stops allocating on every draw.
@@ -341,8 +412,11 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // lifetime id, so a freed-and-reallocated sampler or texture at the same heap address
         // always gets a fresh id and misses (a raw pointer would false-hit that ABA) - so a
         // stale guess can only miss and fall through to the hash, never resolve wrong. Still
-        // reset each frame alongside the descriptor-set cache. Indexed by binding.
+        // reset each frame alongside the descriptor-set cache. Indexed by binding, but the
+        // whole-descriptor entry is additionally keyed by program lifetime: Vulkan binding
+        // numbers are layout-local and unrelated programs routinely reuse binding 0/1.
         struct SamplerResolveMemo {
+            Uint64 infoProgramLifetimeId = 0;
             Uint64 samplerLifetimeId = 0;
             Uint64 textureLifetimeId = 0;
             VkSampler sampler = VK_NULL_HANDLE;

MobileGL/MG_Backend/DirectVulkan/Renderer/VertexInputStateFactory.cpp

@@ -8,6 +8,7 @@
 
 #include "VertexInputStateFactory.h"
 #include "MG_Util/Converters/MGToStr/DataTypeConverter.h"
+#include <MG_Backend/BackendObjects.h>
 #include <utility>
 
 namespace MobileGL::MG_Backend::DirectVulkan {
@@ -107,8 +108,34 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 continue;
             }
 
-            const VkFormat sourceVkFormat =
+            VkFormat sourceVkFormat =
                 ToVkVertexFormat(attr.Type, attr.Size, attr.Normalized, attr.IsInteger, attr.IsBgra, attr.IsLong);
+            VertexStreamConversion conversion = VertexStreamConversion::None;
+            // Gated on the SAME flag ToVkVertexFormat gates its 64-bit path on, and that is
+            // load-bearing rather than belt-and-braces: the narrowing is only correct because the
+            // shader's `dvec` input is a `vec` by the time the pipeline is built, and what
+            // guarantees that is the flag being clear. It is clear on every backend today, and a
+            // program with a 64-bit float vertex input is demoted WHOLE for the same reason even
+            // where the device has native fp64 (ProgramSpirvTask::GenerateSpirv). With the flag
+            // set, a dvec3/dvec4 would be declined by ToVkVertexFormat AND left 64-bit in the
+            // module, so a float32 stream would be fed to a Float64 input.
+            const Bool narrowFloat64Arrays =
+                MG_Backend::pActiveBackendObject == nullptr ||
+                !MG_Backend::pActiveBackendObject->GetDynamicParameters().SupportsFloat64VertexAttributes;
+            if (sourceVkFormat == VK_FORMAT_UNDEFINED && attr.Type == DataType::Float64 && narrowFloat64Arrays) {
+                // No native 64-bit fetch here (see ToVkVertexFormat's Float64 case), but the
+                // source bytes are ordinary IEEE-754 doubles and DemoteFloat64Pass has already
+                // narrowed every dvec input to a vec, so the array is narrowed to match rather
+                // than dropped. Mirrors what DirectGLES does for the same state.
+                const VkFormat narrowedFormat = ToFloat32VertexFormat(attr.Size);
+                if (narrowedFormat != VK_FORMAT_UNDEFINED && SupportsVertexBufferFormat(narrowedFormat)) {
+                    sourceVkFormat = narrowedFormat;
+                    conversion = VertexStreamConversion::Float64ToFloat32;
+                    MGLOG_W_ONCE("Vertex attribute location=%u is a 64-bit (GL_DOUBLE) array; fetching it at "
+                            "float32 precision through format=%d (size=%d long=%s)",
+                            location, static_cast<Int>(narrowedFormat), attr.Size, attr.IsLong ? "true" : "false");
+                }
+            }
             if (sourceVkFormat == VK_FORMAT_UNDEFINED) {
                 MGLOG_E_ONCE("Unsupported vertex attribute layout (location=%u, type=%s, size=%d): the array is "
                         "enabled but cannot be mapped to a VkFormat",
@@ -118,8 +145,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             }
 
             VkFormat vkFormat = sourceVkFormat;
-            VertexStreamConversion conversion = VertexStreamConversion::None;
-            if (!SupportsVertexBufferFormat(vkFormat)) {
+            if (conversion == VertexStreamConversion::None && !SupportsVertexBufferFormat(vkFormat)) {
                 if (IsScaledIntegerVertexFormat(vkFormat)) {
                     const VkFormat fallbackFormat = ToFloat32VertexFormat(attr.Size);
                     if (fallbackFormat != VK_FORMAT_UNDEFINED && SupportsVertexBufferFormat(fallbackFormat)) {
@@ -188,7 +214,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             if (sourceStride != 0) {
                 if (conversion == VertexStreamConversion::Repack) {
                     stride = static_cast<Uint32>(attribByteSize);
-                } else if (conversion == VertexStreamConversion::ScaledIntegerToFloat32) {
+                } else if (conversion == VertexStreamConversion::ScaledIntegerToFloat32 ||
+                           conversion == VertexStreamConversion::Float64ToFloat32) {
                     stride = static_cast<Uint32>(attr.Size * static_cast<Int>(sizeof(Float)));
                 }
             }
@@ -287,8 +314,10 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             if (m_frameBoundaryCounter - it->second->lastUsedFrameBoundary > kRetireAgeBoundaries) {
                 it = m_cache.erase(it);
                 // Invalidate every VAO's state-pointer memo: the erased node's
-                // address may be reused by a future insert.
-                ++m_evictionEpoch;
+                // address may be reused by a future insert. Advance through the
+                // process-wide source so the value stays unique across factory
+                // instances (see the member comment).
+                m_evictionEpoch = ++s_evictionEpochSource;
             } else {
                 ++it;
             }
@@ -328,6 +357,20 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             // for every R64 float format, so a native 64-bit vertex fetch is simply unavailable there
             // while shaderFloat64 is not. Both halves key off nothing but the attribute being long,
             // so they always agree without extra plumbing.
+            //
+            // ... as long as the shader half still runs. It does not when the backend has declared
+            // no 64-bit vertex attribute support: DemoteFloat64Pass has already narrowed every
+            // `dvec` input to a `vec` by then, so PackDoubleVertexInputsPass finds nothing to pack
+            // and a UINT-formatted attribute would be fed to a float input - garbage with no
+            // diagnostic anywhere. Declining here hands the attribute to the caller's
+            // Float64ToFloat32 fallback instead, which narrows the source doubles to match the
+            // demoted `vec` input - the same thing DirectGLES does for the same state. The
+            // frontend RECORDS the format either way, so this gate is the only thing standing
+            // between a legal glVertexAttribLFormat and a mismatched pipeline.
+            if (MG_Backend::pActiveBackendObject == nullptr ||
+                !MG_Backend::pActiveBackendObject->GetDynamicParameters().SupportsFloat64VertexAttributes) {
+                return VK_FORMAT_UNDEFINED;
+            }
             if (!isLong || isInteger || normalized) return VK_FORMAT_UNDEFINED;
             switch (size) {
             case 1: return VK_FORMAT_R32G32_UINT;

MobileGL/MG_Backend/DirectVulkan/Renderer/VertexInputStateFactory.h

@@ -23,6 +23,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             None = 0,
             Repack,
             ScaledIntegerToFloat32,
+            // GL_DOUBLE source data narrowed to a tightly packed float32 stream: the fetch half
+            // of the fp64 demotion the shader side already does unconditionally.
+            Float64ToFloat32,
         };
 
         struct BackendVertexInputState {
@@ -125,7 +128,17 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // construction); a memo is honored only while its recorded epoch
         // matches, so an evicted entry can never be dereferenced through a
         // stale memo.
-        Uint64 m_evictionEpoch = 1;
+        //
+        // Drawn from a process-wide source, never a per-instance counter: the VAO
+        // memos outlive this factory (they live on the frontend context's VAOs, the renderer
+        // is destroyed and recreated on EGL surface release/re-create), so a fresh
+        // factory restarting at a dead factory's epoch value would honor its
+        // dangling entry pointers. The constructor takes a value strictly greater
+        // than anything a predecessor ever stamped, so a dead factory's memo can
+        // never compare equal here - the same never-reused idiom as the lifetime ids.
+        // Single-threaded like the rest of the factory (renderer-thread only).
+        static inline Uint64 s_evictionEpochSource = 0;
+        Uint64 m_evictionEpoch = ++s_evictionEpochSource;
         static inline XXH64_state_t* m_hashState = XXH64_createState();
     };
 } // namespace MobileGL::MG_Backend::DirectVulkan

MobileGL/MG_Backend/DirectVulkan/Renderer/VkBufferManager.cpp

@@ -10,12 +10,23 @@
 #include "../DirectVulkan.h"
 #include "VulkanRenderer.h"
 
+#include "MG_Util/Metrics/PipeStats.h"
+
 namespace MobileGL::MG_Backend::DirectVulkan {
     namespace {
         constexpr VmaAllocationCreateFlags kResidentBufferAllocationFlags =
             VMA_ALLOCATION_CREATE_HOST_ACCESS_SEQUENTIAL_WRITE_BIT;
         constexpr SizeT kLiveResourcePruneThreshold = 256;
 
+        // See VkBufferManager::AcquireUnboundStorageDescriptor. 256 bytes: comfortably past
+        // every minStorageBufferOffsetAlignment in the wild, and free.
+        constexpr VkDeviceSize kUnboundStorageDescriptorBytes = 256;
+        // See VkBufferManager::AcquireUnboundTexelBufferDescriptor. The same 256 bytes, for the
+        // same reason plus one: a texel buffer view's range must be a whole number of texels of
+        // whatever format the placeholder is asked for, and 256 divides by every texel size in
+        // the GL image-format table (1, 2, 4, 8 and 16 bytes).
+        constexpr VkDeviceSize kUnboundTexelBufferDescriptorBytes = 256;
+
         // A zero-copy persistent buffer is created once and never recreated (the app holds
         // its mapped pointer), and may be bound to any role, so it carries every usage.
         // TRANSFER_DST is added by CreateResidentStorage.
@@ -130,6 +141,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             }
         }
         m_transientUploadArena.Shutdown();
+        m_unboundStorageBuffer.Destroy();
+        m_unboundTexelBuffer.Destroy();
         DestroyAllDeferredReleases();
         ReleaseAllLiveResources();
         m_copyProvider = nullptr;
@@ -218,8 +231,38 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
     Bool VkBufferManager::UploadTransient(BufferKind kind, Uint32 frameIndex, const void* data,
                                           VkDeviceSize size, VkDeviceSize alignment, BufferSlice& outSlice) {
-        (void)kind;
-        return m_transientUploadArena.Upload(frameIndex, data, size, alignment, outSlice);
+        if (!m_transientUploadArena.Upload(frameIndex, data, size, alignment, outSlice)) {
+            return false;
+        }
+        if (MG_Util::PipeStats::Enabled()) {
+            // The single chokepoint for Magma's per-draw staging. Uniform is deliberately
+            // absent: its bytes are counted by the caller, which is the only place that
+            // knows whether the payload is the default block (stage-ubo-global) or a named
+            // one repacked into the ring (stage-ubo-named), and counting here as well would
+            // double every uniform byte.
+            switch (kind) {
+            case BufferKind::Vertex:
+                MG_Util::PipeStats::AddBytes(MG_Util::PipeStats::ByteClass::StageVertexClient,
+                                             static_cast<Uint64>(size));
+                break;
+            case BufferKind::Index:
+                MG_Util::PipeStats::AddBytes(MG_Util::PipeStats::ByteClass::StageIndexClient,
+                                             static_cast<Uint64>(size));
+                break;
+            case BufferKind::Indirect:
+                MG_Util::PipeStats::AddBytes(MG_Util::PipeStats::ByteClass::StageIndirectCmd,
+                                             static_cast<Uint64>(size));
+                break;
+            case BufferKind::TextureBuffer:
+            case BufferKind::ShaderStorage:
+                MG_Util::PipeStats::AddBytes(MG_Util::PipeStats::ByteClass::StageBuffer,
+                                             static_cast<Uint64>(size));
+                break;
+            case BufferKind::Uniform:
+                break;
+            }
+        }
+        return true;
     }
 
     Bool VkBufferManager::InitializeTransientArenas() {
@@ -328,6 +371,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             resource.pendingFullUpload = true;
             return false;
         }
+        if (MG_Util::PipeStats::Enabled()) {
+            MG_Util::PipeStats::AddBytes(MG_Util::PipeStats::ByteClass::StageBuffer, static_cast<Uint64>(size));
+        }
         resource.pendingFullUpload = false;
         return true;
     }
@@ -342,6 +388,11 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                                            static_cast<VkDeviceSize>(size), 16, staging)) {
             return false;
         }
+        if (MG_Util::PipeStats::Enabled()) {
+            // The staging fill is the host copy; the vkCmdCopyBuffer below is the device
+            // half of the same bytes and is not counted twice.
+            MG_Util::PipeStats::AddBytes(MG_Util::PipeStats::ByteClass::StageBuffer, static_cast<Uint64>(size));
+        }
         VkCommandBuffer commandBuffer = m_copyProvider->AcquireBufferCopyCommandBuffer();
         if (commandBuffer == VK_NULL_HANDLE) {
             return false;
@@ -411,6 +462,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         if (!resource->buffer.Upload(bufferObject.MappedData(), size, 0)) {
             MGLOG_E_ONCE("VkBufferManager::OnRespecify: in-place upload failed");
             resource->pendingFullUpload = true;
+        } else if (MG_Util::PipeStats::Enabled()) {
+            MG_Util::PipeStats::AddBytes(MG_Util::PipeStats::ByteClass::StageBuffer, static_cast<Uint64>(size));
         }
     }
 
@@ -436,6 +489,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                                          static_cast<VkDeviceSize>(size), static_cast<VkDeviceSize>(offset))) {
                 MGLOG_E_ONCE("VkBufferManager::OnSubData: host upload failed");
                 resource->pendingFullUpload = true;
+            } else if (MG_Util::PipeStats::Enabled()) {
+                MG_Util::PipeStats::AddBytes(MG_Util::PipeStats::ByteClass::StageBuffer,
+                                             static_cast<Uint64>(size));
             }
             return;
         }
@@ -473,6 +529,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                                          static_cast<VkDeviceSize>(size), static_cast<VkDeviceSize>(offset))) {
                 MGLOG_E_ONCE("VkBufferManager::OnFlushMappedRange: host upload failed");
                 resource->pendingFullUpload = true;
+            } else if (MG_Util::PipeStats::Enabled()) {
+                MG_Util::PipeStats::AddBytes(MG_Util::PipeStats::ByteClass::StageBuffer,
+                                             static_cast<Uint64>(size));
             }
             return;
         }
@@ -543,6 +602,13 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         const Uint8* seed = bufferObject.MappedData();
         if (seed != nullptr) {
             resource->buffer.Upload(seed, size, 0);
+            if (MG_Util::PipeStats::Enabled()) {
+                // The one-time seed of a persistent map. Everything the app writes AFTER
+                // this goes straight through the mapping and is persistent-map-push
+                // territory (unwired, D4/D-B4), not this class.
+                MG_Util::PipeStats::AddBytes(MG_Util::PipeStats::ByteClass::StageBuffer,
+                                             static_cast<Uint64>(size));
+            }
         }
         resource->persistentMapped = true;
         resource->pendingFullUpload = false;
@@ -591,6 +657,10 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 resource->usageFlags = 0;
                 return false;
             }
+            if (MG_Util::PipeStats::Enabled()) {
+                MG_Util::PipeStats::AddBytes(MG_Util::PipeStats::ByteClass::StageBuffer,
+                                             static_cast<Uint64>(size));
+            }
             resource->pendingFullUpload = false;
         }
 
@@ -670,6 +740,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                                            outSlice)) {
             return false;
         }
+        if (MG_Util::PipeStats::Enabled()) {
+            MG_Util::PipeStats::AddBytes(MG_Util::PipeStats::ByteClass::StageBuffer, static_cast<Uint64>(size));
+        }
         resource->transientSlice = outSlice;
         resource->transientFrameSerial = m_frameSerial;
         resource->transientChangeSerial = changeSerial;
@@ -706,6 +779,70 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         m_deferredResourceReleases[frameIndex].clear();
     }
 
+    BufferSlice VkBufferManager::AcquireUnboundStorageDescriptor() {
+        if (!m_unboundStorageBuffer.IsValid()) {
+            if (m_initInfo.allocator == nullptr) {
+                return {};
+            }
+            // Host-visible so the zero fill needs no command buffer: this can be reached from
+            // descriptor resolution, which runs inside an already-open recording and must not
+            // start a copy of its own. The size is a whole minStorageBufferOffsetAlignment-safe
+            // block rather than 4 bytes so that a shader which does read the block gets a
+            // plausible unsized-array length instead of one that rounds to zero.
+            const Bool created = m_unboundStorageBuffer.Create({
+                .allocator = m_initInfo.allocator,
+                .size = kUnboundStorageDescriptorBytes,
+                .usage = VK_BUFFER_USAGE_STORAGE_BUFFER_BIT | VK_BUFFER_USAGE_TRANSFER_DST_BIT,
+                .memoryUsage = VMA_MEMORY_USAGE_AUTO,
+                .allocationFlags = VMA_ALLOCATION_CREATE_HOST_ACCESS_SEQUENTIAL_WRITE_BIT |
+                                   VMA_ALLOCATION_CREATE_MAPPED_BIT,
+                .requiredFlags = VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT | VK_MEMORY_PROPERTY_HOST_COHERENT_BIT,
+            });
+            if (!created) {
+                MGLOG_E_ONCE("VkBufferManager::AcquireUnboundStorageDescriptor: placeholder creation failed");
+                m_unboundStorageBuffer.Destroy();
+                return {};
+            }
+            if (void* mapped = m_unboundStorageBuffer.GetMappedData()) {
+                Memset(mapped, 0, static_cast<SizeT>(kUnboundStorageDescriptorBytes));
+            }
+        }
+        return m_unboundStorageBuffer.GetSlice();
+    }
+
+    BufferSlice VkBufferManager::AcquireUnboundTexelBufferDescriptor() {
+        if (!m_unboundTexelBuffer.IsValid()) {
+            if (m_initInfo.allocator == nullptr) {
+                return {};
+            }
+            // A SECOND placeholder rather than more usage bits on the storage-block one. The two
+            // are independent failure domains: a device that refuses this allocation must not
+            // take the storage-block placeholder - and with it the fix this one is a sibling of -
+            // down with it. Host-visible and zero-filled for the same reason as that one: this is
+            // reached from descriptor resolution, inside an already-open recording, which must
+            // not start a copy of its own.
+            const Bool created = m_unboundTexelBuffer.Create({
+                .allocator = m_initInfo.allocator,
+                .size = kUnboundTexelBufferDescriptorBytes,
+                .usage = VK_BUFFER_USAGE_UNIFORM_TEXEL_BUFFER_BIT | VK_BUFFER_USAGE_STORAGE_TEXEL_BUFFER_BIT |
+                         VK_BUFFER_USAGE_TRANSFER_DST_BIT,
+                .memoryUsage = VMA_MEMORY_USAGE_AUTO,
+                .allocationFlags = VMA_ALLOCATION_CREATE_HOST_ACCESS_SEQUENTIAL_WRITE_BIT |
+                                   VMA_ALLOCATION_CREATE_MAPPED_BIT,
+                .requiredFlags = VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT | VK_MEMORY_PROPERTY_HOST_COHERENT_BIT,
+            });
+            if (!created) {
+                MGLOG_E_ONCE("VkBufferManager::AcquireUnboundTexelBufferDescriptor: placeholder creation failed");
+                m_unboundTexelBuffer.Destroy();
+                return {};
+            }
+            if (void* mapped = m_unboundTexelBuffer.GetMappedData()) {
+                Memset(mapped, 0, static_cast<SizeT>(kUnboundTexelBufferDescriptorBytes));
+            }
+        }
+        return m_unboundTexelBuffer.GetSlice();
+    }
+
     VkBufferUsageFlags VkBufferManager::GetVkBufferUsage(BufferKind kind) {
         switch (kind) {
         case BufferKind::Vertex:

MobileGL/MG_Backend/DirectVulkan/Renderer/VkBufferManager.h

@@ -120,6 +120,27 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         Bool UploadTransient(BufferKind kind, Uint32 frameIndex, const void* data, VkDeviceSize size,
                              VkDeviceSize alignment, BufferSlice& outSlice);
 
+        // The descriptor a shader storage block gets when the program declares it and the
+        // application bound no buffer at its GL binding point. GL 4.6 core 7.8 makes that a
+        // legal state - the block simply has no store, so reads are undefined and writes go
+        // nowhere - whereas Vulkan has no such thing as an unwritten descriptor, so something
+        // real has to sit in the set or the whole draw/dispatch is lost. One zero-filled
+        // buffer, created once and shared by every unbound binding: bindings that are only
+        // declared (the case this exists for) never touch it, and one that is actually read
+        // sees zeros, which is inside GL's "undefined". robustBufferAccess bounds anything
+        // that indexes past it.
+        BufferSlice AcquireUnboundStorageDescriptor();
+
+        // The store a texel-buffer descriptor - `samplerBuffer` or `imageBuffer` - gets when the
+        // unit the program's uniform names has no buffer texture on it, or the buffer texture on
+        // it has no GL buffer attached. Both are legal GL states that make a fetch return
+        // undefined values (GL 4.6 core 8.9: a buffer texture with no attached buffer object is
+        // incomplete, and sampling an incomplete texture is undefined - not a lost draw), and both
+        // used to take the whole draw or dispatch with them. The VIEW over this - one per format,
+        // and the descriptor is a VkBufferView, not a buffer - is built by
+        // UniformManager::AcquireUnboundTexelBufferView.
+        BufferSlice AcquireUnboundTexelBufferDescriptor();
+
         // Draw-time acquire for resident (device-storage) buffers: ensures the
         // resource exists and is fully uploaded, marks it used this frame.
         Bool AcquireResidentSlice(BufferKind kind, const SharedPtr<MG_State::GLState::BufferObject>& bufferObject,
@@ -180,6 +201,11 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
         VkBufferManagerInitInfo m_initInfo{};
         BufferArena m_transientUploadArena;
+        // See AcquireUnboundStorageDescriptor. Lazily created, never re-created, torn down
+        // with the manager.
+        VkBufferObject m_unboundStorageBuffer;
+        // See AcquireUnboundTexelBufferDescriptor. Same lifetime rules.
+        VkBufferObject m_unboundTexelBuffer;
         IBufferCopyCommandProvider* m_copyProvider = nullptr;
         Vector<Vector<VkBufferObject>> m_deferredBufferReleases;
         Vector<Vector<SharedPtr<VkBufferResource>>> m_deferredResourceReleases;

MobileGL/MG_Backend/DirectVulkan/Renderer/VkClearManager.cpp

@@ -8,7 +8,12 @@
 
 #include "VkClearManager.h"
 
+// For the shared ResolveAttachmentLayerCount (and the ToVulkanLevelExtent it is built on): the
+// clear key's layer span has to be the same one the render pass builds its attachment view from.
+#include "VkTextureManager.h"
+
 #include "MG_State/GLState/Core.h"
+#include <MG_Pipe/PipeInputsSwitch.h>
 #include "MG_Util/Converters/MGToStr/FramebufferEnumConverter.h"
 #include "MG_Util/Converters/MGToStr/TextureEnumConverter.h"
 
@@ -50,7 +55,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         if (payload.colorEncoding != ClearColorEncoding::Float) return;
         // With GL_FRAMEBUFFER_SRGB enabled GL performs the encoding itself, so the driver doing it
         // is exactly right and there is nothing to undo.
-        if (MG_State::pGLContext->IsCapabilityEnabled(MobileGL::CapabilityInput::FramebufferSrgb)) return;
+        if (MGB_CTX->IsCapabilityEnabled(MobileGL::CapabilityInput::FramebufferSrgb)) return;
         if (ResolveSrgbAttachmentWriteFormat(destinationFormat, false) == destinationFormat) return;
 
         // sRGB -> linear (GL 4.6 core 8.24), applied to the colour channels only: alpha is stored
@@ -100,13 +105,13 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         return ResolveAttachmentBaseArrayLayer(uploadTarget);
     }
 
-    static Uint32 ResolveAttachmentLayerCount(
-        const MG_State::GLState::FramebufferAttachmentObject& attachment) {
-        if (attachment.IsLayered()) {
-            return static_cast<Uint32>(std::max(attachment.GetSize().z(), 1));
-        }
-        return 1u;
-    }
+    // ResolveAttachmentLayerCount used to be duplicated here, reading attachment.GetSize().z()
+    // raw - no ToVulkanLevelExtent remap for a 1D array, no six-faces arm for a cube map. That is
+    // not a cosmetic difference: the count below is not key-only, it is written straight into
+    // VkImageSubresourceRange::layerCount by MaterializePendingClearForTexture, which then POPS
+    // the entry - so a layered cube map's glClear reached one face and the other five were lost
+    // for good, while the very same queued clear cleared all six through the render pass's
+    // LOAD_OP_CLEAR. The helper now lives once, in VkTextureManager.h beside ToVulkanLevelExtent.
 
     static const MG_State::GLState::FramebufferAttachmentObject* GetClearableAttachment(
         const MG_State::GLState::FramebufferObject& drawFbo, FramebufferAttachmentType attachmentType) {
@@ -122,8 +127,30 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         return &attachment;
     }
 
-    PendingClearKey VkClearManager::MakePendingClearKey(MG_State::GLState::ITextureObject* texture, Uint32 mipLevel,
+    // The texture a pending clear is actually ABOUT. A clear issued through a GL texture view
+    // (ARB_texture_view) targets the storage it views, so it must queue against - and be found
+    // by - the storage texture; keying it on the view instead left the clear invisible to every
+    // materialisation done through the parent's name (and vice versa), so the image stayed in
+    // VK_IMAGE_LAYOUT_UNDEFINED and the readback was dropped as unreadable.
+    static MG_State::GLState::ITextureObject* ClearStorageTextureOf(MG_State::GLState::ITextureObject* texture) {
+        if (texture == nullptr) {
+            return nullptr;
+        }
+        const auto& storageOwner = texture->GetViewStorageOwner();
+        return storageOwner ? storageOwner.get() : texture;
+    }
+
+    PendingClearKey VkClearManager::MakePendingClearKey(MG_State::GLState::ITextureObject* rawTexture, Uint32 mipLevel,
                                                         Uint32 baseArrayLayer, Uint32 layerCount) {
+        MG_State::GLState::ITextureObject* texture = ClearStorageTextureOf(rawTexture);
+        if (rawTexture != nullptr && texture != rawTexture) {
+            // The caller named a level and a layer of the VIEW; the key describes the STORAGE, so
+            // both have to be shifted into its numbering (GL 4.6 core 8.18). Without this a clear
+            // of a view's level 0 would collide with a clear of the storage's level 0 even when
+            // the view opened onto level 1.
+            mipLevel += static_cast<Uint32>(rawTexture->GetViewMinLevel());
+            baseArrayLayer += static_cast<Uint32>(rawTexture->GetViewMinLayer());
+        }
         return PendingClearKey {
             .texture = texture,
             .textureLifetimeId = texture ? texture->GetLifetimeId() : 0,
@@ -157,6 +184,15 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     }
 
     TextureIdentity VkClearManager::MakeTextureIdentity(MG_State::GLState::ITextureObject* texture) {
+        // Same rule as VkTextureManager::MakeTextureIdentity: a GL texture view is identified by
+        // the storage it views. A clear posted against a view and one posted against its parent
+        // target the same image, so they have to coalesce rather than queue independently.
+        if (texture != nullptr) {
+            const auto& storageOwner = texture->GetViewStorageOwner();
+            if (storageOwner) {
+                texture = storageOwner.get();
+            }
+        }
         return TextureIdentity {
             .texture = texture,
             .lifetimeId = texture ? texture->GetLifetimeId() : 0,
@@ -166,7 +202,15 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     void VkClearManager::MergeClearPayload(ClearAttachmentPayload& dst, const ClearAttachmentPayload& src) {
         dst.mask |= src.mask;
         if ((src.mask & GL_COLOR_BUFFER_BIT) != 0) {
+            // The whole colour story travels together (same rule as
+            // VkRenderPassManager::QueueRenderbufferClear): a glClearBufferiv/uiv
+            // payload carries its value in colorInt/colorUint and its branch selector
+            // in colorEncoding - dropping them here would leave the pending clear
+            // reading as an all-zero float one.
             dst.color = src.color;
+            dst.colorEncoding = src.colorEncoding;
+            dst.colorInt = src.colorInt;
+            dst.colorUint = src.colorUint;
         }
         if ((src.mask & GL_DEPTH_BUFFER_BIT) != 0) {
             dst.depth = src.depth;
@@ -278,9 +322,11 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             return;
         }
 
-        const PendingClearKey key = MakePendingClearKey(texture.get());
+        const auto& storageOwner = texture->GetViewStorageOwner();
+        const SharedPtr<MG_State::GLState::ITextureObject>& storageTexture = storageOwner ? storageOwner : texture;
+        const PendingClearKey key = MakePendingClearKey(storageTexture.get());
         const std::lock_guard<std::mutex> lock(m_mutex);
-        m_aliveObjects[MakeTextureIdentity(texture.get())] = texture;
+        m_aliveObjects[MakeTextureIdentity(storageTexture.get())] = storageTexture;
         auto& pending = m_pendingClears[key];
         MergeClearPayload(pending, clearPayload);
         m_pendingCount.store(static_cast<Uint32>(m_pendingClears.size()), std::memory_order_relaxed);
@@ -297,8 +343,14 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         }
 
         const PendingClearKey key = MakePendingClearKey(attachment);
+        // The alive entry must hold the STORAGE object, because the key names it:
+        // LockTextureIdentityLocked cross-checks the two, and registering a view here under its
+        // storage's identity made every lookup of this clear fail that check and silently report
+        // "nothing pending" - which is how a clear issued through a view's framebuffer vanished.
+        const auto& storageOwner = texture->GetViewStorageOwner();
+        const SharedPtr<MG_State::GLState::ITextureObject>& storageTexture = storageOwner ? storageOwner : texture;
         const std::lock_guard<std::mutex> lock(m_mutex);
-        m_aliveObjects[MakeTextureIdentity(texture.get())] = texture;
+        m_aliveObjects[MakeTextureIdentity(storageTexture.get())] = storageTexture;
         auto& pending = m_pendingClears[key];
         MergeClearPayload(pending, clearPayload);
         m_pendingCount.store(static_cast<Uint32>(m_pendingClears.size()), std::memory_order_relaxed);
@@ -313,6 +365,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             return false;  // per-draw hot path: nothing pending anywhere
         }
 
+        texture = ClearStorageTextureOf(texture);
         const Uint64 lifetimeId = texture->GetLifetimeId();
         const std::lock_guard<std::mutex> lock(m_mutex);
         for (auto it = m_pendingClears.begin(); it != m_pendingClears.end(); ++it) {
@@ -403,6 +456,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             return false;  // per-draw hot path: nothing pending anywhere
         }
 
+        texture = ClearStorageTextureOf(texture);
         const Uint64 lifetimeId = texture->GetLifetimeId();
         const std::lock_guard<std::mutex> lock(m_mutex);
         SharedPtr<MG_State::GLState::ITextureObject> liveTexture;

MobileGL/MG_Backend/DirectVulkan/Renderer/VkClearManager.h

@@ -114,7 +114,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     class VkClearManager {
     public:
         static PendingClearKey MakePendingClearKey(const MG_State::GLState::FramebufferAttachmentObject& attachment);
-        static PendingClearKey MakePendingClearKey(MG_State::GLState::ITextureObject* texture, Uint32 mipLevel = 0,
+        // Resolves a GL texture view to the storage it views before keying; see the definition.
+        static PendingClearKey MakePendingClearKey(MG_State::GLState::ITextureObject* rawTexture, Uint32 mipLevel = 0,
                                                    Uint32 baseArrayLayer = 0, Uint32 layerCount = 1);
 
         Bool Initialize();

MobileGL/MG_Backend/DirectVulkan/Renderer/VkRenderPassManager.cpp

@@ -13,6 +13,7 @@
 #include "MG_Util/Converters/MGToStr/FramebufferEnumConverter.h"
 #include "MG_Util/Converters/MGToVk/TextureEnumConverter.h"
 #include "MG_Util/Metrics/TextureMetrics.h"
+#include <MG_Pipe/PipeInputsSwitch.h>
 
 namespace MobileGL::MG_Backend::DirectVulkan {
     static Bool TryResolveSampleCountFlagBits(Int requestedSamples, VkSampleCountFlagBits& outSampleCount) {
@@ -67,28 +68,58 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     }
 
     static Uint32 ResolveAttachmentBaseArrayLayer(const MG_State::GLState::FramebufferAttachmentObject& attachment) {
+        // Every branch has to go through ToStorageArrayLayer, including the two that name layer 0
+        // implicitly: a layered attachment of a texture VIEW starts at the view's first layer, not
+        // at the image's, and a cube FACE index is a layer index like any other. Leaving either
+        // unshifted made the render pass write layers [0, n) while the clear key, the blit, the
+        // copy and the readback for the same attachment all addressed [minLayer, minLayer + n) -
+        // they resolve the layer through their own copies of this helper, which do shift.
+        const auto* texture = attachment.GetTexture().get();
         if (attachment.IsLayered()) {
-            return 0;
+            return ToStorageArrayLayer(texture, 0);
         }
         const TextureUploadTarget uploadTarget = attachment.GetTextureUploadTarget();
         if (!IsCubeMapFaceUploadTarget(uploadTarget)) {
-            return static_cast<Uint32>(std::max(attachment.GetTextureLayer(), 0));
+            return ToStorageArrayLayer(texture, attachment.GetTextureLayer());
         }
-        return static_cast<Uint32>(uploadTarget) - static_cast<Uint32>(TextureUploadTarget::CubeMapPositiveX);
+        const Int face =
+            static_cast<Int>(uploadTarget) - static_cast<Int>(TextureUploadTarget::CubeMapPositiveX);
+        return ToStorageArrayLayer(texture, face);
     }
 
-    static Uint32 ResolveAttachmentLayerCount(const MG_State::GLState::FramebufferAttachmentObject& attachment) {
-        if (attachment.IsLayered()) {
-            return static_cast<Uint32>(std::max(attachment.GetSize().z(), 1));
-        }
-        return 1u;
-    }
+    // ResolveAttachmentLayerCount lives in VkTextureManager.h, beside ToVulkanLevelExtent, because
+    // VkClearManager needs the SAME answer: its pending-clear key's layerCount becomes a real
+    // VkImageSubresourceRange when a clear is materialised outside a render pass. See the header.
 
+    // VUID-VkFramebufferCreateInfo-flags-04113: every view handed to vkCreateFramebuffer must have
+    // been created as VK_IMAGE_VIEW_TYPE_2D or VK_IMAGE_VIEW_TYPE_2D_ARRAY. The image's OWN view
+    // type is not a legal answer for several of the targets GL can attach, and returning it
+    // unchanged is what took the process down on every layered 3D / cube-map-array attachment:
+    // a 3D view is refused outright by the layer-span guard in GetOrCreateAttachmentViewAtMipLevel
+    // (3D images have arrayLayers == 1) and a CUBE_ARRAY view is built happily and then rejected -
+    // or dereferenced - by the driver inside vkCreateFramebuffer.
+    //
+    // A 2D_ARRAY view is the legal spelling of all three: over a 2D-array-compatible 3D image its
+    // "layers" are the mip's z slices (VUID-VkImageViewCreateInfo-image-04970), and over a
+    // CUBE_COMPATIBLE 2D image - which is what both cube targets are - its layers are the faces.
+    //
+    // Knowingly NOT remapped: VK_IMAGE_VIEW_TYPE_1D / _1D_ARRAY, which 04113 also forbids. There is
+    // no legal alternative for them (a VK_IMAGE_TYPE_1D image admits no 2D-family view at all), so
+    // the only honest answer would be to decline the attachment - and every driver this has run on,
+    // lavapipe included, accepts them. Declining would turn working GL_TEXTURE_1D[_ARRAY] render
+    // targets into skipped draws to satisfy a VU nothing enforces. Left as-is, deliberately.
     static VkImageViewType ResolveAttachmentViewType(
         const MG_State::GLState::FramebufferAttachmentObject& attachment,
         const VkTextureManager::TextureResource& resource) {
         if (attachment.IsLayered()) {
-            return resource.viewType;
+            switch (resource.viewType) {
+            case VK_IMAGE_VIEW_TYPE_3D:
+            case VK_IMAGE_VIEW_TYPE_CUBE:
+            case VK_IMAGE_VIEW_TYPE_CUBE_ARRAY:
+                return VK_IMAGE_VIEW_TYPE_2D_ARRAY;
+            default:
+                return resource.viewType;
+            }
         }
         // A non-layered attachment names ONE layer, so the view over it is a plain 2D view whatever
         // the image's own view type is. The cube-face upload targets always meant this; a cube map
@@ -96,8 +127,15 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // a single layer is not a legal attachment. The CUBE arm is inert today - no frontend path
         // produces a non-layered cube attachment without a face upload target - and is kept for
         // symmetry with CUBE_ARRAY.
+        //
+        // 3D belongs in the same list and was missing from it, which is why the "per-slice
+        // attachment view is a 2D view whose array layer is the slice" branch in
+        // GetOrCreateAttachmentViewAtMipLevel was unreachable: glFramebufferTextureLayer on a
+        // GL_TEXTURE_3D asked for a 3D view (illegal as an attachment) whose span was then checked
+        // against arrayLayers == 1, so every slice above z = 0 came back VK_NULL_HANDLE.
         if (IsCubeMapFaceUploadTarget(attachment.GetTextureUploadTarget()) ||
-            resource.viewType == VK_IMAGE_VIEW_TYPE_CUBE_ARRAY || resource.viewType == VK_IMAGE_VIEW_TYPE_CUBE) {
+            resource.viewType == VK_IMAGE_VIEW_TYPE_CUBE_ARRAY || resource.viewType == VK_IMAGE_VIEW_TYPE_CUBE ||
+            resource.viewType == VK_IMAGE_VIEW_TYPE_3D) {
             return VK_IMAGE_VIEW_TYPE_2D;
         }
         return resource.viewType;
@@ -318,47 +356,18 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         }
 
         const auto internalFormat = renderbuffer->GetInternalFormat();
-        // Three-channel color formats widen to their RGBA twin exactly like textures do
-        // (VkTextureManager::ResolveTextureFormatInfo): blits/resolves between a
-        // renderbuffer and a texture of the same GL format then see one VkFormat.
-        const VkFormat format = [&]() -> VkFormat {
-            switch (internalFormat) {
-            case TextureInternalFormat::RGB:
-            case TextureInternalFormat::RGB8:
-            case TextureInternalFormat::R3G3B2:
-            case TextureInternalFormat::RGB4:
-            case TextureInternalFormat::RGB5:
-                return VK_FORMAT_R8G8B8A8_UNORM;
-            case TextureInternalFormat::SRGB8:
-                return VK_FORMAT_R8G8B8A8_SRGB;
-            case TextureInternalFormat::RGB8Snorm:
-                return VK_FORMAT_R8G8B8A8_SNORM;
-            case TextureInternalFormat::RGB10:
-            case TextureInternalFormat::RGB12:
-            case TextureInternalFormat::RGB16:
-                return VK_FORMAT_R16G16B16A16_UNORM;
-            case TextureInternalFormat::RGB16Snorm:
-                return VK_FORMAT_R16G16B16A16_SNORM;
-            case TextureInternalFormat::RGB16F:
-                return VK_FORMAT_R16G16B16A16_SFLOAT;
-            case TextureInternalFormat::RGB32F:
-                return VK_FORMAT_R32G32B32A32_SFLOAT;
-            case TextureInternalFormat::RGB8I:
-                return VK_FORMAT_R8G8B8A8_SINT;
-            case TextureInternalFormat::RGB8UI:
-                return VK_FORMAT_R8G8B8A8_UINT;
-            case TextureInternalFormat::RGB16I:
-                return VK_FORMAT_R16G16B16A16_SINT;
-            case TextureInternalFormat::RGB16UI:
-                return VK_FORMAT_R16G16B16A16_UINT;
-            case TextureInternalFormat::RGB32I:
-                return VK_FORMAT_R32G32B32A32_SINT;
-            case TextureInternalFormat::RGB32UI:
-                return VK_FORMAT_R32G32B32A32_UINT;
-            default:
-                return MG_Util::ConvertTextureInternalFormatToVkEnum(internalFormat);
-            }
-        }();
+        // ONE resolver, shared with textures (VkTextureManager::ResolveTextureFormatInfo), so a
+        // renderbuffer and a texture of the same GL format cannot disagree about their VkFormat.
+        // `expandRgbToRgba` / `componentByteCount` / `alphaBytes` describe how to reshape a SHADOW
+        // UPLOAD, and a renderbuffer has none, so only `.format` is taken.
+        //
+        // This used to be a hand-maintained second copy of that table, and it was missing exactly
+        // four rows: RGBA2 and RGBA12 fell through to ConvertTextureInternalFormatToVkEnum's
+        // VK_FORMAT_UNDEFINED (no image at all - bound as a draw buffer the attachment became
+        // VK_ATTACHMENT_UNUSED and every draw into it was dropped), while RGBA4 and RGB5A1 fell
+        // through to the 16-bit packed formats and then faced 32-bit R8G8B8A8_UNORM textures across
+        // a size-incompatible vkCmdCopyImage.
+        const VkFormat format = ResolveTextureFormatInfo(internalFormat).format;
         const VkImageAspectFlags aspect = ResolveImageAspectMaskForFormat(format);
         // Renderbuffers are never sampled (GL has no way to bind one to a sampler), so the
         // usage set is attachment + transfer: transfer covers readback (vkCmdCopyImageToBuffer),
@@ -602,7 +611,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // sRGB attachments switch between their sRGB and UNORM-twin views with this
         // capability (ResolveSrgbAttachmentWriteFormat), changing the render pass formats.
         const Bool framebufferSrgbEnabled =
-            MG_State::pGLContext->IsCapabilityEnabled(MobileGL::CapabilityInput::FramebufferSrgb);
+            MGB_CTX->IsCapabilityEnabled(MobileGL::CapabilityInput::FramebufferSrgb);
         XXHASH_VERIFY(XXH64_update(m_hashState, &framebufferSrgbEnabled, sizeof(framebufferSrgbEnabled)));
         auto& drawBuffers = fbo.GetDrawBuffers();
         XXHASH_VERIFY(XXH64_update(m_hashState, drawBuffers.data(), drawBuffers.size() * sizeof(drawBuffers[0])));
@@ -633,11 +642,13 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             if (att.IsTexture()) {
                 const Uint64 textureLifetimeId = att.GetTexture()->GetLifetimeId();
                 XXHASH_VERIFY(XXH64_update(m_hashState, &textureLifetimeId, sizeof(textureLifetimeId)));
-                const Int textureLevel = att.GetTextureLevel();
+                const Int textureLevel = static_cast<Int>(ToStorageMipLevel(att.GetTexture().get(),
+                                                                             att.GetTextureLevel()));
                 XXHASH_VERIFY(XXH64_update(m_hashState, &textureLevel, sizeof(textureLevel)));
                 const TextureUploadTarget textureUploadTarget = att.GetTextureUploadTarget();
                 XXHASH_VERIFY(XXH64_update(m_hashState, &textureUploadTarget, sizeof(textureUploadTarget)));
-                const Int textureLayer = att.GetTextureLayer();
+                const Int textureLayer = static_cast<Int>(ToStorageArrayLayer(att.GetTexture().get(),
+                                                                              att.GetTextureLayer()));
                 XXHASH_VERIFY(XXH64_update(m_hashState, &textureLayer, sizeof(textureLayer)));
                 const Bool textureLayered = att.IsLayered();
                 XXHASH_VERIFY(XXH64_update(m_hashState, &textureLayered, sizeof(textureLayered)));
@@ -754,7 +765,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         return XXH64_digest(m_hashState);
     }
 
-    RenderPassEntry& VkRenderPassManager::GetOrCreateRenderPass(const MG_State::GLState::FramebufferObject& fbo,
+    RenderPassEntry* VkRenderPassManager::GetOrCreateRenderPass(const MG_State::GLState::FramebufferObject& fbo,
                                                                 Uint32 swapchainImageIndex,
                                                                 Bool drawUsesDepthStencil) {
         // Resolve the default-FBO depth flavor (see the header comment): keep the
@@ -831,6 +842,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // recreated since (texture + renderbuffer image epochs), and no pending clear (which alters
         // load ops). Any of these differing forces the full recompute below. Portable to VK 1.1.
         if (activeRenderPass != nullptr && m_rpFastValid && m_rpFastFbo == &fbo &&
+            m_rpFastFboLifetimeId == fbo.GetLifetimeId() &&
             m_rpFastFboVersion == fbo.GetObjectVersion() && m_rpFastSwapchainIndex == swapchainImageIndex &&
             m_rpFastTexEpoch == m_textureManager.GetTextureImageEpoch() &&
             m_rpFastRbEpoch == m_renderbufferImageEpoch &&
@@ -839,7 +851,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             auto activeIt = m_renderPasses.find(activeRenderPass->hash);
             if (activeIt != m_renderPasses.end()) {
                 activeIt->second.lastUsedFrame = m_frameCounter;
-                return activeIt->second;
+                return &activeIt->second;
             }
         }
 
@@ -855,6 +867,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             // epochs AFTER ComputeHash: its attachment SyncTexture can create an image (bump the epoch).
             m_rpFastValid = true;
             m_rpFastFbo = &fbo;
+            m_rpFastFboLifetimeId = fbo.GetLifetimeId();
             m_rpFastFboVersion = fbo.GetObjectVersion();
             m_rpFastSwapchainIndex = swapchainImageIndex;
             m_rpFastTexEpoch = m_textureManager.GetTextureImageEpoch();
@@ -862,13 +875,13 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             m_rpFastRenderPassHash = activeRenderPass->hash;
             m_rpFastHadDepthStencil = activeIt->second.hasDepthStencilAttachment;
             activeIt->second.lastUsedFrame = m_frameCounter;
-            return activeIt->second;
+            return &activeIt->second;
         }
         auto hash = ComputeHash(fbo, swapchainImageIndex, true, includeDefaultFboDepthStencil);
         auto it = m_renderPasses.find(hash);
         if (it != m_renderPasses.end()) {
             it->second.lastUsedFrame = m_frameCounter;
-            return it->second;
+            return &it->second;
         }
 
         Bool isDefaultFbo = fbo.IsDefaultFramebuffer();
@@ -950,7 +963,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
                     const VkImageLayout trackedRbLayout = rbResource->layout;
                     const Bool rbFramebufferSrgb =
-                        MG_State::pGLContext->IsCapabilityEnabled(MobileGL::CapabilityInput::FramebufferSrgb);
+                        MGB_CTX->IsCapabilityEnabled(MobileGL::CapabilityInput::FramebufferSrgb);
                     const VkFormat rbAttachmentFormat =
                         ResolveSrgbAttachmentWriteFormat(rbResource->format, rbFramebufferSrgb);
                     rbDesc.flags = 0;
@@ -991,8 +1004,12 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                     textureResources.emplace_back(nullptr);
                     attachmentViews.emplace_back(rbAttachmentFormat != rbResource->format ? rbResource->unormTwinView
                                                                                           : rbResource->view);
-                    MOBILEGL_ASSERT(attachmentViews.back() != VK_NULL_HANDLE,
-                                    "GetOrCreateRenderPass: renderbuffer view missing at color attachment %d", i);
+                    if (attachmentViews.back() == VK_NULL_HANDLE) {
+                        MGLOG_E_ONCE("GetOrCreateRenderPass: renderbuffer %u has no usable view for color attachment "
+                                     "%u on FBO %u; declining the render pass",
+                                     renderbuffer->GetExternalIndex(), i, fbo.GetExternalIndex());
+                        return nullptr;
+                    }
 
                     colorAttachmentRefs[i].attachment = rbAttachmentIndex;
                     continue;
@@ -1004,7 +1021,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 continue;
 
             auto& att = fbo.GetAttachment(drawbuf);
-            const Uint32 attachmentMipLevel = static_cast<Uint32>(std::max(att.GetTextureLevel(), 0));
+            const Uint32 attachmentMipLevel = ToStorageMipLevel(att.GetTexture().get(), att.GetTextureLevel());
             const auto textureTarget = texture->GetTarget();
             const Uint32 attachmentIndex = static_cast<Uint32>(attachmentDescriptions.size());
             attachmentDescriptions.emplace_back();
@@ -1047,8 +1064,12 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                             .key = VkClearManager::MakePendingClearKey(att)
                         });
                     }
-                    const IntVec2 attachmentExtent =
-                        ResolveRenderPassFramebufferExtent(isDefaultFbo, att.GetSize(), swapchainExtent);
+                    // Same remap as ResolveAttachmentLayerCount, for the same reason: a
+                    // 1D-array attachment's GL height is its layer count, and using it as the
+                    // framebuffer height asks for a framebuffer taller than the VK_IMAGE_TYPE_1D
+                    // image it is built over.
+                    const IntVec2 attachmentExtent = ResolveRenderPassFramebufferExtent(
+                        isDefaultFbo, ToVulkanLevelExtent(texture->GetTarget(), att.GetSize()), swapchainExtent);
                     if (width == 0)
                         width = attachmentExtent.x();
                     if (height == 0)
@@ -1076,12 +1097,19 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                         attachmentViews.emplace_back(swapchainViews[swapchainImageIndex]);
                     } else {
                         auto* textureResource = m_textureManager.SyncTextureAndGetDescriptor(*texture);
-                        MOBILEGL_ASSERT(textureResource,
-                                        "GetOrCreateRenderPass: SyncTextureAndGetDescriptor failed at color attachment %d", i);
+                        if (textureResource == nullptr) {
+                            // SyncTextureResource legitimately declines - an unsupported format,
+                            // sample count or image-flag combination, or a vkCreateImage the driver
+                            // refused. There is no image to attach, so there is no render pass.
+                            MGLOG_E_ONCE("GetOrCreateRenderPass: textureId=%d could not be backed for color "
+                                         "attachment %u on FBO %u; declining the render pass",
+                                         texture->GetExternalIndex(), i, fbo.GetExternalIndex());
+                            return nullptr;
+                        }
                         textureResources.emplace_back(textureResource);
                         desc.format = ResolveSrgbAttachmentWriteFormat(
                             textureResource->format,
-                            MG_State::pGLContext->IsCapabilityEnabled(MobileGL::CapabilityInput::FramebufferSrgb));
+                            MGB_CTX->IsCapabilityEnabled(MobileGL::CapabilityInput::FramebufferSrgb));
                         attachmentSampleCount = textureResource->sampleCount;
                         trackedColorLayout = textureResource->layout;
                         trackedAttachmentLayouts.emplace_back(TrackedAttachmentLayoutInfo {
@@ -1098,8 +1126,15 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                         attachmentViews.emplace_back(
                             m_textureManager.GetOrCreateAttachmentViewAtMipLevel(
                                 *texture, attachmentMipLevel, baseArrayLayer, layerCount, attachmentViewType));
-                        MOBILEGL_ASSERT(attachmentViews.back() != VK_NULL_HANDLE,
-                                        "GetOrCreateRenderPass: GetOrCreateAttachmentView failed at color attachment %d", i);
+                        if (attachmentViews.back() == VK_NULL_HANDLE) {
+                            MGLOG_E_ONCE("GetOrCreateRenderPass: no attachment view for textureId=%d mip=%u layers "
+                                         "[%u, %u) viewType=%d at color attachment %u on FBO %u; declining the "
+                                         "render pass",
+                                         texture->GetExternalIndex(), attachmentMipLevel, baseArrayLayer,
+                                         baseArrayLayer + layerCount, static_cast<Int>(attachmentViewType), i,
+                                         fbo.GetExternalIndex());
+                            return nullptr;
+                        }
                     }
                     desc.samples = attachmentSampleCount;
                     adoptRenderPassSampleCount(attachmentSampleCount, "color", texture->GetExternalIndex());
@@ -1142,7 +1177,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             if (a.IsTexture() && b.IsTexture()) {
                 return a.GetTexture().get() == b.GetTexture().get() &&
                        a.GetTextureUploadTarget() == b.GetTextureUploadTarget() &&
-                       a.GetTextureLevel() == b.GetTextureLevel();
+                       ToStorageMipLevel(a.GetTexture().get(), a.GetTextureLevel()) ==
+                           ToStorageMipLevel(b.GetTexture().get(), b.GetTextureLevel());
             }
             if (a.IsRenderbuffer() && b.IsRenderbuffer()) {
                 return a.GetRenderbuffer().get() == b.GetRenderbuffer().get();
@@ -1191,20 +1227,29 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             } else if (selectedDepthStencilAttachment->IsTexture()) {
                 auto& texture = *selectedDepthStencilAttachment->GetTexture();
                 depthTextureResource = m_textureManager.SyncTextureAndGetDescriptor(texture);
-                MOBILEGL_ASSERT(depthTextureResource,
-                                "GetOrCreateRenderPass: SyncTextureAndGetDescriptor failed at depth attachment");
+                if (depthTextureResource == nullptr) {
+                    MGLOG_E_ONCE("GetOrCreateRenderPass: textureId=%d could not be backed for the depth/stencil "
+                                 "attachment of FBO %u; declining the render pass",
+                                 texture.GetExternalIndex(), fbo.GetExternalIndex());
+                    return nullptr;
+                }
                 trackedDepthLayout = depthTextureResource->layout;
                 depthAttachmentDescription.format = depthTextureResource->format;
                 depthAttachmentSampleCount = depthTextureResource->sampleCount;
                 depthAttachmentId = static_cast<Int>(texture.GetExternalIndex());
-                attachmentExtent =
-                    ResolveRenderPassFramebufferExtent(isDefaultFbo, selectedDepthStencilAttachment->GetSize(),
-                                                       swapchainExtent);
+                attachmentExtent = ResolveRenderPassFramebufferExtent(
+                    isDefaultFbo,
+                    ToVulkanLevelExtent(texture.GetTarget(), selectedDepthStencilAttachment->GetSize()),
+                    swapchainExtent);
             } else {
                 const auto& renderbuffer = selectedDepthStencilAttachment->GetRenderbuffer();
                 depthRenderbufferResource = GetOrCreateRenderbufferResource(renderbuffer);
-                MOBILEGL_ASSERT(depthRenderbufferResource,
-                                "GetOrCreateRenderPass: GetOrCreateRenderbufferResource failed at depth attachment");
+                if (depthRenderbufferResource == nullptr) {
+                    MGLOG_E_ONCE("GetOrCreateRenderPass: renderbuffer %u could not be backed for the depth/stencil "
+                                 "attachment of FBO %u; declining the render pass",
+                                 renderbuffer->GetExternalIndex(), fbo.GetExternalIndex());
+                    return nullptr;
+                }
                 trackedDepthLayout = depthRenderbufferResource->layout;
                 depthAttachmentDescription.format = depthRenderbufferResource->format;
                 depthAttachmentSampleCount = depthRenderbufferResource->sampleCount;
@@ -1252,7 +1297,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             } else if (selectedDepthStencilAttachment->IsTexture()) {
                 auto& texture = *selectedDepthStencilAttachment->GetTexture();
                 const Uint32 attachmentMipLevel =
-                    static_cast<Uint32>(std::max(selectedDepthStencilAttachment->GetTextureLevel(), 0));
+                    ToStorageMipLevel(selectedDepthStencilAttachment->GetTexture().get(),
+                                      selectedDepthStencilAttachment->GetTextureLevel());
                 MOBILEGL_ASSERT(depthTextureResource->layout != VK_IMAGE_LAYOUT_UNDEFINED ||
                                     depthAttachmentDescription.loadOp != VK_ATTACHMENT_LOAD_OP_LOAD,
                                 "GetOrCreateRenderPass: depth attachment textureId=%d has undefined tracked layout with LOAD_OP_LOAD",
@@ -1277,8 +1323,14 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 attachmentViews.emplace_back(
                     m_textureManager.GetOrCreateAttachmentViewAtMipLevel(
                         texture, attachmentMipLevel, baseArrayLayer, layerCount, attachmentViewType));
-                MOBILEGL_ASSERT(attachmentViews.back() != VK_NULL_HANDLE,
-                                "GetOrCreateRenderPass: GetOrCreateAttachmentView failed at depth attachment");
+                if (attachmentViews.back() == VK_NULL_HANDLE) {
+                    MGLOG_E_ONCE("GetOrCreateRenderPass: no attachment view for textureId=%d mip=%u layers [%u, %u) "
+                                 "viewType=%d at the depth/stencil attachment of FBO %u; declining the render pass",
+                                 texture.GetExternalIndex(), attachmentMipLevel, baseArrayLayer,
+                                 baseArrayLayer + layerCount, static_cast<Int>(attachmentViewType),
+                                 fbo.GetExternalIndex());
+                    return nullptr;
+                }
                 if (width == 0 || height == 0) {
                     width = attachmentExtent.x();
                     height = attachmentExtent.y();
@@ -1300,6 +1352,12 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 });
                 textureResources.emplace_back(nullptr);
                 attachmentViews.emplace_back(depthRenderbufferResource->view);
+                if (attachmentViews.back() == VK_NULL_HANDLE) {
+                    MGLOG_E_ONCE("GetOrCreateRenderPass: renderbuffer %u has no usable view for the depth/stencil "
+                                 "attachment of FBO %u; declining the render pass",
+                                 renderbuffer->GetExternalIndex(), fbo.GetExternalIndex());
+                    return nullptr;
+                }
                 if (width == 0 || height == 0) {
                     width = attachmentExtent.x();
                     height = attachmentExtent.y();
@@ -1397,8 +1455,25 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         renderPassCreateInfo.dependencyCount = 2;
         renderPassCreateInfo.pDependencies = subpassDependencies;
 
+        // NOT VK_VERIFY. VkIncludes.h states the rule this function now lives by: VK_VERIFY is the
+        // INVARIANT check - a should-never-happen state, fatal-logged unlatched and trapped in a
+        // DEBUG build - and "a soft, recoverable failure must therefore NOT be routed through
+        // VK_VERIFY. Check the VkResult directly and report it with MGLOG_E_ONCE". A decline here
+        // is recoverable by construction: the caller drops the draw. Routing it through VK_VERIFY
+        // would have made the recovery dead code in a DEBUG build (the TRAP fires inside the macro,
+        // before the handle is ever examined) and, in an INFO build, printed an UNLATCHED fatal
+        // line on every draw for the life of the process - a decline caches nothing, so every
+        // later draw to the same framebuffer re-enters this path and fails again.
         VkRenderPass renderPass = VK_NULL_HANDLE;
-        VK_VERIFY(vkCreateRenderPass(m_device, &renderPassCreateInfo, nullptr, &renderPass));
+        const VkResult renderPassResult =
+            vkCreateRenderPass(m_device, &renderPassCreateInfo, nullptr, &renderPass);
+        if (renderPassResult != VK_SUCCESS || renderPass == VK_NULL_HANDLE) {
+            MGLOG_E_ONCE("GetOrCreateRenderPass: vkCreateRenderPass failed (%s, %d) for FBO %u; declining the "
+                         "render pass",
+                         VkResultToString(renderPassResult), static_cast<Int>(renderPassResult),
+                         fbo.GetExternalIndex());
+            return nullptr;
+        }
 
         // Framebuffer
         VkFramebufferCreateInfo framebufferCreateInfo;
@@ -1411,8 +1486,21 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         framebufferCreateInfo.width = width;
         framebufferCreateInfo.height = height;
         framebufferCreateInfo.layers = framebufferLayers;
+        // Direct VkResult check, for the same reason as vkCreateRenderPass above.
         VkFramebuffer framebuffer = VK_NULL_HANDLE;
-        VK_VERIFY(vkCreateFramebuffer(m_device, &framebufferCreateInfo, nullptr, &framebuffer));
+        const VkResult framebufferResult =
+            vkCreateFramebuffer(m_device, &framebufferCreateInfo, nullptr, &framebuffer);
+        if (framebufferResult != VK_SUCCESS || framebuffer == VK_NULL_HANDLE) {
+            // The render pass has no entry to own it yet, so it is destroyed here rather than
+            // leaked - RenderPassEntry's destructor is the only other thing that would.
+            MGLOG_E_ONCE("GetOrCreateRenderPass: vkCreateFramebuffer failed (%s, %d) for FBO %u (%dx%d, "
+                         "%u attachments, %u layers); declining the render pass",
+                         VkResultToString(framebufferResult), static_cast<Int>(framebufferResult),
+                         fbo.GetExternalIndex(), width, height,
+                         static_cast<Uint32>(attachmentViews.size()), framebufferLayers);
+            vkDestroyRenderPass(m_device, renderPass, nullptr);
+            return nullptr;
+        }
         IntVec2 extent = {width, height};
         RenderPassEntry renderPassEntry {
             hash,
@@ -1437,7 +1525,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 extent.y());
         auto [insertedIt, _] = m_renderPasses.emplace(hash, Move(renderPassEntry));
         insertedIt->second.lastUsedFrame = m_frameCounter;
-        return insertedIt->second;
+        return &insertedIt->second;
     }
 
     void VkRenderPassManager::OnPresent() {
@@ -1507,7 +1595,23 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             ClearAttachmentPayload clearPayload{};
             SharedPtr<MG_State::GLState::ITextureObject> liveTexture;
             if (pending.hasInlinePayload) {
-                clearPayload = pending.inlinePayload;
+                // The inline payload was snapshotted when the entry was CREATED, but the
+                // clear VALUE is not part of the entry's hash - a cache hit with a newer
+                // glClear would replay the creation-time value and drop the new one (the
+                // texture path below is immune because it re-reads the live payload).
+                // Same defense as ClearAttachmentsOnActiveRenderPass: prefer the live
+                // pending clear, fall back to the snapshot only when none is queued.
+                if (s_renderPassManager != nullptr &&
+                    s_renderPassManager->GetPendingRenderbufferClear(pending.renderbuffer, clearPayload)) {
+                    if ((clearPayload.mask & GL_COLOR_BUFFER_BIT) != 0 && pending.renderbuffer != nullptr &&
+                        MG_Util::GetBaseInternalFormatComponentCount(pending.renderbuffer->GetInternalFormat()) ==
+                            3) {
+                        // RGB renderbuffers are backed by an RGBA image; the missing alpha reads as 1.
+                        ForceOpaqueClearAlpha(clearPayload);
+                    }
+                } else {
+                    clearPayload = pending.inlinePayload;
+                }
             } else {
                 if (pending.key.texture == nullptr ||
                     !s_clearManager->GetPendingClear(pending.key, clearPayload, liveTexture)) {

MobileGL/MG_Backend/DirectVulkan/Renderer/VkRenderPassManager.h

@@ -243,9 +243,24 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // draw against a depth-less active pass resolves to a new (incompatible)
         // entry, which the caller's compatibility check turns into a pass split;
         // the new pass's depth loads DONT_CARE (content was undefined all along).
-        RenderPassEntry& GetOrCreateRenderPass(const MG_State::GLState::FramebufferObject& fbo,
-                                               Uint32 swapchainImageIndex,
-                                               Bool drawUsesDepthStencil = true);
+        //
+        // Returns NULLPTR when this framebuffer cannot be represented as a Vulkan render pass at
+        // all - a texture the texture manager declined to back (an unsupported format or sample
+        // count), or an attachment view it cannot construct (a layer span the image has no room
+        // for, a 3D image whose format was refused 2D-array compatibility). This used to be
+        // unrepresentable: the function returned a reference, so the only thing the two fallible
+        // calls it builds on could do was trip a MOBILEGL_ASSERT - which is compiled out of every
+        // INFO build - and then dereference the null resource, or hand VK_NULL_HANDLE to
+        // vkCreateFramebuffer. That took the whole process down (51 lost CTS records over 21
+        // bodies, one runner restart each) where a declined draw is merely a wrong picture.
+        //
+        // EVERY caller must handle nullptr by dropping the operation, exactly as the draw path
+        // already drops a draw whose sampler descriptor could not be resolved
+        // (UniformManager::BindProgramUniformBuffers). The failure paths log MGLOG_E_ONCE
+        // themselves, so a caller needs no message of its own.
+        [[nodiscard]] RenderPassEntry* GetOrCreateRenderPass(const MG_State::GLState::FramebufferObject& fbo,
+                                                             Uint32 swapchainImageIndex,
+                                                             Bool drawUsesDepthStencil = true);
         void QueueRenderbufferClear(GLbitfield mask, const ClearFramebufferPayload& clearPayload,
                                     const MG_State::GLState::FramebufferObject& drawFbo);
         void QueueRenderbufferClear(const ClearAttachmentPayload& clearPayload,
@@ -289,6 +304,11 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // or a pending clear. Portable to Vulkan 1.1 (no dynamic_rendering / imageless FB needed).
         Bool m_rpFastValid = false;
         const MG_State::GLState::FramebufferObject* m_rpFastFbo = nullptr;
+        // The FBO's never-reused lifetime id joins the raw pointer + Uint16 version:
+        // a deleted FBO reallocated at the same address whose fresh setup performed
+        // the same number of version bumps would otherwise compare equal (both count
+        // from 0), serving the dead framebuffer's pass to the new object.
+        Uint64 m_rpFastFboLifetimeId = 0;
         Uint16 m_rpFastFboVersion = 0;
         Uint32 m_rpFastSwapchainIndex = 0;
         Uint64 m_rpFastTexEpoch = 0;

MobileGL/MG_Backend/DirectVulkan/Renderer/VkSamplerManager.cpp

@@ -21,6 +21,156 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                    sampler.GetWrapR() == SamplerWrapMode::ClampToBorder;
         }
 
+        // The numeric domain the texture is SAMPLED in. Vulkan splits VkBorderColor into a float
+        // family and an integer family and requires the sampler's choice to match the image view's
+        // format (a float border on an integer view, or the reverse, is undefined) - so the domain
+        // comes from the TEXTURE, while the value comes from whichever GL entry point wrote it.
+        enum class BorderColorDomain {
+            Float,
+            SignedInteger,
+            UnsignedInteger
+        };
+
+        BorderColorDomain ResolveBorderColorDomain(TextureInternalFormat format) {
+            switch (format) {
+            case TextureInternalFormat::R8I:
+            case TextureInternalFormat::R16I:
+            case TextureInternalFormat::R32I:
+            case TextureInternalFormat::RG8I:
+            case TextureInternalFormat::RG16I:
+            case TextureInternalFormat::RG32I:
+            case TextureInternalFormat::RGB8I:
+            case TextureInternalFormat::RGB16I:
+            case TextureInternalFormat::RGB32I:
+            case TextureInternalFormat::RGBA8I:
+            case TextureInternalFormat::RGBA16I:
+            case TextureInternalFormat::RGBA32I:
+                return BorderColorDomain::SignedInteger;
+            case TextureInternalFormat::R8UI:
+            case TextureInternalFormat::R16UI:
+            case TextureInternalFormat::R32UI:
+            case TextureInternalFormat::RG8UI:
+            case TextureInternalFormat::RG16UI:
+            case TextureInternalFormat::RG32UI:
+            case TextureInternalFormat::RGB8UI:
+            case TextureInternalFormat::RGB16UI:
+            case TextureInternalFormat::RGB32UI:
+            case TextureInternalFormat::RGBA8UI:
+            case TextureInternalFormat::RGBA16UI:
+            case TextureInternalFormat::RGBA32UI:
+            case TextureInternalFormat::RGB10A2UI:
+                return BorderColorDomain::UnsignedInteger;
+            default:
+                return BorderColorDomain::Float;
+            }
+        }
+
+        Bool IsSignedNormalizedFormat(TextureInternalFormat format) {
+            switch (format) {
+            case TextureInternalFormat::R8Snorm:
+            case TextureInternalFormat::R16Snorm:
+            case TextureInternalFormat::RG8Snorm:
+            case TextureInternalFormat::RG16Snorm:
+            case TextureInternalFormat::RGB8Snorm:
+            case TextureInternalFormat::RGB16Snorm:
+            case TextureInternalFormat::RGBA8Snorm:
+            case TextureInternalFormat::RGBA16Snorm:
+                return true;
+            default:
+                return false;
+            }
+        }
+
+        // GL 4.6 core 8.14.2: "The border values are clamped before they are used, according to the
+        // format in which texture components are stored. For signed and unsigned normalized
+        // fixed-point formats, border values are clamped to [-1,1] and [0,1] respectively. For
+        // floating-point and integer formats, border values are clamped to the representable range of
+        // the format." Every clause of that sentence is a real case here - the clamp is not just the
+        // normalized one.
+        //
+        // Only the 32-bit float formats are genuinely unclamped: every finite float is representable
+        // in them. Half-float has a finite maximum, and the two packed "float" formats are UNSIGNED,
+        // so a negative border on them must come back as 0 rather than as a negative number the
+        // driver delivers verbatim through VK_BORDER_COLOR_FLOAT_CUSTOM_EXT.
+        struct FloatBorderRange {
+            Bool clamped = true;
+            Float minValue = 0.0f;
+            Float maxValue = 1.0f;
+        };
+
+        FloatBorderRange ResolveFloatBorderRange(TextureInternalFormat format, Bool isSignedNormalized) {
+            switch (format) {
+            case TextureInternalFormat::R32F:
+            case TextureInternalFormat::RG32F:
+            case TextureInternalFormat::RGB32F:
+            case TextureInternalFormat::RGBA32F:
+                return {false, 0.0f, 0.0f};
+            case TextureInternalFormat::R16F:
+            case TextureInternalFormat::RG16F:
+            case TextureInternalFormat::RGB16F:
+            case TextureInternalFormat::RGBA16F:
+                return {true, -65504.0f, 65504.0f};
+            // Unsigned packed floats: no sign bit at all. 65024 is the largest 11-bit float; the
+            // 10-bit blue channel tops out lower (64512) and RGB9E5 higher (65408), but the bound
+            // that matters for correctness is the lower one, and a single conservative upper bound
+            // costs nothing a real border colour will ever notice.
+            case TextureInternalFormat::R11FG11FB10F:
+                return {true, 0.0f, 64512.0f};
+            case TextureInternalFormat::RGB9E5:
+                return {true, 0.0f, 65408.0f};
+            default:
+                return {true, isSignedNormalized ? -1.0f : 0.0f, 1.0f};
+            }
+        }
+
+        // Per-component representable range of an integer texture format, as Int64 so that the whole
+        // signed and unsigned 32-bit ranges are expressible in one type and the clamp can be written
+        // once for both domains. Alpha is carried separately because RGB10_A2UI is the one format
+        // whose alpha is narrower than its colour channels.
+        struct IntegerBorderRange {
+            Int64 rgbMin = 0;
+            Int64 rgbMax = 0;
+            Int64 alphaMin = 0;
+            Int64 alphaMax = 0;
+        };
+
+        IntegerBorderRange ResolveIntegerBorderRange(TextureInternalFormat format) {
+            const auto uniform = [](Int64 low, Int64 high) { return IntegerBorderRange{low, high, low, high}; };
+            switch (format) {
+            case TextureInternalFormat::R8I:
+            case TextureInternalFormat::RG8I:
+            case TextureInternalFormat::RGB8I:
+            case TextureInternalFormat::RGBA8I:
+                return uniform(-128, 127);
+            case TextureInternalFormat::R16I:
+            case TextureInternalFormat::RG16I:
+            case TextureInternalFormat::RGB16I:
+            case TextureInternalFormat::RGBA16I:
+                return uniform(-32768, 32767);
+            case TextureInternalFormat::R8UI:
+            case TextureInternalFormat::RG8UI:
+            case TextureInternalFormat::RGB8UI:
+            case TextureInternalFormat::RGBA8UI:
+                return uniform(0, 255);
+            case TextureInternalFormat::R16UI:
+            case TextureInternalFormat::RG16UI:
+            case TextureInternalFormat::RGB16UI:
+            case TextureInternalFormat::RGBA16UI:
+                return uniform(0, 65535);
+            case TextureInternalFormat::R32UI:
+            case TextureInternalFormat::RG32UI:
+            case TextureInternalFormat::RGB32UI:
+            case TextureInternalFormat::RGBA32UI:
+                return uniform(0, 4294967295LL);
+            case TextureInternalFormat::RGB10A2UI:
+                return {0, 1023, 0, 3};
+            default:
+                // The signed 32-bit formats, and anything unexpected: the full int32 range, i.e. a
+                // clamp that cannot alter a value the GL entry points could have carried.
+                return uniform(-2147483648LL, 2147483647LL);
+            }
+        }
+
         Bool IsDepthTextureFormat(TextureInternalFormat format) {
             switch (format) {
             case TextureInternalFormat::DepthComponent:
@@ -72,6 +222,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         m_config = initInfo.config;
         m_samplerAnisotropySupported = initInfo.samplerAnisotropySupported;
         m_maxSamplerAnisotropy = std::max(initInfo.maxSamplerAnisotropy, 1.0f);
+        m_customBorderColorSupported = initInfo.customBorderColorSupported;
+        m_maxCustomBorderColorSamplers = initInfo.maxCustomBorderColorSamplers;
+        m_customBorderColorSamplerCount = 0;
         MOBILEGL_ASSERT(m_device != VK_NULL_HANDLE && m_config != nullptr,
                         "VkSamplerManager::Initialize failed: invalid initialization info");
         return true;
@@ -102,6 +255,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         m_device = VK_NULL_HANDLE;
         m_config = nullptr;
         m_frameBoundaryCounter = 0;
+        m_customBorderColorSupported = false;
+        m_maxCustomBorderColorSamplers = 0;
+        m_customBorderColorSamplerCount = 0;
     }
 
     void VkSamplerManager::OnFrameBoundary() {
@@ -123,6 +279,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 if (m_device != VK_NULL_HANDLE && entry.handle != VK_NULL_HANDLE) {
                     vkDestroySampler(m_device, entry.handle, nullptr);
                 }
+                if (entry.usesCustomBorderColor && m_customBorderColorSamplerCount > 0) {
+                    --m_customBorderColorSamplerCount;
+                }
                 it = m_samplers.erase(it);
             } else {
                 ++it;
@@ -131,8 +290,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     }
 
     Uint64 VkSamplerManager::BuildSamplerKey(const MG_State::GLState::SamplerObject& sampler,
-                                             const MG_State::GLState::ITextureObject& texture,
-                                             Bool forceNearestFiltering, Bool singleLevelView) const {
+                                             Bool forceNearestFiltering, Bool singleLevelView,
+                                             const ResolvedBorderColor& borderColor) const {
         MOBILEGL_ASSERT(m_config != nullptr, "VkSamplerManager::BuildSamplerKey: m_config is null");
         XXHASH_VERIFY(XXH64_reset(m_hashState, m_config->CacheVersion));
 
@@ -166,8 +325,13 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         XXHASH_VERIFY(XXH64_update(m_hashState, &compareMode, sizeof(compareMode)));
         const auto compareFunc = sampler.GetSamplerCompareFunc();
         XXHASH_VERIFY(XXH64_update(m_hashState, &compareFunc, sizeof(compareFunc)));
-        const auto borderColor = ResolveVkBorderColor(sampler, texture);
-        XXHASH_VERIFY(XXH64_update(m_hashState, &borderColor, sizeof(borderColor)));
+        // The resolved enum AND, when it is one of the *_CUSTOM_EXT values, the sixteen bytes of the
+        // colour itself: two samplers that differ only in a custom border colour carry the same enum
+        // and would otherwise collide onto whichever one was created first.
+        XXHASH_VERIFY(XXH64_update(m_hashState, &borderColor.color, sizeof(borderColor.color)));
+        if (borderColor.isCustom) {
+            XXHASH_VERIFY(XXH64_update(m_hashState, &borderColor.customValue, sizeof(borderColor.customValue)));
+        }
         return XXH64_digest(m_hashState);
     }
 
@@ -183,7 +347,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // allocation for a genuinely single-level image) and faults the GPU - the same failure
         // the default-framebuffer blit shader had to work around with an explicit-LOD sample.
         const Bool singleLevelView = viewLevelCount == 1;
-        const Uint64 key = BuildSamplerKey(sampler, texture, forceNearestFiltering, singleLevelView);
+        // Resolved once and used for both the key and the create-info; see ResolvedBorderColor.
+        const ResolvedBorderColor borderColor = ResolveBorderColor(sampler, texture);
+        const Uint64 key = BuildSamplerKey(sampler, forceNearestFiltering, singleLevelView, borderColor);
         auto it = m_samplers.find(key);
         if (it != m_samplers.end()) {
             it->second.lastUsedFrameBoundary = m_frameBoundaryCounter;
@@ -211,9 +377,21 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // Must match BuildSamplerKey's resolution exactly.
         samplerInfo.maxLod = ResolveSingleLevelMaxLod(sampler, singleLevelView);
         samplerInfo.minLod = ResolveEffectiveMinLod(sampler, samplerInfo.maxLod);
-        samplerInfo.borderColor = ResolveVkBorderColor(sampler, texture);
+        samplerInfo.borderColor = borderColor.color;
         samplerInfo.unnormalizedCoordinates = VK_FALSE;
 
+        // VK_EXT_custom_border_color. `format` stays UNDEFINED, which is legal only because
+        // customBorderColorWithoutFormat was required alongside customBorderColors at device
+        // creation - a GL sampler object has no idea which texture it will be paired with.
+        VkSamplerCustomBorderColorCreateInfoEXT customBorderColorInfo{};
+        if (borderColor.isCustom) {
+            customBorderColorInfo.sType = VK_STRUCTURE_TYPE_SAMPLER_CUSTOM_BORDER_COLOR_CREATE_INFO_EXT;
+            customBorderColorInfo.customBorderColor = borderColor.customValue;
+            customBorderColorInfo.format = VK_FORMAT_UNDEFINED;
+            customBorderColorInfo.pNext = samplerInfo.pNext;
+            samplerInfo.pNext = &customBorderColorInfo;
+        }
+
         VkSampler vkSampler = VK_NULL_HANDLE;
         VK_VERIFY(vkCreateSampler(m_device, &samplerInfo, nullptr, &vkSampler), "vkCreateSampler(texture)");
 
@@ -222,6 +400,10 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         entry.externalIndex = sampler.GetExternalIndex();
         entry.version = sampler.GetVersion();
         entry.lastUsedFrameBoundary = m_frameBoundaryCounter;
+        entry.usesCustomBorderColor = borderColor.isCustom;
+        if (entry.usesCustomBorderColor) {
+            ++m_customBorderColorSamplerCount;
+        }
         m_samplers[key] = entry;
         return vkSampler;
     }
@@ -281,39 +463,148 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         }
     }
 
-    VkBorderColor VkSamplerManager::ResolveVkBorderColor(const MG_State::GLState::SamplerObject& sampler,
-                                                         const MG_State::GLState::ITextureObject& texture) {
+    VkSamplerManager::ResolvedBorderColor VkSamplerManager::ResolveBorderColor(
+        const MG_State::GLState::SamplerObject& sampler, const MG_State::GLState::ITextureObject& texture) const {
+        ResolvedBorderColor resolved{};
         if (!UsesBorderColor(sampler)) {
-            return VK_BORDER_COLOR_FLOAT_TRANSPARENT_BLACK;
+            return resolved; // FLOAT_TRANSPARENT_BLACK, never sampled
         }
 
         // Border colour is sampler state: a bound sampler object supplies its own, and a texture
         // with none reaches the very same value through the sampler object it owns.
-        const auto& borderColor = sampler.GetBorderColor();
-        const Bool isDepthTexture = IsDepthTextureFormat(texture.GetFormat());
+        const auto format = texture.GetFormat();
+        const auto domain = ResolveBorderColorDomain(format);
+        const Bool canUseCustom = m_customBorderColorSupported && m_maxCustomBorderColorSamplers > 0 &&
+                                  m_customBorderColorSamplerCount < m_maxCustomBorderColorSamplers;
 
-        if (isDepthTexture) {
+        if (domain != BorderColorDomain::Float) {
+            // An integer image view REQUIRES an integer border colour, whatever the value is - even
+            // (0,0,0,1). The value itself is whichever integer form the application wrote; a float
+            // border on an integer texture is nonsense GL leaves undefined, so the derived integer
+            // representation (a plain cast) is as good an answer as any.
+            //
+            // Clamped to the format's representable range FIRST, per GL 4.6 core 8.14.2, and read
+            // through Int64 so the whole signed and unsigned 32-bit ranges are expressible at once.
+            //
+            // Which representation to start from is the TEXTURE's domain, not the entry-point form
+            // the application used. GL 4.6 core 8.10 stores an "I"-form border colour unmodified with
+            // an integer internal data type and does not define a sign conversion between the two
+            // integer forms, so the stored bits are reinterpreted in the sampled format's own
+            // signedness. Measured, not assumed: a border of -1 written with glTexParameterIiv
+            // against a GL_R8UI texture samples as 255 on the ES driver, i.e. as 0xFFFFFFFF clamped
+            // to the format's maximum - see the IntegerBorderColorScenario case that pins it. Picking
+            // the representation by the FORM instead would answer 0 here, which is a defensible
+            // reading of the same spec text but puts DirectVulkan at odds with DirectGLES - and
+            // DirectGLES cannot deviate, it forwards the value to the driver verbatim. Cross-backend
+            // agreement decides it.
+            const auto range = ResolveIntegerBorderRange(format);
+            const auto& borderColorI = sampler.GetBorderColorI();
+            const auto& borderColorUI = sampler.GetBorderColorUI();
+            const Bool startFromUnsigned = domain == BorderColorDomain::UnsignedInteger;
+            Int64 clamped[4];
+            for (SizeT channel = 0; channel < 4; ++channel) {
+                const Int64 raw = startFromUnsigned ? static_cast<Int64>(borderColorUI[channel])
+                                                    : static_cast<Int64>(borderColorI[channel]);
+                const Int64 low = channel == 3 ? range.alphaMin : range.rgbMin;
+                const Int64 high = channel == 3 ? range.alphaMax : range.rgbMax;
+                clamped[channel] = std::clamp(raw, low, high);
+            }
+
+            // Matched against the CLAMPED value, so a border the format cannot hold still lands on
+            // the palette entry it clamps to rather than missing every one of them.
+            const Bool allZeroRgb = clamped[0] == 0 && clamped[1] == 0 && clamped[2] == 0;
+            if (allZeroRgb && clamped[3] == 0) {
+                resolved.color = VK_BORDER_COLOR_INT_TRANSPARENT_BLACK;
+                return resolved;
+            }
+            if (allZeroRgb && clamped[3] == 1) {
+                resolved.color = VK_BORDER_COLOR_INT_OPAQUE_BLACK;
+                return resolved;
+            }
+            if (clamped[0] == 1 && clamped[1] == 1 && clamped[2] == 1 && clamped[3] == 1) {
+                resolved.color = VK_BORDER_COLOR_INT_OPAQUE_WHITE;
+                return resolved;
+            }
+            if (canUseCustom) {
+                resolved.color = VK_BORDER_COLOR_INT_CUSTOM_EXT;
+                resolved.isCustom = true;
+                for (SizeT channel = 0; channel < 4; ++channel) {
+                    if (domain == BorderColorDomain::UnsignedInteger) {
+                        resolved.customValue.uint32[channel] = static_cast<Uint32>(clamped[channel]);
+                    } else {
+                        resolved.customValue.int32[channel] = static_cast<Int32>(clamped[channel]);
+                    }
+                }
+                return resolved;
+            }
+            // No custom colour available: pick the nearest of the three integer palette entries
+            // rather than always answering transparent black, which is what turned an integer border
+            // of (-1,-1,-1,-1) into 0 and broke the CTS's clamped-texel detection outright.
+            const Bool opaque = clamped[3] != 0;
+            const Bool bright = clamped[0] != 0 || clamped[1] != 0 || clamped[2] != 0;
+            resolved.color = !opaque ? VK_BORDER_COLOR_INT_TRANSPARENT_BLACK
+                                     : (bright ? VK_BORDER_COLOR_INT_OPAQUE_WHITE : VK_BORDER_COLOR_INT_OPAQUE_BLACK);
+            return resolved;
+        }
+
+        // Float domain. GL 4.6 core 8.14.2/8.23: the border colour is interpreted in the texture's
+        // format, so it is clamped to that format's representable range first. Without the clamp the
+        // CTS's border of (255,255,255,255) on a GL_RGBA8 texture matched none of the palette entries
+        // and fell through to transparent black - every border texel sampled 0 where the test wanted
+        // 255. The range is per format class, not just the normalized [0,1] / [-1,1] pair: only the
+        // 32-bit float formats are unclamped.
+        FloatVec4 borderColor = sampler.GetBorderColor();
+        if (const auto range = ResolveFloatBorderRange(format, IsSignedNormalizedFormat(format)); range.clamped) {
+            borderColor = FloatVec4(std::clamp(borderColor.x(), range.minValue, range.maxValue),
+                                    std::clamp(borderColor.y(), range.minValue, range.maxValue),
+                                    std::clamp(borderColor.z(), range.minValue, range.maxValue),
+                                    std::clamp(borderColor.w(), range.minValue, range.maxValue));
+        }
+
+        // A depth texture samples one component, so only x decides - and its alpha reads as 1.
+        if (IsDepthTextureFormat(format)) {
             if (NearlyEqual(borderColor.x(), 1.0f)) {
-                return VK_BORDER_COLOR_FLOAT_OPAQUE_WHITE;
+                resolved.color = VK_BORDER_COLOR_FLOAT_OPAQUE_WHITE;
+                return resolved;
             }
             if (NearlyEqual(borderColor.x(), 0.0f)) {
-                return VK_BORDER_COLOR_FLOAT_OPAQUE_BLACK;
+                resolved.color = VK_BORDER_COLOR_FLOAT_OPAQUE_BLACK;
+                return resolved;
             }
         }
 
         const Bool rgbZero = NearlyEqual(borderColor.x(), 0.0f) && NearlyEqual(borderColor.y(), 0.0f) &&
                              NearlyEqual(borderColor.z(), 0.0f);
         if (rgbZero && NearlyEqual(borderColor.w(), 0.0f)) {
-            return VK_BORDER_COLOR_FLOAT_TRANSPARENT_BLACK;
+            resolved.color = VK_BORDER_COLOR_FLOAT_TRANSPARENT_BLACK;
+            return resolved;
         }
         if (rgbZero && NearlyEqual(borderColor.w(), 1.0f)) {
-            return VK_BORDER_COLOR_FLOAT_OPAQUE_BLACK;
+            resolved.color = VK_BORDER_COLOR_FLOAT_OPAQUE_BLACK;
+            return resolved;
         }
         if (NearlyEqual(borderColor.x(), 1.0f) && NearlyEqual(borderColor.y(), 1.0f) &&
             NearlyEqual(borderColor.z(), 1.0f) && NearlyEqual(borderColor.w(), 1.0f)) {
-            return VK_BORDER_COLOR_FLOAT_OPAQUE_WHITE;
+            resolved.color = VK_BORDER_COLOR_FLOAT_OPAQUE_WHITE;
+            return resolved;
         }
 
-        return VK_BORDER_COLOR_FLOAT_TRANSPARENT_BLACK;
+        if (canUseCustom) {
+            resolved.color = VK_BORDER_COLOR_FLOAT_CUSTOM_EXT;
+            resolved.isCustom = true;
+            resolved.customValue.float32[0] = borderColor.x();
+            resolved.customValue.float32[1] = borderColor.y();
+            resolved.customValue.float32[2] = borderColor.z();
+            resolved.customValue.float32[3] = borderColor.w();
+            return resolved;
+        }
+
+        // Nearest of the three float palette entries. Transparent black stays the answer for a
+        // transparent border, which is what the old unconditional fallback got right by accident.
+        const Bool opaque = borderColor.w() >= 0.5f;
+        const Bool bright = (borderColor.x() + borderColor.y() + borderColor.z()) >= 1.5f;
+        resolved.color = !opaque ? VK_BORDER_COLOR_FLOAT_TRANSPARENT_BLACK
+                                 : (bright ? VK_BORDER_COLOR_FLOAT_OPAQUE_WHITE : VK_BORDER_COLOR_FLOAT_OPAQUE_BLACK);
+        return resolved;
     }
 } // namespace MobileGL::MG_Backend::DirectVulkan

MobileGL/MG_Backend/DirectVulkan/Renderer/VkSamplerManager.h

@@ -28,6 +28,13 @@ public:
         Bool samplerAnisotropySupported = false;
         // VkPhysicalDeviceLimits::maxSamplerAnisotropy.
         Float maxSamplerAnisotropy = 1.0f;
+        // VK_EXT_custom_border_color was enabled with BOTH customBorderColors and
+        // customBorderColorWithoutFormat; see VulkanRenderer::m_customBorderColorFeatureEnabled.
+        Bool customBorderColorSupported = false;
+        // VkPhysicalDeviceCustomBorderColorPropertiesEXT::maxCustomBorderColorSamplers. A hard device
+        // limit on how many LIVE samplers may carry a custom border colour, so the cache counts them
+        // and falls back to the snapped predefined value once it is reached.
+        Uint32 maxCustomBorderColorSamplers = 0;
     };
 
     Bool Initialize(const InitInfo& initInfo);
@@ -52,6 +59,21 @@ public:
     // boundaries.
     void OnFrameBoundary();
 
+    // What GL_TEXTURE_BORDER_COLOR resolves to for one (sampler, texture) pair. `color` is always a
+    // legal VkBorderColor; when `isCustom` it is one of the *_CUSTOM_EXT values and `customValue`
+    // carries the actual components in a VkSamplerCustomBorderColorCreateInfoEXT.
+    //
+    // Resolved ONCE per GetOrCreateSampler call and threaded into both the cache key and the
+    // create-info, so the two cannot disagree - the same discipline the resolved anisotropy needs,
+    // and here it also makes the maxCustomBorderColorSamplers fallback deterministic: whether a
+    // custom colour was affordable is decided before the key is built, not twice with a budget
+    // change in between.
+    struct ResolvedBorderColor {
+        VkBorderColor color = VK_BORDER_COLOR_FLOAT_TRANSPARENT_BLACK;
+        VkClearColorValue customValue{};
+        Bool isCustom = false;
+    };
+
 private:
     struct SamplerCacheEntry {
         VkSampler handle = VK_NULL_HANDLE;
@@ -60,17 +82,18 @@ private:
         // Frame boundary of the last cache hit; entries idle past the
         // OnFrameBoundary retirement age have their VkSampler destroyed.
         Uint64 lastUsedFrameBoundary = 0;
+        // Counted against maxCustomBorderColorSamplers for as long as this entry lives.
+        Bool usesCustomBorderColor = false;
     };
 
-    Uint64 BuildSamplerKey(const MG_State::GLState::SamplerObject& sampler,
-                           const MG_State::GLState::ITextureObject& texture,
-                           Bool forceNearestFiltering, Bool singleLevelView) const;
+    Uint64 BuildSamplerKey(const MG_State::GLState::SamplerObject& sampler, Bool forceNearestFiltering,
+                           Bool singleLevelView, const ResolvedBorderColor& borderColor) const;
     static VkFilter ToVkFilter(SamplerFilterMode mode);
     static VkSamplerMipmapMode ToVkMipmapMode(SamplerMipmapMode mode);
     static VkSamplerAddressMode ToVkAddressMode(SamplerWrapMode mode);
     static VkCompareOp ToVkCompareOp(SamplerCompareFunc func);
-    static VkBorderColor ResolveVkBorderColor(const MG_State::GLState::SamplerObject& sampler,
-                                              const MG_State::GLState::ITextureObject& texture);
+    ResolvedBorderColor ResolveBorderColor(const MG_State::GLState::SamplerObject& sampler,
+                                           const MG_State::GLState::ITextureObject& texture) const;
     // The anisotropy Vulkan will actually apply: 1.0 (i.e. disabled) unless the feature is on and
     // the sampler filters linearly both ways, otherwise the GL request clamped to the device limit.
     // GL happily carries GL_TEXTURE_MAX_ANISOTROPY on a NEAREST sampler (Blaze3D's blocks do exactly
@@ -82,6 +105,12 @@ private:
     const VulkanRendererConfig* m_config = nullptr;
     Bool m_samplerAnisotropySupported = false;
     Float m_maxSamplerAnisotropy = 1.0f;
+    Bool m_customBorderColorSupported = false;
+    Uint32 m_maxCustomBorderColorSamplers = 0;
+    // Live cache entries carrying a custom border colour. Kept in step with the entries themselves
+    // in exactly the three places one can appear or disappear: creation, the OnFrameBoundary sweep,
+    // and Shutdown.
+    Uint32 m_customBorderColorSamplerCount = 0;
     UnorderedMap<Uint64, SamplerCacheEntry> m_samplers;
     // Monotonic frame-boundary counter (bumped in OnFrameBoundary) for cache aging.
     Uint64 m_frameBoundaryCounter = 0;

MobileGL/MG_Backend/DirectVulkan/Renderer/VkTextureManager.cpp

@@ -11,8 +11,10 @@
 #include "ProgramFactory.h"
 
 #include "MG_State/GLState/Core.h"
+#include <MG_Pipe/PipeInputsSwitch.h>
 #include "MG_Util/Converters/MGToStr/TextureEnumConverter.h"
 #include "MG_Util/Converters/MGToVk/TextureEnumConverter.h"
+#include "MG_Util/Metrics/PipeStats.h"
 
 #include <Config.h>
 #include <algorithm>
@@ -46,13 +48,6 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         return mipLevelCount;
     }
 
-    struct TextureFormatInfo {
-        VkFormat format = VK_FORMAT_UNDEFINED;
-        Bool expandRgbToRgba = false;
-        Uint32 componentByteCount = 0;
-        Array<Uint8, 4> alphaBytes = {0, 0, 0, 0};
-    };
-
     struct TextureShapeInfo {
         VkImageType imageType = VK_IMAGE_TYPE_2D;
         VkImageViewType viewType = VK_IMAGE_VIEW_TYPE_2D;
@@ -220,6 +215,22 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
     VkTextureManager::TextureIdentity VkTextureManager::MakeTextureIdentity(
         MG_State::GLState::ITextureObject* texture) {
+        // A GL texture view (ARB_texture_view) is identified by the texture whose STORAGE it
+        // views, not by itself. Everything this identity keys - the TextureResource, the tracked
+        // image layout, the alive-object weak reference, the storage-usage marks, the per-draw
+        // sync memos - is a property of the IMAGE, and a view shares that image exactly. Doing
+        // the resolution here rather than at each call site is what makes it impossible to miss
+        // one: a layout update posted against a view's own identity would have found no resource
+        // at all, which is precisely how an attached view came back blank.
+        //
+        // One hop suffices and cannot recurse: glTextureView composes a view-of-a-view onto the
+        // root at creation, so a storage owner is never itself a view.
+        if (texture != nullptr) {
+            const auto& storageOwner = texture->GetViewStorageOwner();
+            if (storageOwner) {
+                texture = storageOwner.get();
+            }
+        }
         return TextureIdentity{
             .texture = texture,
             .lifetimeId = texture ? texture->GetLifetimeId() : 0,
@@ -364,7 +375,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         return true;
     }
 
-    static TextureFormatInfo ResolveTextureFormatInfo(TextureInternalFormat format) {
+    TextureFormatInfo ResolveTextureFormatInfo(TextureInternalFormat format) {
         switch (format) {
         case TextureInternalFormat::RGB:
         case TextureInternalFormat::RGB8:
@@ -694,6 +705,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     }
 
     void VkTextureManager::EraseTrackedTexture(const TextureIdentity& identity) {
+        m_viewRequestedImageFlags.erase(identity);
+        m_viewRequestedFormats.erase(identity);
         auto resourceIt = m_textureResources.find(identity);
         if (resourceIt != m_textureResources.end()) {
             DeferResourceRelease(Move(resourceIt->second));
@@ -737,9 +750,19 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         m_drawSyncedThisDraw.clear();
     }
 
-    VkTextureManager::TextureResource* VkTextureManager::SyncTextureAndGetDescriptor(MG_State::GLState::ITextureObject& texture) {
+    VkTextureManager::TextureResource* VkTextureManager::SyncTextureAndGetDescriptor(MG_State::GLState::ITextureObject& textureOrView) {
         MOBILEGL_ASSERT(m_device != VK_NULL_HANDLE, "SyncTextureAndGetDescriptor: m_device == VK_NULL_HANDLE");
 
+        // A GL texture view has no image of its own; it resolves to - and shares - the resource
+        // of the texture whose storage it views, so that there is exactly one VkImage, one
+        // tracked layout and one upload path per storage. Everything that makes the view a
+        // different texture (format, level/layer window, sampled aspect) is applied where the
+        // VkImageViews are built, keyed in alternateSampledViews / attachmentViews.
+        MG_State::GLState::ITextureObject& texture = StorageTextureOf(textureOrView);
+        if (&texture != &textureOrView) {
+            NoteTextureViewImageRequirements(textureOrView, texture);
+        }
+
         const TextureIdentity identity = MakeTextureIdentity(&texture);
 
         // Per-draw memo fast path (see BeginDrawSyncScope): a texture already fully
@@ -784,7 +807,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             // sampled-texture sync scan the entire alive-texture map per draw.
             if (aliveIt == m_aliveObjects.end()) {
                 WeakPtr<MG_State::GLState::ITextureObject> aliveTexture;
-                const auto& liveTexture = MG_State::pGLContext->GetTextureObject(texture.GetExternalIndex());
+                const auto& liveTexture = MGB_CTX->GetTextureObject(texture.GetExternalIndex());
                 if (liveTexture && liveTexture.get() == &texture) {
                     aliveTexture = liveTexture;
                 } else {
@@ -838,7 +861,19 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
     VkImageView VkTextureManager::GetOrCreateViewAtMipLevel(MG_State::GLState::ITextureObject& texture, Uint32 mipLevel) {
         TextureResource* resource = SyncTextureAndGetDescriptor(texture);
-        if (resource == nullptr || resource->image == VK_NULL_HANDLE || mipLevel >= resource->mipLevels) {
+        if (resource == nullptr || resource->image == VK_NULL_HANDLE) {
+            return VK_NULL_HANDLE;
+        }
+        // A GL texture view shares this resource with the texture it views, so it must not touch
+        // perMipViews: that vector is indexed by mip level alone and holds views built with the
+        // STORAGE texture's format and full layer range. Route it through the keyed attachment
+        // cache instead, where its own window is part of the key.
+        if (texture.IsTextureView()) {
+            const TextureViewWindow window = ResolveTextureViewWindow(texture, *resource);
+            return GetOrCreateAttachmentViewAtMipLevel(texture, mipLevel, window.baseArrayLayer, window.layerCount,
+                                                       window.viewType);
+        }
+        if (mipLevel >= resource->mipLevels) {
             return VK_NULL_HANDLE;
         }
 
@@ -866,21 +901,43 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                                                                       Uint32 layerCount,
                                                                       VkImageViewType viewType) {
         TextureResource* resource = SyncTextureAndGetDescriptor(texture);
-        if (resource == nullptr || resource->image == VK_NULL_HANDLE || mipLevel >= resource->mipLevels) {
+        if (resource == nullptr || resource->image == VK_NULL_HANDLE) {
             return VK_NULL_HANDLE;
         }
-        // A 3D image has arrayLayers == 1 and keeps its GL layers on the z axis, so a per-slice
-        // attachment view is a 2D view whose "array layer" is the slice - legal only on a
-        // 2D-array-compatible image (VUID-VkImageViewCreateInfo-image-04970), which
-        // SyncTextureResource asks for and may have had refused per format.
-        if (resource->viewType == VK_IMAGE_VIEW_TYPE_3D && viewType == VK_IMAGE_VIEW_TYPE_2D) {
+        // mipLevel and baseArrayLayer arrive in STORAGE space - every caller runs them through
+        // ToStorageMipLevel / ToStorageArrayLayer at the GL attachment boundary. What a GL texture
+        // view still contributes here is its own internal format, which may reinterpret the
+        // storage's (GL 4.6 core table 8.21) and is what the attachment must actually be written
+        // through.
+        VkFormat viewFormatOverride = VK_FORMAT_UNDEFINED;
+        if (texture.IsTextureView()) {
+            viewFormatOverride = ResolveTextureViewWindow(texture, *resource).format;
+        }
+        if (mipLevel >= resource->mipLevels) {
+            return VK_NULL_HANDLE;
+        }
+        // A 3D image has arrayLayers == 1 and keeps its GL layers on the z axis, so an attachment
+        // view over it addresses SLICES through baseArrayLayer/layerCount: one slice for a
+        // non-layered attachment (a 2D view) and the whole span for a layered one (a 2D_ARRAY view,
+        // which is what a layered GL_TEXTURE_3D attachment plus a gl_Layer-writing geometry shader
+        // means). BOTH spellings are legal only on a 2D-array-compatible image
+        // (VUID-VkImageViewCreateInfo-image-04970 / -06723), which SyncTextureResource asks for and
+        // may have had refused per format.
+        //
+        // The span is validated against the MIP's slice count, never against arrayLayers: a 3D
+        // image's arrayLayers is 1 by construction, so measuring a layered span against it rejected
+        // every layered 3D attachment - the null view that used to reach vkCreateFramebuffer.
+        if (resource->viewType == VK_IMAGE_VIEW_TYPE_3D &&
+            (viewType == VK_IMAGE_VIEW_TYPE_2D || viewType == VK_IMAGE_VIEW_TYPE_2D_ARRAY)) {
             const Uint32 sliceCount = std::max(resource->depth >> mipLevel, 1u);
             if ((resource->imageCreateFlags & VK_IMAGE_CREATE_2D_ARRAY_COMPATIBLE_BIT) == 0 ||
                 layerCount == 0 || baseArrayLayer >= sliceCount || baseArrayLayer + layerCount > sliceCount) {
-                MGLOG_D("%s: cannot name slice span [%u, %u) of 3D textureId=%d (mip %u has %u slices, "
-                        "2D-array-compatible=%d)",
+                // Not an error line: the render-pass builder turns the null view into one
+                // MGLOG_E_ONCE and a skipped draw, which is the level this belongs at.
+                MGLOG_D("%s: cannot name slice span [%u, %u) of 3D textureId=%d as viewType=%d (mip %u has %u "
+                        "slices, 2D-array-compatible=%d)",
                         __func__, baseArrayLayer, baseArrayLayer + layerCount, texture.GetExternalIndex(),
-                        mipLevel, sliceCount,
+                        static_cast<Int>(viewType), mipLevel, sliceCount,
                         (int)((resource->imageCreateFlags & VK_IMAGE_CREATE_2D_ARRAY_COMPATIBLE_BIT) != 0));
                 return VK_NULL_HANDLE;
             }
@@ -893,11 +950,16 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         }
 
         const Bool framebufferSrgbEnabled =
-            MG_State::pGLContext->IsCapabilityEnabled(MobileGL::CapabilityInput::FramebufferSrgb);
-        const VkFormat attachmentFormat = ResolveSrgbAttachmentWriteFormat(resource->format, framebufferSrgbEnabled);
+            MGB_CTX->IsCapabilityEnabled(MobileGL::CapabilityInput::FramebufferSrgb);
+        const VkFormat baseAttachmentFormat =
+            viewFormatOverride != VK_FORMAT_UNDEFINED ? viewFormatOverride : resource->format;
+        const VkFormat attachmentFormat =
+            ResolveSrgbAttachmentWriteFormat(baseAttachmentFormat, framebufferSrgbEnabled);
 
-        if (attachmentFormat == resource->format && baseArrayLayer == 0 && layerCount == resource->arrayLayers &&
-            viewType == resource->viewType) {
+        // The shortcut back to the per-mip vector is only sound for the storage texture itself;
+        // for a view every field below is part of what distinguishes it from its parent.
+        if (viewFormatOverride == VK_FORMAT_UNDEFINED && attachmentFormat == resource->format &&
+            baseArrayLayer == 0 && layerCount == resource->arrayLayers && viewType == resource->viewType) {
             return GetOrCreateViewAtMipLevel(texture, mipLevel);
         }
 
@@ -932,7 +994,22 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     VkImageView VkTextureManager::GetOrCreateSampledViewAtMipLevel(MG_State::GLState::ITextureObject& texture,
                                                                    Uint32 mipLevel) {
         TextureResource* resource = SyncTextureAndGetDescriptor(texture);
-        if (resource == nullptr || resource->image == VK_NULL_HANDLE || mipLevel >= resource->mipLevels) {
+        if (resource == nullptr || resource->image == VK_NULL_HANDLE) {
+            return VK_NULL_HANDLE;
+        }
+        // As in GetOrCreateViewAtMipLevel: perMipSampledViews belongs to the storage texture's
+        // own format and aspect, so a GL view has to go to the keyed cache.
+        if (texture.IsTextureView()) {
+            if (mipLevel >= resource->mipLevels) {
+                return VK_NULL_HANDLE;
+            }
+            TextureViewWindow window = ResolveTextureViewWindow(texture, *resource);
+            // Storage space already (see ToStorageMipLevel); only the level COUNT narrows.
+            window.baseMipLevel = mipLevel;
+            window.levelCount = 1;
+            return GetOrCreateWindowedSampledView(texture, *resource, window);
+        }
+        if (mipLevel >= resource->mipLevels) {
             return VK_NULL_HANDLE;
         }
 
@@ -960,14 +1037,76 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         return perMipSampledView;
     }
 
-    VkImageView VkTextureManager::GetOrCreateSampledImageView(MG_State::GLState::ITextureObject& texture,
-                                                               VkFormat format) {
-        TextureResource* resource = SyncTextureAndGetDescriptor(texture);
-        if (resource == nullptr || resource->image == VK_NULL_HANDLE ||
-            resource->sampledView == VK_NULL_HANDLE) {
+    // Builds (and caches) one sampled VkImageView over `resource`'s image for an arbitrary
+    // window - the shared back end of every GL-texture-view sampled path. Keyed by the whole
+    // window, which is what keeps a D24S8's depth-aspect view and its stencil-aspect view apart
+    // in the same cache while both name the same image, the same levels and the same layers.
+    VkImageView VkTextureManager::GetOrCreateWindowedSampledView(MG_State::GLState::ITextureObject& texture,
+                                                                 TextureResource& resource,
+                                                                 const TextureViewWindow& window) {
+        const TextureResource::SampledImageViewKey key{
+            .baseMipLevel = window.baseMipLevel,
+            .levelCount = window.levelCount,
+            .baseArrayLayer = window.baseArrayLayer,
+            .layerCount = window.layerCount,
+            .viewType = window.viewType,
+            .format = window.format,
+            .aspect = window.sampledAspect,
+            .componentSwizzle = PackComponentSwizzle(window.components),
+        };
+        const auto existing = resource.alternateSampledViews.find(key);
+        if (existing != resource.alternateSampledViews.end()) {
+            return existing->second;
+        }
+
+        if (window.format != resource.format &&
+            (resource.imageCreateFlags & VK_IMAGE_CREATE_MUTABLE_FORMAT_BIT) == 0) {
+            MGLOG_E_ONCE("%s: textureId=%d needs a mutable-format image to be viewed as format=%d "
+                         "(image format=%d)",
+                         __func__, texture.GetExternalIndex(), static_cast<Int>(window.format),
+                         static_cast<Int>(resource.format));
             return VK_NULL_HANDLE;
         }
 
+        const VkImageView view =
+            CreateImageView(resource.image, window.format, window.sampledAspect, window.viewType,
+                            window.baseMipLevel, window.levelCount, window.baseArrayLayer, window.layerCount,
+                            &window.components);
+        if (view == VK_NULL_HANDLE) {
+            MGLOG_E_ONCE("%s: failed to create sampled view for textureId=%d format=%d aspect=0x%x "
+                         "mips=[%u,%u) layers=[%u,%u)",
+                         __func__, texture.GetExternalIndex(), static_cast<Int>(window.format),
+                         static_cast<Uint32>(window.sampledAspect), window.baseMipLevel,
+                         window.baseMipLevel + window.levelCount, window.baseArrayLayer,
+                         window.baseArrayLayer + window.layerCount);
+            return VK_NULL_HANDLE;
+        }
+        resource.alternateSampledViews.emplace(key, view);
+        return view;
+    }
+
+    VkImageView VkTextureManager::GetOrCreateSampledImageView(MG_State::GLState::ITextureObject& texture,
+                                                               VkFormat format) {
+        TextureResource* resource = SyncTextureAndGetDescriptor(texture);
+        if (resource == nullptr || resource->image == VK_NULL_HANDLE) {
+            return VK_NULL_HANDLE;
+        }
+
+        // A GL texture view never has a sampledView of its own on this resource - that one
+        // belongs to the storage texture, with the storage texture's format, level range and
+        // depth/stencil aspect. The window is the view's whole identity, so it always goes to the
+        // keyed cache, even when the requested format happens to match the image's.
+        if (texture.IsTextureView()) {
+            TextureViewWindow window = ResolveTextureViewWindow(texture, *resource);
+            if (format != VK_FORMAT_UNDEFINED) {
+                window.format = format;
+            }
+            return GetOrCreateWindowedSampledView(texture, *resource, window);
+        }
+
+        if (resource->sampledView == VK_NULL_HANDLE) {
+            return VK_NULL_HANDLE;
+        }
         if (format == VK_FORMAT_UNDEFINED || format == resource->format) {
             return resource->sampledView;
         }
@@ -987,8 +1126,13 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         const TextureResource::SampledImageViewKey key{
             .baseMipLevel = resource->sampledBaseMipLevel,
             .levelCount = resource->sampledLevelCount,
+            .baseArrayLayer = 0,
+            .layerCount = resource->arrayLayers,
             .viewType = resource->viewType,
             .format = format,
+            .aspect = VK_IMAGE_ASPECT_COLOR_BIT,
+            .componentSwizzle = PackComponentSwizzle(
+                ResolveSampledViewComponents(texture, ResolveTextureFormatInfo(texture.GetFormat()))),
         };
         const auto existing = resource->alternateSampledViews.find(key);
         if (existing != resource->alternateSampledViews.end()) {
@@ -1029,6 +1173,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     VkImageView VkTextureManager::GetOrCreateStorageImageView(MG_State::GLState::ITextureObject& texture,
                                                                Uint32 mipLevel, VkFormat format,
                                                                Bool layered, Int32 layer) {
+        // mipLevel and layer arrive in STORAGE space; ResolveStorageImageDescriptor converts
+        // the glBindImageTexture values with ToStorageMipLevel / ToStorageArrayLayer.
         TextureResource* resource = SyncTextureAndGetDescriptor(texture);
         if (resource == nullptr || resource->image == VK_NULL_HANDLE || mipLevel >= resource->mipLevels ||
             resource->sampleCount != VK_SAMPLE_COUNT_1_BIT ||
@@ -1053,8 +1199,13 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             return VK_NULL_HANDLE;
         }
 
-        Uint32 baseArrayLayer = 0;
-        Uint32 layerCount = resource->arrayLayers;
+        // A GL texture view opens onto a WINDOW of the storage's layers; a layered image
+        // binding of it must not reach past that window into the parent's other layers.
+        Uint32 baseArrayLayer = ToStorageArrayLayer(&texture, 0);
+        Uint32 layerCount = texture.IsTextureView()
+                                ? std::min(static_cast<Uint32>(texture.GetViewNumLayers()),
+                                           resource->arrayLayers - baseArrayLayer)
+                                : resource->arrayLayers;
         VkImageViewType viewType = resource->viewType;
         if (!layered) {
             switch (resource->viewType) {
@@ -1087,7 +1238,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
 
         const Bool isFullResourceView = baseArrayLayer == 0 && layerCount == resource->arrayLayers &&
                                         viewType == resource->viewType;
-        if (format == resource->format && isFullResourceView) {
+        if (format == resource->format && isFullResourceView && !texture.IsTextureView()) {
             return GetOrCreateViewAtMipLevel(texture, mipLevel);
         }
 
@@ -1291,6 +1442,158 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         return ok;
     }
 
+    Bool VkTextureManager::SnapshotTextureForSampling(VkCommandBuffer commandBuffer,
+                                                      MG_State::GLState::ITextureObject& texture,
+                                                      SamplerNumericDomain numericDomain,
+                                                      VkPipelineStageFlags consumerShaderStageMask,
+                                                      SampledTextureSnapshot& outSnapshot) {
+        outSnapshot = {};
+        TextureResource* source = SyncTextureAndGetDescriptor(texture);
+        if (source == nullptr || source->image == VK_NULL_HANDLE || source->sampleCount != VK_SAMPLE_COUNT_1_BIT ||
+            source->sampledLevelCount == 0) {
+            return false;
+        }
+
+        const VkFormat sampledFormat = ResolveSampledImageViewFormat(source->format, numericDomain);
+        if (sampledFormat == VK_FORMAT_UNDEFINED ||
+            !AreSampledImageViewFormatsCompatible(source->format, sampledFormat)) {
+            MGLOG_E_ONCE("SnapshotTextureForSampling: textureId=%d cannot create sampled view format=%d from image format=%d",
+                         texture.GetExternalIndex(), static_cast<Int>(sampledFormat), static_cast<Int>(source->format));
+            return false;
+        }
+        if (sampledFormat != source->format &&
+            (source->imageCreateFlags & VK_IMAGE_CREATE_MUTABLE_FORMAT_BIT) == 0) {
+            MGLOG_E_ONCE("SnapshotTextureForSampling: textureId=%d needs unavailable mutable image format=%d for sampled view=%d",
+                         texture.GetExternalIndex(), static_cast<Int>(source->format), static_cast<Int>(sampledFormat));
+            return false;
+        }
+
+        VkImageType imageType = VK_IMAGE_TYPE_2D;
+        switch (source->viewType) {
+        case VK_IMAGE_VIEW_TYPE_1D:
+        case VK_IMAGE_VIEW_TYPE_1D_ARRAY:
+            imageType = VK_IMAGE_TYPE_1D;
+            break;
+        case VK_IMAGE_VIEW_TYPE_3D:
+            imageType = VK_IMAGE_TYPE_3D;
+            break;
+        default:
+            break;
+        }
+
+        TextureResource snapshot{};
+        VkImageCreateInfo imageInfo{};
+        imageInfo.sType = VK_STRUCTURE_TYPE_IMAGE_CREATE_INFO;
+        imageInfo.flags = source->imageCreateFlags;
+        imageInfo.imageType = imageType;
+        imageInfo.extent = {source->extent.width, source->extent.height, source->depth};
+        imageInfo.mipLevels = source->mipLevels;
+        imageInfo.arrayLayers = source->arrayLayers;
+        imageInfo.format = source->format;
+        imageInfo.tiling = VK_IMAGE_TILING_OPTIMAL;
+        imageInfo.initialLayout = VK_IMAGE_LAYOUT_UNDEFINED;
+        imageInfo.usage = VK_IMAGE_USAGE_TRANSFER_DST_BIT | VK_IMAGE_USAGE_SAMPLED_BIT;
+        imageInfo.samples = VK_SAMPLE_COUNT_1_BIT;
+        imageInfo.sharingMode = VK_SHARING_MODE_EXCLUSIVE;
+
+        // Keep the temporary's view-format list just as narrow as the source's sampler use. This
+        // has no storage-image usage, so unlike an app image binding the exact list is knowable.
+        Vector<VkFormat> viewFormats;
+        VkImageFormatListCreateInfo formatListInfo{};
+        if (m_imageFormatListSupported && (imageInfo.flags & VK_IMAGE_CREATE_MUTABLE_FORMAT_BIT) != 0) {
+            viewFormats.push_back(source->format);
+            if (sampledFormat != source->format) {
+                viewFormats.push_back(sampledFormat);
+            }
+            formatListInfo.sType = VK_STRUCTURE_TYPE_IMAGE_FORMAT_LIST_CREATE_INFO;
+            formatListInfo.viewFormatCount = static_cast<Uint32>(viewFormats.size());
+            formatListInfo.pViewFormats = viewFormats.data();
+            imageInfo.pNext = &formatListInfo;
+        }
+
+        VmaAllocationCreateInfo allocationInfo{};
+        allocationInfo.usage = VMA_MEMORY_USAGE_AUTO_PREFER_DEVICE;
+        allocationInfo.requiredFlags = VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT;
+        const VkResult createResult =
+            vmaCreateImage(m_allocator, &imageInfo, &allocationInfo, &snapshot.image, &snapshot.allocation, nullptr);
+        if (createResult != VK_SUCCESS) {
+            MGLOG_E_ONCE("SnapshotTextureForSampling: vmaCreateImage failed result=%d textureId=%d", createResult,
+                         texture.GetExternalIndex());
+            return false;
+        }
+
+        snapshot.extent = source->extent;
+        snapshot.depth = source->depth;
+        snapshot.arrayLayers = source->arrayLayers;
+        snapshot.mipLevels = source->mipLevels;
+        snapshot.sampledBaseMipLevel = source->sampledBaseMipLevel;
+        snapshot.sampledLevelCount = source->sampledLevelCount;
+        snapshot.format = source->format;
+        snapshot.aspect = source->aspect;
+        snapshot.viewType = source->viewType;
+        snapshot.sampleCount = VK_SAMPLE_COUNT_1_BIT;
+        snapshot.imageCreateFlags = imageInfo.flags;
+        snapshot.usageFlags = imageInfo.usage;
+
+        const TextureFormatInfo formatInfo = ResolveTextureFormatInfo(texture.GetFormat());
+        const VkComponentMapping sampledComponents = ResolveSampledViewComponents(texture, formatInfo);
+        const VkImageAspectFlags sampledAspect =
+            ResolveSampledImageViewAspectMask(snapshot.aspect, texture.GetDepthStencilTextureMode());
+        snapshot.sampledView = CreateImageView(snapshot.image, sampledFormat, sampledAspect, snapshot.viewType,
+                                               snapshot.sampledBaseMipLevel, snapshot.sampledLevelCount, 0,
+                                               snapshot.arrayLayers, &sampledComponents);
+        if (snapshot.sampledView == VK_NULL_HANDLE) {
+            MGLOG_E_ONCE("SnapshotTextureForSampling: failed to create sampled view textureId=%d", texture.GetExternalIndex());
+            return false;
+        }
+
+        VkPipelineStageFlags sourceStageMask = VK_PIPELINE_STAGE_TOP_OF_PIPE_BIT;
+        VkAccessFlags sourceAccessMask = 0;
+        const VkImageLayout sourceLayout = source->layout;
+        GetImageTransitionSourceState(sourceLayout, sourceStageMask, sourceAccessMask);
+        if (!TransitionImageLayout(commandBuffer, source->image, source->layout, VK_IMAGE_LAYOUT_TRANSFER_SRC_OPTIMAL,
+                                   sourceStageMask, VK_PIPELINE_STAGE_TRANSFER_BIT, sourceAccessMask,
+                                   VK_ACCESS_TRANSFER_READ_BIT, source->aspect, 0, source->mipLevels) ||
+            !TransitionImageLayout(commandBuffer, snapshot.image, snapshot.layout, VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL,
+                                   VK_PIPELINE_STAGE_TOP_OF_PIPE_BIT, VK_PIPELINE_STAGE_TRANSFER_BIT, 0,
+                                   VK_ACCESS_TRANSFER_WRITE_BIT, snapshot.aspect, snapshot.sampledBaseMipLevel,
+                                   snapshot.sampledLevelCount)) {
+            return false;
+        }
+
+        Vector<VkImageCopy> copyRegions;
+        copyRegions.reserve(snapshot.sampledLevelCount);
+        for (Uint32 level = snapshot.sampledBaseMipLevel;
+             level < snapshot.sampledBaseMipLevel + snapshot.sampledLevelCount; ++level) {
+            VkImageCopy copy{};
+            copy.srcSubresource = {source->aspect, level, 0, source->arrayLayers};
+            copy.dstSubresource = {snapshot.aspect, level, 0, snapshot.arrayLayers};
+            copy.extent = {std::max(source->extent.width >> level, 1u),
+                           std::max(source->extent.height >> level, 1u),
+                           std::max(source->depth >> level, 1u)};
+            copyRegions.push_back(copy);
+        }
+        vkCmdCopyImage(commandBuffer, source->image, VK_IMAGE_LAYOUT_TRANSFER_SRC_OPTIMAL, snapshot.image,
+                       VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL, static_cast<Uint32>(copyRegions.size()), copyRegions.data());
+
+        if (!TransitionImageLayout(commandBuffer, snapshot.image, snapshot.layout,
+                                   ResolveSampledReadOnlyLayout(snapshot.aspect), VK_PIPELINE_STAGE_TRANSFER_BIT,
+                                   consumerShaderStageMask, VK_ACCESS_TRANSFER_WRITE_BIT,
+                                   VK_ACCESS_SHADER_READ_BIT, snapshot.aspect, snapshot.sampledBaseMipLevel,
+                                   snapshot.sampledLevelCount) ||
+            !TransitionImageLayout(commandBuffer, source->image, source->layout, sourceLayout,
+                                   VK_PIPELINE_STAGE_TRANSFER_BIT, consumerShaderStageMask,
+                                   VK_ACCESS_TRANSFER_READ_BIT, VK_ACCESS_SHADER_READ_BIT | VK_ACCESS_SHADER_WRITE_BIT,
+                                   source->aspect, 0, source->mipLevels)) {
+            return false;
+        }
+
+        StampResourceRecordingUse(*source);
+        outSnapshot = {.imageView = snapshot.sampledView, .layout = snapshot.layout};
+        DeferResourceRelease(Move(snapshot));
+        return true;
+    }
+
     void VkTextureManager::MarkStorageImageTexture(MG_State::GLState::ITextureObject& texture) {
         m_storageImageTextures.insert(MakeTextureIdentity(&texture));
     }
@@ -1342,6 +1645,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         const auto* mipTexture = MG_State::GLState::AsMipmapTexture(&texture);
         const Uint32 mipLevelCount = mipTexture != nullptr ? mipTexture->GetMipmapLevelCount() : 0u;
         return resource.syncedContentVersion != texture.GetContentVersion() ||
+               resource.syncedShapeVersion != texture.GetShapeVersion() ||
                resource.syncedTextureParamsVersion != texture.GetTextureParamsVersion() ||
                resource.syncedMipLevelCount != mipLevelCount;
     }
@@ -1441,11 +1745,16 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     Bool VkTextureManager::SyncTexture(MG_State::GLState::ITextureObject &texture,
                                        TextureResource &outResource) {
         // Cross-draw fast path: if the resource is already built and neither the texture's
-        // pixel content (bumped in MarkStorageDirty) nor its params changed since the last
-        // sync, there is nothing to re-check or re-upload - skip CheckMipmapCompleteness,
-        // SyncTextureResource, SyncTextureViews and the per-level dirty scan. Layout is
-        // maintained separately by the transition path, so the resource still reflects truth.
+        // pixel content (bumped in MarkStorageDirty), its SHAPE (bumped in BumpShapeVersion)
+        // nor its params changed since the last sync, there is nothing to re-check or
+        // re-upload - skip CheckMipmapCompleteness, SyncTextureResource, SyncTextureViews and
+        // the per-level dirty scan. Layout is maintained separately by the transition path, so
+        // the resource still reflects truth. The shape version is NOT redundant with the
+        // content one: glTexImage2D(..., nullptr) re-specifies a level's size or format
+        // without dirtying a texel, which is exactly how a re-specified image-unit texture used
+        // to keep reporting its old imageSize().
         const Uint64 syncingContentVersion = texture.GetContentVersion();
+        const Uint64 syncingShapeVersion = texture.GetShapeVersion();
         const auto* syncingMipTexture = MG_State::GLState::AsMipmapTexture(&texture);
         const Uint32 syncingMipLevelCount =
             syncingMipTexture != nullptr ? syncingMipTexture->GetMipmapLevelCount() : 0u;
@@ -1455,8 +1764,22 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         const Bool storageUpgradePending =
             !outResource.storageUsageResolved &&
             m_storageImageTextures.find(MakeTextureIdentity(&texture)) != m_storageImageTextures.end();
-        if (outResource.image != VK_NULL_HANDLE && !storageUpgradePending &&
+        // Same shape for a GL texture view's demands on the image (MUTABLE_FORMAT for a
+        // format-reinterpreting view, CUBE_COMPATIBLE for a cube view of an array texture):
+        // nothing about the texture itself changed, but the live image cannot carry the view.
+        // Masked by what this format can actually be given: MUTABLE_FORMAT is deliberately
+        // withheld from formats the driver already refused it for (see SyncTextureResource), and
+        // without this mask the "upgrade still pending" test below could never come true again -
+        // costing every later sync of that texture the whole slow path, forever.
+        VkImageCreateFlags requestedViewFlags = GetViewRequestedImageFlags(texture);
+        if (m_mutableFormatUnsupported.find(outResource.format) != m_mutableFormatUnsupported.end()) {
+            requestedViewFlags &= ~VK_IMAGE_CREATE_MUTABLE_FORMAT_BIT;
+        }
+        const Bool viewFlagUpgradePending =
+            (outResource.imageCreateFlags & requestedViewFlags) != requestedViewFlags;
+        if (outResource.image != VK_NULL_HANDLE && !storageUpgradePending && !viewFlagUpgradePending &&
             outResource.syncedContentVersion == syncingContentVersion &&
+            outResource.syncedShapeVersion == syncingShapeVersion &&
             outResource.syncedTextureParamsVersion == texture.GetTextureParamsVersion() &&
             outResource.syncedMipLevelCount == syncingMipLevelCount) {
             return true;
@@ -1477,6 +1800,12 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             return false;
         }
 
+        // From here down the size is VULKAN geometry, not GL's: a 1D array's layer count moves
+        // out of the height it occupies GL-side and into z, which is the slot
+        // TryResolveTextureShapeInfo reads arrayLayers from and the only one that leaves
+        // extent.height at the 1 a VK_IMAGE_TYPE_1D image is required to have.
+        texelSize = ToVulkanLevelExtent(texture.GetTarget(), texelSize);
+
         if (!SyncTextureResource(texture, uploadTarget, texelSize, byteSize, mipLevelCount, outResource)) {
             MGLOG_D("%s: SyncTextureResource failed", __func__);
             return false;
@@ -1508,6 +1837,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         if (!hasDirtyMipLevel) {
             outResource.syncedContentVersion = syncingContentVersion;
             outResource.syncedMipLevelCount = syncingMipLevelCount;
+            outResource.syncedShapeVersion = syncingShapeVersion;
             return true;
         }
 
@@ -1517,6 +1847,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         }
         outResource.syncedContentVersion = syncingContentVersion;
         outResource.syncedMipLevelCount = syncingMipLevelCount;
+        outResource.syncedShapeVersion = syncingShapeVersion;
         return true;
     }
 
@@ -1640,6 +1971,16 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             m_mutableFormatUnsupported.find(format) == m_mutableFormatUnsupported.end()) {
             imageCreateFlags |= VK_IMAGE_CREATE_MUTABLE_FORMAT_BIT;
         }
+        // Flags a GL texture view over this storage asked for (see NoteTextureViewImageRequirements).
+        // MUTABLE_FORMAT is still withheld from formats the driver has already refused it for, so a
+        // reinterpreting view degrades to no view rather than to no texture.
+        const VkImageCreateFlags requestedViewFlags = GetViewRequestedImageFlags(texture);
+        if (requestedViewFlags != 0) {
+            imageCreateFlags |= requestedViewFlags;
+            if (m_mutableFormatUnsupported.find(format) != m_mutableFormatUnsupported.end()) {
+                imageCreateFlags &= ~VK_IMAGE_CREATE_MUTABLE_FORMAT_BIT;
+            }
+        }
         // sRGB color images attach through their UNORM twin while GL_FRAMEBUFFER_SRGB is
         // disabled (see ResolveSrgbAttachmentWriteFormat), which needs format-reinterpreting
         // views - multisample sRGB render targets included.
@@ -1696,6 +2037,19 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                         }
                     }
                 }
+                if (rounded == 0 && (supported & VK_SAMPLE_COUNT_1_BIT) != 0) {
+                    // Nothing at two samples or above. Reachable because the frontend validates
+                    // multisample allocations against the count MobileGL ADVERTISES (GL requires
+                    // GL_MAX_SAMPLES >= 4) rather than against the device's per-format support, so
+                    // a format this device cannot multisample at all now gets here instead of
+                    // being refused up front. Keeping the unsupported count would hand
+                    // vkCreateImage an invalid VkImageCreateInfo; one sample is at least a legal
+                    // image, and the samples-08726 hazard above is the lesser of the two.
+                    MGLOG_W_ONCE("Multisample texture format %d supports no count above one on this device; "
+                                 "backing it with a single sample",
+                                 static_cast<Int>(format));
+                    rounded = static_cast<Uint32>(VK_SAMPLE_COUNT_1_BIT);
+                }
                 if (rounded != 0) {
                     resolvedSampleCount = static_cast<VkSampleCountFlagBits>(rounded);
                 }
@@ -1787,6 +2141,11 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                     viewFormats.push_back(viewFormat);
                 }
             }
+            // ...plus every format a glTextureView over this storage reinterprets it as. Those
+            // are NOT enumerable from ResolveSampledImageViewFormat - an application may name any
+            // member of the format's view class (GL 4.6 core table 8.21) - so without this the
+            // list would forbid the very view the MUTABLE_FORMAT bit was requested for.
+            AppendViewRequestedFormats(texture, viewFormats);
             formatListInfo.sType = VK_STRUCTURE_TYPE_IMAGE_FORMAT_LIST_CREATE_INFO;
             formatListInfo.viewFormatCount = static_cast<Uint32>(viewFormats.size());
             formatListInfo.pViewFormats = viewFormats.data();
@@ -1819,12 +2178,16 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             }
             if (imageFormatResult != VK_SUCCESS && !isMultisampleTexture &&
                 (imageInfo.flags & VK_IMAGE_CREATE_2D_ARRAY_COMPATIBLE_BIT) != 0) {
-                // Losing 2D-array compatibility only costs per-slice framebuffer attachment for this
-                // format; failing creation would lose the texture entirely. Remembered so later syncs
-                // neither reprobe nor flag-mismatch against this image and recreate it.
+                // Losing 2D-array compatibility only costs framebuffer attachment of this format's
+                // 3D images - per-slice AND layered, since both are spelled as a 2D-family view over
+                // the z axis; failing creation would lose the texture entirely. Recorded here (the
+                // per-format set below) so later syncs neither reprobe nor flag-mismatch against this
+                // image and recreate it, and so GetOrCreateAttachmentViewAtMipLevel declines rather
+                // than handing back a view that cannot exist - the render-pass builder then turns
+                // that decline into a skipped draw instead of a null VkImageView in pAttachments.
                 MGLOG_W_ONCE("%s: VK_IMAGE_CREATE_2D_ARRAY_COMPATIBLE_BIT is unsupported for format=%d "
-                        "textureId=%d; creating without it (per-slice framebuffer attachment will be "
-                        "unavailable for it)",
+                        "textureId=%d; creating without it (per-slice and layered framebuffer "
+                        "attachment of 3D textures in this format will be unavailable)",
                         __func__, static_cast<Int>(format), texture.GetExternalIndex());
                 m_2dArrayCompatibleUnsupported.insert(format);
                 imageInfo.flags &= ~VK_IMAGE_CREATE_2D_ARRAY_COMPATIBLE_BIT;
@@ -1841,6 +2204,13 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                         texture.GetExternalIndex(),
                         MG_Util::ConvertTextureUploadTargetToString(uploadTarget).c_str(),
                         static_cast<Int>(format), static_cast<Uint32>(imageInfo.usage));
+                // The preserved image was written by GPU work that may still be in flight
+                // (preserve requires layout != UNDEFINED); park it on the deferred ring
+                // like every other destruction path instead of letting the unique_ptr
+                // destroy it synchronously under the GPU.
+                if (preservedResource) {
+                    DeferResourceRelease(Move(*preservedResource));
+                }
                 return false;
             }
         }
@@ -1863,6 +2233,12 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                     static_cast<Int>(imageInfo.samples), static_cast<Int>(imageInfo.format));
             resource.image = VK_NULL_HANDLE;
             resource.allocation = nullptr;
+            // Same as the probe failure above: the preserved live image must go through
+            // the deferred ring, never a synchronous destructor while frames that
+            // reference it are still in flight.
+            if (preservedResource) {
+                DeferResourceRelease(Move(*preservedResource));
+            }
             return false;
         }
         ++m_textureImageEpoch; // a new attachment image invalidates cached render passes
@@ -2199,6 +2575,164 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         m_deferredViewReleases[m_currentFrameIndex].push_back(view);
     }
 
+    MG_State::GLState::ITextureObject& VkTextureManager::StorageTextureOf(
+        MG_State::GLState::ITextureObject& texture) {
+        const auto& storageOwner = texture.GetViewStorageOwner();
+        return storageOwner ? *storageOwner : texture;
+    }
+
+    // The VkImageViewType a GL texture view's own target asks for. Deliberately derived from the
+    // GL target rather than inherited from the storage image: a 2D view of a 2D-array texture is
+    // a VK_IMAGE_VIEW_TYPE_2D over one layer, and a cube view of the same image is a
+    // VK_IMAGE_VIEW_TYPE_CUBE over six - which is the whole reason table 8.20 lists those pairs.
+    static VkImageViewType ResolveTextureViewImageViewType(TextureTarget target,
+                                                           VkImageViewType storageViewType) {
+        switch (target) {
+        case TextureTarget::Texture1D:
+            return VK_IMAGE_VIEW_TYPE_1D;
+        case TextureTarget::Texture1DArray:
+            return VK_IMAGE_VIEW_TYPE_1D_ARRAY;
+        case TextureTarget::Texture2D:
+        case TextureTarget::TextureRectangle:
+        case TextureTarget::Texture2DMultisample:
+            return VK_IMAGE_VIEW_TYPE_2D;
+        case TextureTarget::Texture2DArray:
+        case TextureTarget::Texture2DMultisampleArray:
+            return VK_IMAGE_VIEW_TYPE_2D_ARRAY;
+        case TextureTarget::TextureCubeMap:
+            return VK_IMAGE_VIEW_TYPE_CUBE;
+        case TextureTarget::TextureCubeMapArray:
+            return VK_IMAGE_VIEW_TYPE_CUBE_ARRAY;
+        default:
+            return storageViewType;
+        }
+    }
+
+    VkTextureManager::TextureViewWindow VkTextureManager::ResolveTextureViewWindow(
+        MG_State::GLState::ITextureObject& texture, const TextureResource& resource) const {
+        TextureViewWindow window{};
+        window.format = resource.format;
+        window.viewType = resource.viewType;
+        window.baseArrayLayer = 0;
+        window.layerCount = resource.arrayLayers;
+        window.sampledAspect =
+            ResolveSampledImageViewAspectMask(resource.aspect, texture.GetDepthStencilTextureMode());
+        window.components = ResolveSampledViewComponents(texture, ResolveTextureFormatInfo(texture.GetFormat()));
+        ResolveViewMipRange(texture, resource.mipLevels, window.baseMipLevel, window.levelCount);
+        if (!texture.IsTextureView()) {
+            return window;
+        }
+
+        window.isTextureView = true;
+        // GL 4.6 core 8.18: the view's TEXTURE_BASE_LEVEL / TEXTURE_MAX_LEVEL are relative to the
+        // view, so ResolveViewMipRange above already clamped them against the view's own level
+        // count (TextureObjectView reports it); shifting by TEXTURE_VIEW_MIN_LEVEL puts them back
+        // into the storage image's numbering.
+        window.baseMipLevel += static_cast<Uint32>(texture.GetViewMinLevel());
+        window.baseArrayLayer = static_cast<Uint32>(texture.GetViewMinLayer());
+        window.layerCount = static_cast<Uint32>(texture.GetViewNumLayers());
+        window.viewType = ResolveTextureViewImageViewType(texture.GetTarget(), resource.viewType);
+        // The view's OWN internalformat, which may reinterpret the storage's (table 8.21).
+        const VkFormat viewFormat = ResolveTextureFormatInfo(texture.GetFormat()).format;
+        if (viewFormat != VK_FORMAT_UNDEFINED) {
+            window.format = viewFormat;
+        }
+        // Recomputed against the view's own format: a depth/stencil storage viewed as
+        // depth/stencil still has to honour the VIEW's DEPTH_STENCIL_TEXTURE_MODE, which is the
+        // one parameter Better Clouds deliberately sets differently on the two names.
+        window.sampledAspect =
+            ResolveSampledImageViewAspectMask(GetAspectMaskForFormat(window.format) != VK_IMAGE_ASPECT_NONE
+                                                  ? GetAspectMaskForFormat(window.format)
+                                                  : resource.aspect,
+                                              texture.GetDepthStencilTextureMode());
+
+        // Clamp to what the image actually has; a malformed view must degrade to an empty range
+        // rather than reach vkCreateImageView with an out-of-bounds subresource.
+        if (window.baseMipLevel >= resource.mipLevels) {
+            window.baseMipLevel = resource.mipLevels - 1;
+            window.levelCount = 1;
+        } else {
+            window.levelCount = std::min(window.levelCount, resource.mipLevels - window.baseMipLevel);
+        }
+        if (window.levelCount == 0) window.levelCount = 1;
+        if (window.baseArrayLayer >= resource.arrayLayers) {
+            window.baseArrayLayer = resource.arrayLayers - 1;
+            window.layerCount = 1;
+        } else {
+            window.layerCount = std::min(window.layerCount, resource.arrayLayers - window.baseArrayLayer);
+        }
+        if (window.layerCount == 0) window.layerCount = 1;
+        return window;
+    }
+
+    // The extra VkImageCreateFlags a GL texture view needs on the image it views. Recorded
+    // BEFORE the storage texture is synced (see SyncTextureAndGetDescriptor) so the very first
+    // resolve of a view already creates - or recreates and copies forward - an image the view can
+    // legally be built over, instead of handing back VK_NULL_HANDLE for a frame.
+    void VkTextureManager::NoteTextureViewImageRequirements(MG_State::GLState::ITextureObject& viewTexture,
+                                                            MG_State::GLState::ITextureObject& storageTexture) {
+        const TextureIdentity storageIdentity = MakeTextureIdentity(&storageTexture);
+        VkImageCreateFlags required = 0;
+        const VkFormat viewFormat = ResolveTextureFormatInfo(viewTexture.GetFormat()).format;
+        const VkFormat storageFormat = ResolveTextureFormatInfo(storageTexture.GetFormat()).format;
+        if (viewFormat != VK_FORMAT_UNDEFINED && storageFormat != VK_FORMAT_UNDEFINED &&
+            viewFormat != storageFormat) {
+            required |= VK_IMAGE_CREATE_MUTABLE_FORMAT_BIT;
+            // The image may be created with a NARROWED format list (see SyncTextureResource), and
+            // that list is a promise about every format the image will ever be viewed as. Record
+            // this one so the promise stays true.
+            m_viewRequestedFormats[storageIdentity].insert(viewFormat);
+        }
+        const TextureTarget viewTarget = viewTexture.GetTarget();
+        if (viewTarget == TextureTarget::TextureCubeMap || viewTarget == TextureTarget::TextureCubeMapArray) {
+            // Only when the storage could legally carry the bit. VK_IMAGE_CREATE_CUBE_COMPATIBLE
+            // demands a 2D image with square levels and at least six array layers
+            // (VUID-VkImageCreateInfo-flags-00954), and asking for it on a storage that has fewer
+            // would fail vkCreateImage - which, because SyncTextureResource has already released
+            // the old resource by then, would leave the PARENT texture with no image at all. A
+            // degenerate view must not be able to destroy the texture it views; let its own view
+            // creation fail instead.
+            const IntVec3 storageSize = storageTexture.GetBaseSize();
+            const Bool storageCanBeCube = storageSize.x() == storageSize.y() &&
+                                          storageTexture.GetViewNumLayers() >= 6 &&
+                                          storageTexture.GetTarget() != TextureTarget::Texture3D;
+            if (storageCanBeCube) {
+                required |= VK_IMAGE_CREATE_CUBE_COMPATIBLE_BIT;
+            } else {
+                MGLOG_W_ONCE("Texture view %d wants a cube view of texture %d, whose storage is %dx%d with %u "
+                             "layers and cannot be cube-compatible; the view will have no image view.",
+                             viewTexture.GetExternalIndex(), storageTexture.GetExternalIndex(), storageSize.x(),
+                             storageSize.y(), storageTexture.GetViewNumLayers());
+            }
+        }
+        if (required == 0) {
+            return;
+        }
+        VkImageCreateFlags& stored = m_viewRequestedImageFlags[storageIdentity];
+        stored |= required;
+    }
+
+    VkImageCreateFlags VkTextureManager::GetViewRequestedImageFlags(
+        const MG_State::GLState::ITextureObject& storageTexture) const {
+        const auto it = m_viewRequestedImageFlags.find(
+            MakeTextureIdentity(const_cast<MG_State::GLState::ITextureObject*>(&storageTexture)));
+        return it == m_viewRequestedImageFlags.end() ? 0 : it->second;
+    }
+
+    void VkTextureManager::AppendViewRequestedFormats(const MG_State::GLState::ITextureObject& storageTexture,
+                                                      Vector<VkFormat>& outFormats) const {
+        const auto it = m_viewRequestedFormats.find(
+            MakeTextureIdentity(const_cast<MG_State::GLState::ITextureObject*>(&storageTexture)));
+        if (it == m_viewRequestedFormats.end()) {
+            return;
+        }
+        for (const VkFormat viewFormat : it->second) {
+            if (std::find(outFormats.begin(), outFormats.end(), viewFormat) == outFormats.end()) {
+                outFormats.push_back(viewFormat);
+            }
+        }
+    }
+
     Bool VkTextureManager::SyncTextureViews(const MG_State::GLState::ITextureObject& texture, TextureResource& resource) {
         MOBILEGL_ASSERT(resource.image != VK_NULL_HANDLE, "SyncTextureViews: image == VK_NULL_HANDLE");
 
@@ -2358,7 +2892,13 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 uploadItem.target = target;
                 uploadItem.level = level;
                 uploadItem.baseArrayLayer = ResolveUploadArrayLayer(target);
-                uploadItem.texelSize = texelSize;
+                // Vulkan geometry, like the image this stages into (see SyncTexture): a 1D
+                // array's layers move from y to z, where the copy loop's depthSelectsArrayLayer
+                // branch turns them into layerCount. The shadow needs no repacking to follow -
+                // one layer of a 1D array IS one row of `width` texels, so the tight-packed
+                // per-layer copy the swapped size describes reads the same bytes in the same
+                // order as the row-major level it replaces.
+                uploadItem.texelSize = ToVulkanLevelExtent(mipmapTexture.GetTarget(), texelSize);
                 uploadItem.source = source;
                 uploadItem.offset = stagingSize;
                 uploadItem.uploadByteSize = byteSize;
@@ -2396,6 +2936,23 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                             }
                             uploadItem.uploadByteSize = rectTexels * uploadItem.texelBytes;
                         }
+                        // The boxes came out of the shadow in GL coordinates, where a 1D
+                        // array's layer is the y. They have to follow texelSize across to z or
+                        // they would address rows of an image that now has exactly one, and
+                        // the staging walk would read the wrong bytes for them. Every byte
+                        // count computed above is a product of the three extents, so moving
+                        // the axes leaves all of them alone - and an OFFSET lands on a zero y,
+                        // not on the extent's one, which is why this is spelled out rather than
+                        // handed to ToVulkanLevelExtent.
+                        if (mipmapTexture.GetTarget() == TextureTarget::Texture1DArray) {
+                            uploadItem.regionLo = {uploadItem.regionLo.x(), 0, uploadItem.regionLo.y()};
+                            uploadItem.regionSize = {uploadItem.regionSize.x(), 1,
+                                                     uploadItem.regionSize.y()};
+                            for (auto& rect : uploadItem.rects) {
+                                rect.lo = {rect.lo.x(), 0, rect.lo.y()};
+                                rect.hi = {rect.hi.x(), 1, rect.hi.y()};
+                            }
+                        }
                     }
                 }
                 if (formatInfo.expandRgbToRgba) {
@@ -2595,6 +3152,32 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             packBox(dst, item.regionLo, item.regionSize);
         }
 
+        if (MG_Util::PipeStats::Enabled()) {
+            // Same shape split as Espryt's: one union box per item, or one job per rect of
+            // a refined rect list. The box/rect decision is invisible to SSIM and is what
+            // the +6 ms/frame Mali cliff of section 7.3 was, so it is counted apart from
+            // the bytes.
+            Uint64 boxEmissions = 0;
+            Uint64 rectEmissions = 0;
+            Uint64 jobs = 0;
+            for (const auto& item : uploadItems) {
+                if (item.rects.empty()) {
+                    ++boxEmissions;
+                    jobs += isCombinedDepthStencil ? 2u : 1u;
+                } else {
+                    ++rectEmissions;
+                    jobs += static_cast<Uint64>(item.rects.size());
+                }
+            }
+            MG_Util::PipeStats::AddBytes(MG_Util::PipeStats::ByteClass::StageTexture,
+                                         static_cast<Uint64>(stagingSize));
+            MG_Util::PipeStats::AddCalls(MG_Util::PipeStats::CallClass::TextureUploadEmissions,
+                                         static_cast<Uint64>(uploadItems.size()));
+            MG_Util::PipeStats::AddCalls(MG_Util::PipeStats::CallClass::TextureUploadBoxEmissions, boxEmissions);
+            MG_Util::PipeStats::AddCalls(MG_Util::PipeStats::CallClass::TextureUploadRectEmissions, rectEmissions);
+            MG_Util::PipeStats::AddCalls(MG_Util::PipeStats::CallClass::TextureUploadJobs, jobs);
+        }
+
         const VkImageAspectFlags aspectMask = GetAspectMaskForFormat(outResource.format);
         VkPipelineStageFlags uploadSrcStageMask = VK_PIPELINE_STAGE_TOP_OF_PIPE_BIT;
         VkAccessFlags uploadSrcAccessMask = 0;

MobileGL/MG_Backend/DirectVulkan/Renderer/VkTextureManager.h

@@ -10,8 +10,10 @@
 
 #include "../VkIncludes.h"
 #include <Includes.h>
+#include <MG_State/GLState/FramebufferState/FramebufferObject.h>
 #include <MG_State/GLState/TextureState/TextureObject.h>
 #include <vk_mem_alloc.h>
+#include <algorithm>
 #include <unordered_map>
 #include <unordered_set>
 
@@ -22,6 +24,102 @@ class ITextureObject;
 namespace MobileGL::MG_Backend::DirectVulkan {
 enum class SamplerNumericDomain : Uint8;
 
+// What VkFormat a GL internal format is BACKED with, and how a shadow upload has to be reshaped to
+// fit it. This is not the same question as "is there an exact VkFormat for this GL format", which is
+// what ConvertTextureInternalFormatToVkEnum answers: several GL formats have no Vulkan twin at all
+// (RGBA2, RGBA12) and several three-channel ones are deliberately widened to their four-channel twin
+// because Vulkan devices rarely support the 3-channel layouts.
+//
+// SHARED, and it must stay the only answer to that question. A renderbuffer and a texture of the
+// same GL format have to resolve to the SAME VkFormat or every blit, resolve and glCopyImageSubData
+// between them crosses a size-incompatible pair, which vkCmdCopyImage leaves undefined
+// (VUID-vkCmdCopyImage-srcImage-01548). The renderbuffer path used to carry a hand-maintained second
+// copy of this table that was missing four rows - RGBA2, RGBA4, RGB5A1 and RGBA12 - so those four
+// renderbuffer formats either got no image at all or a 16-bit-packed one facing a 32-bit texture.
+struct TextureFormatInfo {
+    VkFormat format = VK_FORMAT_UNDEFINED;
+    // The GL format has three channels and is carried in a four-channel image; a shadow upload has
+    // to be expanded, inserting `alphaBytes` after every `componentByteCount * 3` source bytes.
+    Bool expandRgbToRgba = false;
+    Uint32 componentByteCount = 0;
+    Array<Uint8, 4> alphaBytes = {0, 0, 0, 0};
+};
+
+// Callers that only need the backing VkFormat (a renderbuffer has no shadow upload to reshape) take
+// `.format` and ignore the rest.
+TextureFormatInfo ResolveTextureFormatInfo(TextureInternalFormat format);
+
+// A GL 1D-ARRAY level keeps its LAYER COUNT in the state-side HEIGHT: that is what
+// glTexImage2D(GL_TEXTURE_1D_ARRAY, width, layers) means, and the frontend records the level
+// as {width, layers, 1} (see GL_Texture.cpp's AllocateStorage and the completeness walk in
+// TextureObject.cpp, which shrinks only x down the chain). Vulkan packs it the other way: a
+// 1D array is a VK_IMAGE_TYPE_1D image whose extent.height MUST be 1 and whose layers live in
+// arrayLayers - i.e. in the slot this backend reads out of z. So every place that turns a GL
+// level size into Vulkan image geometry has to move the count across first, and every GL-space
+// sub-box that rides along with it has to move its y the same way. DirectGLES performs the
+// identical remap onto the ES 2D array it maps 1D arrays to (GetBackendUploadSize).
+//
+// Applied to nothing else: a 2D array, a cube array and a 3D texture all already carry their
+// depth/layer count in z, which is where the Vulkan side expects it.
+inline IntVec3 ToVulkanLevelExtent(TextureTarget stateTarget, const IntVec3& glTexelSize) {
+    if (stateTarget == TextureTarget::Texture1DArray) {
+        return {glTexelSize.x(), 1, glTexelSize.y()};
+    }
+    return glTexelSize;
+}
+
+// How many Vulkan array layers (or, for a 3D image, z slices) a GL framebuffer attachment spans.
+//
+// THE ONE COPY, deliberately. This used to exist twice - privately in VkRenderPassManager.cpp and
+// again in VkClearManager.cpp - and the two are not independent: the render pass builds the
+// attachment view and VkFramebufferCreateInfo::layers from one, while the CLEAR key built from the
+// other is written verbatim into VkImageSubresourceRange::layerCount when a queued glClear is
+// materialised outside a render pass (MaterializePendingClearForTexture). They are two consumers
+// of the same GL clear, so any disagreement means the same glClear produces two different pictures
+// depending only on which path happens to consume it first - and the materialise path then POPS
+// the entry, so the other one never runs. Fixing one copy and leaving the other is exactly how
+// that split gets introduced; keep them the same function.
+//
+// Two shapes make this more than `size.z()`:
+//   * GL_TEXTURE_1D_ARRAY keeps its layer count in the state-side HEIGHT (see ToVulkanLevelExtent
+//     just above), so z reads 1 and every layer above the first was silently dropped.
+//   * GL_TEXTURE_CUBE_MAP is attached layered as its REPRESENTATIVE upload target, the +X face
+//     (ResolveRepresentableFramebufferTextureUploadTarget), and one face's level size has z = 1 -
+//     but a layered cube attachment names all six faces (GL 4.6 core 9.2.8), which are the image's
+//     six array layers. A cube ARRAY needs no such arm: its representative target carries 6n in z.
+inline Uint32 ResolveAttachmentLayerCount(const MG_State::GLState::FramebufferAttachmentObject& attachment) {
+    if (!attachment.IsLayered()) {
+        return 1u;
+    }
+    const auto& texture = attachment.GetTexture();
+    const TextureTarget target = texture != nullptr ? texture->GetTarget() : TextureTarget::Unknown;
+    if (target == TextureTarget::TextureCubeMap) {
+        return 6u;
+    }
+    return static_cast<Uint32>(std::max(ToVulkanLevelExtent(target, attachment.GetSize()).z(), 1));
+}
+
+// A GL framebuffer attachment's level/layer, and a GL image unit's, are relative to the texture
+// the application NAMED. When that texture was created by glTextureView (ARB_texture_view) they
+// are relative to the VIEW, and have to be shifted into the storage image's numbering before they
+// can index a Vulkan subresource - DirectVulkan gives a view no image of its own, it shares the
+// storage texture's (VkTextureManager::StorageTextureOf).
+//
+// Apply EXACTLY ONCE, at the boundary where a GL level/layer becomes a subresource index. Every
+// GetOrCreate*View entry point below expects values that have already been through here, and so
+// does everything that reads or copies an attachment directly. Both are identity on a plain
+// texture (TEXTURE_VIEW_MIN_LEVEL / MIN_LAYER are 0 there), so the conversion is unconditional
+// and there is no second, view-only code path to keep in step.
+inline Uint32 ToStorageMipLevel(const MG_State::GLState::ITextureObject* texture, Int glLevel) {
+    const Uint32 level = static_cast<Uint32>(glLevel > 0 ? glLevel : 0);
+    return texture != nullptr ? level + static_cast<Uint32>(texture->GetViewMinLevel()) : level;
+}
+
+inline Uint32 ToStorageArrayLayer(const MG_State::GLState::ITextureObject* texture, Int glLayer) {
+    const Uint32 layer = static_cast<Uint32>(glLayer > 0 ? glLayer : 0);
+    return texture != nullptr ? layer + static_cast<Uint32>(texture->GetViewMinLayer()) : layer;
+}
+
 class VkTextureManager {
 public:
     // Monotonic epoch bumped whenever a texture VkImage is (re)created. The render-pass
@@ -120,17 +218,35 @@ public:
             }
         };
 
+        // Layer range and aspect join the key because a GL texture view (ARB_texture_view) can
+        // differ from its storage on either: the Better Clouds shape samples ONE D24S8 image
+        // through two GL names in one draw, the parent with the stencil aspect and the view with
+        // the depth aspect, and a layer-sliced view of an array texture names a sub-range of the
+        // same image. Without these two fields those views would alias each other in the cache.
         struct SampledImageViewKey {
             Uint32 baseMipLevel = 0;
             Uint32 levelCount = 1;
+            Uint32 baseArrayLayer = 0;
+            Uint32 layerCount = 1;
             VkImageViewType viewType = VK_IMAGE_VIEW_TYPE_2D;
             VkFormat format = VK_FORMAT_UNDEFINED;
+            VkImageAspectFlags aspect = VK_IMAGE_ASPECT_COLOR_BIT;
+            // GL_TEXTURE_SWIZZLE_* is per-texture state, so two views over one storage with the
+            // same window but different swizzles are different views. Baked into the key because
+            // a GL texture view's ONLY sampled view lives in this cache: unlike the storage
+            // texture's own sampledView, which SyncTextureViews rebuilds whenever the params
+            // version moves, nothing else would ever notice a swizzle change on a view.
+            Uint32 componentSwizzle = 0;
 
             Bool operator==(const SampledImageViewKey& other) const {
                 return baseMipLevel == other.baseMipLevel &&
                        levelCount == other.levelCount &&
+                       baseArrayLayer == other.baseArrayLayer &&
+                       layerCount == other.layerCount &&
                        viewType == other.viewType &&
-                       format == other.format;
+                       format == other.format &&
+                       aspect == other.aspect &&
+                       componentSwizzle == other.componentSwizzle;
             }
         };
 
@@ -138,10 +254,15 @@ public:
             SizeT operator()(const SampledImageViewKey& key) const {
                 SizeT hash = std::hash<Uint32>{}(key.baseMipLevel);
                 hash ^= std::hash<Uint32>{}(key.levelCount) + 0x9e3779b9u + (hash << 6) + (hash >> 2);
+                hash ^= std::hash<Uint32>{}(key.baseArrayLayer) + 0x9e3779b9u + (hash << 6) + (hash >> 2);
+                hash ^= std::hash<Uint32>{}(key.layerCount) + 0x9e3779b9u + (hash << 6) + (hash >> 2);
                 hash ^= std::hash<Uint32>{}(static_cast<Uint32>(key.viewType)) +
                         0x9e3779b9u + (hash << 6) + (hash >> 2);
                 hash ^= std::hash<Uint32>{}(static_cast<Uint32>(key.format)) +
                         0x9e3779b9u + (hash << 6) + (hash >> 2);
+                hash ^= std::hash<Uint32>{}(static_cast<Uint32>(key.aspect)) +
+                        0x9e3779b9u + (hash << 6) + (hash >> 2);
+                hash ^= std::hash<Uint32>{}(key.componentSwizzle) + 0x9e3779b9u + (hash << 6) + (hash >> 2);
                 return hash;
             }
         };
@@ -206,6 +327,12 @@ public:
         // as defense-in-depth: any path that grows the level set (which resizes the sampled view)
         // busts the skip even if it failed to bump the content version.
         Uint32 syncedMipLevelCount = 0;
+        // Snapshot of ITextureObject::GetShapeVersion() at the last successful sync. The content
+        // version alone does NOT cover a re-specification: glTexImage2D(..., nullptr) on an
+        // already-defined level changes its size or format and dirties no texel, so it moves the
+        // shape version and nothing else. Without this in the early-out key the image, its views
+        // and therefore imageSize() all keep answering with the texture's PREVIOUS shape.
+        Uint64 syncedShapeVersion = 0;
 
         TextureResource() = default;
         TextureResource(const TextureResource&) = delete;
@@ -237,6 +364,7 @@ public:
             std::swap(this->lastRecordingGeneration, that.lastRecordingGeneration);
             std::swap(this->syncedContentVersion, that.syncedContentVersion);
             std::swap(this->syncedMipLevelCount, that.syncedMipLevelCount);
+            std::swap(this->syncedShapeVersion, that.syncedShapeVersion);
         }
 
         void Reset() {
@@ -300,6 +428,7 @@ public:
             syncedTextureParamsVersion = 0;
             syncedContentVersion = 0;
             syncedMipLevelCount = 0;
+            syncedShapeVersion = 0;
         }
 
         ~TextureResource() {
@@ -310,6 +439,11 @@ public:
         static inline VmaAllocator s_allocator = VK_NULL_HANDLE;
     };
 
+    struct SampledTextureSnapshot {
+        VkImageView imageView = VK_NULL_HANDLE;
+        VkImageLayout layout = VK_IMAGE_LAYOUT_UNDEFINED;
+    };
+
     Bool Initialize(const InitInfo& initInfo);
     void Shutdown();
     void BeginFrame(Uint32 frameIndex);
@@ -326,6 +460,58 @@ public:
     // present-less frame-boundary drain.
     void CollectAllDeferredReleases();
 
+    // ---- GL texture views (ARB_texture_view / GL 4.6 core 8.18) ----
+    // The GL texture whose STORAGE backs the given one: itself, or - for a texture created by
+    // glTextureView - the texture it views. Every image-scoped question (which VkImage, its
+    // LAYOUT, its uploads, its extent, its usage) must be asked of this object, because a view
+    // has none of its own; only the VkImageViews differ per GL texture object. Sharing one
+    // TextureResource is not an optimisation, it is the only correct arrangement: layout is a
+    // property of the image, and VulkanRenderer caches raw pointers straight to the resource's
+    // layout field, so a second resource aliasing the same image would desynchronise the moment
+    // either of them transitioned it.
+    static MG_State::GLState::ITextureObject& StorageTextureOf(MG_State::GLState::ITextureObject& texture);
+
+    // The window a GL texture object opens onto its storage image. For a plain texture this is
+    // the resource's own full extent; for a view it is the sub-range, format and aspect
+    // glTextureView gave it. Views built from a non-default window must live in the KEYED caches
+    // (attachmentViews / alternateSampledViews), never in the per-mip vectors, which belong to
+    // the storage texture's own defaults.
+    struct TextureViewWindow {
+        Uint32 baseMipLevel = 0;
+        Uint32 levelCount = 1;
+        Uint32 baseArrayLayer = 0;
+        Uint32 layerCount = 1;
+        VkFormat format = VK_FORMAT_UNDEFINED;
+        VkImageViewType viewType = VK_IMAGE_VIEW_TYPE_2D;
+        VkImageAspectFlags sampledAspect = VK_IMAGE_ASPECT_COLOR_BIT;
+        VkComponentMapping components{VK_COMPONENT_SWIZZLE_R, VK_COMPONENT_SWIZZLE_G, VK_COMPONENT_SWIZZLE_B,
+                                      VK_COMPONENT_SWIZZLE_A};
+        Bool isTextureView = false;
+    };
+
+    // The four component swizzles packed into one value, for the sampled-view cache key.
+    static Uint32 PackComponentSwizzle(const VkComponentMapping& components) {
+        return (static_cast<Uint32>(components.r) & 0xFFu) | ((static_cast<Uint32>(components.g) & 0xFFu) << 8) |
+               ((static_cast<Uint32>(components.b) & 0xFFu) << 16) |
+               ((static_cast<Uint32>(components.a) & 0xFFu) << 24);
+    }
+    TextureViewWindow ResolveTextureViewWindow(MG_State::GLState::ITextureObject& texture,
+                                               const TextureResource& resource) const;
+    // Records what a GL texture view needs of the image it views, so the next sync of the
+    // STORAGE texture creates (or recreates and copies forward) an image the view can be built
+    // over. See m_viewRequestedImageFlags for why this is lazy rather than unconditional.
+    void NoteTextureViewImageRequirements(MG_State::GLState::ITextureObject& viewTexture,
+                                          MG_State::GLState::ITextureObject& storageTexture);
+    VkImageCreateFlags GetViewRequestedImageFlags(const MG_State::GLState::ITextureObject& storageTexture) const;
+    // Appends every format a GL texture view reinterprets this storage as, for the narrowed
+    // VkImageFormatListCreateInfo the image is created with.
+    void AppendViewRequestedFormats(const MG_State::GLState::ITextureObject& storageTexture,
+                                    Vector<VkFormat>& outFormats) const;
+    // Builds (and caches, keyed by the whole window) one sampled VkImageView over a storage
+    // image. Shared back end of every GL-texture-view sampled path.
+    VkImageView GetOrCreateWindowedSampledView(MG_State::GLState::ITextureObject& texture,
+                                               TextureResource& resource, const TextureViewWindow& window);
+
     TextureResource* SyncTextureAndGetDescriptor(
         MG_State::GLState::ITextureObject& texture);
     VkImageView GetOrCreateViewAtMipLevel(MG_State::GLState::ITextureObject& texture, Uint32 mipLevel);
@@ -343,6 +529,13 @@ public:
                                                       VkImageLayout newLayout);
     Bool TransitionTextureForSampling(VkCommandBuffer commandBuffer, MG_State::GLState::ITextureObject& texture);
     Bool TransitionTextureForStorageImage(VkCommandBuffer commandBuffer, MG_State::GLState::ITextureObject& texture);
+    // Copies the complete sampler-visible mip range into a transient sampled image. The source is
+    // restored to its prior layout, so image-store descriptors continue to name the original image.
+    // The transient ownership is tied to the current frame slot and is safe through its submission.
+    Bool SnapshotTextureForSampling(VkCommandBuffer commandBuffer, MG_State::GLState::ITextureObject& texture,
+                                    SamplerNumericDomain numericDomain,
+                                    VkPipelineStageFlags consumerShaderStageMask,
+                                    SampledTextureSnapshot& outSnapshot);
 
     // Recording-generation bookkeeping for the pre-pass command stream. The
     // generation advances every time the frame command buffer (re)begins
@@ -533,6 +726,19 @@ private:
     std::unordered_map<TextureIdentity, TextureResource, TextureIdentityHash> m_textureResources;
     // Textures that have been bound to a GL image unit (see MarkStorageImageTexture).
     std::unordered_set<TextureIdentity, TextureIdentityHash> m_storageImageTextures;
+    // Extra VkImageCreateFlags a GL texture view needs on the storage image it views, keyed by
+    // the STORAGE texture's identity. Requested lazily, exactly like STORAGE usage above and for
+    // the same reason: VK_IMAGE_CREATE_MUTABLE_FORMAT_BIT costs bandwidth compression on tilers
+    // (it is what VK_KHR_image_format_list exists to claw back), so setting it on every
+    // immutable-storage texture would tax every glTexStorage2D render target in a game for a
+    // feature almost none of them use. A SAME-format view - which is the common case, and the
+    // Better Clouds case - needs no flag at all and therefore costs nothing.
+    std::unordered_map<TextureIdentity, VkImageCreateFlags, TextureIdentityHash> m_viewRequestedImageFlags;
+    // Every VkFormat a GL texture view has asked to reinterpret this storage as. The narrowed
+    // VkImageFormatListCreateInfo the image is created with must name them: the list is a promise
+    // that NO other format will ever be viewed, and building a view outside it is
+    // VUID-VkImageViewCreateInfo-pNext-01585. Keyed, like the flags above, by the STORAGE texture.
+    std::unordered_map<TextureIdentity, std::unordered_set<VkFormat>, TextureIdentityHash> m_viewRequestedFormats;
     // Supported multisample counts per format, so repeat texture syncs do not
     // re-query vkGetPhysicalDeviceImageFormatProperties.
     std::unordered_map<VkFormat, VkSampleCountFlags> m_multisampleCountsByFormat;

MobileGL/MG_Backend/DirectVulkan/Renderer/VulkanRenderer.h

@@ -23,6 +23,8 @@
 #include "VkTimerQueryManager.h"
 #include "MG_Util/Math/VectorTypes.h"
 #include <Includes.h>
+#include <MG_Backend/BackendObject.h>
+#include <MG_Util/SelfTest/PrimitivesGeneratedNoXfbProbe.h>
 #include <vk_mem_alloc.h>
 
 #include "../VkIncludes.h"
@@ -197,9 +199,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                                   GLbitfield mask, GLenum filter);
         void CopyTexSubImage2D(GLenum target, GLint level, GLint xoffset, GLint yoffset,
                        GLint x, GLint y, GLsizei width, GLsizei height);
-        void CopyImageSubData(const SharedPtr<MG_State::GLState::ITextureObject>& srcTexture,
+        void CopyImageSubData(const CopyImageEndpoint& srcEndpoint,
                               GLenum srcTarget, GLint srcLevel, GLint srcX, GLint srcY, GLint srcZ,
-                              const SharedPtr<MG_State::GLState::ITextureObject>& dstTexture,
+                              const CopyImageEndpoint& dstEndpoint,
                               GLenum dstTarget, GLint dstLevel, GLint dstX, GLint dstY, GLint dstZ,
                               GLsizei srcWidth, GLsizei srcHeight, GLsizei srcDepth);
         void GenerateMipmap(GLenum target);
@@ -216,10 +218,15 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // depth/stencil image, which this renderer stores display-side-up: the copy rect then
         // has to be mapped out of GL's bottom-origin space and the copied rows re-oriented on
         // the way back, exactly as the colour ReadPixels path does.
+        // `sourceLayerCount` above 1 says the `height` rows the client is owed are stored as that
+        // many ARRAY LAYERS of a one-row image rather than as rows of one layer - the shape a GL
+        // 1D array has in Vulkan. The two produce byte-identical tightly-packed readbacks, so
+        // only the copy region differs; everything after it is written against `height`.
         void ReadDepthStencilImageToClient(VkImage image, VkFormat vkFormat, VkImageLayout* trackedLayout,
                                            VkImageAspectFlags imageAspect, Uint32 mipLevel, Uint32 baseArrayLayer,
                                            GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type,
-                                           void* pixels, Bool defaultFramebufferOrientation = false);
+                                           void* pixels, Bool defaultFramebufferOrientation = false,
+                                           Uint32 sourceLayerCount = 1);
         // Same-extent depth blit between images of different depth formats: host
         // round-trip with a per-texel re-encode (see BlitNamedFramebuffer).
         Bool BlitDepthAcrossFormats(FrameContext::FrameData& frame, VkImage srcImage, VkFormat srcFormat,
@@ -229,6 +236,18 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                                     GLint dstY, GLint width, GLint height, VkImageLayout srcRestoreLayout,
                                     VkImageLayout dstRestoreLayout, Bool stencilAspect);
         static SizeT GetReadbackTexelSize(VkFormat sourceFormat);
+        // Map a GL bottom-left-origin rectangle into the display-oriented swapchain image.
+        // Quarter-turn surface transforms swap the copy extent's axes.
+        static Bool MapDefaultFramebufferReadbackRect(GLint x, GLint y, GLsizei width, GLsizei height,
+                                                      VkExtent2D imageExtent,
+                                                      VkSurfaceTransformFlagBitsKHR preTransform,
+                                                      VkOffset2D* imageOffset, VkExtent2D* imageCopyExtent);
+        // Reorder a tightly packed block copied with MapDefaultFramebufferReadbackRect back into
+        // GL row order. The input block has swapped dimensions for 90/270 degree transforms.
+        static Bool RemapDefaultFramebufferReadback(const Uint8* rawPixels, Uint32 logicalWidth,
+                                                    Uint32 logicalHeight,
+                                                    VkSurfaceTransformFlagBitsKHR preTransform,
+                                                    SizeT texelSize, Uint8* outPixels);
         static Bool ConvertReadbackPixels(const Uint8* sourcePixels, VkFormat sourceFormat,
                                           GLsizei width, GLsizei height, GLenum destinationFormat,
                                           GLenum destinationType, SizeT destinationRowStride,
@@ -298,6 +317,12 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // The samplerAnisotropy device feature was granted, so GL_TEXTURE_MAX_ANISOTROPY_EXT is
         // honored rather than accepted-and-ignored.
         Bool IsSamplerAnisotropySupported() const { return m_samplerAnisotropyFeatureEnabled; }
+        // ARB_base_instance extends indirect command records with a non-zero firstInstance and
+        // requires gl_InstanceID to remain zero-based. Vulkan needs both features to honor that
+        // complete contract: one legalizes the command word, the other enables the shader rebase.
+        Bool IsNonZeroIndirectBaseInstanceSupported() const {
+            return m_drawIndirectFirstInstanceFeatureEnabled && m_shaderDrawParametersFeatureEnabled;
+        }
         // Ensures the frame command buffer is recording (same lazy pattern as
         // SetupDraw) and writes a bottom-of-pipe timestamp into the current
         // frame's pool. Null when unsupported or the pool is exhausted.
@@ -536,7 +561,20 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         Bool m_samplerAnisotropyFeatureEnabled = false;
         Bool m_shaderDrawParametersExtensionEnabled = false;
         Bool m_shaderDrawParametersFeatureEnabled = false;
+        // Native subgroup topology, queried at device creation for the compute-module
+        // subgroup repairs (SubgroupSupportPolicy.h) and the REQUIRE_FULL_SUBGROUPS
+        // stage flag; 0 / false when the device has no usable compute subgroups or
+        // MOBILEGL_MAGMA_DISABLE_SUBGROUP forced them off.
+        Uint32 m_nativeSubgroupSize = 0;
+        Bool m_nativeSubgroupSupported = false;
+        Bool m_computeFullSubgroupsFeatureEnabled = false;
+        // VkPhysicalDeviceSubgroupSizeControlProperties::maxComputeWorkgroupSubgroups;
+        // 0 when the extension (and therefore the full-subgroups flag) is unavailable.
+        Uint32 m_maxComputeWorkgroupSubgroups = 0;
         Bool m_unformattedFloatStorageImagesEnabled = false;
+        // Set only after descriptor-indexing feature AND property queries prove that
+        // update-after-bind is legal for every descriptor category this renderer emits.
+        ProgramFactory::UpdateAfterBindLimits m_updateAfterBindLimits{};
         // fillModeNonSolid gates VK_POLYGON_MODE_LINE/_POINT (glPolygonMode); independentBlend gates
         // per-draw-buffer color write masks (glColorMaski). Both are cached at device creation and
         // drive a runtime fallback when the device lacks them.
@@ -547,6 +585,30 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // needs no feature). Both cached at device creation and drive a hard-fail-at-draw when absent.
         Bool m_dualSrcBlendFeatureEnabled = false;
         Bool m_primitiveTopologyListRestartFeatureEnabled = false;
+        // shaderTessellationAndGeometryPointSize gates the PointSize built-in in a tessellation
+        // or geometry stage, which desktop GL treats as an ordinary per-vertex output (writable,
+        // and capturable by name through transform feedback). Cached at device creation and
+        // handed to ProgramFactory, which refuses a program whose tessellation or geometry module
+        // declares the matching SPIR-V capability while this is false - SetupDraw then skips its
+        // draws (VkProgramObject::pointSizeCapabilityUnsupported) rather than building a pipeline
+        // that is invalid usage.
+        Bool m_tessellationAndGeometryPointSizeFeatureEnabled = false;
+        // VK_EXT_custom_border_color. Vulkan's four predefined VkBorderColor values cover only
+        // transparent/opaque black and opaque white; GL_TEXTURE_BORDER_COLOR is an arbitrary vec4 (or
+        // an arbitrary ivec4/uvec4 through the "I" entry points). Without this extension a border
+        // colour outside the palette has to be snapped to the nearest predefined one. Both features
+        // are required together: customBorderColorWithoutFormat is what lets a sampler carry a custom
+        // colour without naming the image format it will be paired with, which GL's sampler objects
+        // cannot know. maxCustomBorderColorSamplers is a real device limit, so the sampler cache has
+        // to be able to fall back to the snapped value once it is reached.
+        Bool m_customBorderColorFeatureEnabled = false;
+        Uint32 m_maxCustomBorderColorSamplers = 0;
+        // sampleRateShading gates VkPipelineMultisampleStateCreateInfo::sampleShadingEnable, i.e.
+        // glEnable(GL_SAMPLE_SHADING) + glMinSampleShading. Unlike dualSrcBlend this does NOT
+        // hard-fail the draw when absent: sample shading is a rate hint, and every sample-rate
+        // pipeline is still correct (just not per-sample) at the default rate - so the enable is
+        // dropped and the draw proceeds, which is what a GL implementation with SAMPLES=1 does too.
+        Bool m_sampleRateShadingFeatureEnabled = false;
         // multiViewport gates rasterizing into more than one of ARB_viewport_array's 16 viewports
         // (gl_ViewportIndex). m_maxRasterizableViewports is min(MAX_VIEWPORTS, device limit), or 1
         // when the feature is off, and is the viewportCount a gl_ViewportIndex-writing pipeline
@@ -613,8 +675,18 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // per object: one group of four slots each, handed out on first use.
         static constexpr SizeT kXfbCounterObjectSlots = 16;
         VkBufferObject m_xfbCounterBuffer;
-        UnorderedMap<Uint, Uint32> m_xfbCounterSlotByObject;
-        Uint32 m_xfbNextCounterSlot = 0;
+        // Which transform feedback object owns each slot group, by the frontend's never-reused
+        // lifetime id (0 = the slot is free). This used to be an UnorderedMap keyed on the GL
+        // NAME, which is recycled by glGenTransformFeedbacks: a deleted-and-recreated object
+        // inherited the dead one's slot, and since nothing ever removed an entry the map also
+        // grew for the life of the context. A fixed table cannot do either: a group is taken over
+        // only from an owner with no OPEN span (see CurrentXfbCounterSlot), so an object whose
+        // counters can still be resumed never loses them, and a dead object's group comes back.
+        Array<Uint64, kXfbCounterObjectSlots> m_xfbCounterSlotOwner{};
+        // Tie-break among reclaimable groups only; never on its own, because the paused span the
+        // groups exist for is by construction the least recently used one.
+        Array<Uint64, kXfbCounterObjectSlots> m_xfbCounterSlotLastUse{};
+        Uint64 m_xfbCounterSlotUseSerial = 0;
         // Set for a slot once a captured draw has been recorded into its span; selects
         // counter-buffer resume on the next captured draw of the same span.
         Array<Bool, kXfbCounterObjectSlots> m_xfbCountersValid{};
@@ -656,15 +728,74 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         Vector<Uint32> m_xfbQueryActiveSlots[2];
         Bool m_xfbQuerySlotOpen = false;
         Uint32 m_xfbQueryOpenSlot = 0;
+        // GL_PRIMITIVES_GENERATED reroute for draws made while transform feedback is
+        // INACTIVE. The stream pool's primitivesNeeded is defined to count those draws
+        // too, but a Mali driver (and Mesa lavapipe) answers 0 unless a capture span
+        // is open (the CTS's tessellator-measuring shape). Where the bring-up probe
+        // finds that defect with a working control - or
+        // MOBILEGL_MAGMA_PRIMGEN_QUERY_REROUTE forces it - such draws accumulate the
+        // GENERATED count through this pool instead, whose type the arming picks:
+        // VK_QUERY_TYPE_PRIMITIVES_GENERATED_EXT where the device hosts the dedicated
+        // query with its rasterizer-discard feature (exact semantics by definition -
+        // the extension exists because GL needs this count without a capture), else a
+        // VK_QUERY_TYPE_PIPELINE_STATISTICS pool over clipping-stage invocations (one
+        // per primitive reaching primitive clipping - after every vertex processing
+        // stage, before rasterizer discard - which is the same set).
+        // XFB-ACTIVE draws keep the stream slot (exact today, and WRITTEN needs it);
+        // every draw with no open capture - a PAUSED span's draws included - takes a
+        // reroute slot, and the span then ignores the frontend's CPU paused-primitive
+        // counter rather than adding it on top (see IsPrimGenRerouteArmed): that
+        // counter is written by only 3 of the ~15 draw entry points and answers 0 for
+        // GL_PATCHES, so it cannot price the draws this reroute exists to repair. One
+        // GL query span may therefore hold slots of both pools.
+        Bool m_pipelineStatisticsQueryFeatureEnabled = false;
+        // VK_EXT_primitives_generated_query: base feature, and the
+        // ...WithRasterizerDiscard feature without which a discarding draw inside the
+        // query is invalid usage (so the reroute never picks the dedicated pool on a
+        // base-only device - GL applications toggle discard freely).
+        Bool m_primitivesGeneratedQueryFeatureEnabled = false;
+        Bool m_primitivesGeneratedQueryDiscardFeatureEnabled = false;
+        // tessellationShader was enabled at device creation (it is taken whenever the
+        // device advertises it); gates the probe's PATCHES shape.
+        Bool m_tessellationShaderFeatureEnabled = false;
+        MG_Util::SelfTest::PrimGenRerouteKind m_primGenRerouteKind =
+            MG_Util::SelfTest::PrimGenRerouteKind::None;
+        // The bring-up probe measured this device's stream query as counting draws made
+        // with no capture span open (the StreamCounts verdict) - so it counts the
+        // PAUSED-span ones too, through the stream slot they take when nothing is
+        // rerouted. Only the probe can know this, so it stays false wherever the probe
+        // is not consulted (the forced arms), which keeps those lanes' accounting as it
+        // was.
+        Bool m_primGenStreamCountsXfbInactiveDraws = false;
+        VkQueryPool m_primGenReroutePool = VK_NULL_HANDLE;
+        Uint32 m_primGenRerouteSlotCursor = 0;
+        Vector<Uint32> m_primGenRerouteActiveSlots;
+        Bool m_primGenRerouteSlotOpen = false;
+        Uint32 m_primGenRerouteOpenSlot = 0;
+        // Runs the bring-up probe (memoized per process) and decides
+        // m_primGenRerouteKind. Called at the end of device creation: it records on
+        // m_graphicsQueue, which nothing else is using yet.
+        void ArmPrimGenReroute();
 
     public:
+        // Whether a GENERATED span opened now will have the draws made while the GL
+        // span is PAUSED counted on the GPU - through the reroute pool, which takes
+        // every draw with no open capture, or (where the reroute is not armed because
+        // the stream query was measured to count capture-less draws) through the stream
+        // slot such a draw still takes. The frontend's CPU paused-primitive counter
+        // must not be added on top of either: it would double count, and it cannot
+        // price the draws that matter anyway - only 3 of the ~15 draw entry points
+        // write it and it answers 0 for GL_PATCHES. Read once per span, after
+        // StartXfbQueryCapture (whose pool creation may disarm the reroute).
+        Bool ArePausedDrawsGpuCounted() const;
         // kind: 0 = PRIMITIVES_WRITTEN, 1 = PRIMITIVES_GENERATED.
         Bool StartXfbQueryCapture(Uint32 kind);
-        void StopXfbQueryCapture(Uint32 kind, Vector<Uint32>& outSlots);
-        Bool ResolveXfbQueryResult(const Vector<Uint32>& slots, Bool wantGenerated, Uint64& outPrimitives);
+        void StopXfbQueryCapture(Uint32 kind, Vector<Uint32>& outSlots, Vector<Uint32>& outRerouteSlots);
+        Bool ResolveXfbQueryResult(const Vector<Uint32>& slots, const Vector<Uint32>& rerouteSlots,
+                                   Bool wantGenerated, Uint64& outPrimitives);
 
     private:
-        void BeginXfbQueryForDraw(VkCommandBuffer commandBuffer);
+        void BeginXfbQueryForDraw(VkCommandBuffer commandBuffer, Bool xfbActive);
         void EndXfbQueryForDraw(VkCommandBuffer commandBuffer);
 
         VkCommandPool m_commandPool = VK_NULL_HANDLE;
@@ -696,6 +827,12 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             // values the memo already holds.
             Uint64 pipelineStateHash = 0;
             ProgramFactory::CompileOptionFlags transformFlags = {};
+            // Baked into the pipeline (PipelineFactory::ComputeHash mixes it), and NOT derivable
+            // from anything else in this key: it depends on whether the draw is indexed and on the
+            // index type, neither of which the mode/program/state hashes carry. Without it an
+            // indexed and a non-indexed draw over the same program and state collide on one entry
+            // and the second one gets the first one's restart setting.
+            Bool primitiveRestartEnable = false;
             VkPipeline pipeline = VK_NULL_HANDLE;
         };
         static constexpr Uint32 kPipelineMemoSize = 8;
@@ -709,9 +846,18 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // version: the version is monotonic and bumps on every pipeline-state
         // change, so an unchanged (version, colorAttachmentCount) proves the state
         // bytes are unchanged and the hash can be reused without re-reading them.
-        Uint64 ComputePipelineStateHash(Uint32 colorAttachmentCount) const;
+        Uint64 ComputePipelineStateHash(Uint32 colorAttachmentCount,
+                                        VkSampleCountFlagBits rasterizationSamples) const;
+        // The effective GL_SAMPLE_MASK word for a draw at this rasterization sample count; see
+        // the definition for the GL-vs-Vulkan rule it reconciles. Shared by the pipeline payload
+        // and the pipeline-state memo word so the two cannot disagree.
+        Uint32 ResolveEffectiveSampleMask(VkSampleCountFlagBits rasterizationSamples) const;
         Uint m_pipelineStateHashVersion = 0;
         Uint32 m_pipelineStateHashColorCount = 0;
+        // The sample count the cached hash was computed at. A pipeline-state input now depends on
+        // it (the effective sample mask), so a draw that changes only the target's sample count
+        // has to recompute rather than reuse.
+        VkSampleCountFlagBits m_pipelineStateHashSampleCount = VK_SAMPLE_COUNT_1_BIT;
         Uint64 m_pipelineStateHash = 0;
         Bool m_pipelineStateHashValid = false;
         // GetShaderTransformFlags memo. NOT pure in the pre-transform alone: the
@@ -753,7 +899,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // Skip the per-draw CollectSampledTextures walk (~5% of the render thread) when the sampled
         // texture SET is provably unchanged from the previous draw: same program (lifetime id +
         // backend-state version, which covers sampler-uniform reassignment / relink) and transform
-        // flags, and no texture bind/unbind/delete since (GetTextureBindGeneration). On a hit,
+        // flags, no texture bind/unbind/delete since (GetTextureBindGeneration), and nothing that
+        // moves a texture's shape or a sampler's parameters since (GetSamplingResolutionGeneration
+        // - membership depends on mipmap-completeness, which both of those decide). On a hit,
         // m_sampledTexturesScratch still holds the previous draw's list and steps 2-4 (feedback /
         // layout probe / transition) re-run on it, so layout correctness is unaffected - only the GL
         // walk is skipped. The program lifetime id (never reused, unlike the GL name) and the
@@ -764,6 +912,11 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         Uint32 m_lastSampledSetProgramVersion = 0;
         ProgramFactory::CompileOptionFlags m_lastSampledSetTransformFlags = {};
         Uint64 m_lastSampledSetBindGeneration = 0;
+        Uint64 m_lastSampledSetSamplingGeneration = 0;
+        // Set from the draw's resolved VkProgramObject on both the full and the fast setup paths;
+        // read by BeginXfbCaptureForDraw, which has only GL state otherwise. See
+        // VkProgramObject::xfbCaptureDeclined.
+        Bool m_currentDrawXfbCaptureDeclined = false;
 
         // Memo for the per-draw explicit-LOD-0 eligibility probe
         // (ProgramSamplesOnlySingleLevelTextures): same key family as the
@@ -776,6 +929,10 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         Uint32 m_lastLodProgramVersion = 0;
         Uint64 m_lastLodBindGeneration = 0;
         Uint64 m_lastLodParamsSum = 0;
+        // Sampling-resolution generation at probe time. The probe reads the effective
+        // sampler's filters/aniso/LOD range, whose setters bump only this counter -
+        // the params-version sum above never moves for them.
+        Uint64 m_lastLodSamplingGeneration = 0;
         ProgramFactory::CompileOptionFlags m_lastLodBaseFlags = {};
         ProgramFactory::CompileOptionFlags m_lastLodResultFlags = {};
 
@@ -813,12 +970,23 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             Uint64 vaoLifetimeId = 0;
             Uint32 vaoConfigVersion = 0;
             const void* drawFbo = nullptr;
+            // Never-reused lifetime id beside the raw pointer + Uint16 version: a
+            // deleted FBO recycled at the same address with the same fresh version
+            // count would otherwise compare equal (same ABA as the render-pass
+            // manager's fast-path memo).
+            Uint64 drawFboLifetimeId = 0;
             Uint16 fboVersion = 0;
             Bool drawFboIsDefault = false;
             Uint renderStateVersion = 0;
             Uint64 bindGeneration = 0;
             Uint32 baseTransformFlags = 0;
             Uint32 resolvedTransformFlags = 0;
+            // What ResolvePrimitiveRestartEnable answered for the draw this snapshot was taken
+            // from, i.e. what its pipeline's primitiveRestartEnable was built with. `aspects`
+            // already separates indexed from non-indexed draws, but not one index TYPE from
+            // another, and a restart index that fits GL_UNSIGNED_INT but not GL_UNSIGNED_SHORT
+            // makes those two draws want different pipelines.
+            Bool primitiveRestartEnable = false;
             Uint64 renderPassHash = 0;
             Uint32 imageIndex = 0;
             Uint64 textureEraseEpoch = 0;
@@ -847,6 +1015,9 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             // probe the pipeline memo after a state change without re-fetching the
             // render-pass entry (the pass itself is pinned by renderPassHash above).
             Uint32 renderPassColorCount = 0;
+            // Pinned with the colour count and for the same reason: the fast path recomputes the
+            // pipeline-state value hash from the snapshot, and that hash reads the sample count.
+            VkSampleCountFlagBits renderPassSampleCount = VK_SAMPLE_COUNT_1_BIT;
             VkPipeline pipeline = VK_NULL_HANDLE;
             // layoutHash of the snapshotting draw's vertex-input state. The pipeline and
             // the vertex-input pre-flight depend on the VAO only through this (plus the
@@ -908,6 +1079,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // already sampleable.
         Vector<VkTextureManager::TextureResource*> m_sampledResourcesScratch;
         Vector<MG_State::GLState::ITextureObject*> m_storageImageTexturesScratch;
+        Vector<UniformManager::SamplerImageFeedbackBinding> m_samplerImageFeedbackScratch;
+        Vector<UniformManager::SamplerBindingOverride> m_samplerImageBindingOverridesScratch;
         Vector<VkBuffer> m_vertexBuffersScratch;
         Vector<VkDeviceSize> m_vertexOffsetsScratch;
         Vector<VkVertexInputAttributeDescription> m_patchedAttributesScratch;
@@ -1034,6 +1207,14 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             VkBuffer indexVkBuffer = VK_NULL_HANDLE;
             VkDeviceSize indexSliceOffset = 0;
             Uint64 indexFrameSerial = 0;
+            // The EBO carried a host map when the slice was recorded - the mirror of
+            // anyBufferMapped on the vertex half. A shadow-backed (non-adopted)
+            // persistent map mutates its shadow with no API call and no epoch bump, so
+            // the one-compare rescue must decline and re-run the acquire, whose
+            // SyncPersistentMappedRange is the push-down. A map taken AFTER the record
+            // is already covered: AcquirePersistentMap bumps the slice epoch for the
+            // request itself, adopted or declined.
+            Bool indexBufferMapped = false;
 
             // Bound per draw (first bindingCount elements).
             VkBuffer vkBuffers[kMaxBindings] = {};
@@ -1112,13 +1293,23 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         void CreateSwapchain();
         void CreateCommandPool();
 
+        // Whether THIS draw's primitive stream restarts, and therefore what
+        // VkPipelineInputAssemblyStateCreateInfo::primitiveRestartEnable must be. Resolved by the
+        // caller because it needs two facts a pipeline cannot see: whether the draw is indexed at
+        // all (GL primitive restart acts on the index stream, so it is a no-op for glDrawArrays),
+        // and the index TYPE (an application restart index that does not fit the type matches no
+        // index, so that draw restarts nowhere - see UploadAndBindIndexBuffer).
+        Bool ResolvePrimitiveRestartEnable(Flags<DrawSetupAspect> aspects,
+                                           const IndexBufferView* pIndexBufferView) const;
+
         VkPipeline GetOrCreatePipeline(
             GLenum mode,
             const MG_State::GLState::ProgramObject& program,
             const ProgramFactory::VkProgramObject& programObj,
             ProgramFactory::CompileOptionFlags transformFlags,
             const MG_State::GLState::VertexArrayObject& vao,
-            const RenderPassEntry& renderPassEntry);
+            const RenderPassEntry& renderPassEntry,
+            Bool primitiveRestartEnable);
         VkPipeline GetOrCreateComputePipeline(const ProgramFactory::VkProgramObject& programObj);
         void DestroyComputePipelines();
         // Takes the frame rather than a command buffer: a first-time storage-usage upgrade has to
@@ -1127,6 +1318,14 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             FrameContext::FrameData& frame,
             const MG_State::GLState::ProgramObject& program,
             const ProgramFactory::VkProgramObject& programObj);
+        // Vulkan forbids a sampled descriptor and writable storage descriptor from naming the
+        // same image subresource in one shader operation. Snapshot only the sampler side; the
+        // storage descriptor continues to name the application texture.
+        Bool PrepareSamplerImageFeedbackSnapshots(
+            FrameContext::FrameData& frame,
+            const MG_State::GLState::ProgramObject& program,
+            const ProgramFactory::VkProgramObject& programObj,
+            VkPipelineStageFlags consumerShaderStageMask);
 
         // The per-draw dynamic-state tail (viewport, scissor, blend constants, depth
         // bias, line width, stencil), gated behind one render-state-parameters-version
@@ -1177,13 +1376,25 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                                                    GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1,
                                                    GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1,
                                                    GLenum filter);
-        // Clears one z slice of a VK_IMAGE_TYPE_3D colour image. See the call site in
-        // MaterializePendingClearForTexture for why a transfer clear cannot do this.
+        // Clears one layer of a colour image through a throwaway render pass whose entire content
+        // is its LOAD_OP_CLEAR. Two callers, both of which a transfer clear cannot serve: a z
+        // slice of a VK_IMAGE_TYPE_3D image (vkCmdClearColorImage cannot name one), and a
+        // MULTISAMPLE image (which carries no TRANSFER_DST usage at all). `finalLayout` is the
+        // layout the caller already tracks for the whole image, so this never has to touch
+        // resource->layout.
         Bool ClearDepthSliceWithRenderPass(VkCommandBuffer commandBuffer,
                                            MG_State::GLState::ITextureObject& texture, Uint32 mipLevel,
-                                           Uint32 depthSlice, const VkClearValue& clearValue);
+                                           Uint32 depthSlice, const VkClearValue& clearValue,
+                                           VkImageLayout finalLayout = VK_IMAGE_LAYOUT_TRANSFER_DST_OPTIMAL);
         Bool MaterializePendingClearForTexture(VkCommandBuffer commandBuffer,
                                                MG_State::GLState::ITextureObject& texture);
+        // The multisample arm of the above. Split out rather than branched inline because it
+        // shares none of the transfer path: a multisample image carries no TRANSFER_DST usage, so
+        // neither the TRANSFER_DST transition nor vkCmdClearColorImage is legal on one.
+        Bool MaterializeMultisamplePendingClear(VkCommandBuffer commandBuffer,
+                                                MG_State::GLState::ITextureObject& texture,
+                                                VkTextureManager::TextureResource& resource,
+                                                const Vector<PendingClearEntry>& pendingClears);
         Bool MaterializePendingClearForRenderbuffer(
             VkCommandBuffer commandBuffer,
             const SharedPtr<MG_State::GLState::RenderbufferObject>& renderbuffer);

MobileGL/MG_Backend/DirectVulkan/SubgroupSupportPolicy.h

@@ -0,0 +1,63 @@
+// MobileGL - MobileGL/MG_Backend/DirectVulkan/SubgroupSupportPolicy.h
+// Copyright (c) 2026 MobileGL-Dev
+// Licensed under the GNU Lesser General Public License v3.0:
+//   https://www.gnu.org/licenses/gpl-3.0.txt
+//   https://www.gnu.org/licenses/lgpl-3.0.txt
+// SPDX-License-Identifier: LGPL-3.0-only
+// End of Source File Header
+
+#pragma once
+
+#include <Config.h>
+#include <Includes.h>
+
+namespace MobileGL::MG_Backend::DirectVulkan {
+    // The single decision point for how DirectVulkan implements GL_KHR_shader_subgroup,
+    // shared by capability advertisement (BackendObject) and module lowering
+    // (VulkanRenderer / ProgramFactory) so the two can never disagree.
+    //
+    // Native subgroups are the implementation whenever the device has them, whatever
+    // their width - subgroup operations execute on the hardware paths they were made
+    // for. Module-level repairs keep the GL contract intact around them:
+    //   - FixIterationRPSubgroupScratchPass patches the one known pack bug: iterationRP's
+    //     prefixSumCache[32], under-declared for sub-16-lane devices (8-lane lavapipe);
+    //   - FixIterationRPBarrierPass repairs Program 203's race between two reductions
+    //     reusing that scratch, when explicitly enabled;
+    //   - DeriveNumSubgroupsPass replaces the one builtin drivers get wrong
+    //     (gl_NumSubgroups) with the value the rest of the topology implies.
+    // The 32-lane shared-memory emulation (EmulateSubgroupsPass) is a LAST RESORT for
+    // devices with no subgroup support at all, and only when the user opts in with
+    // MOBILEGL_MAGMA_EMULATE_SUBGROUP=1; it never replaces available native operations.
+
+    inline constexpr Uint32 kEmulatedSubgroupSize = 32u;
+    inline constexpr Uint32 kEmulatedSubgroupStages = GL_COMPUTE_SHADER_BIT;
+    inline constexpr Uint32 kEmulatedSubgroupFeatures =
+        GL_SUBGROUP_FEATURE_BASIC_BIT_KHR | GL_SUBGROUP_FEATURE_VOTE_BIT_KHR |
+        GL_SUBGROUP_FEATURE_ARITHMETIC_BIT_KHR | GL_SUBGROUP_FEATURE_BALLOT_BIT_KHR |
+        GL_SUBGROUP_FEATURE_SHUFFLE_BIT_KHR | GL_SUBGROUP_FEATURE_SHUFFLE_RELATIVE_BIT_KHR |
+        GL_SUBGROUP_FEATURE_CLUSTERED_BIT_KHR | GL_SUBGROUP_FEATURE_QUAD_BIT_KHR;
+
+    inline Bool ShouldEmulateSubgroups(const Bool nativeSubgroupSupported) {
+        return MG_Config::Features.MagmaEmulateSubgroup && !nativeSubgroupSupported &&
+               !MG_Config::Features.MagmaDisableSubgroup;
+    }
+
+    inline Bool ShouldFixIterationRPSubgroupScratch() {
+        // Auto is ON: the patch is fingerprint-gated to iterationRP's reduction and
+        // grows one under-declared array; every other module passes through untouched.
+        return MG_Config::Features.MagmaFixIterationRPSubgroupScratch !=
+               MG_Config::QuirkOverride::ForceOff;
+    }
+
+    inline Bool ShouldFixIterationRPBarrier() {
+        return MG_Config::Features.MagmaIterationRPFixBarrier;
+    }
+
+    inline Bool ShouldDeriveNumSubgroups() {
+        // Auto is ON: gl_NumSubgroups must agree with the gl_SubgroupID range for the GL
+        // contract to hold, and the derived ceil() value is the one the renderer can pin
+        // with REQUIRE_FULL_SUBGROUPS - the driver builtin is the value with no
+        // cross-driver guarantee (Adreno returns 1 for an 8-subgroup dispatch).
+        return MG_Config::Features.MagmaDeriveNumSubgroups != MG_Config::QuirkOverride::ForceOff;
+    }
+} // namespace MobileGL::MG_Backend::DirectVulkan

MobileGL/MG_Backend/MGPipe/PipeInputs.cpp

@@ -0,0 +1,179 @@
+// MobileGL - MobileGL/MG_Backend/MGPipe/PipeInputs.cpp
+// Copyright (c) 2025-2026 MobileGL-Dev
+// Licensed under the GNU Lesser General Public License v3.0:
+//   https://www.gnu.org/licenses/gpl-3.0.txt
+//   https://www.gnu.org/licenses/lgpl-3.0.txt
+// SPDX-License-Identifier: LGPL-3.0-only
+// End of Source File Header
+
+// The backend-side half of the PipeInputs block: the poison Fatal with its verb name, the
+// name lookups the runtime knobs need, and - in a verify build - the per-field equality,
+// the entry comparator and the corruption injector. Compiled only under MOBILEGL_PIPE_PUSH
+// (CMakeLists.txt appends it to SOURCE_FILES there), so the pull build never sees it. Spells
+// no MG_State global: everything that reads the live context lives in MG_Impl/Pipe/PipeFill.cpp.
+#include <MG_Backend/MGPipe/PipeInputs.h>
+
+#include <cstdint>
+#include <cstring>
+
+namespace MobileGL::MG_Pipe {
+    const char* MGPipeVerbName(MGPipeVerb verb) {
+        const auto index = static_cast<SizeT>(verb);
+        return index < kMGPipeVerbCount ? kMGPipeVerbNames[index] : "<none>";
+    }
+
+    [[noreturn]] void MGPipeInputPoisonFatalForVerb(MGPipeInputField field, MGPipeVerb verb) {
+        MGPipeInputPoisonFatal(field, MGPipeVerbName(verb));
+    }
+
+    Optional<MGPipeInputField> MGPipeFindInputField(const char* name) {
+        if (name == nullptr) return std::nullopt;
+        for (SizeT i = 0; i < kMGPipeInputFieldCount; ++i) {
+            if (std::strcmp(kMGPipeInputFieldNames[i], name) == 0) return static_cast<MGPipeInputField>(i);
+        }
+        return std::nullopt;
+    }
+
+    Optional<MGPipeVerb> MGPipeFindVerb(const char* name) {
+        if (name == nullptr) return std::nullopt;
+        for (SizeT i = 0; i < kMGPipeVerbCount; ++i) {
+            if (std::strcmp(kMGPipeVerbNames[i], name) == 0) return static_cast<MGPipeVerb>(i);
+        }
+        return std::nullopt;
+    }
+
+#if MOBILEGL_PIPE_VERIFY
+    namespace {
+        using CurrentVertexAttributeValue = PipeInputs::CurrentVertexAttributeValue;
+
+        // Every overload is declared up front: the array overloads recurse into their element
+        // type, and a call inside a template only sees what was declared before the template.
+        template <class T>
+        Bool StorageEqual(const T& a, const T& b);
+        template <class T>
+        Bool StorageEqual(T* const& a, T* const& b);
+        template <class T>
+        Bool StorageEqual(const SharedPtr<T>& a, const SharedPtr<T>& b);
+        template <class T, SizeT N>
+        Bool StorageEqual(const T (&a)[N], const T (&b)[N]);
+        Bool StorageEqual(const PipeInputs::IndexedCapabilities& a, const PipeInputs::IndexedCapabilities& b);
+        Bool StorageEqual(const CurrentVertexAttributeValue& a, const CurrentVertexAttributeValue& b);
+        template <class T>
+        void CorruptStorage(T& v);
+        template <class T>
+        void CorruptStorage(T*& p);
+        template <class T>
+        void CorruptStorage(SharedPtr<T>& p);
+        template <class T, SizeT N>
+        void CorruptStorage(T (&a)[N]);
+        void CorruptStorage(PipeInputs::IndexedCapabilities& c);
+        void CorruptStorage(CurrentVertexAttributeValue& v);
+
+        // ---- equality over one field's storage ----
+        // O-class storage compares by identity: a raw pointer into the context, or the object a
+        // SharedPtr owns. Everything else goes through G4's MGPipeFieldEqual, recursing through
+        // C arrays element-wise.
+        template <class T>
+        Bool StorageEqual(T* const& a, T* const& b) {
+            return a == b;
+        }
+        template <class T>
+        Bool StorageEqual(const SharedPtr<T>& a, const SharedPtr<T>& b) {
+            return a.get() == b.get();
+        }
+        template <class T, SizeT N>
+        Bool StorageEqual(const T (&a)[N], const T (&b)[N]) {
+            for (SizeT i = 0; i < N; ++i) {
+                if (!StorageEqual(a[i], b[i])) return false;
+            }
+            return true;
+        }
+        Bool StorageEqual(const PipeInputs::IndexedCapabilities& a, const PipeInputs::IndexedCapabilities& b) {
+            return StorageEqual(a.Blend, b.Blend) && StorageEqual(a.ScissorTest, b.ScissorTest);
+        }
+        // Three scalar arrays and nothing else (Core.h), so a bitwise compare has no padding to
+        // false-differ on and keeps a NaN float attribute equal to itself. The size assertion is
+        // what turns a fourth member into a build break rather than a blind spot.
+        Bool StorageEqual(const CurrentVertexAttributeValue& a, const CurrentVertexAttributeValue& b) {
+            static_assert(sizeof(CurrentVertexAttributeValue) == 3 * 4 * 4,
+                          "CurrentVertexAttributeValue grew a member; update the comparator");
+            return std::memcmp(&a, &b, sizeof(CurrentVertexAttributeValue)) == 0;
+        }
+        template <class T>
+        Bool StorageEqual(const T& a, const T& b) {
+            return MGPipeFieldEqual(a, b);
+        }
+
+        // ---- corruption of one field's storage ----
+        // Every shape is perturbed in a way the comparator above must see: a Bool flips, a
+        // scalar or enum moves by one, a pointer's low bits are flipped (never dereferenced:
+        // the snapshot is only ever compared), a SharedPtr becomes an aliasing pointer to a
+        // flipped address with no control block, an array corrupts its first element, and any
+        // other struct has its first byte XOR'ed with 0x5A.
+        template <class T>
+        T* FlipPointer(T* p) {
+            return reinterpret_cast<T*>(reinterpret_cast<std::uintptr_t>(p) ^ 0x5A);
+        }
+        template <class T>
+        void CorruptStorage(T*& p) {
+            p = FlipPointer(p);
+        }
+        template <class T>
+        void CorruptStorage(SharedPtr<T>& p) {
+            p = SharedPtr<T>(SharedPtr<T>(), FlipPointer(p.get()));
+        }
+        template <class T, SizeT N>
+        void CorruptStorage(T (&a)[N]) {
+            CorruptStorage(a[0]);
+        }
+        void CorruptStorage(PipeInputs::IndexedCapabilities& c) {
+            CorruptStorage(c.Blend);
+        }
+        void CorruptStorage(CurrentVertexAttributeValue& v) {
+            v.floatValue[0] += 1.f;
+        }
+        template <class T>
+        void CorruptStorage(T& v) {
+            if constexpr (std::is_same_v<T, Bool>) {
+                v = !v;
+            } else if constexpr (std::is_enum_v<T>) {
+                v = static_cast<T>(static_cast<std::underlying_type_t<T>>(v) + 1);
+            } else if constexpr (std::is_arithmetic_v<T>) {
+                v = static_cast<T>(v + 1);
+            } else {
+                static_assert(std::is_trivially_copyable_v<T>, "PipeInputs storage must be trivially copyable");
+                unsigned char first = 0;
+                std::memcpy(&first, &v, 1);
+                first ^= 0x5A;
+                std::memcpy(&v, &first, 1);
+            }
+        }
+    } // namespace
+
+    Bool MGPipeInputsFieldEqual(MGPipeInputField field, const PipeInputs& a, const PipeInputs& b) {
+        // A forwarded field has no storage and is equal by definition; VisitStorage answers
+        // false for it, hence the explicit sticky test first.
+        if (kMGPipeInputFieldSticky[static_cast<SizeT>(field)]) return true;
+        return PipeInputs::VisitStorage(field, a, b, [](const auto& x, const auto& y) { return StorageEqual(x, y); });
+    }
+
+    Bool MGPipeVerifyInputs(const PipeInputs& pushed, const PipeInputs& snapshot, const MGPipeFieldMask& mask,
+                            MGPipeInputField* outField) {
+        for (SizeT i = 0; i < kMGPipeInputFieldCount; ++i) {
+            const auto field = static_cast<MGPipeInputField>(i);
+            if (!MGPipeFieldMaskHas(mask, field)) continue;
+            if (MGPipeInputsFieldEqual(field, pushed, snapshot)) continue;
+            if (outField != nullptr) *outField = field;
+            return false;
+        }
+        return true;
+    }
+
+    Bool MGPipeApplyVerifyCorruption(PipeInputs& snapshot, MGPipeInputField field) {
+        return PipeInputs::VisitStorage(field, snapshot, snapshot, [](auto& x, auto&) {
+            CorruptStorage(x);
+            return true;
+        });
+    }
+#endif // MOBILEGL_PIPE_VERIFY
+} // namespace MobileGL::MG_Pipe

MobileGL/MG_Backend/MGPipe/PipeInputs.h

@@ -0,0 +1,714 @@
+// MobileGL - MobileGL/MG_Backend/MGPipe/PipeInputs.h
+// Copyright (c) 2025-2026 MobileGL-Dev
+// Licensed under the GNU Lesser General Public License v3.0:
+//   https://www.gnu.org/licenses/gpl-3.0.txt
+//   https://www.gnu.org/licenses/lgpl-3.0.txt
+// SPDX-License-Identifier: LGPL-3.0-only
+// End of Source File Header
+
+#pragma once
+#include <MG_Pipe/MGPipe.h>
+// The frontend types the accessors return. Allowed here: P13 keeps this include for the
+// verify arm (ARCHITECTURE.md 9.5). This header spells no MG_State global - every read of
+// the live context happens on the client side, in MG_Impl/Pipe/PipeFill.cpp.
+#include <MG_State/GLState/Core.h>
+
+// MOBILEGL_PIPE_POISON: the per-verb generation stamps and the read-side
+// Fatal{UnmigratedPipeInput} check. Derived here, once. The repository's debug gate is
+// MOBILEGL_LOG_ACTIVE_LEVEL <= MOBILEGL_LOG_LEVEL_DEBUG (Defines.h); the verify CI build is
+// Release/INFO with MOBILEGL_BUILD_DISAGGREGATED=OFF, so the third arm is what arms the poison
+// there without dragging MG_Remote in.
+#if MOBILEGL_PIPE_PUSH && (MOBILEGL_LOG_ACTIVE_LEVEL <= MOBILEGL_LOG_LEVEL_DEBUG || MOBILEGL_BUILD_DISAGGREGATED || \
+                           MOBILEGL_PIPE_VERIFY)
+#define MOBILEGL_PIPE_POISON 1
+#else
+#define MOBILEGL_PIPE_POISON 0
+#endif
+
+namespace MobileGL::MG_Pipe {
+    // PipeInputs.cpp. The poison Fatal with the verb's name ("<none>" before the first
+    // verb): MGLOG_F + std::abort(), live at every log level on purpose - this is not
+    // MOBILEGL_ASSERT, which is inert in INFO builds.
+    [[noreturn]] void MGPipeInputPoisonFatalForVerb(MGPipeInputField field, MGPipeVerb verb);
+    // kMGPipeVerbNames[verb], or "<none>" for kVerbCount (no verb has been filled yet).
+    const char* MGPipeVerbName(MGPipeVerb verb);
+    // Name lookups for the runtime knobs (MOBILEGL_PIPE_VERIFY_CORRUPT names a field,
+    // MOBILEGL_PIPE_POISON_OMIT a Verb:Field pair). Empty on an unknown name.
+    Optional<MGPipeInputField> MGPipeFindInputField(const char* name);
+    Optional<MGPipeVerb> MGPipeFindVerb(const char* name);
+
+    // The read-side poison check, on every non-forwarded accessor. Under MOBILEGL_PIPE_POISON
+    // a read of a field whose stamp is older than the current verb serial is
+    // Fatal{UnmigratedPipeInput, "Field@Verb"}; otherwise the accessor is a plain load.
+#if MOBILEGL_PIPE_POISON
+#define MGP_INPUT_CHECK(Field)                                                                                         \
+    do {                                                                                                               \
+        if (!::MobileGL::MG_Pipe::MGPipeInputFieldIsFresh(m_filled, (Field))) {                                        \
+            ::MobileGL::MG_Pipe::MGPipeInputPoisonFatalForVerb((Field), m_currentVerb);                                \
+        }                                                                                                              \
+    } while (0)
+#else
+#define MGP_INPUT_CHECK(Field) ((void)0)
+#endif
+    // The compare-at-read hook of the MOBILEGL_PIPE_VERIFY comparator (P1 brief D8), defined
+    // in MG_Impl/Pipe/PipeFill.cpp: re-reads the field from the live context and compares it
+    // against the stored value, and reports the FIRST divergence as
+    // Fatal{PipeVerifyDiffer, "Field@Verb", verb=<serial>, where=read} (the indices go in a
+    // preceding MGLOG_E). Only the live block (gPipeInputs) is verified; a snapshot's own
+    // accessors are plain loads. Off in every other build.
+    struct PipeInputs;
+#if MOBILEGL_PIPE_VERIFY
+    void MGPipeVerifyReadHook(const PipeInputs& self, MGPipeInputField field, Uint index0, Uint index1);
+#define MGP_INPUT_VERIFY_READ(Field, Index0, Index1)                                                                   \
+    ::MobileGL::MG_Pipe::MGPipeVerifyReadHook(*this, (Field), static_cast<Uint>(Index0), static_cast<Uint>(Index1))
+#else
+#define MGP_INPUT_VERIFY_READ(Field, Index0, Index1) ((void)0)
+#endif
+
+    // The V/O storage of every field that has storage, by field id. The seven F-class
+    // (forwarded) fields have none. PipeInputs::VisitStorage dispatches on this list, which
+    // is what keeps the comparator and the corruption injector one function each instead of
+    // two sixty-way switches.
+    // clang-format off
+#define MGP_INPUT_STORAGE_LIST(X)                                                    \
+    X(GetActiveTextureUnit,                        m_activeTextureUnit)              \
+    X(GetBlendColor,                               m_blendColor)                     \
+    X(GetBlendEquationIndexed,                     m_blendEquation)                  \
+    X(GetBlendFuncIndexed,                         m_blendFunc)                      \
+    X(GetBoundTransformFeedbackName,               m_boundTransformFeedbackName)     \
+    X(GetBoundVertexArray,                         m_boundVertexArray)               \
+    X(GetBufferBindingSlot,                        m_bufferBindingSlot)              \
+    X(GetBufferBindingPoint,                       m_bufferBindingPointBase)         \
+    X(GetTouchedBufferBindingPointCount,           m_touchedBindingPointCount)       \
+    X(GetClampReadColor,                           m_clampReadColor)                 \
+    X(GetClearColor,                               m_clearColor)                     \
+    X(GetClearDepth,                               m_clearDepth)                     \
+    X(GetClearStencil,                             m_clearStencil)                   \
+    X(GetColorMaskIndexed,                         m_colorMask)                      \
+    X(GetCullFaceMode,                             m_cullFaceMode)                   \
+    X(GetCurrentVertexAttribute,                   m_currentVertexAttribute)         \
+    X(GetDepthFunc,                                m_depthFunc)                      \
+    X(GetDepthMask,                                m_depthMask)                      \
+    X(GetDepthRangeIndexed,                        m_depthRange)                     \
+    X(GetFramebufferBindingSlot,                   m_framebufferBindingSlot)         \
+    X(GetImageTextureBinding,                      m_imageTextureBindingBase)        \
+    X(GetLineWidth,                                m_lineWidth)                      \
+    X(GetLogicOp,                                  m_logicOp)                        \
+    X(GetMaxTouchedTextureUnit,                    m_maxTouchedTextureUnit)          \
+    X(GetMinSampleShadingValue,                    m_minSampleShadingValue)          \
+    X(GetPatchDefaultInnerLevel,                   m_patchDefaultInnerLevel)         \
+    X(GetPatchDefaultOuterLevel,                   m_patchDefaultOuterLevel)         \
+    X(GetPatchVertices,                            m_patchVertices)                  \
+    X(GetPipelineStateVersion,                     m_pipelineStateVersion)           \
+    X(GetPixelStoreParameters,                     m_pixelStore)                     \
+    X(GetPolygonModeFront,                         m_polygonModeFront)               \
+    X(GetPolygonOffsetFactor,                      m_polygonOffsetFactor)            \
+    X(GetPolygonOffsetUnits,                       m_polygonOffsetUnits)             \
+    X(GetPrimitiveRestartIndex,                    m_primitiveRestartIndex)          \
+    X(GetProgramForDispatch,                       m_programForDispatch)             \
+    X(GetProgramForDraw,                           m_programForDraw)                 \
+    X(GetProvokingVertexMode,                      m_provokingVertexMode)            \
+    X(GetRenderStateParameters,                    m_renderState)                    \
+    X(GetRenderStateParametersVersion,             m_renderStateParametersVersion)   \
+    X(GetSamplingResolutionGeneration,             m_samplingResolutionGeneration)   \
+    X(GetScissorBox,                               m_scissorBox)                     \
+    X(GetStencilState,                             m_stencil)                        \
+    X(GetTextureBindGeneration,                    m_textureBindGeneration)          \
+    X(GetTextureContextId,                         m_textureContextId)               \
+    X(GetTextureUnitObject,                        m_textureUnitBase)                \
+    X(GetTransformFeedbackCapturedVertices,        m_transformFeedbackCapturedVertices) \
+    X(GetTransformFeedbackGeneration,              m_transformFeedbackGeneration)    \
+    X(GetTransformFeedbackPausedPrimitiveCounter,  m_transformFeedbackPausedPrimitiveCounter) \
+    X(GetTransformFeedbackProgram,                 m_transformFeedbackProgram)       \
+    X(GetViewport,                                 m_viewport)                       \
+    X(GetViewportIndexed,                          m_viewportIndexed)                \
+    X(IsCapabilityEnabled,                         m_capability)                     \
+    X(IsCapabilityEnabledIndexed,                  m_capabilityIndexed)              \
+    X(IsTransformFeedbackActive,                   m_transformFeedbackActive)        \
+    X(IsTransformFeedbackPaused,                   m_transformFeedbackPaused)        \
+    X(GetBoundTransformFeedbackLifetimeId,         m_boundTransformFeedbackLifetimeId)
+    // clang-format on
+
+    // The seven F-class fields, for the arithmetic below and for the sticky table's proof.
+    // The forwarded set IS the sticky set (PipeFields.def marks the same seven rows F and
+    // sticky), so an eighth sticky row without a forwarder is refused here, not by a test.
+    inline constexpr SizeT kMGPipeForwardedFieldCount = 7;
+    static_assert(kMGPipeForwardedFieldCount == kMGPipeInputStickyFieldCount,
+                  "the forwarded (F-class) fields and the sticky fields of PipeFields.def are the same seven rows");
+
+    // The block the backends read instead of GLContext (ARCHITECTURE.md 9.2 phase A, P1 brief
+    // D4). One struct, three storage classes, and every accessor keeps the NAME, PARAMETERS
+    // and RETURN TYPE of its GLContext counterpart (MG_State/GLState/Core.h) so the strangler
+    // sed is type-neutral:
+    //
+    //   V (value)            copied out of GLContext at fill time by calling the same accessor;
+    //                        no derivation logic is re-implemented here, which is what keeps the
+    //                        copy semantically identical by construction.
+    //   O (object reference) a SharedPtr copy, or a raw pointer to the live GLContext-owned
+    //                        slot/array for the accessors that return a non-const reference into
+    //                        the context. Identity is what phase C turns into a handle.
+    //   F (forwarded)        argument-keyed lookups and reverse-channel calls, defined out of
+    //                        line in MG_Impl/Pipe/PipeFill.cpp (the client side, where the live
+    //                        context may be spelled). Sticky: stamped once by the first fill that
+    //                        sees a live context.
+    //
+    // Every non-forwarded accessor is MGP_INPUT_CHECK (poison) -> MGP_INPUT_VERIFY_READ
+    // (compare-at-read) -> the storage. Both macros expand to nothing when their switch is
+    // off, so a plain MOBILEGL_PIPE_PUSH build's accessor is a load.
+    struct PipeInputs {
+        using GLContext = MG_State::GLState::GLContext;
+        using BufferObject = MG_State::GLState::BufferObject;
+        using BufferTarget = ::MobileGL::BufferTarget;
+        using FramebufferObject = MG_State::GLState::FramebufferObject;
+        using FramebufferTarget = ::MobileGL::FramebufferTarget;
+        using VertexArrayObject = MG_State::GLState::VertexArrayObject;
+        using ProgramObject = MG_State::GLState::ProgramObject;
+        using ITextureObject = MG_State::GLState::ITextureObject;
+        using TextureUnit = MG_State::GLState::TextureUnit;
+        using ImageTextureBinding = MG_State::GLState::ImageTextureBinding;
+        using CurrentVertexAttributeValue = MG_State::GLState::CurrentVertexAttributeValue;
+
+        static constexpr SizeT kBufferTargetCount = static_cast<SizeT>(BufferTarget::BufferTargetCount);
+        static constexpr SizeT kFramebufferTargetCount = static_cast<SizeT>(FramebufferTarget::FramebufferTargetCount);
+        static constexpr SizeT kCapabilityCount = static_cast<SizeT>(CapabilityInput::CapabilityInputCount);
+        static constexpr SizeT kMaxViewports = RenderStateParameters::MAX_VIEWPORTS;
+        static constexpr SizeT kMaxVertexAttribs = VertexArrayObject::MAX_VERTEX_ATTRIBS;
+        static constexpr SizeT kStencilFaceCount = static_cast<SizeT>(StencilFace::StencilFaceCount);
+
+        // IsCapabilityEnabledIndexed's two indexed capabilities, the only ones GLContext keeps
+        // indexed state for (RenderState::IsCapabilityEnabledIndexed).
+        struct IndexedCapabilities {
+            Bool Blend[kMGMaxDrawBuffers];
+            Bool ScissorTest[kMaxViewports];
+        };
+
+        // ---- identity / liveness (not fields) ----
+        // Whether a live GLContext exists. Forwarded (PipeFill.cpp): under push MGB_CTX_LIVE
+        // must be true as soon as a context exists, fill or no fill, which is what today's
+        // null-context guards test.
+        Bool IsLive() const;
+        // The live GLContext's address at the last fill; serves MGB_CTX_IDENTITY.
+        const void* ContextIdentity() const { return m_contextIdentity; }
+        // The verb of the last fill, kVerbCount before the first one.
+        MGPipeVerb CurrentVerb() const { return m_currentVerb; }
+#if MOBILEGL_PIPE_POISON
+        const MGPipeFilledState& FilledState() const { return m_filled; }
+#endif
+
+        // ---- V: values ----
+        Int GetActiveTextureUnit() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetActiveTextureUnit);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetActiveTextureUnit, 0, 0);
+            return m_activeTextureUnit;
+        }
+        const FloatVec4& GetBlendColor() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetBlendColor);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetBlendColor, 0, 0);
+            return m_blendColor;
+        }
+        void GetBlendEquationIndexed(Uint index, BlendEquation& color, BlendEquation& alpha) const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetBlendEquationIndexed);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetBlendEquationIndexed, index, 0);
+            if (index >= kMGMaxDrawBuffers) {
+                MOBILEGL_ASSERT(false, "Blend equation index out of range: %u", index);
+                return;
+            }
+            color = m_blendEquation[index][0];
+            alpha = m_blendEquation[index][1];
+        }
+        void GetBlendFuncIndexed(Uint index, BlendFactor& srcRGB, BlendFactor& dstRGB, BlendFactor& srcAlpha,
+                                 BlendFactor& dstAlpha) const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetBlendFuncIndexed);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetBlendFuncIndexed, index, 0);
+            if (index >= kMGMaxDrawBuffers) {
+                MOBILEGL_ASSERT(false, "Blend func index out of range: %u", index);
+                return;
+            }
+            srcRGB = m_blendFunc[index][0];
+            dstRGB = m_blendFunc[index][1];
+            srcAlpha = m_blendFunc[index][2];
+            dstAlpha = m_blendFunc[index][3];
+        }
+        // Dead field: filled, read by no backend since the D21 XFB counter-slot rekey; kept so
+        // the vendored inventory row keeps its mapping (Coverage.def).
+        Uint GetBoundTransformFeedbackName() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetBoundTransformFeedbackName);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetBoundTransformFeedbackName, 0, 0);
+            return m_boundTransformFeedbackName;
+        }
+        SizeT GetTouchedBufferBindingPointCount(BufferTarget target) const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetTouchedBufferBindingPointCount);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetTouchedBufferBindingPointCount, static_cast<Uint>(target), 0);
+            return m_touchedBindingPointCount[static_cast<SizeT>(target)];
+        }
+        GLenum GetClampReadColor() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetClampReadColor);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetClampReadColor, 0, 0);
+            return m_clampReadColor;
+        }
+        const FloatVec4& GetClearColor() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetClearColor);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetClearColor, 0, 0);
+            return m_clearColor;
+        }
+        Float GetClearDepth() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetClearDepth);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetClearDepth, 0, 0);
+            return m_clearDepth;
+        }
+        Uint32 GetClearStencil() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetClearStencil);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetClearStencil, 0, 0);
+            return m_clearStencil;
+        }
+        BoolVec4 GetColorMaskIndexed(Uint index) const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetColorMaskIndexed);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetColorMaskIndexed, index, 0);
+            return m_colorMask[index];
+        }
+        CullFaceMode GetCullFaceMode() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetCullFaceMode);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetCullFaceMode, 0, 0);
+            return m_cullFaceMode;
+        }
+        const CurrentVertexAttributeValue& GetCurrentVertexAttribute(Uint index) const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetCurrentVertexAttribute);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetCurrentVertexAttribute, index, 0);
+            if (index >= kMaxVertexAttribs) {
+                static const CurrentVertexAttributeValue defaultValue{};
+                MGLOG_E_ONCE("PipeInputs::GetCurrentVertexAttribute: index %u is out of range", index);
+                return defaultValue;
+            }
+            return m_currentVertexAttribute[index];
+        }
+        DepthTestFunc GetDepthFunc() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetDepthFunc);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetDepthFunc, 0, 0);
+            return m_depthFunc;
+        }
+        Bool GetDepthMask() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetDepthMask);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetDepthMask, 0, 0);
+            return m_depthMask;
+        }
+        const FloatVec2& GetDepthRangeIndexed(Uint index) const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetDepthRangeIndexed);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetDepthRangeIndexed, index, 0);
+            if (index >= kMaxViewports) {
+                MOBILEGL_ASSERT(false, "Depth range index out of range: %u", index);
+                return m_depthRange[0];
+            }
+            return m_depthRange[index];
+        }
+        Float GetLineWidth() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetLineWidth);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetLineWidth, 0, 0);
+            return m_lineWidth;
+        }
+        LogicOperation GetLogicOp() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetLogicOp);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetLogicOp, 0, 0);
+            return m_logicOp;
+        }
+        Int GetMaxTouchedTextureUnit() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetMaxTouchedTextureUnit);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetMaxTouchedTextureUnit, 0, 0);
+            return m_maxTouchedTextureUnit;
+        }
+        Float GetMinSampleShadingValue() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetMinSampleShadingValue);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetMinSampleShadingValue, 0, 0);
+            return m_minSampleShadingValue;
+        }
+        const FloatVec2& GetPatchDefaultInnerLevel() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetPatchDefaultInnerLevel);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetPatchDefaultInnerLevel, 0, 0);
+            return m_patchDefaultInnerLevel;
+        }
+        const FloatVec4& GetPatchDefaultOuterLevel() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetPatchDefaultOuterLevel);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetPatchDefaultOuterLevel, 0, 0);
+            return m_patchDefaultOuterLevel;
+        }
+        Uint GetPatchVertices() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetPatchVertices);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetPatchVertices, 0, 0);
+            return m_patchVertices;
+        }
+        Uint GetPipelineStateVersion() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetPipelineStateVersion);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetPipelineStateVersion, 0, 0);
+            return m_pipelineStateVersion;
+        }
+        Uint GetRenderStateParametersVersion() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetRenderStateParametersVersion);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetRenderStateParametersVersion, 0, 0);
+            return m_renderStateParametersVersion;
+        }
+        PixelStoreParameters GetPixelStoreParameters(Bool isUnpack) const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetPixelStoreParameters);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetPixelStoreParameters, isUnpack ? 1u : 0u, 0);
+            return m_pixelStore[isUnpack ? 1 : 0];
+        }
+        GLenum GetPolygonModeFront() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetPolygonModeFront);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetPolygonModeFront, 0, 0);
+            return m_polygonModeFront;
+        }
+        Float GetPolygonOffsetFactor() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetPolygonOffsetFactor);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetPolygonOffsetFactor, 0, 0);
+            return m_polygonOffsetFactor;
+        }
+        Float GetPolygonOffsetUnits() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetPolygonOffsetUnits);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetPolygonOffsetUnits, 0, 0);
+            return m_polygonOffsetUnits;
+        }
+        Uint32 GetPrimitiveRestartIndex() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetPrimitiveRestartIndex);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetPrimitiveRestartIndex, 0, 0);
+            return m_primitiveRestartIndex;
+        }
+        ProvokingVertexMode GetProvokingVertexMode() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetProvokingVertexMode);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetProvokingVertexMode, 0, 0);
+            return m_provokingVertexMode;
+        }
+        const RenderStateParameters& GetRenderStateParameters() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetRenderStateParameters);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetRenderStateParameters, 0, 0);
+            return m_renderState;
+        }
+        Uint64 GetSamplingResolutionGeneration() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetSamplingResolutionGeneration);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetSamplingResolutionGeneration, 0, 0);
+            return m_samplingResolutionGeneration;
+        }
+        const IntVec4& GetScissorBox() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetScissorBox);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetScissorBox, 0, 0);
+            return m_scissorBox;
+        }
+        const StencilFaceState& GetStencilState(StencilFace face) const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetStencilState);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetStencilState, static_cast<Uint>(face), 0);
+            return m_stencil[face == StencilFace::Back ? 1 : 0];
+        }
+        Uint64 GetTextureBindGeneration() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetTextureBindGeneration);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetTextureBindGeneration, 0, 0);
+            return m_textureBindGeneration;
+        }
+        Uint64 GetTextureContextId() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetTextureContextId);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetTextureContextId, 0, 0);
+            return m_textureContextId;
+        }
+        Uint64 GetTransformFeedbackCapturedVertices() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetTransformFeedbackCapturedVertices);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetTransformFeedbackCapturedVertices, 0, 0);
+            return m_transformFeedbackCapturedVertices;
+        }
+        Uint64 GetTransformFeedbackGeneration() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetTransformFeedbackGeneration);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetTransformFeedbackGeneration, 0, 0);
+            return m_transformFeedbackGeneration;
+        }
+        Uint64 GetTransformFeedbackPausedPrimitiveCounter() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetTransformFeedbackPausedPrimitiveCounter);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetTransformFeedbackPausedPrimitiveCounter, 0, 0);
+            return m_transformFeedbackPausedPrimitiveCounter;
+        }
+        Uint64 GetBoundTransformFeedbackLifetimeId() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetBoundTransformFeedbackLifetimeId);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetBoundTransformFeedbackLifetimeId, 0, 0);
+            return m_boundTransformFeedbackLifetimeId;
+        }
+        IntVec4 GetViewport() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetViewport);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetViewport, 0, 0);
+            return m_viewport;
+        }
+        const FloatVec4& GetViewportIndexed(Uint index) const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetViewportIndexed);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetViewportIndexed, index, 0);
+            if (index >= kMaxViewports) {
+                MOBILEGL_ASSERT(false, "Viewport index out of range: %u", index);
+                return m_viewportIndexed[0];
+            }
+            return m_viewportIndexed[index];
+        }
+        Bool IsCapabilityEnabled(CapabilityInput cap) const {
+            MGP_INPUT_CHECK(MGPipeInputField::IsCapabilityEnabled);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::IsCapabilityEnabled, static_cast<Uint>(cap), 0);
+            const auto index = static_cast<SizeT>(cap);
+            return index < kCapabilityCount ? m_capability[index] : false;
+        }
+        // Blend and ScissorTest are the only indexed capabilities GLContext keeps; no backend
+        // asks for another (VulkanRenderer asks Blend). Any other cap is a read the fill cannot
+        // have served: Fatal{UnmigratedPipeInput} naming the field and the verb, the cap in a
+        // preceding MGLOG_E.
+        Bool IsCapabilityEnabledIndexed(CapabilityInput cap, Uint index) const {
+            MGP_INPUT_CHECK(MGPipeInputField::IsCapabilityEnabledIndexed);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::IsCapabilityEnabledIndexed, static_cast<Uint>(cap), index);
+            if (cap == CapabilityInput::Blend) {
+                return index < kMGMaxDrawBuffers ? m_capabilityIndexed.Blend[index] : false;
+            }
+            if (cap == CapabilityInput::ScissorTest) {
+                return index < kMaxViewports ? m_capabilityIndexed.ScissorTest[index] : false;
+            }
+            MGLOG_E("PipeInputs::IsCapabilityEnabledIndexed: no indexed storage for cap=%d (index=%u)",
+                    static_cast<int>(cap), index);
+            MGPipeInputPoisonFatalForVerb(MGPipeInputField::IsCapabilityEnabledIndexed, m_currentVerb);
+        }
+        Bool IsTransformFeedbackActive() const {
+            MGP_INPUT_CHECK(MGPipeInputField::IsTransformFeedbackActive);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::IsTransformFeedbackActive, 0, 0);
+            return m_transformFeedbackActive;
+        }
+        Bool IsTransformFeedbackPaused() const {
+            MGP_INPUT_CHECK(MGPipeInputField::IsTransformFeedbackPaused);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::IsTransformFeedbackPaused, 0, 0);
+            return m_transformFeedbackPaused;
+        }
+
+        // ---- O: object references ----
+        const SharedPtr<VertexArrayObject>& GetBoundVertexArray() {
+            MGP_INPUT_CHECK(MGPipeInputField::GetBoundVertexArray);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetBoundVertexArray, 0, 0);
+            return m_boundVertexArray;
+        }
+        // A target the fill left null (one outside GlobalBufferTargets / BufferBindPointTargets,
+        // or a read before any fill) is a read the fill cannot have served: the poison Fatal,
+        // the target in a preceding MGLOG_E.
+        BindingSlot<BufferObject>& GetBufferBindingSlot(BufferTarget target) {
+            MGP_INPUT_CHECK(MGPipeInputField::GetBufferBindingSlot);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetBufferBindingSlot, static_cast<Uint>(target), 0);
+            const auto index = static_cast<SizeT>(target);
+            if (index >= kBufferTargetCount || m_bufferBindingSlot[index] == nullptr) {
+                MGLOG_E("PipeInputs::GetBufferBindingSlot: no slot for target=%d", static_cast<int>(target));
+                MGPipeInputPoisonFatalForVerb(MGPipeInputField::GetBufferBindingSlot, m_currentVerb);
+            }
+            return *m_bufferBindingSlot[index];
+        }
+        BindingSlotRange1D<BufferObject>& GetBufferBindingPoint(BufferTarget target, Uint index) {
+            MGP_INPUT_CHECK(MGPipeInputField::GetBufferBindingPoint);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetBufferBindingPoint, static_cast<Uint>(target), index);
+            const auto targetIndex = static_cast<SizeT>(target);
+            if (targetIndex >= kBufferTargetCount || m_bufferBindingPointBase[targetIndex] == nullptr) {
+                MGLOG_E("PipeInputs::GetBufferBindingPoint: no binding points for target=%d (index=%u)",
+                        static_cast<int>(target), index);
+                MGPipeInputPoisonFatalForVerb(MGPipeInputField::GetBufferBindingPoint, m_currentVerb);
+            }
+            // The live storage is Array<Array<BindingSlotRange1D, BufferBindingPointCount>, N>
+            // (BufferState.h), so base[index] is the live slot GLContext would hand out.
+            return m_bufferBindingPointBase[targetIndex][index];
+        }
+        BindingSlot<FramebufferObject>& GetFramebufferBindingSlot(FramebufferTarget target) {
+            MGP_INPUT_CHECK(MGPipeInputField::GetFramebufferBindingSlot);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetFramebufferBindingSlot, static_cast<Uint>(target), 0);
+            const auto index = static_cast<SizeT>(target);
+            if (index >= kFramebufferTargetCount || m_framebufferBindingSlot[index] == nullptr) {
+                MGLOG_E("PipeInputs::GetFramebufferBindingSlot: no slot for target=%d", static_cast<int>(target));
+                MGPipeInputPoisonFatalForVerb(MGPipeInputField::GetFramebufferBindingSlot, m_currentVerb);
+            }
+            return *m_framebufferBindingSlot[index];
+        }
+        ImageTextureBinding& GetImageTextureBinding(Int unit) {
+            MGP_INPUT_CHECK(MGPipeInputField::GetImageTextureBinding);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetImageTextureBinding, static_cast<Uint>(unit), 0);
+            if (m_imageTextureBindingBase == nullptr) {
+                MGPipeInputPoisonFatalForVerb(MGPipeInputField::GetImageTextureBinding, m_currentVerb);
+            }
+            return m_imageTextureBindingBase[unit];
+        }
+        const ImageTextureBinding& GetImageTextureBinding(Int unit) const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetImageTextureBinding);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetImageTextureBinding, static_cast<Uint>(unit), 0);
+            if (m_imageTextureBindingBase == nullptr) {
+                MGPipeInputPoisonFatalForVerb(MGPipeInputField::GetImageTextureBinding, m_currentVerb);
+            }
+            return m_imageTextureBindingBase[unit];
+        }
+        const SharedPtr<ProgramObject>& GetProgramForDispatch() {
+            MGP_INPUT_CHECK(MGPipeInputField::GetProgramForDispatch);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetProgramForDispatch, 0, 0);
+            return m_programForDispatch;
+        }
+        const SharedPtr<ProgramObject>& GetProgramForDraw() {
+            MGP_INPUT_CHECK(MGPipeInputField::GetProgramForDraw);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetProgramForDraw, 0, 0);
+            return m_programForDraw;
+        }
+        const SharedPtr<ProgramObject>& GetTransformFeedbackProgram() const {
+            MGP_INPUT_CHECK(MGPipeInputField::GetTransformFeedbackProgram);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetTransformFeedbackProgram, 0, 0);
+            return m_transformFeedbackProgram;
+        }
+        TextureUnit& GetTextureUnitObject(Int unit) {
+            MGP_INPUT_CHECK(MGPipeInputField::GetTextureUnitObject);
+            MGP_INPUT_VERIFY_READ(MGPipeInputField::GetTextureUnitObject, static_cast<Uint>(unit), 0);
+            if (m_textureUnitBase == nullptr) {
+                MGPipeInputPoisonFatalForVerb(MGPipeInputField::GetTextureUnitObject, m_currentVerb);
+            }
+            return m_textureUnitBase[unit];
+        }
+
+        // ---- F: forwarded to the live context (MG_Impl/Pipe/PipeFill.cpp); sticky ----
+        // Each takes an argument that is not verb state - a GL name, a lifetime id, a target -
+        // i.e. it is a lookup or a reverse-channel write, not a state read; there is no value
+        // the filler could copy and no verb whose fill could make it stale. Phase C replaces
+        // them with handle tables and callbacks.
+        // They carry no MGP_INPUT_CHECK / MGP_INPUT_VERIFY_READ (the declared exception to
+        // P1 brief D4's "every accessor body"): a forward is a live call, not a stored value,
+        // and InvalidateCompileEnv is reached from backend initialisation before any verb has
+        // filled, where a check would be Fatal{...@<none>} on every start. Their sticky stamp
+        // is therefore consulted by no accessor; the tests pin it through
+        // MGPipeInputFieldIsFresh directly.
+        SizeT GetBufferBindingPointCount(BufferTarget target) const;
+        const SharedPtr<ProgramObject>& GetProgramObject(Uint index);
+        const SharedPtr<ITextureObject>& GetTextureObject(Uint index);
+        Bool HasOpenTransformFeedbackSpan(Uint64 lifetimeId) const;
+        void InvalidateCompileEnv();
+        Bool ValidateProgramName(Uint index) const;
+        // Dropped with an MGLOG_E_ONCE when no context is live; today's guarded sites never
+        // reach it without one.
+        void RecordError(ErrorCode code, UniquePtr<ErrorInfo> info);
+
+        // ---- the storage visitor ----
+        // Calls fn(a.<member>, b.<member>) for the field's storage and returns its result; returns
+        // false without calling fn for a forwarded field, which has none. The comparator's
+        // per-field equality and the verify corruption injector are both one call of this.
+        template <class Fn>
+        static Bool VisitStorage(MGPipeInputField field, PipeInputs& a, PipeInputs& b, Fn&& fn) {
+            switch (field) {
+#define MGP_INPUT_VISIT(Field, Member)                                                                                 \
+    case MGPipeInputField::Field:                                                                                      \
+        return fn(a.Member, b.Member);
+                MGP_INPUT_STORAGE_LIST(MGP_INPUT_VISIT)
+#undef MGP_INPUT_VISIT
+            default:
+                return false;
+            }
+        }
+        template <class Fn>
+        static Bool VisitStorage(MGPipeInputField field, const PipeInputs& a, const PipeInputs& b, Fn&& fn) {
+            switch (field) {
+#define MGP_INPUT_VISIT(Field, Member)                                                                                 \
+    case MGPipeInputField::Field:                                                                                      \
+        return fn(a.Member, b.Member);
+                MGP_INPUT_STORAGE_LIST(MGP_INPUT_VISIT)
+#undef MGP_INPUT_VISIT
+            default:
+                return false;
+            }
+        }
+
+    private:
+        // The one door into the storage from the client side (MG_Impl/Pipe/PipeFill.cpp):
+        // the filler's per-field copies and stamps, and the verify snapshot.
+        friend struct MGPipeFillAccess;
+
+        // ---- identity ----
+        const void* m_contextIdentity = nullptr;
+        Bool m_live = false;
+        MGPipeVerb m_currentVerb = MGPipeVerb::kVerbCount;
+#if MOBILEGL_PIPE_POISON
+        MGPipeFilledState m_filled{};
+#endif
+
+        // ---- V ----
+        Int m_activeTextureUnit = 0;
+        FloatVec4 m_blendColor{};
+        BlendEquation m_blendEquation[kMGMaxDrawBuffers][2]{};
+        BlendFactor m_blendFunc[kMGMaxDrawBuffers][4]{};
+        Uint m_boundTransformFeedbackName = 0;
+        SizeT m_touchedBindingPointCount[kBufferTargetCount]{};
+        GLenum m_clampReadColor = 0;
+        FloatVec4 m_clearColor{};
+        Float m_clearDepth = 0.f;
+        Uint32 m_clearStencil = 0;
+        BoolVec4 m_colorMask[kMGMaxDrawBuffers]{};
+        CullFaceMode m_cullFaceMode{};
+        CurrentVertexAttributeValue m_currentVertexAttribute[kMaxVertexAttribs]{};
+        DepthTestFunc m_depthFunc{};
+        Bool m_depthMask = false;
+        FloatVec2 m_depthRange[kMaxViewports]{};
+        Float m_lineWidth = 0.f;
+        LogicOperation m_logicOp{};
+        Int m_maxTouchedTextureUnit = -1;
+        Float m_minSampleShadingValue = 0.f;
+        FloatVec2 m_patchDefaultInnerLevel{};
+        FloatVec4 m_patchDefaultOuterLevel{};
+        Uint m_patchVertices = 0;
+        Uint m_pipelineStateVersion = 0;
+        Uint m_renderStateParametersVersion = 0;
+        PixelStoreParameters m_pixelStore[2]{}; // [0] = pack, [1] = unpack
+        GLenum m_polygonModeFront = 0;
+        Float m_polygonOffsetFactor = 0.f;
+        Float m_polygonOffsetUnits = 0.f;
+        Uint32 m_primitiveRestartIndex = 0;
+        ProvokingVertexMode m_provokingVertexMode{};
+        RenderStateParameters m_renderState{};
+        Uint64 m_samplingResolutionGeneration = 0;
+        Uint64 m_textureBindGeneration = 0;
+        Uint64 m_textureContextId = 0;
+        IntVec4 m_scissorBox{};
+        StencilFaceState m_stencil[kStencilFaceCount]{};
+        Uint64 m_transformFeedbackCapturedVertices = 0;
+        Uint64 m_transformFeedbackGeneration = 0;
+        Uint64 m_transformFeedbackPausedPrimitiveCounter = 0;
+        Uint64 m_boundTransformFeedbackLifetimeId = 0;
+        IntVec4 m_viewport{};
+        FloatVec4 m_viewportIndexed[kMaxViewports]{};
+        Bool m_capability[kCapabilityCount]{};
+        IndexedCapabilities m_capabilityIndexed{};
+        Bool m_transformFeedbackActive = false;
+        Bool m_transformFeedbackPaused = false;
+
+        // ---- O ----
+        SharedPtr<VertexArrayObject> m_boundVertexArray;
+        BindingSlot<BufferObject>* m_bufferBindingSlot[kBufferTargetCount]{};
+        BindingSlotRange1D<BufferObject>* m_bufferBindingPointBase[kBufferTargetCount]{};
+        BindingSlot<FramebufferObject>* m_framebufferBindingSlot[kFramebufferTargetCount]{};
+        ImageTextureBinding* m_imageTextureBindingBase = nullptr;
+        SharedPtr<ProgramObject> m_programForDispatch;
+        SharedPtr<ProgramObject> m_programForDraw;
+        SharedPtr<ProgramObject> m_transformFeedbackProgram;
+        TextureUnit* m_textureUnitBase = nullptr;
+    };
+
+    // The single global the backends read through MGB_CTX (ARCHITECTURE.md 9.2). An inline
+    // variable: no .cpp is needed for the definition.
+    inline PipeInputs gPipeInputs{};
+
+    // Every field has storage or is forwarded, and nothing else.
+#define MGP_INPUT_COUNT_ONE(Field, Member) +1
+    static_assert(0 MGP_INPUT_STORAGE_LIST(MGP_INPUT_COUNT_ONE) + kMGPipeForwardedFieldCount == kMGPipeInputFieldCount,
+                  "MGP_INPUT_STORAGE_LIST plus the seven forwarded fields is not the PipeInputs field set");
+#undef MGP_INPUT_COUNT_ONE
+    // The docs budget ~20 KB; the block is a few KB.
+    static_assert(sizeof(PipeInputs) < 20 * 1024, "PipeInputs outgrew its budget");
+
+#if MOBILEGL_PIPE_VERIFY
+    // PipeInputs.cpp. Per-field equality for the entry compare (P1 brief D8): V by value
+    // through G4's MGPipeFieldEqual (bitwise floats, field-wise structs), O by identity, F
+    // always equal (no storage).
+    Bool MGPipeInputsFieldEqual(MGPipeInputField field, const PipeInputs& a, const PipeInputs& b);
+    // PipeInputs.cpp. The entry compare: every field in `mask` of the pushed block against the
+    // snapshot, first differing field out. Exported from the shared library on purpose - the
+    // retrace-verify CI job proves it swapped in a verify build by finding this symbol with
+    // nm -D, so a "green" run against a library without the comparator cannot happen.
+#if defined(__GNUC__) || defined(__clang__)
+    __attribute__((visibility("default")))
+#endif
+    Bool MGPipeVerifyInputs(const PipeInputs& pushed, const PipeInputs& snapshot, const MGPipeFieldMask& mask,
+                            MGPipeInputField* outField);
+    // PipeInputs.cpp. Negative control A: perturbs one field's storage (flip a Bool, +1 a
+    // scalar, ^0x5A the first byte of a struct, flip a pointer's low bits - never
+    // dereferenced, the snapshot is only ever compared). Returns false for a forwarded field,
+    // which has nothing to corrupt.
+    Bool MGPipeApplyVerifyCorruption(PipeInputs& snapshot, MGPipeInputField field);
+#endif
+} // namespace MobileGL::MG_Pipe

MobileGL/MG_Benchmark/CMakeLists.txt

@@ -43,4 +43,6 @@ set_tests_properties(SanityBench PROPERTIES LABELS benchmark)
 add_subdirectory(Program)
 add_subdirectory(Buffer)
 add_subdirectory(Driver)
-add_subdirectory(Container)
+add_subdirectory(Container)
+add_subdirectory(ShaderCache)
+add_subdirectory(Transpile)

MobileGL/MG_Benchmark/ShaderCache/CMakeLists.txt

@@ -0,0 +1,21 @@
+cmake_minimum_required(VERSION 3.24)
+
+add_executable(
+        TranslationCacheBench
+        TranslationCacheBench.cpp
+)
+
+target_include_directories(TranslationCacheBench PRIVATE
+        ${MGL_ROOT}/include
+        ${MGL_ROOT}/MobileGL
+        ${MGL_ROOT}/3rdparty/SPIRV-Reflect
+)
+
+target_link_libraries(
+        TranslationCacheBench PRIVATE
+        benchmark::benchmark
+        ${LINK_LIBRARIES}
+)
+
+add_test(NAME TranslationCacheBench COMMAND TranslationCacheBench --benchmark_counters_tabular=true)
+set_tests_properties(TranslationCacheBench PROPERTIES LABELS benchmark)

MobileGL/MG_Benchmark/ShaderCache/TranslationCacheBench.cpp

@@ -0,0 +1,457 @@
+// MobileGL - MobileGL/MG_Benchmark/ShaderCache/TranslationCacheBench.cpp
+// Copyright (c) 2025-2026 MobileGL-Dev
+// Licensed under the GNU Lesser General Public License v3.0:
+//   https://www.gnu.org/licenses/gpl-3.0.txt
+//   https://www.gnu.org/licenses/lgpl-3.0.txt
+// SPDX-License-Identifier: LGPL-3.0-only
+// End of Source File Header
+
+// What the two-level shader translation memo is worth, measured on the workload that
+// motivated it: the KHR-GL33.texture_swizzle.smoke_* shape, where one case builds 2592
+// programs out of a handful of distinct sources.
+//
+// Four pairs of cases, each Off/On:
+//
+//   ProgramLink      - the whole glCompileShader + glLinkProgram path for one program, with
+//                      FRESH SHADER OBJECTS every iteration. This is the CTS shape exactly,
+//                      and it is the headline case now. It used to be the PESSIMISTIC one:
+//                      a hit still paid for both glslang parses, because the parse happens
+//                      at glCompileShader - a different entry point from the one L1
+//                      memoizes - and fresh shader objects meant ShaderCompileAdoptionMap
+//                      could not hand the earlier parse over either. L1c is what closed
+//                      that: the compile half of the memo recognises each stage's source
+//                      and publishes its verdict without parsing, so on a hit this case now
+//                      constructs no glslang object at all.
+//
+//   SharedShaderLink - the same program population with the shader objects KEPT ALIVE, so
+//                      the parses happen once outside the measured loop whatever the cache
+//                      does. That makes it the CONTROL for L1c rather than a target: its
+//                      numbers should not move, and if they do, L1c has added cost to a
+//                      path it was supposed to leave alone.
+//
+//   DeferredParseLink - the shape where L1c could LOSE: a constant vertex source (which
+//                      hits L1c and therefore skips its parse) against a fresh fragment
+//                      source every iteration (which makes the PROGRAM key miss, so the
+//                      skipped parse has to happen inside the link after all). Same parse
+//                      count either way, so the pair should land within noise; see its own
+//                      header below.
+//
+//   EsslTranspile    - the DirectGLES backend segment: the SPIR-V pass chain plus
+//                      SPIRV-Cross. Runs the driver-INDEPENDENT half of the real chain (the
+//                      passes SyncToBackend runs unconditionally, plus the two stage-gated
+//                      ones a fragment module reaches) so the miss path costs what
+//                      production costs; the capability-gated passes need a live ES driver
+//                      and are not reachable from a benchmark process.
+//
+// Every On case runs with a warm cache: the first iteration misses and every one after it
+// hits, which is exactly the steady state of a 2592-program smoke case.
+
+#include <benchmark/benchmark.h>
+
+#include <string>
+
+#include "Config.h"
+#include "Includes.h"
+#include "Init.h"
+#include "MG_Impl/GLImpl/Program/GL_Program.h"
+#include "MG_State/GLState/Core.h"
+#include "MG_State/GLState/ProgramState/ProgramTranslationCache.h"
+#include "MG_Util/ShaderTranspiler/ShaderCompiler.h"
+#include "MG_Util/ShaderTranspiler/SpvcSession.h"
+#include "MG_Util/ShaderTranspiler/TranslationCache.h"
+#include "MG_Util/ShaderTranspiler/Types.h"
+
+using namespace MobileGL;
+using namespace MobileGL::MG_Util::ShaderTranspiler;
+
+namespace {
+    const char* kVertexSource = R"(#version 460
+layout(location = 0) in vec3 aPos;
+out vec3 vPos;
+out vec2 vUv;
+void main() {
+    vPos = aPos;
+    vUv = aPos.xy * 0.5 + 0.5;
+    gl_Position = vec4(aPos, 1.0);
+}
+)";
+
+    // Shaped after gl3cTextureSwizzleTests.cpp's template: a sampler of one type, one
+    // TEXTURE_ACCESS, one CHANNEL, and an output whose BASIC_TYPE is the only thing that
+    // varies within a case. Padded with enough real arithmetic that the translation chain
+    // is doing work rather than measuring fixed overheads.
+    // `padLines` = 0 is the honest CTS size: gl3cTextureSwizzleTests' smoke template is a
+    // handful of lines, and that is the workload the memo exists for. The padded variant is
+    // kept alongside it because a shaderpack stage is orders of magnitude bigger, and the
+    // two bracket the ratio the cache is worth in practice.
+    String SwizzleLikeFragment(const String& prefix, const int padLines) {
+        String source = "#version 460\n";
+        source += "in vec3 vPos;\n";
+        source += "in vec2 vUv;\n";
+        source += "layout(location = 0) out " + prefix + "vec4 fragColor;\n";
+        source += "uniform sampler2D uTex;\n";
+        source += "uniform vec4 uTint;\n";
+        source += "uniform mat4 uModel;\n";
+        source += "uniform float uArr[8];\n";
+        source += "void main() {\n";
+        source += "    vec4 s = texture(uTex, vUv);\n";
+        source += "    float acc = s.r;\n";
+        for (int i = 0; i < padLines; ++i) {
+            source += "    acc = acc * 1.0001 + sin(acc + " + std::to_string(i) + ".0) * cos(acc);\n";
+        }
+        source += "    for (int i = 0; i < 8; ++i) acc += uArr[i];\n";
+        source += "    vec4 p = uModel * vec4(vPos, 1.0);\n";
+        source += "    fragColor = " + prefix + "vec4((s + uTint) * acc + p);\n";
+        source += "}\n";
+        return source;
+    }
+
+    class CacheModeScope {
+    public:
+        explicit CacheModeScope(const Bool enabled)
+            : m_saved(MG_Config::Features.ShaderTranslationCache) {
+            MG_Config::Features.ShaderTranslationCache =
+                enabled ? MG_Config::QuirkOverride::ForceOn : MG_Config::QuirkOverride::ForceOff;
+        }
+        ~CacheModeScope() { MG_Config::Features.ShaderTranslationCache = m_saved; }
+
+    private:
+        const MG_Config::QuirkOverride m_saved;
+    };
+
+    class SyncCompileScope {
+    public:
+        SyncCompileScope() : m_saved(MG_Config::Features.AsyncShaderCompile) {
+            MG_Config::Features.AsyncShaderCompile = MG_Config::QuirkOverride::ForceOff;
+        }
+        ~SyncCompileScope() { MG_Config::Features.AsyncShaderCompile = m_saved; }
+
+    private:
+        const MG_Config::QuirkOverride m_saved;
+    };
+
+    // One program, built the way the CTS builds one: fresh shader objects every time.
+    void LinkOneProgram(const String& vertexSource, const String& fragmentSource) {
+        using namespace MG_Impl::GLImpl;
+        const GLuint vs = CreateShader(GL_VERTEX_SHADER);
+        const char* vsText = vertexSource.c_str();
+        ShaderSource(vs, 1, &vsText, nullptr);
+        CompileShader(vs);
+
+        const GLuint fs = CreateShader(GL_FRAGMENT_SHADER);
+        const char* fsText = fragmentSource.c_str();
+        ShaderSource(fs, 1, &fsText, nullptr);
+        CompileShader(fs);
+
+        const GLuint program = CreateProgram();
+        AttachShader(program, vs);
+        AttachShader(program, fs);
+        LinkProgram(program);
+        benchmark::DoNotOptimize(program);
+
+        DeleteProgram(program);
+        DeleteShader(vs);
+        DeleteShader(fs);
+    }
+
+    Vector<Uint32> BuildSanitizedFragmentSpirv(const String& fragmentSource) {
+        ShaderAttrib attrib{.shaderType = GL_FRAGMENT_SHADER, .sourceStr = fragmentSource};
+        auto shader = ShaderCompiler::CompileShader(attrib);
+        if (!shader) return {};
+        ProgramAttrib programAttrib{.shaders = {shader.value()}};
+        auto program = ShaderCompiler::LinkProgram(programAttrib);
+        if (!program) return {};
+        ProgramBinaryAttrib binaryAttrib{.shaderTypes = {GL_FRAGMENT_SHADER}, .program = *program.value()};
+        auto binary = ShaderCompiler::GetSpirvBinaryFromProgram(binaryAttrib);
+        if (!binary || binary->empty()) return {};
+        Vector<Uint32> sanitized;
+        if (!ShaderCompiler::SanitizeAndOptimizeBinary(binary->front(), sanitized)) return {};
+        return sanitized;
+    }
+
+    // The driver-independent part of BackendProgramObjectImpl::TranspileSpirvToEssl, in the
+    // same order. What is missing is only the capability-gated passes (viewport lowering,
+    // multisample clamping, noperspective emulation, the image-format bake), which cannot
+    // fire without a live ES driver to arm them.
+    Bool TranspileLikeDirectGles(const Vector<Uint32>& spirv, const Uint esslVersion, String& outEssl) {
+        Vector<Uint32> a;
+        const Vector<Uint32>* effective = &spirv;
+        if (ShaderCompiler::StripUboMemberRelaxedPrecisionForEssl(*effective, a, false) && !a.empty()) {
+            effective = &a;
+        }
+        Vector<Uint32> b;
+        if (ShaderCompiler::LowerRectImages(*effective, b, false) && !b.empty()) effective = &b;
+        Vector<Uint32> c;
+        if (ShaderCompiler::Lower1DArrayImagesForEssl(*effective, c, false) && !c.empty()) effective = &c;
+        Vector<Uint32> d;
+        if (ShaderCompiler::LegalizeFragmentOutputIndexingForEssl(*effective, d, false) && !d.empty()) {
+            effective = &d;
+        }
+
+        SpvcSession session(*effective, SessionUsageBit::Transpile);
+        spvc_compiler_options options;
+        if (session.CreateOptions(&options) != SPVC_SUCCESS) return false;
+        spvc_compiler_options_set_uint(options, SPVC_COMPILER_OPTION_GLSL_VERSION, esslVersion);
+        spvc_compiler_options_set_bool(options, SPVC_COMPILER_OPTION_GLSL_ES, SPVC_TRUE);
+        spvc_compiler_options_set_bool(options, SPVC_COMPILER_OPTION_GLSL_VULKAN_SEMANTICS, SPVC_FALSE);
+        session.SetOptions(options);
+        const char* result = nullptr;
+        session.Compile(&result);
+        if (!result) return false;
+        outEssl = result;
+        return true;
+    }
+
+    EsslTranslationKeyInputs EsslInputsFor(const Vector<Uint32>& spirv) {
+        EsslTranslationKeyInputs inputs;
+        inputs.spirv = &spirv;
+        inputs.shaderType = GL_FRAGMENT_SHADER;
+        inputs.maxColorTextureSamples = 4;
+        inputs.maxIntegerSamples = 1;
+        inputs.maxDepthTextureSamples = 4;
+        inputs.advertisedMaxSamples = 4;
+        inputs.esslVersion = 320;
+        return inputs;
+    }
+} // namespace
+
+// ---------------------------------------------------------------------------------------
+// L1, in situ: the full glCompileShader + glLinkProgram path for a repeated program.
+// ---------------------------------------------------------------------------------------
+// Arg(0) = the CTS smoke size; Arg(120) = a heavy stage, bracketing the ratio.
+static void BM_ProgramLink_CacheOff(benchmark::State& state) {
+    MobileGL::Initialize();
+    const SyncCompileScope sync;
+    const CacheModeScope cache(false);
+    const String vs = kVertexSource;
+    const String fs = SwizzleLikeFragment("", static_cast<int>(state.range(0)));
+    for (auto _ : state) {
+        LinkOneProgram(vs, fs);
+    }
+    state.SetLabel("MOBILEGL_SHADER_CACHE=0");
+}
+BENCHMARK(BM_ProgramLink_CacheOff)->Arg(0)->Arg(120)->Unit(benchmark::kMicrosecond);
+
+static void BM_ProgramLink_CacheOn(benchmark::State& state) {
+    MobileGL::Initialize();
+    const SyncCompileScope sync;
+    const CacheModeScope cache(true);
+    const String vs = kVertexSource;
+    const String fs = SwizzleLikeFragment("", static_cast<int>(state.range(0)));
+    LinkOneProgram(vs, fs); // prime, so the measured loop is the steady state
+    const TranslationCacheStats before = MG_State::GLState::GetProgramTranslationCache().Stats();
+    const TranslationCacheStats parseBefore = GetShaderParseVerdictCache().Stats();
+    for (auto _ : state) {
+        LinkOneProgram(vs, fs);
+    }
+    const TranslationCacheStats stats = MG_State::GLState::GetProgramTranslationCache().Stats();
+    const TranslationCacheStats parseStats = GetShaderParseVerdictCache().Stats();
+    state.counters["L1_hits"] = static_cast<double>(stats.hits - before.hits);
+    state.counters["L1_misses"] = static_cast<double>(stats.misses - before.misses);
+    // Two stages per iteration, so a clean run shows L1c_hits == 2 * iterations and zero
+    // misses: every glCompileShader in the loop skipped its parse.
+    state.counters["L1c_hits"] = static_cast<double>(parseStats.hits - parseBefore.hits);
+    state.counters["L1c_misses"] = static_cast<double>(parseStats.misses - parseBefore.misses);
+}
+BENCHMARK(BM_ProgramLink_CacheOn)->Arg(0)->Arg(120)->Unit(benchmark::kMicrosecond);
+
+// ---------------------------------------------------------------------------------------
+// L1, the shape the memo actually exists for: MANY PROGRAMS OUT OF THE SAME SHADERS.
+//
+// The pair above deletes its shader objects every iteration, which forces a fresh glslang
+// parse per iteration no matter what the link does - glCompileShader parses, and that is a
+// DIFFERENT entry point from the one L1 memoizes. It is a real workload (what an application
+// that never reuses a shader object pays) but it is the pessimistic one, and the residual it
+// leaves is the parse, not the link.
+//
+// This pair keeps the shader objects alive, so the parses happen once before the measured
+// loop and the L1 hit then skips the link, mapIO, the SPIR-V, the reflection and the routing
+// outright.
+//
+// SINCE L1c THIS IS THE CONTROL, NOT THE TARGET. Nothing inside the measured loop calls
+// glCompileShader, so L1c cannot fire here at all - which is exactly what makes the pair
+// useful: it is the shape that says whether the compile-side memo has slowed the LINK path
+// down. Its numbers should be indistinguishable from the pre-L1c ones.
+// ---------------------------------------------------------------------------------------
+namespace {
+    struct SharedShaders {
+        GLuint vs = 0;
+        GLuint fs = 0;
+    };
+
+    SharedShaders MakeSharedShaders(const String& vertexSource, const String& fragmentSource) {
+        using namespace MG_Impl::GLImpl;
+        SharedShaders shaders;
+        shaders.vs = CreateShader(GL_VERTEX_SHADER);
+        const char* vsText = vertexSource.c_str();
+        ShaderSource(shaders.vs, 1, &vsText, nullptr);
+        CompileShader(shaders.vs);
+        shaders.fs = CreateShader(GL_FRAGMENT_SHADER);
+        const char* fsText = fragmentSource.c_str();
+        ShaderSource(shaders.fs, 1, &fsText, nullptr);
+        CompileShader(shaders.fs);
+        return shaders;
+    }
+
+    void LinkFromSharedShaders(const SharedShaders& shaders) {
+        using namespace MG_Impl::GLImpl;
+        const GLuint program = CreateProgram();
+        AttachShader(program, shaders.vs);
+        AttachShader(program, shaders.fs);
+        LinkProgram(program);
+        benchmark::DoNotOptimize(program);
+        DeleteProgram(program);
+    }
+} // namespace
+
+static void BM_SharedShaderLink_CacheOff(benchmark::State& state) {
+    MobileGL::Initialize();
+    const SyncCompileScope sync;
+    const CacheModeScope cache(false);
+    const SharedShaders shaders =
+        MakeSharedShaders(kVertexSource, SwizzleLikeFragment("", static_cast<int>(state.range(0))));
+    for (auto _ : state) {
+        LinkFromSharedShaders(shaders);
+    }
+    state.SetLabel("MOBILEGL_SHADER_CACHE=0");
+}
+BENCHMARK(BM_SharedShaderLink_CacheOff)->Arg(0)->Arg(120)->Unit(benchmark::kMicrosecond);
+
+static void BM_SharedShaderLink_CacheOn(benchmark::State& state) {
+    MobileGL::Initialize();
+    const SyncCompileScope sync;
+    const CacheModeScope cache(true);
+    const SharedShaders shaders =
+        MakeSharedShaders(kVertexSource, SwizzleLikeFragment("", static_cast<int>(state.range(0))));
+    LinkFromSharedShaders(shaders); // prime, so the measured loop is the steady state
+    const TranslationCacheStats before = MG_State::GLState::GetProgramTranslationCache().Stats();
+    for (auto _ : state) {
+        LinkFromSharedShaders(shaders);
+    }
+    const TranslationCacheStats stats = MG_State::GLState::GetProgramTranslationCache().Stats();
+    state.counters["L1_hits"] = static_cast<double>(stats.hits - before.hits);
+    state.counters["L1_misses"] = static_cast<double>(stats.misses - before.misses);
+}
+BENCHMARK(BM_SharedShaderLink_CacheOn)->Arg(0)->Arg(120)->Unit(benchmark::kMicrosecond);
+
+// ---------------------------------------------------------------------------------------
+// L2, component: the DirectGLES SPIR-V pass chain plus SPIRV-Cross for one stage.
+// ---------------------------------------------------------------------------------------
+static void BM_EsslTranspile_CacheOff(benchmark::State& state) {
+    MobileGL::Initialize();
+    const Vector<Uint32> spirv =
+        BuildSanitizedFragmentSpirv(SwizzleLikeFragment("", static_cast<int>(state.range(0))));
+    if (spirv.empty()) {
+        state.SkipWithError("could not build the fragment module");
+        return;
+    }
+    String essl;
+    for (auto _ : state) {
+        if (!TranspileLikeDirectGles(spirv, 320, essl)) {
+            state.SkipWithError("transpile failed");
+            break;
+        }
+        benchmark::DoNotOptimize(essl.data());
+    }
+    state.SetLabel("MOBILEGL_SHADER_CACHE=0");
+}
+BENCHMARK(BM_EsslTranspile_CacheOff)->Arg(0)->Arg(120)->Unit(benchmark::kMicrosecond);
+
+static void BM_EsslTranspile_CacheOn(benchmark::State& state) {
+    MobileGL::Initialize();
+    const Vector<Uint32> spirv =
+        BuildSanitizedFragmentSpirv(SwizzleLikeFragment("", static_cast<int>(state.range(0))));
+    if (spirv.empty()) {
+        state.SkipWithError("could not build the fragment module");
+        return;
+    }
+    BoundedTranslationCache<EsslTranslationResult> cache("bench L2", 64, 8u << 20);
+    const EsslTranslationKeyInputs inputs = EsslInputsFor(spirv);
+    for (auto _ : state) {
+        const TranslationCacheKey key = BuildEsslTranslationKey(inputs);
+        EsslTranslationResultPtr hit = cache.Find(key);
+        if (!hit) {
+            auto payload = MakeShared<EsslTranslationResult>();
+            if (!TranspileLikeDirectGles(spirv, inputs.esslVersion, payload->essl)) {
+                state.SkipWithError("transpile failed");
+                break;
+            }
+            cache.Insert(key, EsslTranslationResultPtr(payload), EsslTranslationResultBytes(*payload));
+            hit = payload;
+        }
+        benchmark::DoNotOptimize(hit->essl.data());
+    }
+    const TranslationCacheStats stats = cache.Stats();
+    state.counters["L2_hits"] = static_cast<double>(stats.hits);
+    state.counters["L2_misses"] = static_cast<double>(stats.misses);
+}
+BENCHMARK(BM_EsslTranspile_CacheOn)->Arg(0)->Arg(120)->Unit(benchmark::kMicrosecond);
+
+// ---------------------------------------------------------------------------------------
+// L1c, the shape where it could LOSE rather than win: the DEFERRED PARSE.
+// ---------------------------------------------------------------------------------------
+// A stage whose compile hits L1c holds no AST, so if the program-level key then MISSES, the
+// parse it skipped has to happen anyway - inside the link, via ClaimParsedShader. The parse
+// is moved, not removed, and this pair is what says whether moving it costs anything.
+//
+// The shape forces exactly that, every iteration: one CONSTANT vertex source (hits L1c after
+// the first iteration) linked against a FRESH fragment source each time (misses L1c, and
+// makes the program key miss too). So:
+//
+//   cache off - two parses at glCompileShader, then the link.
+//   cache on  - one parse at glCompileShader (the fragment), one deferred parse inside the
+//               link (the vertex), then the link.
+//
+// The parse count is identical, so these two should land within noise of each other. If the
+// On arm is materially SLOWER, L1c is charging for something - the per-compile key build and
+// hash over the full preprocessed source, or the loss of the claim-CAS reuse - and that cost
+// shows up here and nowhere else.
+//
+// The distinct fragment sources also churn both front-end levels through their FIFO caps,
+// which is the eviction behaviour a real shaderpack load produces; over a long run the
+// constant vertex entry is occasionally evicted by that churn and re-inserted, so the L1c
+// hit rate reported below is high but not exactly 1.0 per iteration.
+namespace {
+    String UniqueFragmentSource(const Uint64 serial, const int padLines) {
+        return SwizzleLikeFragment("", padLines) +
+               "\n// unique-" + std::to_string(serial) + "\n";
+    }
+} // namespace
+
+static void BM_DeferredParseLink_CacheOff(benchmark::State& state) {
+    MobileGL::Initialize();
+    const SyncCompileScope sync;
+    const CacheModeScope cache(false);
+    const String vs = kVertexSource;
+    Uint64 serial = 0;
+    for (auto _ : state) {
+        LinkOneProgram(vs, UniqueFragmentSource(serial++, static_cast<int>(state.range(0))));
+    }
+    state.SetLabel("MOBILEGL_SHADER_CACHE=0");
+}
+BENCHMARK(BM_DeferredParseLink_CacheOff)->Arg(0)->Arg(120)->Unit(benchmark::kMicrosecond);
+
+static void BM_DeferredParseLink_CacheOn(benchmark::State& state) {
+    MobileGL::Initialize();
+    const SyncCompileScope sync;
+    const CacheModeScope cache(true);
+    const String vs = kVertexSource;
+    Uint64 serial = 0;
+    LinkOneProgram(vs, UniqueFragmentSource(~0ull, static_cast<int>(state.range(0)))); // prime the vertex entry
+    const TranslationCacheStats before = MG_State::GLState::GetProgramTranslationCache().Stats();
+    const TranslationCacheStats parseBefore = GetShaderParseVerdictCache().Stats();
+    for (auto _ : state) {
+        LinkOneProgram(vs, UniqueFragmentSource(serial++, static_cast<int>(state.range(0))));
+    }
+    const TranslationCacheStats stats = MG_State::GLState::GetProgramTranslationCache().Stats();
+    const TranslationCacheStats parseStats = GetShaderParseVerdictCache().Stats();
+    // Expected shape: L1 all misses (every program is new), L1c one hit (vertex) and one miss
+    // (fragment) per iteration.
+    state.counters["L1_hits"] = static_cast<double>(stats.hits - before.hits);
+    state.counters["L1_misses"] = static_cast<double>(stats.misses - before.misses);
+    state.counters["L1c_hits"] = static_cast<double>(parseStats.hits - parseBefore.hits);
+    state.counters["L1c_misses"] = static_cast<double>(parseStats.misses - parseBefore.misses);
+}
+BENCHMARK(BM_DeferredParseLink_CacheOn)->Arg(0)->Arg(120)->Unit(benchmark::kMicrosecond);
+
+BENCHMARK_MAIN();

MobileGL/MG_Benchmark/Transpile/CMakeLists.txt

@@ -0,0 +1,20 @@
+cmake_minimum_required(VERSION 3.24)
+
+# Deliberately NOT a google-benchmark target: the interesting quantity is a per-stage
+# breakdown of one program build, which needs its own clock around sub-steps that share
+# set-up, and a plain main() keeps the output a table this can be read straight out of.
+add_executable(
+        TranspileProfile
+        TranspileProfile.cpp
+)
+
+target_include_directories(TranspileProfile PRIVATE
+        ${MGL_ROOT}/include
+        ${MGL_ROOT}/MobileGL
+        ${MGL_ROOT}/3rdparty/SPIRV-Reflect
+)
+
+target_link_libraries(
+        TranspileProfile PRIVATE
+        ${LINK_LIBRARIES}
+)

MobileGL/MG_Impl/GLImpl/Buffer/GL_Buffer.cpp

@@ -18,6 +18,7 @@
 #include <MG_Util/Converters/GLToStr/GLEnumConverter.h>
 #include <MG_Util/Converters/GLToMG/BufferEnumConverter.h>
 #include <MG_Util/Converters/MGToGL/BufferEnumConverter.h>
+#include <MG_Util/Texture/PixelStoreProcessor.h>
 
 namespace MobileGL::MG_Impl::GLImpl {
     namespace {
@@ -31,6 +32,8 @@ namespace MobileGL::MG_Impl::GLImpl {
             NamedBufferData,
             NamedBufferSubData,
             CopyNamedBufferSubData,
+            ClearBufferData,
+            ClearBufferSubData,
             ClearNamedBufferData,
             ClearNamedBufferSubData,
             MapBufferRange,
@@ -65,6 +68,10 @@ namespace MobileGL::MG_Impl::GLImpl {
                 return "NamedBufferSubData";
             case BufferOp::CopyNamedBufferSubData:
                 return "CopyNamedBufferSubData";
+            case BufferOp::ClearBufferData:
+                return "ClearBufferData";
+            case BufferOp::ClearBufferSubData:
+                return "ClearBufferSubData";
             case BufferOp::ClearNamedBufferData:
                 return "ClearNamedBufferData";
             case BufferOp::ClearNamedBufferSubData:
@@ -143,16 +150,6 @@ namespace MobileGL::MG_Impl::GLImpl {
                 return 0;
             }
 
-            // The pattern is replicated verbatim, which is only the whole story while the client
-            // layout already matches the internal format - the case every entry point in practice
-            // uses, and the only one the conversion machinery here can express. Say so rather than
-            // quietly writing a differently-sized pattern.
-            const SizeT sourceSize = MG_Util::GetInputBytesPerPixel(inputFormat, pixelType);
-            if (sourceSize != elementSize) {
-                MGLOG_W_ONCE("%s: clear pattern is %zu bytes but internalformat 0x%X stores %zu; "
-                        "converting between them is not implemented",
-                        GetBufferOpName(op), sourceSize, internalformat, elementSize);
-            }
             return elementSize;
         }
 
@@ -194,27 +191,59 @@ namespace MobileGL::MG_Impl::GLImpl {
             return true;
         }
 
-        void ClearNamedBufferRange_State(GLuint buffer, GLenum internalformat, GLintptr offset, GLsizeiptr size,
-                                         GLenum format, GLenum type, const void* data, BufferOp op) {
+        Bool BuildClearPattern(GLenum internalformat, GLenum format, GLenum type, const void* data,
+                               SizeT patternSize, BufferOp op, Vector<Uint8>& pattern) {
+            const TextureInternalFormat internal = MG_Util::ConvertGLEnumToTextureInternalFormat(internalformat);
+            const TextureInputFormat inputFormat = MG_Util::ConvertGLEnumToTextureInputFormat(format);
+            const TexturePixelDataType inputType = MG_Util::ConvertGLEnumToTexturePixelDataType(type);
+
+            Vector<Uint8> zeroInput;
+            const void* inputPixel = data;
+            if (inputPixel == nullptr) {
+                const SizeT inputSize = MG_Util::GetInputBytesPerPixel(inputFormat, inputType);
+                if (inputSize == 0) {
+                    MG_State::pGLContext->RecordError(
+                        ErrorCode::InvalidValue,
+                        MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", GetBufferOpName(op),
+                                                     "format and type do not describe a source pixel."));
+                    return false;
+                }
+                zeroInput.resize(inputSize);
+                inputPixel = zeroInput.data();
+            }
+
+            if (!MG_Util::PixelStoreProcessor::ConvertOnePixelToInternal(
+                    internal, inputFormat, inputType, inputPixel, pattern)) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidValue,
+                    MakeUnique<GenericErrorInfo>(
+                        "MG_Impl/GLImpl", GetBufferOpName(op),
+                        std::format("Cannot convert one ({}, {}) pixel into internalformat 0x{:X}.",
+                                    MG_Util::ConvertGLEnumToString(format), MG_Util::ConvertGLEnumToString(type),
+                                    internalformat)));
+                return false;
+            }
+
+            if (data == nullptr) {
+                // GL defines a null clear value as all zero bits in the destination store, while
+                // retaining the format/type validation above.
+                pattern.assign(patternSize, 0);
+            }
+            return true;
+        }
+
+        void ClearBufferRange_State(const SharedPtr<MG_State::GLState::BufferObject>& bufferObject,
+                                    GLenum internalformat, GLintptr offset, GLsizeiptr size,
+                                    GLenum format, GLenum type, const void* data, BufferOp op) {
             const SizeT patternSize = GetClearPatternSize(internalformat, format, type, op);
             if (patternSize == 0) return;
-
-            auto bufferObject = GetNamedBufferObject(buffer, op);
-            if (!bufferObject) return;
             if (!ValidateBufferClearRange(bufferObject, offset, size, patternSize, op)) return;
             if (size == 0) return;
 
-            Vector<Uint8> clearData(static_cast<SizeT>(size));
-            if (data) {
-                const auto* pattern = static_cast<const Uint8*>(data);
-                for (SizeT at = 0; at < clearData.size(); at += patternSize) {
-                    Memcpy(clearData.data() + at, pattern, patternSize);
-                }
-            } else {
-                Memset(clearData.data(), 0, clearData.size());
-            }
-
-            bufferObject->UploadSubData({clearData.data(), clearData.size()}, static_cast<SizeT>(offset));
+            Vector<Uint8> pattern;
+            if (!BuildClearPattern(internalformat, format, type, data, patternSize, op, pattern)) return;
+            bufferObject->FillSubData({pattern.data(), pattern.size()}, static_cast<SizeT>(offset),
+                                      static_cast<SizeT>(size));
         }
 
         auto& GetBufferBindingSlot(BufferTarget target) {
@@ -1197,17 +1226,34 @@ namespace MobileGL::MG_Impl::GLImpl {
                                         static_cast<SizeT>(writeOffset), static_cast<SizeT>(size));
     }
 
+    void ClearBufferData_State(GLenum target, GLenum internalformat, GLenum format, GLenum type, const void* data) {
+        auto bufferObject = GetBoundBufferObject(target, BufferOp::ClearBufferData);
+        if (!bufferObject) return;
+        ClearBufferRange_State(bufferObject, internalformat, 0, static_cast<GLsizeiptr>(bufferObject->GetSize()), format,
+                               type, data, BufferOp::ClearBufferData);
+    }
+
+    void ClearBufferSubData_State(GLenum target, GLenum internalformat, GLintptr offset, GLsizeiptr size,
+                                  GLenum format, GLenum type, const void* data) {
+        auto bufferObject = GetBoundBufferObject(target, BufferOp::ClearBufferSubData);
+        if (!bufferObject) return;
+        ClearBufferRange_State(bufferObject, internalformat, offset, size, format, type, data,
+                               BufferOp::ClearBufferSubData);
+    }
+
     void ClearNamedBufferData_State(GLuint buffer, GLenum internalformat, GLenum format, GLenum type, const void* data) {
         auto bufferObject = GetNamedBufferObject(buffer, BufferOp::ClearNamedBufferData);
         if (!bufferObject) return;
-        ClearNamedBufferRange_State(buffer, internalformat, 0, static_cast<GLsizeiptr>(bufferObject->GetSize()), format,
-                                    type, data, BufferOp::ClearNamedBufferData);
+        ClearBufferRange_State(bufferObject, internalformat, 0, static_cast<GLsizeiptr>(bufferObject->GetSize()), format,
+                               type, data, BufferOp::ClearNamedBufferData);
     }
 
     void ClearNamedBufferSubData_State(GLuint buffer, GLenum internalformat, GLintptr offset, GLsizeiptr size,
                                        GLenum format, GLenum type, const void* data) {
-        ClearNamedBufferRange_State(buffer, internalformat, offset, size, format, type, data,
-                                    BufferOp::ClearNamedBufferSubData);
+        auto bufferObject = GetNamedBufferObject(buffer, BufferOp::ClearNamedBufferSubData);
+        if (!bufferObject) return;
+        ClearBufferRange_State(bufferObject, internalformat, offset, size, format, type, data,
+                               BufferOp::ClearNamedBufferSubData);
     }
 
     void* MapNamedBuffer_State(GLuint buffer, GLenum access) {
@@ -1662,6 +1708,15 @@ namespace MobileGL::MG_Impl::GLImpl {
         CopyNamedBufferSubData_State(readBuffer, writeBuffer, readOffset, writeOffset, size);
     }
 
+    void ClearBufferData(GLenum target, GLenum internalformat, GLenum format, GLenum type, const void* data) {
+        ClearBufferData_State(target, internalformat, format, type, data);
+    }
+
+    void ClearBufferSubData(GLenum target, GLenum internalformat, GLintptr offset, GLsizeiptr size, GLenum format,
+                            GLenum type, const void* data) {
+        ClearBufferSubData_State(target, internalformat, offset, size, format, type, data);
+    }
+
     void ClearNamedBufferData(GLuint buffer, GLenum internalformat, GLenum format, GLenum type, const void* data) {
         ClearNamedBufferData_State(buffer, internalformat, format, type, data);
     }

MobileGL/MG_Impl/GLImpl/Buffer/GL_Buffer.h

@@ -27,6 +27,9 @@ namespace MobileGL::MG_Impl::GLImpl {
     void NamedBufferSubData(GLuint buffer, GLintptr offset, GLsizeiptr size, const void* data);
     void CopyNamedBufferSubData(GLuint readBuffer, GLuint writeBuffer, GLintptr readOffset, GLintptr writeOffset,
                                 GLsizeiptr size);
+    void ClearBufferData(GLenum target, GLenum internalformat, GLenum format, GLenum type, const void* data);
+    void ClearBufferSubData(GLenum target, GLenum internalformat, GLintptr offset, GLsizeiptr size, GLenum format,
+                            GLenum type, const void* data);
     void ClearNamedBufferData(GLuint buffer, GLenum internalformat, GLenum format, GLenum type, const void* data);
     void ClearNamedBufferSubData(GLuint buffer, GLenum internalformat, GLintptr offset, GLsizeiptr size, GLenum format,
                                  GLenum type, const void* data);

MobileGL/MG_Impl/GLImpl/Buffer/Validators.cpp

@@ -13,6 +13,7 @@
 #include <MG_Util/Converters/GLToStr/GLEnumConverter.h>
 #include <MG_Util/Converters/MGToGL/BufferEnumConverter.h>
 #include <MG_Util/Converters/MGToStr/BufferEnumConverter.h>
+#include <MG_Util/ShaderTranspiler/Types.h>
 
 namespace MobileGL::MG_Impl::GLImpl::BufferImpl {
     Bool ValidateBufferTarget(BufferTarget target) {
@@ -67,6 +68,13 @@ namespace MobileGL::MG_Impl::GLImpl::BufferImpl {
                 // binding points in GL 3.3 (no ARB_transform_feedback3).
                 pointCount = std::min<SizeT>(pointCount, 4);
             }
+            if (target == BufferTarget::AtomicCounter) {
+                // GL_MAX_ATOMIC_COUNTER_BUFFER_BINDINGS, which is NOT the state layer's array
+                // size: a counter buffer reaches a shader only as a lowered storage block, so the
+                // reserved range is the ceiling, and glGetIntegerv advertises the same number.
+                pointCount = std::min<SizeT>(
+                    pointCount, static_cast<SizeT>(MG_Util::ShaderTranspiler::MAX_ATOMIC_COUNTER_BUFFER_BINDINGS));
+            }
             return pointCount;
         }
     } // namespace

MobileGL/MG_Impl/GLImpl/Debug/GL_Debug.cpp

@@ -0,0 +1,271 @@
+// MobileGL - MobileGL/MG_Impl/GLImpl/Debug/GL_Debug.cpp
+// Copyright (c) 2025-2026 MobileGL-Dev
+// Licensed under the GNU Lesser General Public License v3.0:
+//   https://www.gnu.org/licenses/gpl-3.0.txt
+//   https://www.gnu.org/licenses/lgpl-3.0.txt
+// SPDX-License-Identifier: LGPL-3.0-only
+// End of Source File Header
+
+#include "GL_Debug.h"
+
+#include <cstring>
+
+#include <MG_State/GLState/Core.h>
+#include <MG_State/GLState/ErrorState/Error.h>
+#include <MG_Impl/GLImpl/Query/GL_Query.h>
+#include <MG_Util/Converters/GLToStr/GLEnumConverter.h>
+
+namespace MobileGL::MG_Impl::GLImpl {
+    namespace {
+        // Must agree with what GL_Getter answers for GL_MAX_DEBUG_GROUP_STACK_DEPTH and
+        // GL_MAX_DEBUG_MESSAGE_LENGTH / GL_MAX_LABEL_LENGTH; an application that sizes a buffer
+        // off the query and then trips a different limit here would have no way to explain it.
+        constexpr SizeT kMaxDebugGroupStackDepth = 64;
+        constexpr GLsizei kMaxDebugMessageLength = 1024;
+        constexpr GLsizei kMaxLabelLength = 256;
+
+        // The debug state KHR_debug makes per-context. Held here rather than on GLContext because
+        // nothing else in MobileGL reads it, and it is keyed on the context id so a
+        // destroyed-and-recreated context starts with an empty stack and no labels - which the
+        // unit tests, which recreate the context between cases, depend on.
+        struct DebugState {
+            Uint64 contextId = 0;
+            // The messages pushed with glPushDebugGroup, innermost last. The base group GL creates
+            // the context with is implicit and is what makes the reported depth start at 1.
+            Vector<String> groupStack;
+            // Keyed by (identifier, name); see MakeObjectLabelKey.
+            UnorderedMap<Uint64, String> objectLabels;
+        };
+
+        DebugState& State() {
+            static DebugState state;
+            const Uint64 contextId = MG_State::pGLContext ? MG_State::pGLContext->GetTextureContextId() : 0;
+            if (state.contextId != contextId) {
+                state.contextId = contextId;
+                state.groupStack.clear();
+                state.objectLabels.clear();
+            }
+            return state;
+        }
+
+        Uint64 MakeObjectLabelKey(GLenum identifier, GLuint name) {
+            return (static_cast<Uint64>(identifier) << 32) | static_cast<Uint64>(name);
+        }
+
+        void RecordDebugError(ErrorCode code, const char* caller, const String& message) {
+            MG_State::pGLContext->RecordError(code, MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", caller, message));
+        }
+
+        // GL 4.6 core 20.2: only an APPLICATION or THIRD_PARTY source may be injected; the rest
+        // are reserved for the implementation itself.
+        Bool ValidateInjectedSource(GLenum source, const char* caller) {
+            if (source == GL_DEBUG_SOURCE_APPLICATION || source == GL_DEBUG_SOURCE_THIRD_PARTY) {
+                return true;
+            }
+            RecordDebugError(ErrorCode::InvalidEnum, caller,
+                             std::format("source {} is not GL_DEBUG_SOURCE_APPLICATION or "
+                                         "GL_DEBUG_SOURCE_THIRD_PARTY.",
+                                         MG_Util::ConvertGLEnumToString(source)));
+            return false;
+        }
+
+        // A negative length means the string is NUL-terminated (GL 4.6 core 20.2), which is how
+        // every one of these entry points spells "just use the whole thing".
+        Bool ValidateDebugStringLength(GLsizei length, const GLchar* text, GLsizei limit, const char* caller,
+                                       const char* what) {
+            const GLsizei effective =
+                length < 0 ? static_cast<GLsizei>(text != nullptr ? std::strlen(text) : 0) : length;
+            if (effective < limit) {
+                return true;
+            }
+            RecordDebugError(ErrorCode::InvalidValue, caller,
+                             std::format("{} length {} is not less than the {} limit of {}.", what, effective, what,
+                                         limit));
+            return false;
+        }
+
+        String MakeDebugString(GLsizei length, const GLchar* text) {
+            if (text == nullptr) return {};
+            return length < 0 ? String(text) : String(text, static_cast<SizeT>(length));
+        }
+
+        // Whether `name` currently names an object of `identifier`'s type. GL 4.6 core 20.5 makes
+        // labelling something that does not exist INVALID_VALUE, and every type KHR_debug lists
+        // has a frontend name check - so this is answered exactly rather than waved through.
+        // GL_DISPLAY_LIST is deliberately absent: it exists only in the compatibility profile,
+        // which MobileGL does not expose, so it falls to the INVALID_ENUM path below.
+        Bool ValidateLabelledObject(GLenum identifier, GLuint name, Bool& outIdentifierKnown) {
+            outIdentifierKnown = true;
+            auto* context = MG_State::pGLContext.get();
+            switch (identifier) {
+            case GL_BUFFER:
+                return context->ValidateBufferName(name);
+            case GL_SHADER:
+                return context->ValidateShaderName(name);
+            case GL_PROGRAM:
+                return context->ValidateProgramName(name);
+            case GL_VERTEX_ARRAY:
+                return context->ValidateVertexArrayName(name);
+            case GL_QUERY:
+                return IsQuery(name) == GL_TRUE;
+            case GL_PROGRAM_PIPELINE:
+                return context->ValidateProgramPipelineName(name);
+            case GL_TRANSFORM_FEEDBACK:
+                return context->ValidateTransformFeedbackName(name);
+            case GL_SAMPLER:
+                return context->ValidateSamplerName(name);
+            case GL_TEXTURE:
+                return context->ValidateTextureName(name);
+            case GL_RENDERBUFFER:
+                return context->ValidateRenderbufferName(name);
+            case GL_FRAMEBUFFER:
+                // Name 0 is the default framebuffer, which is a real, labellable object.
+                return name == 0 || context->ValidateFramebufferName(name);
+            default:
+                outIdentifierKnown = false;
+                return false;
+            }
+        }
+    } // namespace
+
+    GLint GetDebugGroupStackDepth() {
+        // GL 4.6 core 20.6: the context is created with one group already on the stack, so the
+        // reported depth is one more than the number of pushes the application has made.
+        return static_cast<GLint>(State().groupStack.size()) + 1;
+    }
+
+    void PushDebugGroup(GLenum source, GLuint id, GLsizei length, const GLchar* message) {
+        static_cast<void>(id);
+        if (!ValidateInjectedSource(source, __func__)) return;
+        if (!ValidateDebugStringLength(length, message, kMaxDebugMessageLength, __func__, "message")) return;
+
+        auto& state = State();
+        if (state.groupStack.size() + 1 >= kMaxDebugGroupStackDepth) {
+            // Not INVALID_*: KHR_debug gives the group stack its own error code.
+            RecordDebugError(ErrorCode::StackOverflow, __func__,
+                             std::format("the debug group stack is already {} deep, which is its maximum.",
+                                         kMaxDebugGroupStackDepth));
+            return;
+        }
+        state.groupStack.push_back(MakeDebugString(length, message));
+        MGLOG_D("glPushDebugGroup(%s) -> depth %d", state.groupStack.back().c_str(), GetDebugGroupStackDepth());
+    }
+
+    void PopDebugGroup() {
+        auto& state = State();
+        if (state.groupStack.empty()) {
+            // The base group the context was created with may not be popped (GL 4.6 core 20.6).
+            RecordDebugError(ErrorCode::StackUnderflow, __func__,
+                             "the debug group stack holds only the group the context was created with.");
+            return;
+        }
+        MGLOG_D("glPopDebugGroup(%s)", state.groupStack.back().c_str());
+        state.groupStack.pop_back();
+    }
+
+    void DebugMessageInsert(GLenum source, GLenum type, GLuint id, GLenum severity, GLsizei length,
+                            const GLchar* buf) {
+        static_cast<void>(id);
+        if (!ValidateInjectedSource(source, __func__)) return;
+        switch (type) {
+        case GL_DEBUG_TYPE_ERROR:
+        case GL_DEBUG_TYPE_DEPRECATED_BEHAVIOR:
+        case GL_DEBUG_TYPE_UNDEFINED_BEHAVIOR:
+        case GL_DEBUG_TYPE_PORTABILITY:
+        case GL_DEBUG_TYPE_PERFORMANCE:
+        case GL_DEBUG_TYPE_MARKER:
+        case GL_DEBUG_TYPE_PUSH_GROUP:
+        case GL_DEBUG_TYPE_POP_GROUP:
+        case GL_DEBUG_TYPE_OTHER:
+            break;
+        default:
+            RecordDebugError(ErrorCode::InvalidEnum, __func__,
+                             std::format("type {} is not a debug message type.",
+                                         MG_Util::ConvertGLEnumToString(type)));
+            return;
+        }
+        switch (severity) {
+        case GL_DEBUG_SEVERITY_HIGH:
+        case GL_DEBUG_SEVERITY_MEDIUM:
+        case GL_DEBUG_SEVERITY_LOW:
+        case GL_DEBUG_SEVERITY_NOTIFICATION:
+            break;
+        default:
+            RecordDebugError(ErrorCode::InvalidEnum, __func__,
+                             std::format("severity {} is not a debug message severity.",
+                                         MG_Util::ConvertGLEnumToString(severity)));
+            return;
+        }
+        if (!ValidateDebugStringLength(length, buf, kMaxDebugMessageLength, __func__, "message")) return;
+
+        // No callback is ever invoked and the message log is empty by construction
+        // (GL_MAX_DEBUG_LOGGED_MESSAGES is 1 and glGetDebugMessageLog returns nothing), so the
+        // application-visible effect is exactly the error checking above. The text still reaches
+        // MobileGL's own log, where it is worth having next to the calls it annotates - at debug
+        // level, so an application that inserts a message per draw costs nothing in a release build.
+        MGLOG_D("glDebugMessageInsert: %s", MakeDebugString(length, buf).c_str());
+    }
+
+    void ObjectLabel(GLenum identifier, GLuint name, GLsizei length, const GLchar* label) {
+        Bool identifierKnown = false;
+        const Bool objectExists = ValidateLabelledObject(identifier, name, identifierKnown);
+        if (!identifierKnown) {
+            RecordDebugError(ErrorCode::InvalidEnum, __func__,
+                             std::format("identifier {} is not a labellable object type.",
+                                         MG_Util::ConvertGLEnumToString(identifier)));
+            return;
+        }
+        if (!objectExists) {
+            RecordDebugError(ErrorCode::InvalidValue, __func__,
+                             std::format("{} {} is not the name of an existing object.",
+                                         MG_Util::ConvertGLEnumToString(identifier), name));
+            return;
+        }
+        if (!ValidateDebugStringLength(length, label, kMaxLabelLength, __func__, "label")) return;
+
+        auto& labels = State().objectLabels;
+        const Uint64 key = MakeObjectLabelKey(identifier, name);
+        if (label == nullptr) {
+            // GL 4.6 core 20.5: a NULL label removes any label the object had.
+            labels.erase(key);
+            return;
+        }
+        labels[key] = MakeDebugString(length, label);
+    }
+
+    void GetObjectLabel(GLenum identifier, GLuint name, GLsizei bufSize, GLsizei* length, GLchar* label) {
+        if (bufSize < 0) {
+            RecordDebugError(ErrorCode::InvalidValue, __func__, "bufSize must not be negative.");
+            return;
+        }
+        Bool identifierKnown = false;
+        const Bool objectExists = ValidateLabelledObject(identifier, name, identifierKnown);
+        if (!identifierKnown) {
+            RecordDebugError(ErrorCode::InvalidEnum, __func__,
+                             std::format("identifier {} is not a labellable object type.",
+                                         MG_Util::ConvertGLEnumToString(identifier)));
+            return;
+        }
+        if (!objectExists) {
+            RecordDebugError(ErrorCode::InvalidValue, __func__,
+                             std::format("{} {} is not the name of an existing object.",
+                                         MG_Util::ConvertGLEnumToString(identifier), name));
+            return;
+        }
+
+        const auto& labels = State().objectLabels;
+        const auto it = labels.find(MakeObjectLabelKey(identifier, name));
+        const String& text = it != labels.end() ? it->second : String{};
+        // GL 4.6 core 20.5: the returned length excludes the NUL, and an unlabelled object hands
+        // back an empty string with length 0 rather than an error.
+        SizeT copied = 0;
+        if (label != nullptr && bufSize > 0) {
+            copied = std::min(text.size(), static_cast<SizeT>(bufSize) - 1);
+            std::memcpy(label, text.data(), copied);
+            label[copied] = '\0';
+        }
+        if (length != nullptr) {
+            *length = static_cast<GLsizei>(copied);
+        }
+    }
+} // namespace MobileGL::MG_Impl::GLImpl

MobileGL/MG_Impl/GLImpl/Debug/GL_Debug.h

@@ -0,0 +1,42 @@
+// MobileGL - MobileGL/MG_Impl/GLImpl/Debug/GL_Debug.h
+// Copyright (c) 2025-2026 MobileGL-Dev
+// Licensed under the GNU Lesser General Public License v3.0:
+//   https://www.gnu.org/licenses/gpl-3.0.txt
+//   https://www.gnu.org/licenses/lgpl-3.0.txt
+// SPDX-License-Identifier: LGPL-3.0-only
+// End of Source File Header
+
+#pragma once
+#include <Includes.h>
+
+namespace MobileGL::MG_Impl::GLImpl {
+    // KHR_debug, core since GL 4.3 (GL 4.6 core 20). Applications use these to annotate a capture
+    // and to name their objects; Better Clouds calls all four for exactly that.
+    //
+    // MobileGL implements the STATE and the ERRORS, and deliberately does not forward the calls to
+    // the host driver. Two independent reasons:
+    //
+    //  * glObjectLabel names a FRONTEND object. MobileGL's texture 5 is not the ES driver's
+    //    texture 5 (and under DirectVulkan it is not a driver object at all), so forwarding the
+    //    pair verbatim would label an unrelated object or a nonexistent one - worse than not
+    //    labelling.
+    //  * A debug GROUP is only meaningful if it brackets the commands the application issued
+    //    inside it. Neither backend emits its work at the moment the GL call arrives: DirectGLES
+    //    defers and reorders state sync and uploads around draws, and DirectVulkan is usually not
+    //    even recording a command buffer here. A forwarded push/pop would therefore enclose the
+    //    wrong commands, which is a misleading capture rather than a helpful one.
+    //
+    // What the application can rely on is the observable contract: the group stack depth is real
+    // (GL_DEBUG_GROUP_STACK_DEPTH tracks it, and over/underflow raise the errors KHR_debug
+    // specifies), and a label written with glObjectLabel comes back from glGetObjectLabel.
+    void PushDebugGroup(GLenum source, GLuint id, GLsizei length, const GLchar* message);
+    void PopDebugGroup();
+    void DebugMessageInsert(GLenum source, GLenum type, GLuint id, GLenum severity, GLsizei length,
+                            const GLchar* buf);
+    void ObjectLabel(GLenum identifier, GLuint name, GLsizei length, const GLchar* label);
+    void GetObjectLabel(GLenum identifier, GLuint name, GLsizei bufSize, GLsizei* length, GLchar* label);
+
+    // Current depth of the debug group stack, for GL_DEBUG_GROUP_STACK_DEPTH. The base group the
+    // context is created with counts, so this is never below 1 (GL 4.6 core 20.6).
+    GLint GetDebugGroupStackDepth();
+} // namespace MobileGL::MG_Impl::GLImpl

MobileGL/MG_Impl/GLImpl/Drawing/GL_Drawing.cpp

@@ -11,6 +11,7 @@
 #include <MG_State/GLState/Core.h>
 #include <MG_State/EGLState/Core.h>
 #include <MG_Backend/BackendObjects.h>
+#include <MG_Impl/Pipe/PipeFill.h>
 #include "../Getter/GL_Getter.h"
 
 namespace MobileGL::MG_Impl::GLImpl {
@@ -34,8 +35,81 @@ namespace MobileGL::MG_Impl::GLImpl {
         return true;
     }
 
-    static Bool ValidateCurrentProgramForExecution(const char* functionName) {
-        return ValidateProgramForExecution(MG_State::pGLContext->GetProgramForDraw(), functionName);
+    // Takes the ALREADY-RESOLVED draw program rather than looking it up: GLContext::GetProgramForDraw
+    // is not a plain getter (it settles the program's link and SPIR-V jobs so every version a
+    // backend samples during this draw describes the program it is drawing), so the draw funnel
+    // below resolves it exactly once and hands it to both users.
+    static Bool ValidateResolvedProgramForDraw(const SharedPtr<MG_State::GLState::ProgramObject>& currentProgram,
+                                               const char* functionName) {
+        // "If there is no current program object or bound program pipeline object, the results of
+        // a draw are UNDEFINED" - and undefined is not an error (GL 4.6 core 7.3, ES 3.1 7.3).
+        // The draw is dropped, silently, which is one of the shapes "undefined" is allowed to
+        // take; recording INVALID_OPERATION here is not, and es31cSeparateShaderObjsTests'
+        // StateInteraction reads exactly that error back after useProgram(0) + bindProgramPipeline(0).
+        // A DISPATCH is the opposite rule ("INVALID_OPERATION if there is no active program for
+        // the compute shader stage"), which is why this lives on the draw path and not in the
+        // shared ValidateProgramForExecution below.
+        if (!currentProgram) return false;
+        if (!ValidateProgramForExecution(currentProgram, functionName)) return false;
+
+        // GL 4.6 core 7.4.1, the pipeline validation rule every vertex-transferring command
+        // inherits: it is an INVALID_OPERATION when a tessellation control, tessellation
+        // evaluation or geometry stage has an executable but no program supplies an executable
+        // VERTEX shader. A non-separable program cannot reach this - the link rule forbids the
+        // shape - so in practice it catches a program pipeline assembled out of stage programs,
+        // which today draws happily and renders nothing.
+        //
+        // Asked of the EXECUTABLE, like the compute check below: for a pipeline the resolved
+        // program is the graphics composite, whose linked-shader snapshot is built out of exactly
+        // the pipeline's own graphics stage programs (GLContext::GetProgramForDraw), and the only
+        // stage compositing ever invents is a default FRAGMENT shader. A fragment-only pipeline is
+        // deliberately NOT rejected: the rule above names the three pre-rasterization stages, and
+        // nothing else here should start refusing draws GL accepts.
+        //
+        // On the DRAW path only, never in ValidateProgramForExecution itself, so a dispatch -
+        // which shares that helper and legitimately has no vertex stage - is untouched.
+        const Bool hasPreRasterizationStage = currentProgram->HasLinkedShaderStage(ShaderStage::Geometry) ||
+                                              currentProgram->HasLinkedShaderStage(ShaderStage::TessControl) ||
+                                              currentProgram->HasLinkedShaderStage(ShaderStage::TessEval);
+        if (hasPreRasterizationStage && !currentProgram->HasLinkedShaderStage(ShaderStage::Vertex)) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>(
+                    "MG_Impl/GLImpl", functionName,
+                    "The program in use runs a geometry or tessellation stage but has no vertex shader stage."));
+            return false;
+        }
+
+        return true;
+    }
+
+    // gl_NumSamples has no SPIR-V built-in, so the source pipeline lowers it onto a reserved
+    // default-block uniform (see InjectNumSamplesBuiltinShim). This is where that uniform is paid
+    // for: the value is a property of the DRAW FRAMEBUFFER, not of the program, so one program
+    // drawn into a 4x target and then into the default framebuffer must see 4 and then 1 - which
+    // rules out baking it at link time.
+    //
+    // Per draw rather than on framebuffer changes because the pair (program, framebuffer) is what
+    // decides the value and either half can move between draws. It costs a phase-A flag read for
+    // every program that has no shim, and a 4-byte compare for the ones that do: the write only
+    // bumps the UBO content version when the number actually changes, so a run of draws into one
+    // framebuffer re-uploads nothing.
+    static void PublishDrawFramebufferSampleCount(const SharedPtr<MG_State::GLState::ProgramObject>& program) {
+        if (!program || !program->UsesReservedNumSamples()) return;
+        // GL 4.6 core 15.2.2: gl_NumSamples is the number of samples in the framebuffer, or ONE
+        // when the target is not multisampled - where glGetIntegerv(GL_SAMPLES) answers zero.
+        program->WriteReservedNumSamples(static_cast<Int>(std::max<GLint>(ResolveDrawFramebufferSampleCount(), 1)));
+    }
+
+    // The one funnel every drawing command passes through. Order is load-bearing: validate first
+    // (a rejected draw must leave state alone), then publish the sample count - which reads the
+    // DRAW FRAMEBUFFER binding, so it has to run after the caller's framebuffer state is settled
+    // and before the backend consumes the program's UBO content version.
+    static Bool PrepareCurrentProgramForDraw(const char* functionName) {
+        const auto& currentProgram = MG_State::pGLContext->GetProgramForDraw();
+        if (!ValidateResolvedProgramForDraw(currentProgram, functionName)) return false;
+        PublishDrawFramebufferSampleCount(currentProgram);
+        return true;
     }
 
     // A dispatch resolves its program through the DISPATCH accessor: with a pipeline bound
@@ -45,7 +119,11 @@ namespace MobileGL::MG_Impl::GLImpl {
         const auto& currentProgram = MG_State::pGLContext->GetProgramForDispatch();
         if (!ValidateProgramForExecution(currentProgram, functionName)) return false;
 
-        if (currentProgram->GetShaderIndexByStage(ShaderStage::Compute) < 0) {
+        // Of the EXECUTABLE, not the live attach list: attaching a compute shader to an
+        // already-linked graphics program does not give that program a compute stage to
+        // dispatch (GL 4.6 core 7.3), and letting the dispatch through on the strength of the
+        // attach hands the backend a program whose SPIR-V has no compute module in it.
+        if (!currentProgram->HasLinkedShaderStage(ShaderStage::Compute)) {
             MG_State::pGLContext->RecordError(
                 ErrorCode::InvalidOperation,
                 MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName,
@@ -69,6 +147,20 @@ namespace MobileGL::MG_Impl::GLImpl {
         case GL_TRIANGLES: return static_cast<Uint64>(count / 3);
         case GL_TRIANGLE_STRIP:
         case GL_TRIANGLE_FAN: return count >= 3 ? static_cast<Uint64>(count - 2) : 0;
+        // Adjacency primitives (GL 4.6 core table 10.1). Only a geometry stage can consume
+        // them, and it is the ADJACENT-free primitive count that reaches it: 4 vertices per
+        // line, 6 per triangle, one per step for the strips. Answering 0 here - which is what
+        // the default arm did - made AccountTransformFeedbackPrimitives bail before it had
+        // recorded anything, so an adjacency capture advanced neither the captured-vertex
+        // counter the scattered-capture path is bounded by nor the geometry-capture-draw flag
+        // that routes the transform feedback queries to the driver's own counter.
+        case GL_LINES_ADJACENCY: return static_cast<Uint64>(count / 4);
+        case GL_LINE_STRIP_ADJACENCY: return count >= 4 ? static_cast<Uint64>(count - 3) : 0;
+        case GL_TRIANGLES_ADJACENCY: return static_cast<Uint64>(count / 6);
+        case GL_TRIANGLE_STRIP_ADJACENCY: return count >= 6 ? static_cast<Uint64>((count - 4) / 2) : 0;
+        // GL_PATCHES is deliberately absent: the tessellator's amplification is not knowable
+        // on the CPU, and answering 0 is what defers GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN
+        // to the driver's own counter, which is the only correct source for a patch capture.
         default: return 0;
         }
     }
@@ -95,11 +187,17 @@ namespace MobileGL::MG_Impl::GLImpl {
         case GL_LINES:
         case GL_LINE_STRIP:
         case GL_LINE_LOOP:
+        // An adjacency primitive delivers the same line/triangle to the geometry stage; the
+        // adjacent vertices are context, not part of the primitive.
+        case GL_LINES_ADJACENCY:
+        case GL_LINE_STRIP_ADJACENCY:
             verticesPerPrimitive = 2;
             break;
         case GL_TRIANGLES:
         case GL_TRIANGLE_STRIP:
         case GL_TRIANGLE_FAN:
+        case GL_TRIANGLES_ADJACENCY:
+        case GL_TRIANGLE_STRIP_ADJACENCY:
             verticesPerPrimitive = 3;
             break;
         default:
@@ -108,6 +206,12 @@ namespace MobileGL::MG_Impl::GLImpl {
 
         const auto& program = MG_State::pGLContext->GetTransformFeedbackProgram();
         if (program != nullptr) {
+            // A geometry stage writes what it emits, not what the draw assembled, and the
+            // amplification factor lives in the shader. Record that this span contained such
+            // a draw so the transform feedback queries keep their backend result for it.
+            if (program->HasLinkedShaderStage(ShaderStage::Geometry)) {
+                MG_State::pGLContext->AddTransformFeedbackGeometryCaptureDraw();
+            }
             // Capacity in captured vertices = the tightest bound buffer.
             Uint64 capacityVertices = ~0ull;
             for (SizeT i = 0; i < program->GetTransformFeedbackBufferCount(); ++i) {
@@ -127,6 +231,11 @@ namespace MobileGL::MG_Impl::GLImpl {
         }
         MG_State::pGLContext->AddTransformFeedbackPrimitives(primitives);
         MG_State::pGLContext->AddTransformFeedbackCapturedVertices(primitives * verticesPerPrimitive);
+        // Only draws that get this far are in the written counter at all. The instanced and
+        // indirect entry points never call this function, so a span that contains one is NOT
+        // fully accounted, and the queries must be able to tell: they compare this counter's
+        // delta against zero before standing in for the backend's own result.
+        MG_State::pGLContext->AddTransformFeedbackAccountedCaptureDraw();
     }
 
     // Every primitive mode a draw command accepts (GL 4.6 core table 10.1, plus
@@ -151,11 +260,23 @@ namespace MobileGL::MG_Impl::GLImpl {
         }
     }
 
+    // The `mode` INVALID_ENUM in isolation, so a draw entry point can raise it BEFORE any of the
+    // state-dependent INVALID_OPERATIONs below. GL 4.6 core 10.4 makes a bad mode INVALID_ENUM
+    // unconditionally, while "no current program" is not even a spec-listed draw error - it is
+    // MobileGL's own null-dereference guard - so it must never shadow the enum check
+    // (KHR-GL31.api.coverage calls glDrawArraysInstanced/glDrawElementsInstanced with mode
+    // GL_POINTS-1 against a bare context and pins GL_INVALID_ENUM).
+    static Bool ValidatePrimitiveModeEnum(const char* functionName, GLenum mode) {
+        if (IsAcceptedPrimitiveMode(mode)) return true;
+
+        MG_State::pGLContext->RecordError(
+            ErrorCode::InvalidEnum,
+            MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName, "mode is not an accepted primitive type."));
+        return false;
+    }
+
     static Bool ValidatePrimitiveModeForBackend(const char* functionName, GLenum mode) {
-        if (!IsAcceptedPrimitiveMode(mode)) {
-            MG_State::pGLContext->RecordError(
-                ErrorCode::InvalidEnum,
-                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName, "mode is not an accepted primitive type."));
+        if (!ValidatePrimitiveModeEnum(functionName, mode)) {
             return false;
         }
 
@@ -176,13 +297,58 @@ namespace MobileGL::MG_Impl::GLImpl {
             return false;
         }
 
+        const auto& currentProgram = MG_State::pGLContext->GetProgramForDraw();
+
+        // GL 4.6 core 10.1: the tessellation pipeline's only input primitive is GL_PATCHES, and
+        // GL_PATCHES has no meaning without it. Both directions are INVALID_OPERATION, and
+        // neither was implemented - which is two of the four sites
+        // KHR-GL43.transform_feedback.api_errors_test checks with one shared message string.
+        // The EVALUATION stage is what decides: a control stage cannot run without one, and a
+        // program carrying only an evaluation stage still tessellates, through GL's
+        // fixed-function pass-through control stage (11.2.2).
+        // Asked of the LAST LINK, not the live attach list (GL 4.6 core 7.3): attaching a
+        // tessellation evaluation shader to an already-linked program does not put it in the
+        // executable, so reading the live list here would reject every non-GL_PATCHES draw
+        // against a program that does not tessellate - and keep rejecting them, since a detach
+        // is likewise deferred to the next link.
+        const Bool tessellationActive = currentProgram && currentProgram->HasLinkedShaderStage(ShaderStage::TessEval);
+        if (tessellationActive && mode != GL_PATCHES) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>(
+                    "MG_Impl/GLImpl", functionName,
+                    "A program with a tessellation evaluation shader can only be drawn with GL_PATCHES."));
+            return false;
+        }
+        if (!tessellationActive && mode == GL_PATCHES) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName,
+                                             "GL_PATCHES requires an active tessellation evaluation shader."));
+            return false;
+        }
+
         // A geometry stage only accepts the primitive types that decompose into its declared
         // input primitive (GL 4.6 core 11.3.1); anything else is INVALID_OPERATION. GL_PATCHES
         // is the tessellation pipeline's input and reaches the geometry stage already
         // converted, so it is not constrained here.
-        const auto& currentProgram = MG_State::pGLContext->GetProgramForDraw();
-        const GLenum gsInput = currentProgram ? currentProgram->GetGeometryInputType() : GL_NONE;
-        if (gsInput != GL_NONE && mode != GL_PATCHES) {
+        //
+        // "Is there a geometry stage at all" has to be asked of the STAGE, never of the input
+        // primitive: GL_NONE and GL_POINTS are both 0, so a `layout(points) in` geometry shader
+        // is indistinguishable from no geometry shader by its reflected input type alone. The
+        // sentinel test this replaces therefore skipped the whole rule for exactly the geometry
+        // shaders whose input is the most restrictive one - every mode but GL_POINTS was
+        // accepted (KHR-GL43.transform_feedback.api_errors_test draws a points-in geometry
+        // program with GL_LINES and requires INVALID_OPERATION).
+        //
+        // And it has to be asked of the LAST LINK: gsInputPrimitive is a link artifact, so
+        // pairing it with the live attach list would re-point the very same 0-aliasing rather
+        // than remove it. In the window after glAttachShader(GS) on a linked program the live
+        // list says "geometry present" while the artifact still reads GL_NONE == GL_POINTS, and
+        // the switch below would silently reject every mode but GL_POINTS.
+        const Bool geometryActive = currentProgram && currentProgram->HasLinkedShaderStage(ShaderStage::Geometry);
+        const GLenum gsInput = geometryActive ? currentProgram->GetGeometryInputType() : GL_NONE;
+        if (geometryActive && mode != GL_PATCHES) {
             Bool compatible = false;
             switch (gsInput) {
             case GL_POINTS:
@@ -216,24 +382,41 @@ namespace MobileGL::MG_Impl::GLImpl {
         // While transform feedback is active the draw's primitive type must match
         // the feedback primitive mode (GL 3.3 core 13.2.2). With a geometry shader
         // the constraint moves to the shader's output primitive type instead, so
-        // the draw mode itself is unconstrained here. A paused span is exempt: it
-        // captures nothing, so there is nothing for the mode to be incompatible with
-        // (GL 4.6 core 13.2.3).
+        // the draw mode itself is unconstrained here - and a TESSELLATION EVALUATION
+        // stage relocates it exactly the same way (GL 4.6 core 13.2.2 names both):
+        // what is captured is the tessellator's output primitive, and the draw mode
+        // can only ever be GL_PATCHES. A paused span is exempt: it captures nothing,
+        // so there is nothing for the mode to be incompatible with (GL 4.6 core 13.2.3).
+        const auto& feedbackProgram = MG_State::pGLContext->GetTransformFeedbackProgram();
+        // Both stage tests are asked of the last link, for the same reason as the two guards
+        // above: what relocates the constraint is a stage the program actually RUNS, and an
+        // attach that has not been linked in yet gives it none.
+        const Bool feedbackModeIsProgramDriven =
+            feedbackProgram && (feedbackProgram->HasLinkedShaderStage(ShaderStage::Geometry) ||
+                                feedbackProgram->HasLinkedShaderStage(ShaderStage::TessEval));
         if (MG_State::pGLContext->IsTransformFeedbackActive() &&
-            !MG_State::pGLContext->IsTransformFeedbackPaused() &&
-            !(MG_State::pGLContext->GetTransformFeedbackProgram() &&
-              MG_State::pGLContext->GetTransformFeedbackProgram()->GetShaderIndexByStage(ShaderStage::Geometry) >= 0)) {
+            !MG_State::pGLContext->IsTransformFeedbackPaused() && !feedbackModeIsProgramDriven) {
             const GLenum feedbackMode = MG_State::pGLContext->GetTransformFeedbackPrimitiveMode();
             Bool compatible = false;
             switch (feedbackMode) {
             case GL_POINTS:
                 compatible = mode == GL_POINTS;
                 break;
+            // The adjacency modes belong here too (GL 4.6 core table 13.1, ES 3.2 table 12.1).
+            // This arm is only reached when the program has NO geometry or tessellation
+            // evaluation stage, and without a geometry stage the adjacent vertices are simply
+            // ignored (GL 4.6 core 10.1) - the primitive assembled IS a plain line or triangle,
+            // so the combination is legal and must capture. Omitting them raised a spurious
+            // GL_INVALID_OPERATION and dropped the draw entirely, leaving the capture buffer
+            // with its pre-draw bytes. The geometry-stage input table above already carries the
+            // same four arms; this is the second table catching up with it.
             case GL_LINES:
-                compatible = mode == GL_LINES || mode == GL_LINE_STRIP || mode == GL_LINE_LOOP;
+                compatible = mode == GL_LINES || mode == GL_LINE_STRIP || mode == GL_LINE_LOOP ||
+                             mode == GL_LINES_ADJACENCY || mode == GL_LINE_STRIP_ADJACENCY;
                 break;
             case GL_TRIANGLES:
-                compatible = mode == GL_TRIANGLES || mode == GL_TRIANGLE_STRIP || mode == GL_TRIANGLE_FAN;
+                compatible = mode == GL_TRIANGLES || mode == GL_TRIANGLE_STRIP || mode == GL_TRIANGLE_FAN ||
+                             mode == GL_TRIANGLES_ADJACENCY || mode == GL_TRIANGLE_STRIP_ADJACENCY;
                 break;
             default:
                 break;
@@ -303,10 +486,49 @@ namespace MobileGL::MG_Impl::GLImpl {
         }
     }
 
+    // GL 4.6 core 10.3.9: every DrawElements-family count is a sizei and "if count is negative, an
+    // INVALID_VALUE error is generated". The same sentence covers instancecount and the
+    // MultiDraw* drawcount, so one helper serves all of them; the parameter is named for the
+    // caller so the message says which argument the application actually got wrong.
+    static Bool ValidateNonNegativeDrawArgument(const char* functionName, const char* argumentName, GLsizei value) {
+        if (value >= 0) return true;
+        MG_State::pGLContext->RecordError(
+            ErrorCode::InvalidValue,
+            MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName,
+                                         String(argumentName) + " must be non-negative."));
+        return false;
+    }
+
+    // GL 4.6 core 10.3.9 for DrawRangeElements*: "if end < start, an INVALID_VALUE error is
+    // generated". Both are uints, so a caller that passes -1 for start arrives here as
+    // 0xFFFFFFFF and is caught by the same comparison - which is exactly what
+    // KHR-GL4x.draw_elements_base_vertex_tests.invalid_count_argument checks.
+    static Bool ValidateDrawElementsRange(const char* functionName, GLuint start, GLuint end) {
+        if (end >= start) return true;
+        MG_State::pGLContext->RecordError(
+            ErrorCode::InvalidValue,
+            MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName, "end must not be less than start."));
+        return false;
+    }
+
+    // GL 4.6 core 10.9: inside a conditional block whose predicate did not pass, the drawing
+    // commands, Clear, ClearBuffer* and the compute dispatches are DISCARDED. The gate sits on the
+    // wrappers that ISSUE the backend call rather than at the top of each entry point, so that
+    // everything a real driver would still do inside the block - argument validation and the
+    // errors it raises - happens exactly as it does outside one, and only the command itself is
+    // dropped. It is deliberately not on the frontend's transform-feedback accounting either:
+    // that mirrors what the capture stage would have written, and a conditional block around a
+    // capturing draw has no test coverage in either direction.
+    static Bool ConditionalRenderDiscardsCommand() {
+        return MG_State::pGLContext->ConditionalRenderDiscardsCommands();
+    }
+
     void Clear_Backend(GLbitfield mask) {
 #ifdef TRACY_ENABLE
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
+        if (ConditionalRenderDiscardsCommand()) return;
+        MGP_FILL(Clear);
         MG_Backend::gBackendFunctionsTable.GL.Clear(mask);
     }
 
@@ -314,6 +536,8 @@ namespace MobileGL::MG_Impl::GLImpl {
 #ifdef TRACY_ENABLE
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
+        if (ConditionalRenderDiscardsCommand()) return;
+        MGP_FILL(DrawElements);
         MG_Backend::gBackendFunctionsTable.GL.DrawElements(mode, count, type, indices);
     }
 
@@ -322,6 +546,8 @@ namespace MobileGL::MG_Impl::GLImpl {
 #ifdef TRACY_ENABLE
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
+        if (ConditionalRenderDiscardsCommand()) return;
+        MGP_FILL(MultiDrawElements);
         MG_Backend::gBackendFunctionsTable.GL.MultiDrawElements(mode, count, type, indices, drawcount);
     }
 
@@ -330,6 +556,8 @@ namespace MobileGL::MG_Impl::GLImpl {
 #ifdef TRACY_ENABLE
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
+        if (ConditionalRenderDiscardsCommand()) return;
+        MGP_FILL(MultiDrawElementsBaseVertex);
         MG_Backend::gBackendFunctionsTable.GL.MultiDrawElementsBaseVertex(mode, count, type, indices, drawcount,
                                                                           basevertex);
     }
@@ -338,6 +566,8 @@ namespace MobileGL::MG_Impl::GLImpl {
 #ifdef TRACY_ENABLE
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
+        if (ConditionalRenderDiscardsCommand()) return;
+        MGP_FILL(DrawArrays);
         MG_Backend::gBackendFunctionsTable.GL.DrawArrays(mode, first, count);
     }
 
@@ -345,6 +575,8 @@ namespace MobileGL::MG_Impl::GLImpl {
 #ifdef TRACY_ENABLE
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
+        if (ConditionalRenderDiscardsCommand()) return;
+        MGP_FILL(MultiDrawArrays);
         MG_Backend::gBackendFunctionsTable.GL.MultiDrawArrays(mode, first, count, drawcount);
     }
 
@@ -353,6 +585,8 @@ namespace MobileGL::MG_Impl::GLImpl {
 #ifdef TRACY_ENABLE
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
+        if (ConditionalRenderDiscardsCommand()) return;
+        MGP_FILL(DrawElementsBaseVertex);
         MG_Backend::gBackendFunctionsTable.GL.DrawElementsBaseVertex(mode, count, type, indices, basevertex);
     }
 
@@ -361,6 +595,8 @@ namespace MobileGL::MG_Impl::GLImpl {
 #ifdef TRACY_ENABLE
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
+        if (ConditionalRenderDiscardsCommand()) return;
+        MGP_FILL(MultiDrawElementsIndirect);
         MG_Backend::gBackendFunctionsTable.GL.MultiDrawElementsIndirect(mode, type, indirect, drawcount, stride);
     }
 
@@ -368,6 +604,8 @@ namespace MobileGL::MG_Impl::GLImpl {
 #ifdef TRACY_ENABLE
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
+        if (ConditionalRenderDiscardsCommand()) return;
+        MGP_FILL(MultiDrawArraysIndirect);
         MG_Backend::gBackendFunctionsTable.GL.MultiDrawArraysIndirect(mode, indirect, drawcount, stride);
     }
 
@@ -376,6 +614,8 @@ namespace MobileGL::MG_Impl::GLImpl {
 #ifdef TRACY_ENABLE
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
+        if (ConditionalRenderDiscardsCommand()) return;
+        MGP_FILL(MultiDrawElementsIndirectCount);
         MG_Backend::gBackendFunctionsTable.GL.MultiDrawElementsIndirectCount(mode, type, indirect, drawcount,
                                                                              maxdrawcount, stride);
     }
@@ -385,6 +625,8 @@ namespace MobileGL::MG_Impl::GLImpl {
 #ifdef TRACY_ENABLE
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
+        if (ConditionalRenderDiscardsCommand()) return;
+        MGP_FILL(MultiDrawArraysIndirectCount);
         MG_Backend::gBackendFunctionsTable.GL.MultiDrawArraysIndirectCount(mode, indirect, drawcount, maxdrawcount,
                                                                            stride);
     }
@@ -394,6 +636,8 @@ namespace MobileGL::MG_Impl::GLImpl {
 #ifdef TRACY_ENABLE
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
+        if (ConditionalRenderDiscardsCommand()) return;
+        MGP_FILL(DrawRangeElementsBaseVertex);
         MG_Backend::gBackendFunctionsTable.GL.DrawRangeElementsBaseVertex(mode, start, end, count, type, indices,
                                                                           basevertex);
     }
@@ -403,6 +647,8 @@ namespace MobileGL::MG_Impl::GLImpl {
 #ifdef TRACY_ENABLE
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
+        if (ConditionalRenderDiscardsCommand()) return;
+        MGP_FILL(DrawRangeElements);
         MG_Backend::gBackendFunctionsTable.GL.DrawRangeElements(mode, start, end, count, type, indices);
     }
 
@@ -412,6 +658,8 @@ namespace MobileGL::MG_Impl::GLImpl {
 #ifdef TRACY_ENABLE
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
+        if (ConditionalRenderDiscardsCommand()) return;
+        MGP_FILL(DrawElementsInstancedBaseVertexBaseInstance);
         MG_Backend::gBackendFunctionsTable.GL.DrawElementsInstancedBaseVertexBaseInstance(
             mode, count, type, indices, instancecount, basevertex, baseinstance);
     }
@@ -421,6 +669,8 @@ namespace MobileGL::MG_Impl::GLImpl {
 #ifdef TRACY_ENABLE
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
+        if (ConditionalRenderDiscardsCommand()) return;
+        MGP_FILL(DrawElementsInstancedBaseVertex);
         MG_Backend::gBackendFunctionsTable.GL.DrawElementsInstancedBaseVertex(mode, count, type, indices, instancecount,
                                                                               basevertex);
     }
@@ -430,6 +680,8 @@ namespace MobileGL::MG_Impl::GLImpl {
 #ifdef TRACY_ENABLE
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
+        if (ConditionalRenderDiscardsCommand()) return;
+        MGP_FILL(DrawElementsInstancedBaseInstance);
         MG_Backend::gBackendFunctionsTable.GL.DrawElementsInstancedBaseInstance(mode, count, type, indices,
                                                                                 instancecount, baseinstance);
     }
@@ -439,6 +691,8 @@ namespace MobileGL::MG_Impl::GLImpl {
 #ifdef TRACY_ENABLE
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
+        if (ConditionalRenderDiscardsCommand()) return;
+        MGP_FILL(DrawElementsInstanced);
         MG_Backend::gBackendFunctionsTable.GL.DrawElementsInstanced(mode, count, type, indices, instancecount);
     }
 
@@ -446,6 +700,8 @@ namespace MobileGL::MG_Impl::GLImpl {
 #ifdef TRACY_ENABLE
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
+        if (ConditionalRenderDiscardsCommand()) return;
+        MGP_FILL(DrawElementsIndirect);
         MG_Backend::gBackendFunctionsTable.GL.DrawElementsIndirect(mode, type, indirect);
     }
     void DrawArraysInstancedBaseInstance_Backend(GLenum mode, GLint first, GLsizei count, GLsizei instancecount,
@@ -453,6 +709,8 @@ namespace MobileGL::MG_Impl::GLImpl {
 #ifdef TRACY_ENABLE
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
+        if (ConditionalRenderDiscardsCommand()) return;
+        MGP_FILL(DrawArraysInstancedBaseInstance);
         MG_Backend::gBackendFunctionsTable.GL.DrawArraysInstancedBaseInstance(mode, first, count, instancecount,
                                                                               baseinstance);
     }
@@ -461,6 +719,8 @@ namespace MobileGL::MG_Impl::GLImpl {
 #ifdef TRACY_ENABLE
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
+        if (ConditionalRenderDiscardsCommand()) return;
+        MGP_FILL(DrawArraysInstanced);
         MG_Backend::gBackendFunctionsTable.GL.DrawArraysInstanced(mode, first, count, instancecount);
     }
 
@@ -468,6 +728,8 @@ namespace MobileGL::MG_Impl::GLImpl {
 #ifdef TRACY_ENABLE
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
+        if (ConditionalRenderDiscardsCommand()) return;
+        MGP_FILL(DrawArraysIndirect);
         MG_Backend::gBackendFunctionsTable.GL.DrawArraysIndirect(mode, indirect);
     }
 
@@ -496,6 +758,10 @@ namespace MobileGL::MG_Impl::GLImpl {
                 return;
             }
         }
+        // GL 4.3 added both dispatches to the conditional-render set (GL 4.6 core 10.9), which is
+        // exactly what KHR-GL43.compute_shader.conditional-dispatching checks.
+        if (ConditionalRenderDiscardsCommand()) return;
+        MGP_FILL(DispatchCompute);
         dispatchCompute(numGroupsX, numGroupsY, numGroupsZ);
     }
 
@@ -547,6 +813,8 @@ namespace MobileGL::MG_Impl::GLImpl {
             return;
         }
         if (!ValidateCurrentProgramForCompute(__func__)) return;
+        if (ConditionalRenderDiscardsCommand()) return;
+        MGP_FILL(DispatchComputeIndirect);
         dispatchComputeIndirect(indirect);
     }
 
@@ -568,11 +836,70 @@ namespace MobileGL::MG_Impl::GLImpl {
         }
         MG_State::pGLContext->SetPatchVertices(static_cast<Uint>(value));
         if (const auto patchParameteri = MG_Backend::gBackendFunctionsTable.GL.PatchParameteri) {
+            MGP_FILL(PatchParameteri);
             patchParameteri(pname, value);
         }
     }
 
+    // GL 4.6 core 11.2.2. The default tessellation levels a program with an evaluation stage and
+    // NO control stage tessellates at; both backends have to synthesize that control stage
+    // themselves (ES 3.2 and Vulkan both require one), and they compile these numbers into it, so
+    // there is no backend entry point to forward to - ES has none at all. INVALID_ENUM on a bad
+    // pname is the only error the spec lists: any float values are accepted, negatives and NaN
+    // included, and it is the tessellator that clamps them.
+    //
+    // This used to be a stub, which is why the two synthesizers hardcoded 1.0.
+    void PatchParameterfv(GLenum pname, const GLfloat* values) {
+        if (pname != GL_PATCH_DEFAULT_OUTER_LEVEL && pname != GL_PATCH_DEFAULT_INNER_LEVEL) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidEnum,
+                MakeUnique<GenericErrorInfo>(
+                    "MG_Impl/GLImpl", __func__,
+                    "pname must be GL_PATCH_DEFAULT_OUTER_LEVEL or GL_PATCH_DEFAULT_INNER_LEVEL."));
+            return;
+        }
+        if (!values) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "values pointer cannot be null"));
+            return;
+        }
+        if (pname == GL_PATCH_DEFAULT_OUTER_LEVEL) {
+            MG_State::pGLContext->SetPatchDefaultOuterLevel(
+                FloatVec4(values[0], values[1], values[2], values[3]));
+        } else {
+            MG_State::pGLContext->SetPatchDefaultInnerLevel(FloatVec2(values[0], values[1]));
+        }
+    }
+
+    namespace {
+        // GL 4.6 core 7.11.2 (and ARB_shader_image_load_store, which introduced the call): the
+        // barrier bitfield is INVALID_VALUE unless every bit is one of the defined ones, with
+        // GL_ALL_BARRIER_BITS - which is 0xFFFFFFFF, not the union of the list - accepted whole.
+        // Forwarding an undefined bit to the host driver let a caller that had computed its mask
+        // wrongly (or reused an ES-only bit) get silence instead of the error the spec promises.
+        constexpr GLbitfield kAllDefinedBarrierBits =
+            GL_VERTEX_ATTRIB_ARRAY_BARRIER_BIT | GL_ELEMENT_ARRAY_BARRIER_BIT | GL_UNIFORM_BARRIER_BIT |
+            GL_TEXTURE_FETCH_BARRIER_BIT | GL_SHADER_IMAGE_ACCESS_BARRIER_BIT | GL_COMMAND_BARRIER_BIT |
+            GL_PIXEL_BUFFER_BARRIER_BIT | GL_TEXTURE_UPDATE_BARRIER_BIT | GL_BUFFER_UPDATE_BARRIER_BIT |
+            GL_FRAMEBUFFER_BARRIER_BIT | GL_TRANSFORM_FEEDBACK_BARRIER_BIT | GL_ATOMIC_COUNTER_BARRIER_BIT |
+            GL_SHADER_STORAGE_BARRIER_BIT | GL_CLIENT_MAPPED_BUFFER_BARRIER_BIT | GL_QUERY_BUFFER_BARRIER_BIT;
+
+        Bool ValidateMemoryBarrierBits(const char* function, GLbitfield barriers) {
+            if (barriers == GL_ALL_BARRIER_BITS) return true;
+            if ((barriers & ~kAllDefinedBarrierBits) != 0) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidValue,
+                    MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", function,
+                                                 "barriers contains bits that are not defined barrier bits."));
+                return false;
+            }
+            return true;
+        }
+    } // namespace
+
     void MemoryBarrier(GLbitfield barriers) {
+        if (!ValidateMemoryBarrierBits(__func__, barriers)) return;
         auto memoryBarrier = MG_Backend::gBackendFunctionsTable.GL.MemoryBarrier;
         if (!memoryBarrier) {
             MG_State::pGLContext->RecordError(
@@ -580,10 +907,34 @@ namespace MobileGL::MG_Impl::GLImpl {
                 MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "Backend does not support memory barriers."));
             return;
         }
+        MGP_FILL(MemoryBarrier);
         memoryBarrier(barriers);
     }
 
+    void TextureBarrier() {
+        // GL 4.5 core 8.26 / GL_ARB_texture_barrier: order every write the fixed-function
+        // framebuffer has already issued ahead of every subsequent texture fetch, so a shader may
+        // read texels of a texture that is also attached to the current framebuffer.
+        //
+        // Both backends serve this through their existing memory-barrier hook rather than a new
+        // entry point of their own: GL_FRAMEBUFFER_BARRIER_BIT is the source half (framebuffer
+        // writes) and GL_TEXTURE_FETCH_BARRIER_BIT the destination half (texture fetches), which
+        // is exactly the dependency ARB_texture_barrier defines - just expressed with the wider
+        // scope glMemoryBarrier gives it. That is a superset of the required ordering, never a
+        // subset, so it cannot under-synchronize.
+        auto memoryBarrier = MG_Backend::gBackendFunctionsTable.GL.MemoryBarrier;
+        if (!memoryBarrier) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "Backend does not support memory barriers."));
+            return;
+        }
+        MGP_FILL(MemoryBarrier);
+        memoryBarrier(GL_TEXTURE_FETCH_BARRIER_BIT | GL_FRAMEBUFFER_BARRIER_BIT);
+    }
+
     void MemoryBarrierByRegion(GLbitfield barriers) {
+        if (!ValidateMemoryBarrierBits(__func__, barriers)) return;
         auto memoryBarrierByRegion = MG_Backend::gBackendFunctionsTable.GL.MemoryBarrierByRegion;
         if (!memoryBarrierByRegion) {
             MG_State::pGLContext->RecordError(
@@ -592,17 +943,20 @@ namespace MobileGL::MG_Impl::GLImpl {
                                              "Backend does not support regional memory barriers."));
             return;
         }
+        MGP_FILL(MemoryBarrierByRegion);
         memoryBarrierByRegion(barriers);
     }
 
     void MultiDrawElementsIndirect(GLenum mode, GLenum type, const void* indirect, GLsizei drawcount, GLsizei stride) {
-        if (!ValidateCurrentProgramForExecution(__func__)) return;
+        if (!ValidatePrimitiveModeEnum(__func__, mode)) return;
+        if (!PrepareCurrentProgramForDraw(__func__)) return;
         if (!ValidatePrimitiveModeForBackend(__func__, mode)) return;
         MultiDrawElementsIndirect_Backend(mode, type, indirect, drawcount, stride);
     }
 
     void MultiDrawArraysIndirect(GLenum mode, const void* indirect, GLsizei drawcount, GLsizei stride) {
-        if (!ValidateCurrentProgramForExecution(__func__)) return;
+        if (!ValidatePrimitiveModeEnum(__func__, mode)) return;
+        if (!PrepareCurrentProgramForDraw(__func__)) return;
         if (!ValidatePrimitiveModeForBackend(__func__, mode)) return;
         MultiDrawArraysIndirect_Backend(mode, indirect, drawcount, stride);
     }
@@ -680,7 +1034,7 @@ namespace MobileGL::MG_Impl::GLImpl {
         // NegativeApiErrorsTest.IndirectParameterDrawsCheckBothBuffers pins the INVALID_VALUE
         // they produce for a call made with no program bound. Same precedence decision, and
         // the same reason, as DispatchComputeIndirect above.
-        if (!ValidateCurrentProgramForExecution(__func__)) return;
+        if (!PrepareCurrentProgramForDraw(__func__)) return;
         auto multiDrawElementsIndirectCount = MG_Backend::gBackendFunctionsTable.GL.MultiDrawElementsIndirectCount;
         if (!multiDrawElementsIndirectCount) {
             MG_State::pGLContext->RecordError(
@@ -701,7 +1055,7 @@ namespace MobileGL::MG_Impl::GLImpl {
             return;
         }
         // See MultiDrawElementsIndirectCount, including why this one goes last.
-        if (!ValidateCurrentProgramForExecution(__func__)) return;
+        if (!PrepareCurrentProgramForDraw(__func__)) return;
         auto multiDrawArraysIndirectCount = MG_Backend::gBackendFunctionsTable.GL.MultiDrawArraysIndirectCount;
         if (!multiDrawArraysIndirectCount) {
             MG_State::pGLContext->RecordError(
@@ -715,20 +1069,26 @@ namespace MobileGL::MG_Impl::GLImpl {
 
     void DrawRangeElementsBaseVertex(GLenum mode, GLuint start, GLuint end, GLsizei count, GLenum type,
                                      const void* indices, GLint basevertex) {
-        if (!ValidateCurrentProgramForExecution(__func__)) return;
+        if (!ValidatePrimitiveModeEnum(__func__, mode)) return;
+        if (!PrepareCurrentProgramForDraw(__func__)) return;
         if (!ValidatePrimitiveModeForBackend(__func__, mode)) return;
+        if (!ValidateDrawElementsIndexType(__func__, type)) return;
+        if (!ValidateNonNegativeDrawArgument(__func__, "count", count)) return;
+        if (!ValidateDrawElementsRange(__func__, start, end)) return;
         DrawRangeElementsBaseVertex_Backend(mode, start, end, count, type, indices, basevertex);
     }
 
     void DrawRangeElements(GLenum mode, GLuint start, GLuint end, GLsizei count, GLenum type, const void* indices) {
-        if (!ValidateCurrentProgramForExecution(__func__)) return;
+        if (!ValidatePrimitiveModeEnum(__func__, mode)) return;
+        if (!PrepareCurrentProgramForDraw(__func__)) return;
         if (!ValidatePrimitiveModeForBackend(__func__, mode)) return;
         DrawRangeElements_Backend(mode, start, end, count, type, indices);
     }
 
     void DrawElementsInstancedBaseVertexBaseInstance(GLenum mode, GLsizei count, GLenum type, const void* indices,
                                                      GLsizei instancecount, GLint basevertex, GLuint baseinstance) {
-        if (!ValidateCurrentProgramForExecution(__func__)) return;
+        if (!ValidatePrimitiveModeEnum(__func__, mode)) return;
+        if (!PrepareCurrentProgramForDraw(__func__)) return;
         if (!ValidatePrimitiveModeForBackend(__func__, mode)) return;
         DrawElementsInstancedBaseVertexBaseInstance_Backend(mode, count, type, indices, instancecount, basevertex,
                                                             baseinstance);
@@ -736,26 +1096,33 @@ namespace MobileGL::MG_Impl::GLImpl {
 
     void DrawElementsInstancedBaseVertex(GLenum mode, GLsizei count, GLenum type, const void* indices,
                                          GLsizei instancecount, GLint basevertex) {
-        if (!ValidateCurrentProgramForExecution(__func__)) return;
+        if (!ValidatePrimitiveModeEnum(__func__, mode)) return;
+        if (!PrepareCurrentProgramForDraw(__func__)) return;
         if (!ValidatePrimitiveModeForBackend(__func__, mode)) return;
+        if (!ValidateDrawElementsIndexType(__func__, type)) return;
+        if (!ValidateNonNegativeDrawArgument(__func__, "count", count)) return;
+        if (!ValidateNonNegativeDrawArgument(__func__, "instancecount", instancecount)) return;
         DrawElementsInstancedBaseVertex_Backend(mode, count, type, indices, instancecount, basevertex);
     }
 
     void DrawElementsInstancedBaseInstance(GLenum mode, GLsizei count, GLenum type, const void* indices,
                                            GLsizei instancecount, GLuint baseinstance) {
-        if (!ValidateCurrentProgramForExecution(__func__)) return;
+        if (!ValidatePrimitiveModeEnum(__func__, mode)) return;
+        if (!PrepareCurrentProgramForDraw(__func__)) return;
         if (!ValidatePrimitiveModeForBackend(__func__, mode)) return;
         DrawElementsInstancedBaseInstance_Backend(mode, count, type, indices, instancecount, baseinstance);
     }
 
     void DrawElementsInstanced(GLenum mode, GLsizei count, GLenum type, const void* indices, GLsizei instancecount) {
-        if (!ValidateCurrentProgramForExecution(__func__)) return;
+        if (!ValidatePrimitiveModeEnum(__func__, mode)) return;
+        if (!PrepareCurrentProgramForDraw(__func__)) return;
         if (!ValidatePrimitiveModeForBackend(__func__, mode)) return;
         DrawElementsInstanced_Backend(mode, count, type, indices, instancecount);
     }
 
     void DrawElementsIndirect(GLenum mode, GLenum type, const void* indirect) {
-        if (!ValidateCurrentProgramForExecution(__func__)) return;
+        if (!ValidatePrimitiveModeEnum(__func__, mode)) return;
+        if (!PrepareCurrentProgramForDraw(__func__)) return;
         if (!ValidatePrimitiveModeForBackend(__func__, mode)) return;
         if (!ValidateDrawElementsIndexType(__func__, type)) return;
         if (!ValidateIndirectDrawSource(__func__, indirect, kDrawElementsIndirectCommandBytes)) return;
@@ -764,40 +1131,48 @@ namespace MobileGL::MG_Impl::GLImpl {
 
     void DrawArraysInstancedBaseInstance(GLenum mode, GLint first, GLsizei count, GLsizei instancecount,
                                          GLuint baseinstance) {
-        if (!ValidateCurrentProgramForExecution(__func__)) return;
+        if (!ValidatePrimitiveModeEnum(__func__, mode)) return;
+        if (!PrepareCurrentProgramForDraw(__func__)) return;
         if (!ValidatePrimitiveModeForBackend(__func__, mode)) return;
         DrawArraysInstancedBaseInstance_Backend(mode, first, count, instancecount, baseinstance);
     }
 
     void DrawArraysInstanced(GLenum mode, GLint first, GLsizei count, GLsizei instancecount) {
-        if (!ValidateCurrentProgramForExecution(__func__)) return;
+        if (!ValidatePrimitiveModeEnum(__func__, mode)) return;
+        if (!PrepareCurrentProgramForDraw(__func__)) return;
         if (!ValidatePrimitiveModeForBackend(__func__, mode)) return;
         DrawArraysInstanced_Backend(mode, first, count, instancecount);
     }
 
     void DrawArraysIndirect(GLenum mode, const void* indirect) {
-        if (!ValidateCurrentProgramForExecution(__func__)) return;
+        if (!ValidatePrimitiveModeEnum(__func__, mode)) return;
+        if (!PrepareCurrentProgramForDraw(__func__)) return;
         if (!ValidatePrimitiveModeForBackend(__func__, mode)) return;
         if (!ValidateIndirectDrawSource(__func__, indirect, kDrawArraysIndirectCommandBytes)) return;
         DrawArraysIndirect_Backend(mode, indirect);
     }
 
     void DrawElementsBaseVertex(GLenum mode, GLsizei count, GLenum type, const void* indices, GLint basevertex) {
-        if (!ValidateCurrentProgramForExecution(__func__)) return;
+        if (!ValidatePrimitiveModeEnum(__func__, mode)) return;
+        if (!PrepareCurrentProgramForDraw(__func__)) return;
         if (!ValidatePrimitiveModeForBackend(__func__, mode)) return;
+        if (!ValidateDrawElementsIndexType(__func__, type)) return;
+        if (!ValidateNonNegativeDrawArgument(__func__, "count", count)) return;
         AccountTransformFeedbackPrimitives(mode, count);
         DrawElementsBaseVertex_Backend(mode, count, type, indices, basevertex);
     }
 
     void DrawArrays(GLenum mode, GLint first, GLsizei count) {
-        if (!ValidateCurrentProgramForExecution(__func__)) return;
+        if (!ValidatePrimitiveModeEnum(__func__, mode)) return;
+        if (!PrepareCurrentProgramForDraw(__func__)) return;
         if (!ValidatePrimitiveModeForBackend(__func__, mode)) return;
         AccountTransformFeedbackPrimitives(mode, count);
         DrawArrays_Backend(mode, first, count);
     }
 
     void MultiDrawArrays(GLenum mode, const GLint* first, const GLsizei* count, GLsizei drawcount) {
-        if (!ValidateCurrentProgramForExecution(__func__)) return;
+        if (!ValidatePrimitiveModeEnum(__func__, mode)) return;
+        if (!PrepareCurrentProgramForDraw(__func__)) return;
         if (!ValidatePrimitiveModeForBackend(__func__, mode)) return;
         if (drawcount < 0) {
             MG_State::pGLContext->RecordError(
@@ -810,15 +1185,30 @@ namespace MobileGL::MG_Impl::GLImpl {
 
     void MultiDrawElements(GLenum mode, const GLsizei* count, GLenum type, const void* const* indices,
                            GLsizei drawcount) {
-        if (!ValidateCurrentProgramForExecution(__func__)) return;
+        if (!ValidatePrimitiveModeEnum(__func__, mode)) return;
+        if (!PrepareCurrentProgramForDraw(__func__)) return;
         if (!ValidatePrimitiveModeForBackend(__func__, mode)) return;
         MultiDrawElements_Backend(mode, count, type, indices, drawcount);
     }
 
     void MultiDrawElementsBaseVertex(GLenum mode, const GLsizei* count, GLenum type, const void* const* indices,
                                      GLsizei drawcount, const GLint* basevertex) {
-        if (!ValidateCurrentProgramForExecution(__func__)) return;
+        if (!ValidatePrimitiveModeEnum(__func__, mode)) return;
+        if (!PrepareCurrentProgramForDraw(__func__)) return;
         if (!ValidatePrimitiveModeForBackend(__func__, mode)) return;
+        if (!ValidateDrawElementsIndexType(__func__, type)) return;
+        if (!ValidateNonNegativeDrawArgument(__func__, "drawcount", drawcount)) return;
+        // GL 4.6 core 10.5 defines MultiDrawElementsBaseVertex as drawcount separate
+        // DrawElementsBaseVertex calls, so each element of the count array carries the same
+        // non-negative requirement the single-draw entry point applies to its own count. The
+        // whole call is rejected before any sub-draw is issued, which is what makes the error
+        // observable at all - a driver that drew the valid prefix first would leave the
+        // framebuffer half-written.
+        if (count != nullptr) {
+            for (GLsizei draw = 0; draw < drawcount; ++draw) {
+                if (!ValidateNonNegativeDrawArgument(__func__, "every element of count", count[draw])) return;
+            }
+        }
         MultiDrawElementsBaseVertex_Backend(mode, count, type, indices, drawcount, basevertex);
     }
 
@@ -827,7 +1217,8 @@ namespace MobileGL::MG_Impl::GLImpl {
     }
 
     void DrawElements(GLenum mode, GLsizei count, GLenum type, const void* indices) {
-        if (!ValidateCurrentProgramForExecution(__func__)) return;
+        if (!ValidatePrimitiveModeEnum(__func__, mode)) return;
+        if (!PrepareCurrentProgramForDraw(__func__)) return;
         if (!ValidatePrimitiveModeForBackend(__func__, mode)) return;
         AccountTransformFeedbackPrimitives(mode, count);
         DrawElements_Backend(mode, count, type, indices);
@@ -875,6 +1266,7 @@ namespace MobileGL::MG_Impl::GLImpl {
         }
         MG_State::pGLContext->BeginTransformFeedback(primitiveMode, program);
         if (const auto beginXfb = MG_Backend::gBackendFunctionsTable.GL.BeginTransformFeedback) {
+            MGP_FILL(BeginTransformFeedback);
             beginXfb(primitiveMode);
         }
     }
@@ -957,6 +1349,7 @@ namespace MobileGL::MG_Impl::GLImpl {
         // Closed while the capture state is still active: a backend that captures
         // through its own driver reads the capture program and buffer bindings here.
         if (const auto endXfb = MG_Backend::gBackendFunctionsTable.GL.EndTransformFeedback) {
+            MGP_FILL(EndTransformFeedback);
             endXfb();
         }
         MG_State::pGLContext->EndTransformFeedback();
@@ -965,9 +1358,12 @@ namespace MobileGL::MG_Impl::GLImpl {
         // the GPU work is all that is required.
         auto& backendGL = MG_Backend::gBackendFunctionsTable.GL;
         if (backendGL.FenceSync && backendGL.ClientWaitSync) {
+            MGP_FILL(FenceSync);
             if (auto sync = backendGL.FenceSync()) {
+                MGP_FILL(ClientWaitSync);
                 backendGL.ClientWaitSync(sync, GL_SYNC_FLUSH_COMMANDS_BIT, ~0ull);
                 if (backendGL.DeleteSync) {
+                    MGP_FILL(DeleteSync);
                     backendGL.DeleteSync(sync);
                 }
             }
@@ -986,6 +1382,7 @@ namespace MobileGL::MG_Impl::GLImpl {
         }
         MG_State::pGLContext->SetTransformFeedbackPaused(true);
         if (const auto pauseXfb = MG_Backend::gBackendFunctionsTable.GL.PauseTransformFeedback) {
+            MGP_FILL(PauseTransformFeedback);
             pauseXfb();
         }
     }
@@ -1000,6 +1397,7 @@ namespace MobileGL::MG_Impl::GLImpl {
         }
         MG_State::pGLContext->SetTransformFeedbackPaused(false);
         if (const auto resumeXfb = MG_Backend::gBackendFunctionsTable.GL.ResumeTransformFeedback) {
+            MGP_FILL(ResumeTransformFeedback);
             resumeXfb();
         }
     }
@@ -1205,6 +1603,7 @@ namespace MobileGL::MG_Impl::GLImpl {
                 continue;
             }
             if (const auto deleteXfb = MG_Backend::gBackendFunctionsTable.GL.DeleteTransformFeedback) {
+                MGP_FILL(DeleteTransformFeedback);
                 deleteXfb(id);
             }
             MG_State::pGLContext->MarkTransformFeedbackObjectForDeletion(id);
@@ -1236,6 +1635,7 @@ namespace MobileGL::MG_Impl::GLImpl {
         }
         MG_State::pGLContext->BindTransformFeedbackObject(id);
         if (const auto bindXfb = MG_Backend::gBackendFunctionsTable.GL.BindTransformFeedback) {
+            MGP_FILL(BindTransformFeedback);
             bindXfb(id);
         }
     }
@@ -1251,7 +1651,7 @@ namespace MobileGL::MG_Impl::GLImpl {
     // (GL 4.6 core 10.3.7).
     static void DrawTransformFeedbackImpl(const char* functionName, GLenum mode, GLuint id, GLuint stream,
                                           GLsizei instancecount) {
-        if (!ValidateCurrentProgramForExecution(functionName)) return;
+        if (!PrepareCurrentProgramForDraw(functionName)) return;
         if (!ValidatePrimitiveModeForBackend(functionName, mode)) return;
         if (instancecount < 0) {
             MG_State::pGLContext->RecordError(
@@ -1274,8 +1674,13 @@ namespace MobileGL::MG_Impl::GLImpl {
                                              std::to_string(id) + " is not a transform feedback object name."));
             return;
         }
-        // GL_MAX_VERTEX_STREAMS is 1, so stream 0 is the only one that exists.
-        if (stream != 0) {
+        // GL 4.6 core 10.3.7 bounds `stream` by GL_MAX_VERTEX_STREAMS, which this implementation
+        // answers as 1 - so stream 0 is the only one that exists and anything else is
+        // INVALID_VALUE. Read from the getter rather than written as `stream != 0` so the two can
+        // never drift: if vertex-stream support ever lands, this bound moves with the limit.
+        GLint maxVertexStreams = 1;
+        GetIntegerv(GL_MAX_VERTEX_STREAMS, &maxVertexStreams);
+        if (stream >= static_cast<GLuint>(std::max(maxVertexStreams, 1))) {
             MG_State::pGLContext->RecordError(
                 ErrorCode::InvalidValue,
                 MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName,
@@ -1293,6 +1698,7 @@ namespace MobileGL::MG_Impl::GLImpl {
             return;
         }
 
+        // `stream` is provably 0 here (the bound above is 1), so this is stream 0's record.
         const Uint64 vertices = MG_State::pGLContext->GetTransformFeedbackRecordedVertices(id);
         if (vertices == 0) return;
         const auto count = static_cast<GLsizei>(vertices);

MobileGL/MG_Impl/GLImpl/Drawing/GL_Drawing.h

@@ -32,8 +32,10 @@ namespace MobileGL::MG_Impl::GLImpl {
     void DispatchCompute(GLuint numGroupsX, GLuint numGroupsY, GLuint numGroupsZ);
     void DispatchComputeIndirect(GLintptr indirect);
     void PatchParameteri(GLenum pname, GLint value);
+    void PatchParameterfv(GLenum pname, const GLfloat* values);
     void MemoryBarrier(GLbitfield barriers);
     void MemoryBarrierByRegion(GLbitfield barriers);
+    void TextureBarrier();
     void MultiDrawElementsIndirect(GLenum mode, GLenum type, const void* indirect, GLsizei drawcount, GLsizei stride);
     void MultiDrawArraysIndirect(GLenum mode, const void* indirect, GLsizei drawcount, GLsizei stride);
     void MultiDrawElementsIndirectCount(GLenum mode, GLenum type, const void* indirect, GLintptr drawcount,

MobileGL/MG_Impl/GLImpl/Exporting/Definitions.cpp

@@ -20,6 +20,7 @@
 #include "../Framebuffer/GL_Framebuffer.h"
 #include "../VertexArray/GL_VertexArray.h"
 #include "../Sync/GL_Sync.h"
+#include "../Debug/GL_Debug.h"
 #include <MG_State/GLState/Core.h>
 
 #define DECLARE_GL_FUNCTION_STUB_HEAD(type, name, ...) MOBILEGL_GL_API type gl##name(__VA_ARGS__) {
@@ -159,7 +160,7 @@ DECLARE_GL_FUNCTION_STUB_HEAD(void, ReleaseShaderCompiler) DECLARE_GL_FUNCTION_S
 DECLARE_GL_FUNCTION_HEAD(void, RenderbufferStorage, GLenum target, GLenum internalformat, GLsizei width, GLsizei height) DECLARE_GL_FUNCTION_END_NO_RETURN(void, RenderbufferStorage, target, internalformat, width, height)
 DECLARE_GL_FUNCTION_HEAD(void, SampleCoverage, GLfloat value, GLboolean invert) DECLARE_GL_FUNCTION_END_NO_RETURN(void, SampleCoverage, value, invert)
 DECLARE_GL_FUNCTION_HEAD(void, Scissor, GLint x, GLint y, GLsizei width, GLsizei height) DECLARE_GL_FUNCTION_END_NO_RETURN(void, Scissor, x, y, width, height)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ShaderBinary, GLsizei count, const GLuint* shaders, GLenum binaryformat, const void* binary, GLsizei length) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ShaderBinary, count, shaders, binaryformat, binary, length)
+DECLARE_GL_FUNCTION_HEAD(void, ShaderBinary, GLsizei count, const GLuint* shaders, GLenum binaryformat, const void* binary, GLsizei length) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ShaderBinary, count, shaders, binaryformat, binary, length)
 DECLARE_GL_FUNCTION_HEAD(void, ShaderSource, GLuint shader, GLsizei count, const GLchar* const* string, const GLint* length) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ShaderSource, shader, count, string, length)
 DECLARE_GL_FUNCTION_HEAD(void, StencilFunc, GLenum func, GLint ref, GLuint mask) DECLARE_GL_FUNCTION_END_NO_RETURN(void, StencilFunc, func, ref, mask)
 DECLARE_GL_FUNCTION_HEAD(void, StencilFuncSeparate, GLenum face, GLenum func, GLint ref, GLuint mask) DECLARE_GL_FUNCTION_END_NO_RETURN(void, StencilFuncSeparate, face, func, ref, mask)
@@ -378,27 +379,13 @@ DECLARE_GL_FUNCTION_HEAD(void, VertexBindingDivisor, GLuint bindingindex, GLuint
 DECLARE_GL_FUNCTION_STUB_HEAD(void, BlendBarrier) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, BlendBarrier)
 DECLARE_GL_FUNCTION_HEAD(void, CopyImageSubData, GLuint srcName, GLenum srcTarget, GLint srcLevel, GLint srcX, GLint srcY, GLint srcZ, GLuint dstName, GLenum dstTarget, GLint dstLevel, GLint dstX, GLint dstY, GLint dstZ, GLsizei srcWidth, GLsizei srcHeight, GLsizei srcDepth) DECLARE_GL_FUNCTION_END_NO_RETURN(void, CopyImageSubData, srcName, srcTarget, srcLevel, srcX, srcY, srcZ, dstName, dstTarget, dstLevel, dstX, dstY, dstZ, srcWidth, srcHeight, srcDepth)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, DebugMessageControl, GLenum source, GLenum type, GLenum severity, GLsizei count, const GLuint* ids, GLboolean enabled) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, DebugMessageControl, source, type, severity, count, ids, enabled)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, DebugMessageInsert, GLenum source, GLenum type, GLuint id, GLenum severity, GLsizei length, const GLchar* buf) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, DebugMessageInsert, source, type, id, severity, length, buf)
+DECLARE_GL_FUNCTION_HEAD(void, DebugMessageInsert, GLenum source, GLenum type, GLuint id, GLenum severity, GLsizei length, const GLchar* buf) DECLARE_GL_FUNCTION_END_NO_RETURN(void, DebugMessageInsert, source, type, id, severity, length, buf)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, DebugMessageCallback, GLDEBUGPROC callback, const void* userParam) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, DebugMessageCallback, callback, userParam)
 DECLARE_GL_FUNCTION_STUB_HEAD(GLuint, GetDebugMessageLog, GLuint count, GLsizei bufSize, GLenum* sources, GLenum* types, GLuint* ids, GLenum* severities, GLsizei* lengths, GLchar* messageLog) DECLARE_GL_FUNCTION_STUB_END(GLuint, GetDebugMessageLog, count, bufSize, sources, types, ids, severities, lengths, messageLog)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, PushDebugGroup, GLenum source, GLuint id, GLsizei length, const GLchar* message) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, PushDebugGroup, source, id, length, message)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, PopDebugGroup) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, PopDebugGroup)
-MOBILEGL_GL_API void glObjectLabel(GLenum identifier, GLuint name, GLsizei length, const GLchar* label) {
-    (void)identifier;
-    (void)name;
-    (void)length;
-    (void)label;
-}
-MOBILEGL_GL_API void glGetObjectLabel(GLenum identifier, GLuint name, GLsizei bufSize, GLsizei* length, GLchar* label) {
-    (void)identifier;
-    (void)name;
-    if (length) {
-        *length = 0;
-    }
-    if (label && bufSize > 0) {
-        label[0] = '\0';
-    }
-}
+DECLARE_GL_FUNCTION_HEAD(void, PushDebugGroup, GLenum source, GLuint id, GLsizei length, const GLchar* message) DECLARE_GL_FUNCTION_END_NO_RETURN(void, PushDebugGroup, source, id, length, message)
+DECLARE_GL_FUNCTION_HEAD(void, PopDebugGroup) DECLARE_GL_FUNCTION_END_NO_RETURN(void, PopDebugGroup)
+DECLARE_GL_FUNCTION_HEAD(void, ObjectLabel, GLenum identifier, GLuint name, GLsizei length, const GLchar* label) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ObjectLabel, identifier, name, length, label)
+DECLARE_GL_FUNCTION_HEAD(void, GetObjectLabel, GLenum identifier, GLuint name, GLsizei bufSize, GLsizei* length, GLchar* label) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GetObjectLabel, identifier, name, bufSize, length, label)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, ObjectPtrLabel, const void* ptr, GLsizei length, const GLchar* label) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ObjectPtrLabel, ptr, length, label)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetObjectPtrLabel, const void* ptr, GLsizei bufSize, GLsizei* length, GLchar* label) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetObjectPtrLabel, ptr, bufSize, length, label)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetPointerv, GLenum pname, void** params) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetPointerv, pname, params)
@@ -424,7 +411,7 @@ DECLARE_GL_FUNCTION_HEAD(void, ReadnPixels, GLint x, GLint y, GLsizei width, GLs
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetnUniformfv, GLuint program, GLint location, GLsizei bufSize, GLfloat* params) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetnUniformfv, program, location, bufSize, params)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetnUniformiv, GLuint program, GLint location, GLsizei bufSize, GLint* params) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetnUniformiv, program, location, bufSize, params)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetnUniformuiv, GLuint program, GLint location, GLsizei bufSize, GLuint* params) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetnUniformuiv, program, location, bufSize, params)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, MinSampleShading, GLfloat value) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, MinSampleShading, value)
+DECLARE_GL_FUNCTION_HEAD(void, MinSampleShading, GLfloat value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, MinSampleShading, value)
 DECLARE_GL_FUNCTION_HEAD(void, PatchParameteri, GLenum pname, GLint value) DECLARE_GL_FUNCTION_END_NO_RETURN(void, PatchParameteri, pname, value)
 DECLARE_GL_FUNCTION_HEAD(void, TexParameterIiv, GLenum target, GLenum pname, const GLint* params) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TexParameterIiv, target, pname, params)
 DECLARE_GL_FUNCTION_HEAD(void, TexParameterIuiv, GLenum target, GLenum pname, const GLuint* params) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TexParameterIuiv, target, pname, params)
@@ -725,8 +712,8 @@ DECLARE_GL_FUNCTION_STUB_HEAD(void, LoadName, GLuint name) DECLARE_GL_FUNCTION_S
 DECLARE_GL_FUNCTION_STUB_HEAD(void, PushName, GLuint name) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, PushName, name)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, PopName) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, PopName)
 DECLARE_GL_FUNCTION_HEAD(void, ClampColor, GLenum target, GLenum clamp) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ClampColor, target, clamp)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, BeginConditionalRender, GLuint id, GLenum mode) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, BeginConditionalRender, id, mode)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, EndConditionalRender, void) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, EndConditionalRender)
+DECLARE_GL_FUNCTION_HEAD(void, BeginConditionalRender, GLuint id, GLenum mode) DECLARE_GL_FUNCTION_END_NO_RETURN(void, BeginConditionalRender, id, mode)
+DECLARE_GL_FUNCTION_HEAD(void, EndConditionalRender) DECLARE_GL_FUNCTION_END_NO_RETURN(void, EndConditionalRender)
 DECLARE_GL_FUNCTION_HEAD(void, VertexAttribI1i, GLuint index, GLint x) DECLARE_GL_FUNCTION_END_NO_RETURN(void, VertexAttribI1i, index, x)
 DECLARE_GL_FUNCTION_HEAD(void, VertexAttribI2i, GLuint index, GLint x, GLint y) DECLARE_GL_FUNCTION_END_NO_RETURN(void, VertexAttribI2i, index, x, y)
 DECLARE_GL_FUNCTION_HEAD(void, VertexAttribI3i, GLuint index, GLint x, GLint y, GLint z) DECLARE_GL_FUNCTION_END_NO_RETURN(void, VertexAttribI3i, index, x, y, z)
@@ -936,7 +923,7 @@ DECLARE_GL_FUNCTION_STUB_HEAD(void, GetActiveSubroutineName, GLuint program, GLe
 DECLARE_GL_FUNCTION_STUB_HEAD(void, UniformSubroutinesuiv, GLenum shadertype, GLsizei count, const GLuint* indices) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, UniformSubroutinesuiv, shadertype, count, indices)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetUniformSubroutineuiv, GLenum shadertype, GLint location, GLuint* params) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetUniformSubroutineuiv, shadertype, location, params)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetProgramStageiv, GLuint program, GLenum shadertype, GLenum pname, GLint* values) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetProgramStageiv, program, shadertype, pname, values)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, PatchParameterfv, GLenum pname, const GLfloat* values) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, PatchParameterfv, pname, values)
+DECLARE_GL_FUNCTION_HEAD(void, PatchParameterfv, GLenum pname, const GLfloat* values) DECLARE_GL_FUNCTION_END_NO_RETURN(void, PatchParameterfv, pname, values)
 DECLARE_GL_FUNCTION_HEAD(void, DrawTransformFeedback, GLenum mode, GLuint id) DECLARE_GL_FUNCTION_END_NO_RETURN(void, DrawTransformFeedback, mode, id)
 DECLARE_GL_FUNCTION_HEAD(void, DrawTransformFeedbackStream, GLenum mode, GLuint id, GLuint stream) DECLARE_GL_FUNCTION_END_NO_RETURN(void, DrawTransformFeedbackStream, mode, id, stream)
 DECLARE_GL_FUNCTION_HEAD(void, BeginQueryIndexed, GLenum target, GLuint index, GLuint id) DECLARE_GL_FUNCTION_END_NO_RETURN(void, BeginQueryIndexed, target, index, id)
@@ -982,11 +969,11 @@ DECLARE_GL_FUNCTION_HEAD(void, GetDoublei_v, GLenum target, GLuint index, GLdoub
 DECLARE_GL_FUNCTION_HEAD(void, DrawArraysInstancedBaseInstance, GLenum mode, GLint first, GLsizei count, GLsizei instancecount, GLuint baseinstance) DECLARE_GL_FUNCTION_END_NO_RETURN(void, DrawArraysInstancedBaseInstance, mode, first, count, instancecount, baseinstance)
 DECLARE_GL_FUNCTION_HEAD(void, DrawElementsInstancedBaseInstance, GLenum mode, GLsizei count, GLenum type, const void* indices, GLsizei instancecount, GLuint baseinstance) DECLARE_GL_FUNCTION_END_NO_RETURN(void, DrawElementsInstancedBaseInstance, mode, count, type, indices, instancecount, baseinstance)
 DECLARE_GL_FUNCTION_HEAD(void, DrawElementsInstancedBaseVertexBaseInstance, GLenum mode, GLsizei count, GLenum type, const void* indices, GLsizei instancecount, GLint basevertex, GLuint baseinstance) DECLARE_GL_FUNCTION_END_NO_RETURN(void, DrawElementsInstancedBaseVertexBaseInstance, mode, count, type, indices, instancecount, basevertex, baseinstance)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, GetActiveAtomicCounterBufferiv, GLuint program, GLuint bufferIndex, GLenum pname, GLint* params) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetActiveAtomicCounterBufferiv, program, bufferIndex, pname, params)
+DECLARE_GL_FUNCTION_HEAD(void, GetActiveAtomicCounterBufferiv, GLuint program, GLuint bufferIndex, GLenum pname, GLint* params) DECLARE_GL_FUNCTION_END_NO_RETURN(void, GetActiveAtomicCounterBufferiv, program, bufferIndex, pname, params)
 DECLARE_GL_FUNCTION_HEAD(void, DrawTransformFeedbackInstanced, GLenum mode, GLuint id, GLsizei instancecount) DECLARE_GL_FUNCTION_END_NO_RETURN(void, DrawTransformFeedbackInstanced, mode, id, instancecount)
 DECLARE_GL_FUNCTION_HEAD(void, DrawTransformFeedbackStreamInstanced, GLenum mode, GLuint id, GLuint stream, GLsizei instancecount) DECLARE_GL_FUNCTION_END_NO_RETURN(void, DrawTransformFeedbackStreamInstanced, mode, id, stream, instancecount)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ClearBufferData, GLenum target, GLenum internalformat, GLenum format, GLenum type, const void* data) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ClearBufferData, target, internalformat, format, type, data)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ClearBufferSubData, GLenum target, GLenum internalformat, GLintptr offset, GLsizeiptr size, GLenum format, GLenum type, const void* data) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ClearBufferSubData, target, internalformat, offset, size, format, type, data)
+DECLARE_GL_FUNCTION_HEAD(void, ClearBufferData, GLenum target, GLenum internalformat, GLenum format, GLenum type, const void* data) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ClearBufferData, target, internalformat, format, type, data)
+DECLARE_GL_FUNCTION_HEAD(void, ClearBufferSubData, GLenum target, GLenum internalformat, GLintptr offset, GLsizeiptr size, GLenum format, GLenum type, const void* data) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ClearBufferSubData, target, internalformat, offset, size, format, type, data)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetInternalformati64v, GLenum target, GLenum internalformat, GLenum pname, GLsizei count, GLint64* params) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetInternalformati64v, target, internalformat, pname, count, params)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, InvalidateTexSubImage, GLuint texture, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, InvalidateTexSubImage, texture, level, xoffset, yoffset, zoffset, width, height, depth)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, InvalidateTexImage, GLuint texture, GLint level) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, InvalidateTexImage, texture, level)
@@ -996,7 +983,7 @@ DECLARE_GL_FUNCTION_HEAD(void, MultiDrawArraysIndirect, GLenum mode, const void*
 DECLARE_GL_FUNCTION_HEAD(void, MultiDrawElementsIndirect, GLenum mode, GLenum type, const void* indirect, GLsizei drawcount, GLsizei stride) DECLARE_GL_FUNCTION_END_NO_RETURN(void, MultiDrawElementsIndirect, mode, type, indirect, drawcount, stride)
 DECLARE_GL_FUNCTION_HEAD(GLint, GetProgramResourceLocationIndex, GLuint program, GLenum programInterface, const GLchar* name) DECLARE_GL_FUNCTION_END(GLint, GetProgramResourceLocationIndex, program, programInterface, name)
 DECLARE_GL_FUNCTION_HEAD(void, ShaderStorageBlockBinding, GLuint program, GLuint storageBlockIndex, GLuint storageBlockBinding) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ShaderStorageBlockBinding, program, storageBlockIndex, storageBlockBinding)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, TextureView, GLuint texture, GLenum target, GLuint origtexture, GLenum internalformat, GLuint minlevel, GLuint numlevels, GLuint minlayer, GLuint numlayers) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, TextureView, texture, target, origtexture, internalformat, minlevel, numlevels, minlayer, numlayers)
+DECLARE_GL_FUNCTION_HEAD(void, TextureView, GLuint texture, GLenum target, GLuint origtexture, GLenum internalformat, GLuint minlevel, GLuint numlevels, GLuint minlayer, GLuint numlayers) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TextureView, texture, target, origtexture, internalformat, minlevel, numlevels, minlayer, numlayers)
 DECLARE_GL_FUNCTION_HEAD(void, VertexAttribLFormat, GLuint attribindex, GLint size, GLenum type, GLuint relativeoffset) DECLARE_GL_FUNCTION_END_NO_RETURN(void, VertexAttribLFormat, attribindex, size, type, relativeoffset)
 DECLARE_GL_FUNCTION_HEAD(void, BufferStorage, GLenum target, GLsizeiptr size, const void* data, GLbitfield flags) DECLARE_GL_FUNCTION_END_NO_RETURN(void, BufferStorage, target, size, data, flags)
 DECLARE_GL_FUNCTION_HEAD(void, ClearTexImage, GLuint texture, GLint level, GLenum format, GLenum type, const void* data) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ClearTexImage, texture, level, format, type, data)
@@ -1007,7 +994,7 @@ DECLARE_GL_FUNCTION_HEAD(void, BindTextures, GLuint first, GLsizei count, const
 DECLARE_GL_FUNCTION_HEAD(void, BindSamplers, GLuint first, GLsizei count, const GLuint* samplers) DECLARE_GL_FUNCTION_END_NO_RETURN(void, BindSamplers, first, count, samplers)
 DECLARE_GL_FUNCTION_HEAD(void, BindImageTextures, GLuint first, GLsizei count, const GLuint* textures) DECLARE_GL_FUNCTION_END_NO_RETURN(void, BindImageTextures, first, count, textures)
 DECLARE_GL_FUNCTION_HEAD(void, BindVertexBuffers, GLuint first, GLsizei count, const GLuint* buffers, const GLintptr* offsets, const GLsizei* strides) DECLARE_GL_FUNCTION_END_NO_RETURN(void, BindVertexBuffers, first, count, buffers, offsets, strides)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, ClipControl, GLenum origin, GLenum depth) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ClipControl, origin, depth)
+DECLARE_GL_FUNCTION_HEAD(void, ClipControl, GLenum origin, GLenum depth) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ClipControl, origin, depth)
 DECLARE_GL_FUNCTION_HEAD(void, CreateTransformFeedbacks, GLsizei n, GLuint* ids) DECLARE_GL_FUNCTION_END_NO_RETURN(void, CreateTransformFeedbacks, n, ids)
 DECLARE_GL_FUNCTION_HEAD(void, TransformFeedbackBufferBase, GLuint xfb, GLuint index, GLuint buffer) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TransformFeedbackBufferBase, xfb, index, buffer)
 DECLARE_GL_FUNCTION_HEAD(void, TransformFeedbackBufferRange, GLuint xfb, GLuint index, GLuint buffer, GLintptr offset, GLsizeiptr size) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TransformFeedbackBufferRange, xfb, index, buffer, offset, size)
@@ -1060,9 +1047,9 @@ DECLARE_GL_FUNCTION_HEAD(void, TextureStorage3DMultisample, GLuint texture, GLsi
 DECLARE_GL_FUNCTION_HEAD(void, TextureSubImage1D, GLuint texture, GLint level, GLint xoffset, GLsizei width, GLenum format, GLenum type, const void* pixels) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TextureSubImage1D, texture, level, xoffset, width, format, type, pixels)
 DECLARE_GL_FUNCTION_HEAD(void, TextureSubImage2D, GLuint texture, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, GLenum format, GLenum type, const void* pixels) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TextureSubImage2D, texture, level, xoffset, yoffset, width, height, format, type, pixels)
 DECLARE_GL_FUNCTION_HEAD(void, TextureSubImage3D, GLuint texture, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLenum type, const void* pixels) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TextureSubImage3D, texture, level, xoffset, yoffset, zoffset, width, height, depth, format, type, pixels)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, CompressedTextureSubImage1D, GLuint texture, GLint level, GLint xoffset, GLsizei width, GLenum format, GLsizei imageSize, const void* data) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, CompressedTextureSubImage1D, texture, level, xoffset, width, format, imageSize, data)
+DECLARE_GL_FUNCTION_HEAD(void, CompressedTextureSubImage1D, GLuint texture, GLint level, GLint xoffset, GLsizei width, GLenum format, GLsizei imageSize, const void* data) DECLARE_GL_FUNCTION_END_NO_RETURN(void, CompressedTextureSubImage1D, texture, level, xoffset, width, format, imageSize, data)
 DECLARE_GL_FUNCTION_HEAD(void, CompressedTextureSubImage2D, GLuint texture, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, GLenum format, GLsizei imageSize, const void* data) DECLARE_GL_FUNCTION_END_NO_RETURN(void, CompressedTextureSubImage2D, texture, level, xoffset, yoffset, width, height, format, imageSize, data)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, CompressedTextureSubImage3D, GLuint texture, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const void* data) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, CompressedTextureSubImage3D, texture, level, xoffset, yoffset, zoffset, width, height, depth, format, imageSize, data)
+DECLARE_GL_FUNCTION_HEAD(void, CompressedTextureSubImage3D, GLuint texture, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const void* data) DECLARE_GL_FUNCTION_END_NO_RETURN(void, CompressedTextureSubImage3D, texture, level, xoffset, yoffset, zoffset, width, height, depth, format, imageSize, data)
 DECLARE_GL_FUNCTION_HEAD(void, CopyTextureSubImage1D, GLuint texture, GLint level, GLint xoffset, GLint x, GLint y, GLsizei width) DECLARE_GL_FUNCTION_END_NO_RETURN(void, CopyTextureSubImage1D, texture, level, xoffset, x, y, width)
 DECLARE_GL_FUNCTION_HEAD(void, CopyTextureSubImage2D, GLuint texture, GLint level, GLint xoffset, GLint yoffset, GLint x, GLint y, GLsizei width, GLsizei height) DECLARE_GL_FUNCTION_END_NO_RETURN(void, CopyTextureSubImage2D, texture, level, xoffset, yoffset, x, y, width, height)
 DECLARE_GL_FUNCTION_HEAD(void, CopyTextureSubImage3D, GLuint texture, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLint x, GLint y, GLsizei width, GLsizei height) DECLARE_GL_FUNCTION_END_NO_RETURN(void, CopyTextureSubImage3D, texture, level, xoffset, yoffset, zoffset, x, y, width, height)
@@ -1120,11 +1107,11 @@ DECLARE_GL_FUNCTION_STUB_HEAD(void, GetnConvolutionFilter, GLenum target, GLenum
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetnSeparableFilter, GLenum target, GLenum format, GLenum type, GLsizei rowBufSize, void* row, GLsizei columnBufSize, void* column, void* span) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetnSeparableFilter, target, format, type, rowBufSize, row, columnBufSize, column, span)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetnHistogram, GLenum target, GLboolean reset, GLenum format, GLenum type, GLsizei bufSize, void* values) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetnHistogram, target, reset, format, type, bufSize, values)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetnMinmax, GLenum target, GLboolean reset, GLenum format, GLenum type, GLsizei bufSize, void* values) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetnMinmax, target, reset, format, type, bufSize, values)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, TextureBarrier, void) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, TextureBarrier, )
-DECLARE_GL_FUNCTION_STUB_HEAD(void, SpecializeShader, GLuint shader, const GLchar* pEntryPoint, GLuint numSpecializationConstants, const GLuint* pConstantIndex, const GLuint* pConstantValue) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, SpecializeShader, shader, pEntryPoint, numSpecializationConstants, pConstantIndex, pConstantValue)
+DECLARE_GL_FUNCTION_HEAD(void, TextureBarrier, void) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TextureBarrier, )
+DECLARE_GL_FUNCTION_HEAD(void, SpecializeShader, GLuint shader, const GLchar* pEntryPoint, GLuint numSpecializationConstants, const GLuint* pConstantIndex, const GLuint* pConstantValue) DECLARE_GL_FUNCTION_END_NO_RETURN(void, SpecializeShader, shader, pEntryPoint, numSpecializationConstants, pConstantIndex, pConstantValue)
 DECLARE_GL_FUNCTION_HEAD(void, MultiDrawArraysIndirectCount, GLenum mode, const void* indirect, GLintptr drawcount, GLsizei maxdrawcount, GLsizei stride) DECLARE_GL_FUNCTION_END_NO_RETURN(void, MultiDrawArraysIndirectCount, mode, indirect, drawcount, maxdrawcount, stride)
 DECLARE_GL_FUNCTION_HEAD(void, MultiDrawElementsIndirectCount, GLenum mode, GLenum type, const void* indirect, GLintptr drawcount, GLsizei maxdrawcount, GLsizei stride) DECLARE_GL_FUNCTION_END_NO_RETURN(void, MultiDrawElementsIndirectCount, mode, type, indirect, drawcount, maxdrawcount, stride)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, PolygonOffsetClamp, GLfloat factor, GLfloat units, GLfloat clamp) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, PolygonOffsetClamp, factor, units, clamp)
+DECLARE_GL_FUNCTION_HEAD(void, PolygonOffsetClamp, GLfloat factor, GLfloat units, GLfloat clamp) DECLARE_GL_FUNCTION_END_NO_RETURN(void, PolygonOffsetClamp, factor, units, clamp)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, PrimitiveBoundingBoxARB, GLfloat minX, GLfloat minY, GLfloat minZ, GLfloat minW, GLfloat maxX, GLfloat maxY, GLfloat maxZ, GLfloat maxW) DECLARE_GL_FUNCTION_STUB_END(void, PrimitiveBoundingBoxARB, minX, minY, minZ, minW, maxX, maxY, maxZ, maxW)
 DECLARE_GL_FUNCTION_STUB_HEAD(GLuint64, GetTextureHandleARB, GLuint texture) DECLARE_GL_FUNCTION_STUB_END(GLuint64, GetTextureHandleARB, texture)
 DECLARE_GL_FUNCTION_STUB_HEAD(GLuint64, GetTextureSamplerHandleARB, GLuint texture, GLuint sampler) DECLARE_GL_FUNCTION_STUB_END(GLuint64, GetTextureSamplerHandleARB, texture, sampler)
@@ -1163,7 +1150,7 @@ DECLARE_GL_FUNCTION_STUB_HEAD(void, GetProgramLocalParameterdvARB, GLenum target
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetProgramLocalParameterfvARB, GLenum target, GLuint index, GLfloat* params) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetProgramLocalParameterfvARB, target, index, params)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetProgramStringARB, GLenum target, GLenum pname, void* string) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetProgramStringARB, target, pname, string)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, FramebufferTextureFaceARB, GLenum target, GLenum attachment, GLuint texture, GLint level, GLenum face) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, FramebufferTextureFaceARB, target, attachment, texture, level, face)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, SpecializeShaderARB, GLuint shader, const GLchar* pEntryPoint, GLuint numSpecializationConstants, const GLuint* pConstantIndex, const GLuint* pConstantValue) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, SpecializeShaderARB, shader, pEntryPoint, numSpecializationConstants, pConstantIndex, pConstantValue)
+DECLARE_GL_FUNCTION_HEAD(void, SpecializeShaderARB, GLuint shader, const GLchar* pEntryPoint, GLuint numSpecializationConstants, const GLuint* pConstantIndex, const GLuint* pConstantValue) DECLARE_GL_FUNCTION_END_NO_RETURN(void, SpecializeShader, shader, pEntryPoint, numSpecializationConstants, pConstantIndex, pConstantValue)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, Uniform1i64ARB, GLint location, GLint64 x) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, Uniform1i64ARB, location, x)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, Uniform2i64ARB, GLint location, GLint64 x, GLint64 y) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, Uniform2i64ARB, location, x, y)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, Uniform3i64ARB, GLint location, GLint64 x, GLint64 y, GLint64 z) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, Uniform3i64ARB, location, x, y, z)
@@ -1848,9 +1835,9 @@ DECLARE_GL_FUNCTION_STUB_HEAD(void, GetBooleanIndexedvEXT, GLenum target, GLuint
 DECLARE_GL_FUNCTION_STUB_HEAD(void, CompressedTextureImage3DEXT, GLuint texture, GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLsizei imageSize, const void* bits) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, CompressedTextureImage3DEXT, texture, target, level, internalformat, width, height, depth, border, imageSize, bits)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, CompressedTextureImage2DEXT, GLuint texture, GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLint border, GLsizei imageSize, const void* bits) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, CompressedTextureImage2DEXT, texture, target, level, internalformat, width, height, border, imageSize, bits)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, CompressedTextureImage1DEXT, GLuint texture, GLenum target, GLint level, GLenum internalformat, GLsizei width, GLint border, GLsizei imageSize, const void* bits) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, CompressedTextureImage1DEXT, texture, target, level, internalformat, width, border, imageSize, bits)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, CompressedTextureSubImage3DEXT, GLuint texture, GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const void* bits) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, CompressedTextureSubImage3DEXT, texture, target, level, xoffset, yoffset, zoffset, width, height, depth, format, imageSize, bits)
+DECLARE_GL_FUNCTION_HEAD(void, CompressedTextureSubImage3DEXT, GLuint texture, GLenum target, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize, const void* bits) DECLARE_GL_FUNCTION_END_NO_RETURN(void, CompressedTextureSubImage3D, texture, level, xoffset, yoffset, zoffset, width, height, depth, format, imageSize, bits)
 DECLARE_GL_FUNCTION_HEAD(void, CompressedTextureSubImage2DEXT, GLuint texture, GLenum target, GLint level, GLint xoffset, GLint yoffset, GLsizei width, GLsizei height, GLenum format, GLsizei imageSize, const void* bits) DECLARE_GL_FUNCTION_END_NO_RETURN(void, CompressedTextureSubImage2D, texture, level, xoffset, yoffset, width, height, format, imageSize, bits)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, CompressedTextureSubImage1DEXT, GLuint texture, GLenum target, GLint level, GLint xoffset, GLsizei width, GLenum format, GLsizei imageSize, const void* bits) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, CompressedTextureSubImage1DEXT, texture, target, level, xoffset, width, format, imageSize, bits)
+DECLARE_GL_FUNCTION_HEAD(void, CompressedTextureSubImage1DEXT, GLuint texture, GLenum target, GLint level, GLint xoffset, GLsizei width, GLenum format, GLsizei imageSize, const void* bits) DECLARE_GL_FUNCTION_END_NO_RETURN(void, CompressedTextureSubImage1D, texture, level, xoffset, width, format, imageSize, bits)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetCompressedTextureImageEXT, GLuint texture, GLenum target, GLint lod, void* img) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetCompressedTextureImageEXT, texture, target, lod, img)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, CompressedMultiTexImage3DEXT, GLenum texunit, GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLsizei depth, GLint border, GLsizei imageSize, const void* bits) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, CompressedMultiTexImage3DEXT, texunit, target, level, internalformat, width, height, depth, border, imageSize, bits)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, CompressedMultiTexImage2DEXT, GLenum texunit, GLenum target, GLint level, GLenum internalformat, GLsizei width, GLsizei height, GLint border, GLsizei imageSize, const void* bits) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, CompressedMultiTexImage2DEXT, texunit, target, level, internalformat, width, height, border, imageSize, bits)
@@ -2062,7 +2049,7 @@ DECLARE_GL_FUNCTION_STUB_HEAD(void, GetPixelTransformParameterivEXT, GLenum targ
 DECLARE_GL_FUNCTION_STUB_HEAD(void, GetPixelTransformParameterfvEXT, GLenum target, GLenum pname, GLfloat* params) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, GetPixelTransformParameterfvEXT, target, pname, params)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, PointParameterfEXT, GLenum pname, GLfloat param) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, PointParameterfEXT, pname, param)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, PointParameterfvEXT, GLenum pname, const GLfloat* params) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, PointParameterfvEXT, pname, params)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, PolygonOffsetClampEXT, GLfloat factor, GLfloat units, GLfloat clamp) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, PolygonOffsetClampEXT, factor, units, clamp)
+DECLARE_GL_FUNCTION_HEAD(void, PolygonOffsetClampEXT, GLfloat factor, GLfloat units, GLfloat clamp) DECLARE_GL_FUNCTION_END_NO_RETURN(void, PolygonOffsetClamp, factor, units, clamp)
 DECLARE_GL_FUNCTION_HEAD(void, ProvokingVertexEXT, GLenum mode) DECLARE_GL_FUNCTION_END_NO_RETURN(void, ProvokingVertex, mode)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, RasterSamplesEXT, GLuint samples, GLboolean fixedsamplelocations) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, RasterSamplesEXT, samples, fixedsamplelocations)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, SecondaryColor3bEXT, GLbyte red, GLbyte green, GLbyte blue) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, SecondaryColor3bEXT, red, green, blue)
@@ -2559,7 +2546,7 @@ DECLARE_GL_FUNCTION_STUB_HEAD(void, ShadingRateImageBarrierNV, GLboolean synchro
 DECLARE_GL_FUNCTION_STUB_HEAD(void, ShadingRateImagePaletteNV, GLuint viewport, GLuint first, GLsizei count, const GLenum* rates) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ShadingRateImagePaletteNV, viewport, first, count, rates)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, ShadingRateSampleOrderNV, GLenum order) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ShadingRateSampleOrderNV, order)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, ShadingRateSampleOrderCustomNV, GLenum rate, GLuint samples, const GLint* locations) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, ShadingRateSampleOrderCustomNV, rate, samples, locations)
-DECLARE_GL_FUNCTION_STUB_HEAD(void, TextureBarrierNV, void) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, TextureBarrierNV, )
+DECLARE_GL_FUNCTION_HEAD(void, TextureBarrierNV, void) DECLARE_GL_FUNCTION_END_NO_RETURN(void, TextureBarrier, )
 DECLARE_GL_FUNCTION_STUB_HEAD(void, TexImage2DMultisampleCoverageNV, GLenum target, GLsizei coverageSamples, GLsizei colorSamples, GLint internalFormat, GLsizei width, GLsizei height, GLboolean fixedSampleLocations) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, TexImage2DMultisampleCoverageNV, target, coverageSamples, colorSamples, internalFormat, width, height, fixedSampleLocations)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, TexImage3DMultisampleCoverageNV, GLenum target, GLsizei coverageSamples, GLsizei colorSamples, GLint internalFormat, GLsizei width, GLsizei height, GLsizei depth, GLboolean fixedSampleLocations) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, TexImage3DMultisampleCoverageNV, target, coverageSamples, colorSamples, internalFormat, width, height, depth, fixedSampleLocations)
 DECLARE_GL_FUNCTION_STUB_HEAD(void, TextureImage2DMultisampleNV, GLuint texture, GLenum target, GLsizei samples, GLint internalFormat, GLsizei width, GLsizei height, GLboolean fixedSampleLocations) DECLARE_GL_FUNCTION_STUB_END_NO_RETURN(void, TextureImage2DMultisampleNV, texture, target, samples, internalFormat, width, height, fixedSampleLocations)

MobileGL/MG_Impl/GLImpl/Framebuffer/GL_Framebuffer.cpp

@@ -13,7 +13,9 @@
 #include <MG_Backend/BackendObjects.h>
 #include <MG_Util/Metrics/TextureMetrics.h>
 #include <MG_Impl/GLImpl/Texture/Validators.h>
+#include <MG_Impl/GLImpl/Getter/GL_Getter.h>
 #include <MG_State/GLState/ErrorState/Error.h>
+#include <MG_Impl/Pipe/PipeFill.h>
 #include <MG_Util/Converters/GLToStr/GLEnumConverter.h>
 #include <MG_Util/Converters/GLToMG/TextureEnumConverter.h>
 #include <MG_Util/Converters/MGToMG/TextureEnumConverter.h>
@@ -473,6 +475,75 @@ namespace MobileGL::MG_Impl::GLImpl {
             }
         }
 
+        // GL 4.6 core 9.2.8 conditions that depend only on the framebuffer and the attachment
+        // point. Shared, because glFramebufferTexture / 1D / 2D / 3D / TextureLayer are aliases of
+        // one another in that section and a CTS case that walks the family must not get five
+        // different answers - which is exactly what happened when these lived in one helper that
+        // only two of the five went through.
+        Bool ValidateFramebufferTextureAttachmentPoint(const char* functionName,
+                                                       const SharedPtr<MG_State::GLState::FramebufferObject>&
+                                                           framebufferObject,
+                                                       FramebufferAttachmentType attachmentType) {
+            // "An INVALID_OPERATION error is generated if COLOR_ATTACHMENTm is used with m greater
+            // than or equal to MAX_COLOR_ATTACHMENTS."
+            if (!FramebufferImpl::ValidateColorAttachmentInRange(attachmentType, functionName)) return false;
+            // "An INVALID_OPERATION error is generated if zero is bound to target." MobileGL keeps
+            // a real FramebufferObject for framebuffer 0, so a null test can never see this - the
+            // object is always there, and framebuffer 0 has to be recognised by identity instead,
+            // the same comparison DrawBuffers_State makes. Without this an attach onto the default
+            // framebuffer silently REPLACED its colour attachment, permanently desynchronising it
+            // from what the swapchain keeps publishing.
+            const auto& defaultFramebufferInfo = FramebufferImpl::pDefaultFramebufferInfo;
+            if (!framebufferObject ||
+                (defaultFramebufferInfo && framebufferObject == defaultFramebufferInfo->defaultFBO)) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidOperation,
+                    MakeUnique<GenericErrorInfo>(
+                        "MG_Impl/GLImpl", functionName,
+                        "No framebuffer object is bound to the target; the default framebuffer's attachments "
+                        "cannot be named."));
+                return false;
+            }
+            return true;
+        }
+
+        // The other half of 9.2.8: "level must be greater than or equal to zero", and for a
+        // texture with immutable storage it "must be smaller than the number of levels the texture
+        // has". Split from the attachment-point half because the caller only has a texture object
+        // once the detach (texture == 0) case is behind it.
+        Bool ValidateFramebufferTextureLevel(const char* functionName,
+                                             const SharedPtr<MG_State::GLState::ITextureObject>& textureObject,
+                                             GLint level) {
+            if (level < 0) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidValue,
+                    MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName,
+                                                 "Texture level must be non-negative."));
+                return false;
+            }
+            if (!textureObject || !textureObject->IsImmutable()) {
+                // A mutable texture has no level bound here: a level it has not specified yet is
+                // not an error, it just leaves the framebuffer incomplete.
+                return true;
+            }
+            // GetAddressableLevelCount(), NOT GetImmutableLevels(): for a VIEW the latter is
+            // deliberately the ORIGINAL texture's count (GL 4.6 core 8.18 defines
+            // TEXTURE_IMMUTABLE_LEVELS on a view that way), which is far too large a bound - a
+            // two-level view onto a ten-level texture would accept level 5 and attach an image
+            // nothing can draw into.
+            const Uint levelBound = textureObject->GetAddressableLevelCount();
+            if (static_cast<Uint>(level) >= levelBound) {
+                MG_State::pGLContext->RecordError(
+                    ErrorCode::InvalidValue,
+                    MakeUnique<GenericErrorInfo>(
+                        "MG_Impl/GLImpl", functionName,
+                        std::format("Texture level {} is beyond the {} level(s) this texture has.", level,
+                                    levelBound)));
+                return false;
+            }
+            return true;
+        }
+
         void AttachFramebufferTextureWithUploadTarget(const char* functionName, GLenum target, GLenum attachment,
                                                       GLuint texture, GLint level,
                                                       TextureUploadTarget textureUploadTarget, Bool layered = false) {
@@ -481,10 +552,24 @@ namespace MobileGL::MG_Impl::GLImpl {
             }
 
             if (attachment == GL_DEPTH_STENCIL_ATTACHMENT) {
+                // `layered` has to travel with the split. GL_DEPTH_STENCIL_ATTACHMENT is only a
+                // shorthand for attaching the same image to both halves (GL 4.6 core 9.2.6), so
+                // whether glFramebufferTexture made it LAYERED is a property of the call, not of
+                // which half is being recorded - and dropping it here (the parameter defaults to
+                // false) recorded a non-layered depth/stencil attachment beside a layered colour
+                // one for every layered target. That is an inconsistent framebuffer by 9.4.1's
+                // own rule, and downstream it means the depth/stencil attachment covers layer 0
+                // alone: DirectVulkan built its view with layerCount 1 under a framebuffer
+                // declaring N layers (VUID-VkFramebufferCreateInfo-flags-04535), and DirectGLES
+                // attached one layer of it beside a layered colour target, which the driver
+                // answers with GL_FRAMEBUFFER_INCOMPLETE_LAYER_TARGETS - every draw silently
+                // produced nothing. This is the shape
+                // texture_cube_map_array.stencil_attachments_*_layered and
+                // geometry_shader.layered_framebuffer.stencil_support are built on.
                 AttachFramebufferTextureWithUploadTarget(functionName, target, GL_DEPTH_ATTACHMENT, texture, level,
-                                                        textureUploadTarget);
+                                                        textureUploadTarget, layered);
                 AttachFramebufferTextureWithUploadTarget(functionName, target, GL_STENCIL_ATTACHMENT, texture, level,
-                                                        textureUploadTarget);
+                                                        textureUploadTarget, layered);
                 return;
             }
 
@@ -496,13 +581,7 @@ namespace MobileGL::MG_Impl::GLImpl {
 
             auto& bindingSlot = MG_State::pGLContext->GetFramebufferBindingSlot(framebufferTarget);
             auto& framebufferObject = bindingSlot.GetBoundObject();
-            if (!framebufferObject) {
-                MG_State::pGLContext->RecordError(
-                    ErrorCode::InvalidOperation,
-                    MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName,
-                                                 "Framebuffer target is bound to no framebuffer object."));
-                return;
-            }
+            if (!ValidateFramebufferTextureAttachmentPoint(functionName, framebufferObject, attachmentType)) return;
 
             if (texture == 0) {
                 framebufferObject->Detach(attachmentType);
@@ -517,6 +596,7 @@ namespace MobileGL::MG_Impl::GLImpl {
                                                  std::format("Texture object {} is not valid.", texture)));
                 return;
             }
+            if (!ValidateFramebufferTextureLevel(functionName, textureObject, level)) return;
 
             const auto expectedTextureTarget = MG_Util::ConvertTextureUploadTargetToTextureTarget(textureUploadTarget);
             if (expectedTextureTarget == TextureTarget::Unknown ||
@@ -537,6 +617,7 @@ namespace MobileGL::MG_Impl::GLImpl {
 
     void BlitFramebuffer_Backend(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0,
                                  GLint dstX1, GLint dstY1, GLbitfield mask, GLenum filter) {
+        MGP_FILL(BlitFramebuffer);
         MG_Backend::gBackendFunctionsTable.GL.BlitFramebuffer(srcX0, srcY0, srcX1, srcY1, dstX0, dstY0, dstX1, dstY1,
                                                               mask, filter);
     }
@@ -550,6 +631,7 @@ namespace MobileGL::MG_Impl::GLImpl {
             MGLOG_E_ONCE("glBlitNamedFramebuffer skipped: backend does not implement explicit framebuffer blit.");
             return;
         }
+        MGP_FILL(BlitNamedFramebuffer);
         blitNamedFramebuffer(readFramebuffer, drawFramebuffer, srcX0, srcY0, srcX1, srcY1, dstX0, dstY0, dstX1,
                              dstY1, mask, filter);
     }
@@ -561,6 +643,7 @@ namespace MobileGL::MG_Impl::GLImpl {
             MGLOG_E_ONCE("glClearNamedFramebufferfv skipped: backend does not implement explicit framebuffer clear.");
             return;
         }
+        MGP_FILL(ClearNamedFramebufferfv);
         clearNamedFramebufferfv(framebuffer, buffer, drawbuffer, value);
     }
 
@@ -571,6 +654,7 @@ namespace MobileGL::MG_Impl::GLImpl {
             MGLOG_E_ONCE("glClearNamedFramebufferfi skipped: backend does not implement explicit framebuffer clear.");
             return;
         }
+        MGP_FILL(ClearNamedFramebufferfi);
         clearNamedFramebufferfi(framebuffer, buffer, drawbuffer, depth, stencil);
     }
 
@@ -581,6 +665,7 @@ namespace MobileGL::MG_Impl::GLImpl {
             MGLOG_E_ONCE("glClearNamedFramebufferiv skipped: backend does not implement explicit framebuffer clear.");
             return;
         }
+        MGP_FILL(ClearNamedFramebufferiv);
         clearNamedFramebufferiv(framebuffer, buffer, drawbuffer, value);
     }
 
@@ -591,6 +676,7 @@ namespace MobileGL::MG_Impl::GLImpl {
             MGLOG_E_ONCE("glClearNamedFramebufferuiv skipped: backend does not implement explicit framebuffer clear.");
             return;
         }
+        MGP_FILL(ClearNamedFramebufferuiv);
         clearNamedFramebufferuiv(framebuffer, buffer, drawbuffer, value);
     }
 
@@ -617,21 +703,39 @@ namespace MobileGL::MG_Impl::GLImpl {
         if (MG_Backend::pActiveBackendObject == nullptr) {
             return std::numeric_limits<Int>::max();
         }
-        return std::max(MG_Backend::pActiveBackendObject->GetDynamicParameters().MaxSamples, 1);
+        return GetAdvertisedMaxSamples();
+    }
+
+    // GL_MAX_SAMPLES is the ceiling over all formats; an integer format has its own
+    // (GL_MAX_INTEGER_SAMPLES) and GL 4.6 core 9.2.4 makes exceeding it INVALID_OPERATION.
+    // The multisample TEXTURE path resolves the limit per format the same way
+    // (GL_Texture.cpp, GetMaxSupportedTextureSamples), and both now enforce exactly what their
+    // pname advertises. The integer ceiling used to be floored at GL_MAX_SAMPLES so that the
+    // frontend would accept a count it had advertised globally - but on Adreno and Mali the
+    // integer path is genuinely one sample, and accepting four only moved the failure from an
+    // honest INVALID_OPERATION here to a silently under-allocated renderbuffer.
+    // The head of the per-format renderbuffer sample list the backend probed, or 0 when nothing
+    // was probed for it. Same shape as GetProbedMaxTextureSamples in GL_Texture.cpp, and reads
+    // the same cache glGetInternalformativ(GL_RENDERBUFFER, ..., GL_SAMPLES) answers from.
+    static Int GetProbedMaxRenderbufferSamples(TextureInternalFormat format) {
+        if (MG_Backend::pActiveBackendObject == nullptr) {
+            return 0;
+        }
+        const SizeT targetIndex = MG_Backend::GetRenderbufferFormatCapabilityTargetIndex();
+        const SizeT formatIndex = static_cast<SizeT>(format);
+        if (targetIndex >= MG_Backend::kFormatCapabilityTargetCount ||
+            formatIndex >= MG_Backend::kFormatCapabilityFormatCount) {
+            return 0;
+        }
+        const auto& sampleCounts =
+            MG_Backend::pActiveBackendObject->GetFormatCapabilities().SampleCounts[targetIndex][formatIndex];
+        return sampleCounts.empty() ? 0 : sampleCounts.front();
     }
 
-    // GL_MAX_SAMPLES is the ceiling over all formats; an integer format has its own, lower
-    // one (GL_MAX_INTEGER_SAMPLES) and GL 4.6 core 9.2.4 makes exceeding it INVALID_OPERATION.
-    // The multisample TEXTURE path already resolves the limit per format
-    // (GL_Texture.cpp, GetMaxTextureSamplesForFormat); renderbuffers only ever compared
-    // against GL_MAX_SAMPLES, so on a driver where the two differ - Adreno reports
-    // GL_MAX_SAMPLES 4 and GL_MAX_INTEGER_SAMPLES 1 - an integer renderbuffer accepted a
-    // sample count the format cannot deliver, and said GL_NO_ERROR about it.
     Int GetMaxRenderbufferSamplesForFormat_State(TextureInternalFormat format) {
         if (MG_Backend::pActiveBackendObject == nullptr) {
             return std::numeric_limits<Int>::max();
         }
-        const auto& dynamicParameters = MG_Backend::pActiveBackendObject->GetDynamicParameters();
 
         GLenum normalizedInternalFormat = MG_Util::ConvertTextureInternalFormatToGLEnum(format);
         GLenum normalizedFormat = GL_RGBA;
@@ -642,10 +746,24 @@ namespace MobileGL::MG_Impl::GLImpl {
                                                               &normalizedType);
         const Bool isIntegerFormat = normalizedFormat == GL_RED_INTEGER || normalizedFormat == GL_RG_INTEGER ||
                                      normalizedFormat == GL_RGB_INTEGER || normalizedFormat == GL_RGBA_INTEGER;
+        // The per-format probe first, for the same reason the texture path takes it first: GL 4.6
+        // core 9.2.4 words the error as "samples is greater than the maximum number of samples
+        // supported for internalformat (see GetInternalformativ)", and
+        // glGetInternalformativ(GL_RENDERBUFFER, ..., GL_SAMPLES) is answered from exactly this
+        // list. It was never consulted here - the TODO that deferred it was written before the
+        // query was backed and had gone stale - so a format whose multisample probes fail inside
+        // a category that allows four was accepted at four, quietly allocated at one by
+        // ClampSamplesToBackendSupport, and then reported as four by
+        // glGetRenderbufferParameteriv(GL_RENDERBUFFER_SAMPLES).
+        const Int probedMaxSamples = GetProbedMaxRenderbufferSamples(format);
+        if (probedMaxSamples > 0) {
+            return probedMaxSamples;
+        }
         if (!isIntegerFormat) {
             return GetMaxRenderbufferSamples_State();
         }
-        return std::max(dynamicParameters.MaxIntegerSamples, 1);
+        // Exactly what glGetIntegerv(GL_MAX_INTEGER_SAMPLES) reports.
+        return GetAdvertisedIntegerMaxSamples();
     }
 
     Bool ValidateRenderbufferStorageSize_State(GLsizei width, GLsizei height, const char* caller) {
@@ -677,8 +795,10 @@ namespace MobileGL::MG_Impl::GLImpl {
             return false;
         }
 
-        // TODO: Resolve the remaining per-internalformat renderbuffer sample limits once
-        // glGetInternalformativ is backed; integer formats are handled below.
+        // Per-internalformat, from the probe list glGetInternalformativ answers with, falling back
+        // to the format's category pname where nothing was probed. (This carried a TODO deferring
+        // the per-format resolution "once glGetInternalformativ is backed"; it has been backed for
+        // both renderbuffers and multisample textures since, so the deferral was collected.)
         const Int maxSamples = GetMaxRenderbufferSamplesForFormat_State(format);
         if (samples > maxSamples) {
             // GL 4.6 core 9.2.4 makes asking for more samples than the format supports
@@ -1043,13 +1163,7 @@ namespace MobileGL::MG_Impl::GLImpl {
 
         auto& bindingSlot = MG_State::pGLContext->GetFramebufferBindingSlot(framebufferTarget);
         auto& framebufferObject = bindingSlot.GetBoundObject();
-        if (!framebufferObject) {
-            MG_State::pGLContext->RecordError(
-                ErrorCode::InvalidOperation,
-                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName,
-                                             "Framebuffer target is bound to no framebuffer object."));
-            return;
-        }
+        if (!ValidateFramebufferTextureAttachmentPoint(functionName, framebufferObject, attachmentType)) return;
 
         if (texture == 0) {
             framebufferObject->Detach(attachmentType);
@@ -1064,6 +1178,7 @@ namespace MobileGL::MG_Impl::GLImpl {
                                              std::format("Texture object {} is not valid.", texture)));
             return;
         }
+        if (!ValidateFramebufferTextureLevel(functionName, textureObject, level)) return;
         if (layer < 0) {
             MG_State::pGLContext->RecordError(
                 ErrorCode::InvalidValue,
@@ -1186,6 +1301,13 @@ namespace MobileGL::MG_Impl::GLImpl {
                                              "Framebuffer target is bound to no framebuffer object."));
             return;
         }
+        // glFramebufferTexture2D is by far the most-used member of the family and the only one
+        // that inlines its own logic instead of going through the shared helper, so the 9.2.8
+        // conditions have to be asked here explicitly.
+        if (!ValidateFramebufferTextureAttachmentPoint("FramebufferTexture2D_State", framebufferObject,
+                                                       attachmentType)) {
+            return;
+        }
 
         if (texture == 0) {
             framebufferObject->Detach(attachmentType);
@@ -1200,6 +1322,7 @@ namespace MobileGL::MG_Impl::GLImpl {
                                              std::format("Texture object {} is not valid.", texture)));
             return;
         }
+        if (!ValidateFramebufferTextureLevel("FramebufferTexture2D_State", textureObject, level)) return;
 
         const auto expectedTextureTarget = MG_Util::ConvertTextureUploadTargetToTextureTarget(textureUploadTarget);
         if (expectedTextureTarget == TextureTarget::Unknown ||
@@ -1236,6 +1359,12 @@ namespace MobileGL::MG_Impl::GLImpl {
             return;
         }
 
+        // The name's validity is an INVALID_VALUE condition (GL 4.6 core 9.2.8), and it has to be
+        // asked BEFORE the object is resolved: reporting the miss as the INVALID_OPERATION below
+        // pre-empted the shared helper's ValidateTextureName and answered the wrong error code for
+        // every texture name that was never generated.
+        if (!TextureImpl::ValidateTextureName(texture, true)) return;
+
         auto& textureObject = MG_State::pGLContext->GetTextureObject(texture);
         if (!textureObject) {
             MG_State::pGLContext->RecordError(
@@ -1286,13 +1415,10 @@ namespace MobileGL::MG_Impl::GLImpl {
                                              std::format("Texture object {} is not valid.", texture)));
             return;
         }
-        if (level < 0) {
-            MG_State::pGLContext->RecordError(
-                ErrorCode::InvalidValue,
-                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", "NamedFramebufferTexture_State",
-                                             "Texture level must be non-negative."));
-            return;
-        }
+        // The whole level condition, not just its negative half: glNamedFramebufferTexture and
+        // glFramebufferTexture are equivalent in 9.2.8, so an out-of-range immutable level has to
+        // be rejected on both or a CTS case gets two answers for one rule.
+        if (!ValidateFramebufferTextureLevel("NamedFramebufferTexture_State", textureObject, level)) return;
 
         TextureUploadTarget textureUploadTarget = TextureUploadTarget::Unknown;
         Bool layered = false;
@@ -2608,18 +2734,30 @@ namespace MobileGL::MG_Impl::GLImpl {
     }
 
     void ClearBufferfi_Backend(GLenum buffer, GLint drawbuffer, GLfloat depth, GLint stencil) {
+        // GL 4.6 core 10.9 makes ClearBuffer* conditional alongside the drawing commands.
+        if (MG_State::pGLContext->ConditionalRenderDiscardsCommands()) return;
+        MGP_FILL(ClearBufferfi);
         MG_Backend::gBackendFunctionsTable.GL.ClearBufferfi(buffer, drawbuffer, depth, stencil);
     }
 
     void ClearBufferfv_Backend(GLenum buffer, GLint drawbuffer, const GLfloat* value) {
+        // GL 4.6 core 10.9 makes ClearBuffer* conditional alongside the drawing commands.
+        if (MG_State::pGLContext->ConditionalRenderDiscardsCommands()) return;
+        MGP_FILL(ClearBufferfv);
         MG_Backend::gBackendFunctionsTable.GL.ClearBufferfv(buffer, drawbuffer, value);
     }
 
     void ClearBufferuiv_Backend(GLenum buffer, GLint drawbuffer, const GLuint* value) {
+        // GL 4.6 core 10.9 makes ClearBuffer* conditional alongside the drawing commands.
+        if (MG_State::pGLContext->ConditionalRenderDiscardsCommands()) return;
+        MGP_FILL(ClearBufferuiv);
         MG_Backend::gBackendFunctionsTable.GL.ClearBufferuiv(buffer, drawbuffer, value);
     }
 
     void ClearBufferiv_Backend(GLenum buffer, GLint drawbuffer, const GLint* value) {
+        // GL 4.6 core 10.9 makes ClearBuffer* conditional alongside the drawing commands.
+        if (MG_State::pGLContext->ConditionalRenderDiscardsCommands()) return;
+        MGP_FILL(ClearBufferiv);
         MG_Backend::gBackendFunctionsTable.GL.ClearBufferiv(buffer, drawbuffer, value);
     }
 
@@ -2867,6 +3005,7 @@ namespace MobileGL::MG_Impl::GLImpl {
     }
 
     void ReadPixels_Backend(GLint x, GLint y, GLsizei width, GLsizei height, GLenum format, GLenum type, void* pixels) {
+        MGP_FILL(ReadPixels);
         MG_Backend::gBackendFunctionsTable.GL.ReadPixels(x, y, width, height, format, type, pixels);
     }
 
@@ -3148,15 +3287,55 @@ namespace MobileGL::MG_Impl::GLImpl {
         GetNamedFramebufferAttachmentParameteriv_State(framebuffer, attachment, pname, params);
     }
 
+    // The three argument errors GL 4.6 core 18.3.1 asks a blit for. They have to be raised here,
+    // in the backend-independent frontend: DirectGLES drains the driver's error queue around the
+    // blit on purpose (that is how the resolve fallback probes the driver), so an ES-side
+    // rejection never reaches the application and glGetError() answered GL_NO_ERROR for a call
+    // the spec requires to fail (KHR-GL30.api.coverage's glBlitFramebuffer sub-check). DirectVulkan
+    // already dropped the bad-filter and LINEAR-with-depth/stencil calls on the floor with a log
+    // line (VulkanRenderer::BlitFramebuffer), so the only thing that changes for it is that the
+    // error is now visible where the spec says it should be.
+    static Bool ValidateBlitMaskAndFilter(const char* functionName, GLbitfield mask, GLenum filter) {
+        constexpr GLbitfield kBlitMaskBits = GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT | GL_STENCIL_BUFFER_BIT;
+        if ((mask & ~kBlitMaskBits) != 0) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName,
+                                             "mask contains bits other than GL_COLOR_BUFFER_BIT, "
+                                             "GL_DEPTH_BUFFER_BIT and GL_STENCIL_BUFFER_BIT."));
+            return false;
+        }
+        if (filter != GL_NEAREST && filter != GL_LINEAR) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidEnum,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName,
+                                             "filter must be GL_NEAREST or GL_LINEAR."));
+            return false;
+        }
+        // Depth and stencil have no meaningful interpolation, so GL_LINEAR is rejected outright
+        // rather than downgraded - even when the mask also carries the colour bit.
+        if (filter == GL_LINEAR && (mask & (GL_DEPTH_BUFFER_BIT | GL_STENCIL_BUFFER_BIT)) != 0) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", functionName,
+                                             "GL_LINEAR filtering is not allowed when mask includes "
+                                             "GL_DEPTH_BUFFER_BIT or GL_STENCIL_BUFFER_BIT."));
+            return false;
+        }
+        return true;
+    }
+
     void BlitNamedFramebuffer(GLuint readFramebuffer, GLuint drawFramebuffer, GLint srcX0, GLint srcY0, GLint srcX1,
                               GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1, GLint dstY1, GLbitfield mask,
                               GLenum filter) {
+        if (!ValidateBlitMaskAndFilter(__func__, mask, filter)) return;
         BlitNamedFramebuffer_State(readFramebuffer, drawFramebuffer, srcX0, srcY0, srcX1, srcY1, dstX0, dstY0, dstX1,
                                    dstY1, mask, filter);
     }
 
     void BlitFramebuffer(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0, GLint dstX1,
                          GLint dstY1, GLbitfield mask, GLenum filter) {
+        if (!ValidateBlitMaskAndFilter(__func__, mask, filter)) return;
         BlitFramebuffer_Backend(srcX0, srcY0, srcX1, srcY1, dstX0, dstY0, dstX1, dstY1, mask, filter);
     }
 

MobileGL/MG_Impl/GLImpl/Getter/GL_Getter.cpp

@@ -7,9 +7,12 @@
 // End of Source File Header
 
 #include "GL_Getter.h"
+#include <algorithm>
 #include <cmath>
+#include <limits>
 #include <Config.h>
 #include <MGGitHash.h>
+#include <MG_Impl/GLImpl/Debug/GL_Debug.h>
 #include <MG_Impl/GLImpl/VertexArray/Validators.h>
 #include <MG_State/EGLState/Core.h>
 #include <MG_State/GLState/Core.h>
@@ -25,7 +28,9 @@
 #include <MG_State/GLState/FramebufferState/FramebufferObject.h>
 #include <MG_Util/Texture/TextureFormatProcessor.h>
 #include <MG_Util/Async/ShaderCompilePool.h>
+#include <MG_Util/ShaderTranspiler/Types.h>
 #include <MG_Backend/BackendObjects.h>
+#include <MG_Impl/Pipe/PipeFill.h>
 
 namespace MobileGL::MG_Impl::GLImpl {
     // Declared rather than #included from GL_RenderState.h on purpose: that header also declares
@@ -46,13 +51,29 @@ namespace MobileGL::MG_Impl::GLImpl {
             }
         }
 
-        constexpr GLint kFrontendMaxComputeUniformComponents = 1024;
-        constexpr GLint kFrontendMaxComputeAtomicCounters = 8;
-        constexpr GLint kFrontendMaxComputeAtomicCounterBuffers = 8;
+        // Shared with the glslang resource table for the same reason as the atomic-counter
+        // limits below: gl_MaxComputeUniformComponents expands from BuildTBuiltInResource.
+        constexpr GLint kFrontendMaxComputeUniformComponents =
+            static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_COMPUTE_UNIFORM_COMPONENTS);
+        // Every atomic-counter limit is shared with the glslang resource table
+        // (BuildTBuiltInResource) through MG_Util/ShaderTranspiler/Types.h: GL 4.6 requires
+        // glGetIntegerv and the gl_MaxAtomicCounter* built-in constants to agree, and the two
+        // used to be independent tables that disagreed on both the binding count and the buffer
+        // size. Never move one of these without the other.
+        constexpr GLint kFrontendMaxComputeAtomicCounters =
+            static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_ATOMIC_COUNTERS_PER_STAGE);
+        constexpr GLint kFrontendMaxComputeAtomicCounterBuffers =
+            static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_ATOMIC_COUNTER_BUFFERS_PER_STAGE);
         constexpr GLint kFrontendMaxComputeSharedMemorySize = 32768;
         constexpr GLint kFrontendMaxComputeWorkGroupInvocations = 1024;
-        constexpr GLint kFrontendMaxCombinedAtomicCounters = 8;
-        constexpr GLint kFrontendMaxFragmentAtomicCounters = 8;
+        constexpr GLint kFrontendMaxCombinedAtomicCounters =
+            static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_ATOMIC_COUNTERS_PER_STAGE);
+        constexpr GLint kFrontendMaxCombinedAtomicCounterBuffers =
+            static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_ATOMIC_COUNTER_BUFFERS_PER_STAGE);
+        constexpr GLint kFrontendMaxFragmentAtomicCounters =
+            static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_ATOMIC_COUNTERS_PER_STAGE);
+        constexpr GLint kFrontendMaxFragmentAtomicCounterBuffers =
+            static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_ATOMIC_COUNTER_BUFFERS_PER_STAGE);
         constexpr GLint kFrontendMaxGeometryAtomicCounters = 0;
         constexpr GLint kFrontendMaxTessControlAtomicCounters = 0;
         constexpr GLint kFrontendMaxTessEvaluationAtomicCounters = 0;
@@ -66,16 +87,24 @@ namespace MobileGL::MG_Impl::GLImpl {
         constexpr GLint kFrontendMaxTessControlAtomicCounterBuffers = 0;
         constexpr GLint kFrontendMaxTessEvaluationAtomicCounterBuffers = 0;
         constexpr GLint kFrontendMaxVertexAtomicCounterBuffers = 0;
-        // One atomic counter is a uint, and a buffer never has to hold more counters than the
-        // combined limit the frontend advertises. GL 4.6 table 23.63 floors this at 32 bytes.
+        // GL_MAX_ATOMIC_COUNTER_BUFFER_SIZE: the byte offset ceiling a counter may be declared
+        // at. The matching binding count is applied in GetIndexedBufferQueryPointCount, so that
+        // the getter, the indexed queries and glBindBufferBase all share one ceiling.
         constexpr GLint kFrontendMaxAtomicCounterBufferSize =
-            kFrontendMaxCombinedAtomicCounters * static_cast<GLint>(sizeof(GLuint));
+            static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_ATOMIC_COUNTER_BUFFER_SIZE);
         // KHR_debug minima (GL 4.6 table 23.66); the debug entry points are stubs, but the
         // limits they advertise still have to be legal.
         constexpr GLint kFrontendMaxDebugGroupStackDepth = 64;
         constexpr GLint kFrontendMaxDebugLoggedMessages = 1;
-        constexpr GLint kFrontendMaxVertexUniformComponents = 4096;
-        constexpr GLint kFrontendMaxVertexUniformVectors = 128;
+        // The *_VECTORS answers are the *_COMPONENTS ones divided by four, never a second
+        // literal: they used to be independent (4096 components against 128 vectors, 64 varying
+        // components against 8 varying vectors) and could not both be describing the same
+        // capacity. Both are shared with BuildTBuiltInResource through Types.h, because
+        // gl_MaxVertexUniformVectors and gl_MaxVaryingVectors expand from the same numbers.
+        constexpr GLint kFrontendMaxVertexUniformComponents =
+            static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_VERTEX_UNIFORM_COMPONENTS);
+        constexpr GLint kFrontendMaxVertexUniformVectors =
+            static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_VERTEX_UNIFORM_VECTORS);
         constexpr GLint kFrontendMaxVertexUniformBlocks = 14;
         constexpr GLint kFrontendMaxVertexOutputComponents = 64;
         constexpr GLint kFrontendMaxFragmentInputComponents = 128;
@@ -87,33 +116,87 @@ namespace MobileGL::MG_Impl::GLImpl {
         constexpr GLint kFrontendMaxGeometryTextureImageUnits = 16;
         constexpr GLint kFrontendMaxGeometryUniformComponents = 1024;
         constexpr GLint kFrontendMaxGeometryUniformBlocks = 14;
-        constexpr GLint kFrontendMaxCombinedUniformBlocks = kFrontendMaxVertexUniformBlocks +
-                                                            kFrontendMaxGeometryUniformBlocks +
-                                                            kFrontendMaxFragmentUniformBlocks;
-        constexpr GLint kFrontendMaxVaryingComponents = 64;
-        constexpr GLint kFrontendMaxVaryingVectors = 8;
+        // ARB_geometry_shader4's per-invocation count. No TBuiltInResource field and no
+        // gl_MaxGeometryShaderInvocations built-in exists to keep in step, so this is a getter
+        // answer only; 32 is the GL 4.6 core minimum (table 23.57).
+        constexpr GLint kFrontendMaxGeometryShaderInvocations = 32;
+        constexpr GLint kFrontendMaxTessControlUniformBlocks = 14;
+        constexpr GLint kFrontendMaxTessEvaluationUniformBlocks = 14;
+        // The compute stage's share of the combined sum below. Compute's own per-stage answer is
+        // backend-derived (GL_MAX_COMPUTE_UNIFORM_BLOCKS reads dynamicParameters), so this is not
+        // what that query returns - it is the GL 4.3 core minimum, present here only so the
+        // combined total covers all SIX stages.
+        constexpr GLint kFrontendMaxComputeUniformBlocksShare = 14;
+        // GL 4.6 table 23.64 orders MAX_UNIFORM_BUFFER_BINDINGS >= MAX_COMBINED_UNIFORM_BLOCKS >=
+        // every per-stage count, and the sum has to run over SIX stages, not three and not five.
+        // Three (42) was the original bug. Five (70) replaced it and broke the middle term the
+        // other way: compute's per-stage count is backend-derived and clamps at the binding count,
+        // so a device reporting descriptor-indexing-scale uniform buffers (Adreno reports
+        // maxPerStageDescriptorUniformBuffers = 16777216) advertised 84 compute blocks against a
+        // combined 70. Six stages x 14 = 84, which is also exactly the binding-point count and the
+        // arithmetic the GL 4.5 minimum of 84 bindings is built from, so the ordering is now tight
+        // rather than accidental.
+        constexpr GLint kFrontendMaxCombinedUniformBlocks =
+            kFrontendMaxVertexUniformBlocks + kFrontendMaxTessControlUniformBlocks +
+            kFrontendMaxTessEvaluationUniformBlocks + kFrontendMaxGeometryUniformBlocks +
+            kFrontendMaxFragmentUniformBlocks + kFrontendMaxComputeUniformBlocksShare;
+        constexpr GLint kFrontendMaxVaryingComponents =
+            static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_VARYING_COMPONENTS);
+        constexpr GLint kFrontendMaxVaryingVectors =
+            static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_VARYING_VECTORS);
         constexpr GLint kFrontendMaxProgramTexelOffset = 7;
         constexpr GLint kFrontendMinProgramTexelOffset = -8;
         constexpr GLint kFrontendMaxTransformFeedbackInterleavedComponents = 64;
         constexpr GLint kFrontendMaxTransformFeedbackSeparateAttribs = 4;
         constexpr GLint kFrontendMaxTransformFeedbackSeparateComponents = 4;
+        // ARB_transform_feedback3's vertex-stream count. One is what this implementation can
+        // actually emit to; see the GL_MAX_VERTEX_STREAMS case for why it is not four.
+        constexpr GLint kFrontendMaxVertexStreams = 1;
         constexpr GLint kFrontendMaxGeometryOutputVertices = 256;
         constexpr GLint kFrontendMaxGeometryTotalOutputComponents = 1024;
-        constexpr GLint kFrontendMinUniformBufferBindings = 36;
+        // GL 4.5 core table 23.64 requires 84 indexed uniform binding points, and that is exactly
+        // how wide the state layer's array is (BufferState::BufferBindingPointCount) - see the
+        // GL_MAX_UNIFORM_BUFFER_BINDINGS case for why the ES driver's own, smaller count is not
+        // the ceiling here.
+        constexpr GLint kFrontendMinUniformBufferBindings = 84;
         constexpr GLint kFrontendSubpixelBits = 4;
-        constexpr GLint kFrontendMaxSamples = 4;
+        constexpr GLint kFrontendMaxSamples =
+            static_cast<GLint>(MG_Util::ShaderTranspiler::MIN_ADVERTISED_MAX_SAMPLES);
+        // ARB_shader_subroutine's two limits. NOTHING IMPLEMENTS SUBROUTINES: there is no
+        // glGetSubroutineIndex / glUniformSubroutinesuiv, only the program-interface enum
+        // plumbing. These are answered - with the GL 4.5 core minimums - because the conformance
+        // suite queries them before it checks for the feature and an INVALID_ENUM both leaves the
+        // caller reading its own uninitialised stack slot and strands an error for the next
+        // unrelated call to trip over. The extension is deliberately NOT advertised, so the
+        // numbers are a table entry, not a capability claim.
+        constexpr GLint kFrontendMaxSubroutines = 256;
+        constexpr GLint kFrontendMaxSubroutineUniformLocations = 1024;
 
+        // The floors under GL_MAX_COMPUTE_WORK_GROUP_COUNT / _SIZE. Shared with the compile
+        // pipeline (CaptureCompileEnv floors the same driver answers at them, and
+        // BuildTBuiltInResource expands gl_MaxComputeWorkGroup* from the result), because a
+        // shader is allowed to compare the built-in constant against this query.
         constexpr GLint GetMinComputeWorkGroupCount(GLuint index) {
-            return index < 3 ? 65535 : 0;
+            return index < 3 ? static_cast<GLint>(MG_Util::ShaderTranspiler::MIN_COMPUTE_WORK_GROUP_COUNT[index]) : 0;
         }
 
         constexpr GLint GetMinComputeWorkGroupSize(GLuint index) {
-            return index < 2 ? 1024 : (index == 2 ? 64 : 0);
+            return index < 3 ? static_cast<GLint>(MG_Util::ShaderTranspiler::MIN_COMPUTE_WORK_GROUP_SIZE[index]) : 0;
         }
 
+        // GL 4.6 core table 23.64: components + blocks * (blockSize / 4). The product has to be
+        // formed in 64 bits and saturated on the way out - it overflowed a signed 32-bit int on
+        // every Vulkan host that reports a large maxUniformBufferRange. A Mali driver answering
+        // 0xFFFFFFFF saturates to INT32_MAX in the loader, and 14 * (2147483647 / 4) + 4096 wraps
+        // to -1073737742, which the conformance suite read back as a limit "smaller than 58368".
+        // Saturating instead of wrapping is also the only honest answer: an implementation that
+        // can serve more components than a GLint holds still has to report a GLint.
         GLint GetMaxCombinedUniformComponents(GLint maxDefaultUniformComponents, GLint maxUniformBlocks,
                                               GLint maxUniformBlockSizeBytes) {
-            return maxDefaultUniformComponents + maxUniformBlocks * (maxUniformBlockSizeBytes / 4);
+            const Int64 blocks = std::max<Int64>(static_cast<Int64>(maxUniformBlocks), 0);
+            const Int64 componentsPerBlock = std::max<Int64>(static_cast<Int64>(maxUniformBlockSizeBytes), 0) / 4;
+            const Int64 total = static_cast<Int64>(maxDefaultUniformComponents) + blocks * componentsPerBlock;
+            return static_cast<GLint>(std::min<Int64>(total, std::numeric_limits<GLint>::max()));
         }
 
         bool TryDecodeIndexedBufferQuery(GLenum pname, BufferTarget& bufferTarget, IndexedBufferQueryKind& queryKind) {
@@ -186,6 +269,16 @@ namespace MobileGL::MG_Impl::GLImpl {
                     MG_Backend::pActiveBackendObject->GetDynamicParameters().MaxShaderStorageBufferBindings;
                 return std::min(frontendCount, static_cast<SizeT>(std::max(backendCount, 0)));
             }
+            if (bufferTarget == BufferTarget::AtomicCounter) {
+                // The counter family's binding count is NOT the state layer's array size: a
+                // counter buffer only reaches a shader as a lowered storage block, so what an
+                // implementation can serve is the reserved range, and that number is also what
+                // glslang compiles a layout(binding = N) atomic_uint against. Clamped here so
+                // GL_MAX_ATOMIC_COUNTER_BUFFER_BINDINGS, the indexed getters' index check and
+                // glBindBufferBase's all report the same ceiling.
+                return std::min(frontendCount,
+                                static_cast<SizeT>(MG_Util::ShaderTranspiler::MAX_ATOMIC_COUNTER_BUFFER_BINDINGS));
+            }
             return frontendCount;
         }
 
@@ -213,6 +306,23 @@ namespace MobileGL::MG_Impl::GLImpl {
             return ClampBlockCountToBindingPoints(blockCount, BufferTarget::ShaderStorage);
         }
 
+        // The per-stage GL_MAX_*_SHADER_STORAGE_BLOCKS answers. Backend-derived, and NOT a
+        // constant to be "restored" - these used to return a flat 16 for vertex, geometry and
+        // both tessellation stages, which is wrong on any host that does not serve storage
+        // blocks in those stages. Zero is a legal answer: GL 4.6 table 23.64 and ES 3.2 table
+        // 21.44 both set the minimum at 0 for every graphics stage except fragment, which is
+        // why the conformance suite gates each such test on the query instead of assuming it.
+        // ARM's GLES driver reports 0 for all four (a Mali-G925 does), and advertising 16 there
+        // bought nothing: the program still failed to link inside the backend, the frontend
+        // still reported LINK_STATUS as true, and every draw with it silently rendered nothing.
+        GLint StageStorageBlockCount(Int MG_Backend::DynamicBackendParameters::*stageLimit) {
+            static const MG_Backend::DynamicBackendParameters kBackendlessDefaults{};
+            const MG_Backend::DynamicBackendParameters& parameters =
+                MG_Backend::pActiveBackendObject ? MG_Backend::pActiveBackendObject->GetDynamicParameters()
+                                                 : kBackendlessDefaults;
+            return ClampStorageBlockCount(static_cast<GLint>(parameters.*stageLimit));
+        }
+
         bool TryDecodeDrawBufferQuery(GLenum pname, SizeT& drawBufferIndex) {
             if (pname == GL_DRAW_BUFFER) {
                 drawBufferIndex = 0;
@@ -254,24 +364,6 @@ namespace MobileGL::MG_Impl::GLImpl {
             return true;
         }
 
-        GLint ResolveDrawFramebufferSampleCount() {
-            const auto& drawFbo =
-                MG_State::pGLContext->GetFramebufferBindingSlot(FramebufferTarget::Draw).GetBoundObject();
-            if (!drawFbo) return 0;
-
-            GLint maxSamples = 0;
-            for (const auto& attachment : drawFbo->GetAllAttachmentObjects()) {
-                if (attachment.IsRenderbuffer() && attachment.GetRenderbuffer()) {
-                    maxSamples = std::max(maxSamples, static_cast<GLint>(attachment.GetRenderbuffer()->GetSamples()));
-                } else if (attachment.IsTexture() && attachment.GetTexture()) {
-                    // Multisample texture attachments count too (GL_SAMPLE_BUFFERS must
-                    // report 1 for any multisampled draw framebuffer).
-                    maxSamples = std::max(maxSamples, static_cast<GLint>(attachment.GetTexture()->GetSamples()));
-                }
-            }
-            return maxSamples;
-        }
-
         void RecordIndexedOnlyGetterError(const char* functionName, GLenum pname) {
             MG_State::pGLContext->RecordError(
                 ErrorCode::InvalidEnum,
@@ -422,6 +514,70 @@ namespace MobileGL::MG_Impl::GLImpl {
         }
     } // namespace
 
+    // GL 4.6 core table 23.53 requires GL_MAX_SAMPLES >= 4, so the driver's value is floored
+    // before it is advertised. gl_MaxSamples expands from the same floored number
+    // (BuildTBuiltInResource), which is also what sizes gl_SampleMask[].
+    //
+    // THE FLOOR STOPS HERE, and that is the point. It used to be applied to
+    // GL_MAX_INTEGER_SAMPLES, GL_MAX_COLOR_TEXTURE_SAMPLES and GL_MAX_DEPTH_TEXTURE_SAMPLES too,
+    // on the reasoning that an application reads GL_MAX_SAMPLES once and hands that count to
+    // every glTexStorage*Multisample. Table 23.53 gives those three a minimum of ONE, and the
+    // reasoning had it backwards: Adreno and Mali back an integer multisample texture with a
+    // single sample, so flooring the query at 4 did not make four samples exist - it made the
+    // backend silently under-allocate (ClampSamplesToBackendSupport) while the application wrote
+    // per-sample data it could never read back. Reporting what was probed turns that into an
+    // honest "unsupported" the application can branch on.
+    GLint GetAdvertisedMaxSamples() {
+        if (MG_Backend::pActiveBackendObject == nullptr) {
+            return kFrontendMaxSamples;
+        }
+        return std::max(MG_Backend::pActiveBackendObject->GetDynamicParameters().MaxSamples, kFrontendMaxSamples);
+    }
+
+    // GL 4.6 core table 23.53 minimum for the per-category multisample ceilings. One, not four:
+    // see the note on GetAdvertisedMaxSamples. A zero would be a probe that never ran, so it is
+    // floored rather than trusted.
+    namespace {
+        GLint AdvertisedCategoryMaxSamples(Int MG_Backend::DynamicBackendParameters::*categoryLimit) {
+            if (MG_Backend::pActiveBackendObject == nullptr) {
+                return 1;
+            }
+            return std::max(MG_Backend::pActiveBackendObject->GetDynamicParameters().*categoryLimit, 1);
+        }
+    } // namespace
+
+    GLint GetAdvertisedColorTextureMaxSamples() {
+        return AdvertisedCategoryMaxSamples(&MG_Backend::DynamicBackendParameters::MaxColorTextureSamples);
+    }
+
+    GLint GetAdvertisedDepthTextureMaxSamples() {
+        return AdvertisedCategoryMaxSamples(&MG_Backend::DynamicBackendParameters::MaxDepthTextureSamples);
+    }
+
+    GLint GetAdvertisedIntegerMaxSamples() {
+        return AdvertisedCategoryMaxSamples(&MG_Backend::DynamicBackendParameters::MaxIntegerSamples);
+    }
+
+    // Declared in GL_Getter.h, so that the draw path can feed the same number to the reserved
+    // gl_NumSamples stand-in that glGetIntegerv(GL_SAMPLES) reports.
+    GLint ResolveDrawFramebufferSampleCount() {
+        const auto& drawFbo =
+            MG_State::pGLContext->GetFramebufferBindingSlot(FramebufferTarget::Draw).GetBoundObject();
+        if (!drawFbo) return 0;
+
+        GLint maxSamples = 0;
+        for (const auto& attachment : drawFbo->GetAllAttachmentObjects()) {
+            if (attachment.IsRenderbuffer() && attachment.GetRenderbuffer()) {
+                maxSamples = std::max(maxSamples, static_cast<GLint>(attachment.GetRenderbuffer()->GetSamples()));
+            } else if (attachment.IsTexture() && attachment.GetTexture()) {
+                // Multisample texture attachments count too (GL_SAMPLE_BUFFERS must
+                // report 1 for any multisampled draw framebuffer).
+                maxSamples = std::max(maxSamples, static_cast<GLint>(attachment.GetTexture()->GetSamples()));
+            }
+        }
+        return maxSamples;
+    }
+
     /* @INSERTION_POINT:FUNCTION_IMPLEMENTATION@ */
     const GLubyte* GetString(GLenum name) {
         static String vendorString;
@@ -618,12 +774,30 @@ namespace MobileGL::MG_Impl::GLImpl {
             return;
         case GL_MIN_FRAGMENT_INTERPOLATION_OFFSET:
         case GL_MAX_FRAGMENT_INTERPOLATION_OFFSET:
-        case GL_FRAGMENT_INTERPOLATION_OFFSET_BITS: {
+        case GL_FRAGMENT_INTERPOLATION_OFFSET_BITS:
+        // Same reason as the three above: the integer fallback would round the fraction to 0
+        // or 1 first, so a 0.25 sample-shading rate would answer GL_FALSE.
+        case GL_MIN_SAMPLE_SHADING_VALUE: {
             GLfloat value = 0.0f;
             GetFloatv(pname, &value);
             *params = value != 0.0f ? GL_TRUE : GL_FALSE;
             return;
         }
+        // Float-native state, so GL 4.6 core 2.2.2's "zero becomes FALSE, every other value
+        // becomes TRUE" has to be applied to the VALUE. Answering these through the integer getter
+        // below instead - which rounds - reported GL_FALSE for a perfectly non-zero level of 0.25,
+        // and every other float state in this function already reads through GetFloatv for exactly
+        // that reason.
+        case GL_PATCH_DEFAULT_OUTER_LEVEL:
+        case GL_PATCH_DEFAULT_INNER_LEVEL: {
+            const GLsizei componentCount = pname == GL_PATCH_DEFAULT_OUTER_LEVEL ? 4 : 2;
+            GLfloat levels[4] = {};
+            GetFloatv(pname, levels);
+            for (GLsizei i = 0; i < componentCount; ++i) {
+                params[i] = levels[i] != 0.0f ? GL_TRUE : GL_FALSE;
+            }
+            return;
+        }
         default:
             break;
         }
@@ -673,6 +847,22 @@ namespace MobileGL::MG_Impl::GLImpl {
             params[1] = depthRange.y();
             return;
         }
+        // glPatchParameterfv's two states. Float-native, so they are answered here rather than
+        // through the integer fallback below - which rounds, and would report 0 for a level of 0.5.
+        case GL_PATCH_DEFAULT_OUTER_LEVEL: {
+            const FloatVec4& outer = MG_State::pGLContext->GetPatchDefaultOuterLevel();
+            params[0] = outer.x();
+            params[1] = outer.y();
+            params[2] = outer.z();
+            params[3] = outer.w();
+            return;
+        }
+        case GL_PATCH_DEFAULT_INNER_LEVEL: {
+            const FloatVec2& inner = MG_State::pGLContext->GetPatchDefaultInnerLevel();
+            params[0] = inner.x();
+            params[1] = inner.y();
+            return;
+        }
         case GL_VIEWPORT_BOUNDS_RANGE: {
             const auto& dynamicParameters = MG_Backend::pActiveBackendObject->GetDynamicParameters();
             params[0] = dynamicParameters.ViewportBoundsRangeMin;
@@ -738,6 +928,11 @@ namespace MobileGL::MG_Impl::GLImpl {
         case GL_POLYGON_OFFSET_UNITS:
             params[0] = MG_State::pGLContext->GetPolygonOffsetUnits();
             return;
+        case GL_POLYGON_OFFSET_CLAMP:
+            // Float-native state, so it is answered here rather than through the integer
+            // fallback: glPolygonOffsetClamp(1, 1, 0.5) must read back as 0.5, not as 0.
+            params[0] = MG_State::pGLContext->GetPolygonOffsetClamp();
+            return;
         case GL_SMOOTH_LINE_WIDTH_RANGE: {
             const auto& dynamicParameters = MG_Backend::pActiveBackendObject->GetDynamicParameters();
             params[0] = dynamicParameters.SmoothLineWidthRangeMin;
@@ -753,6 +948,11 @@ namespace MobileGL::MG_Impl::GLImpl {
         case GL_SAMPLE_COVERAGE_VALUE:
             params[0] = MG_State::pGLContext->GetSampleCoverageValue();
             return;
+        case GL_MIN_SAMPLE_SHADING_VALUE:
+            // Float state, so it has to be answered here rather than through the integer
+            // fallback: glMinSampleShading(0.5) must read back as 0.5 and not as 0.
+            params[0] = MG_State::pGLContext->GetMinSampleShadingValue();
+            return;
         case GL_POINT_FADE_THRESHOLD_SIZE:
             // Float state: read it directly so the fractional part is not lost to the integer path.
             params[0] = MG_State::pGLContext->GetPointFadeThresholdSize();
@@ -974,6 +1174,7 @@ namespace MobileGL::MG_Impl::GLImpl {
                 : GetMinComputeWorkGroupSize(index);
             GLint backendValue = 0;
             if (getIntegeri) {
+                MGP_FILL(GetIntegeri_v);
                 getIntegeri(target, index, &backendValue);
             }
             *data = std::max(backendValue, minimum);
@@ -1124,6 +1325,13 @@ namespace MobileGL::MG_Impl::GLImpl {
         }
 
         switch (pname) {
+        case GL_MAX_ELEMENT_INDEX:
+            // The largest value a GL_UNSIGNED_INT index may take. It has to be answered HERE and
+            // not left to the 32-bit fallback below: the conformance suite reads it with
+            // glGetInteger64v, and widening the saturated GLint would report INT32_MAX where the
+            // spec requires 2^32-1.
+            params[0] = 0xFFFFFFFFLL;
+            return;
         case GL_MAX_SHADER_STORAGE_BLOCK_SIZE:
             if (MG_Backend::pActiveBackendObject) {
                 params[0] = static_cast<GLint64>(
@@ -1147,6 +1355,7 @@ namespace MobileGL::MG_Impl::GLImpl {
             Int64 timestamp = 0;
             if (!MG_Config::Features.DisableTimerQuery) {
                 if (const auto getGpuTimestampNs = MG_Backend::gBackendFunctionsTable.GL.GetGpuTimestampNs) {
+                    MGP_FILL(GetGpuTimestampNs);
                     timestamp = getGpuTimestampNs();
                 }
             }
@@ -1160,12 +1369,17 @@ namespace MobileGL::MG_Impl::GLImpl {
         GLint ints[4] = {};
         GetIntegerv(pname, ints);
 
+        // GL 4.6 core 22.1 gives glGetInteger64v the same accepted-pname set as glGetIntegerv, so
+        // every pname the integer getter answers with several components owes them all here too.
+        // A pname that reaches the `default:` arm writes params[0] and leaves the caller's other
+        // components holding whatever they held, with no error to say so.
         switch (pname) {
         case GL_BLEND_COLOR:
         case GL_COLOR_CLEAR_VALUE:
         case GL_COLOR_WRITEMASK:
         case GL_SCISSOR_BOX:
         case GL_VIEWPORT:
+        case GL_PATCH_DEFAULT_OUTER_LEVEL:
             for (int i = 0; i < 4; ++i) {
                 params[i] = static_cast<GLint64>(ints[i]);
             }
@@ -1175,6 +1389,7 @@ namespace MobileGL::MG_Impl::GLImpl {
         case GL_MAX_VIEWPORT_DIMS:
         case GL_POINT_SIZE_RANGE:
         case GL_VIEWPORT_BOUNDS_RANGE:
+        case GL_PATCH_DEFAULT_INNER_LEVEL:
             params[0] = static_cast<GLint64>(ints[0]);
             params[1] = static_cast<GLint64>(ints[1]);
             return;
@@ -1206,6 +1421,7 @@ namespace MobileGL::MG_Impl::GLImpl {
         case GL_POINT_SIZE_RANGE:
         case GL_SMOOTH_LINE_WIDTH_RANGE:
         case GL_MAX_VIEWPORT_DIMS:
+        case GL_PATCH_DEFAULT_INNER_LEVEL:
             count = 2;
             break;
         case GL_BLEND_COLOR:
@@ -1213,6 +1429,7 @@ namespace MobileGL::MG_Impl::GLImpl {
         case GL_VIEWPORT:
         case GL_SCISSOR_BOX:
         case GL_COLOR_WRITEMASK:
+        case GL_PATCH_DEFAULT_OUTER_LEVEL:
             count = 4;
             break;
         default:
@@ -1252,6 +1469,15 @@ namespace MobileGL::MG_Impl::GLImpl {
                 *params = 0;
             return;
         }
+        // GL_TEXTURE_BUFFER_BINDING and GL_TEXTURE_BUFFER are the same token (0x8C2A): as a
+        // glGetIntegerv pname it asks which BUFFER object is bound to the buffer-texture target,
+        // not which texture is (that one is GL_TEXTURE_BINDING_BUFFER, handled by the texture-unit
+        // decoder above).
+        case GL_TEXTURE_BUFFER_BINDING: {
+            auto& obj = MG_State::pGLContext->GetBufferBindingSlot(BufferTarget::Texture).GetBoundObject();
+            *params = obj ? static_cast<GLint>(obj->GetExternalIndex()) : 0;
+            return;
+        }
         case GL_BLEND:
             *params = MG_State::pGLContext->IsCapabilityEnabled(CapabilityInput::Blend) ? GL_TRUE : GL_FALSE;
             return;
@@ -1307,6 +1533,16 @@ namespace MobileGL::MG_Impl::GLImpl {
             // this single case serves every getter flavor.
             *params = static_cast<GLint>(MG_State::pGLContext->GetClampReadColor());
             return;
+        // glClipControl's two state variables (GL 4.5 core table 23.7). They answer from the
+        // state the entry point records, which is what the conformance suite's initial-value and
+        // set-then-get cases read - the RASTERIZATION half of clip control is a separate,
+        // backend-side question and does not gate the query.
+        case GL_CLIP_ORIGIN:
+            *params = static_cast<GLint>(MG_State::pGLContext->GetClipOrigin());
+            return;
+        case GL_CLIP_DEPTH_MODE:
+            *params = static_cast<GLint>(MG_State::pGLContext->GetClipDepthMode());
+            return;
         case GL_COLOR_CLEAR_VALUE: {
             const FloatVec4& clearColor = MG_State::pGLContext->GetClearColor();
             params[0] = static_cast<GLint>(clearColor.x());
@@ -1366,19 +1602,21 @@ namespace MobileGL::MG_Impl::GLImpl {
                           : 0;
             return;
         case GL_MAX_DEBUG_GROUP_STACK_DEPTH:
-            // KHR_debug floors this at 64 even when the group entry points are stubs: the
-            // limit describes how deep glPushDebugGroup may nest, and 0 is not a legal answer.
+            // KHR_debug floors this at 64. It must agree with what GL_Debug.cpp actually enforces,
+            // or an application that nests to the reported limit would take a STACK_OVERFLOW.
             *params = kFrontendMaxDebugGroupStackDepth;
             return;
         case GL_MAX_DEBUG_MESSAGE_LENGTH:
-            *params = 1024; // debug-message entrypoints are stubbed, but KHR_debug requires a valid limit
+            *params = 1024; // agrees with GL_Debug.cpp's kMaxDebugMessageLength
             return;
         case GL_MAX_DEBUG_LOGGED_MESSAGES:
             // Size of the message log ring; KHR_debug requires at least 1.
             *params = kFrontendMaxDebugLoggedMessages;
             return;
         case GL_DEBUG_GROUP_STACK_DEPTH:
-            *params = 0; // debug-group entrypoints are stubbed
+            // The live depth, which is never 0: GL 4.6 core 20.6 creates the context with one
+            // group already on the stack, and that is the one glPopDebugGroup may not pop.
+            *params = GetDebugGroupStackDepth();
             return;
         case GL_CONTEXT_FLAGS: {
             *params = MG_State::pEGLContext ? MG_State::pEGLContext->GetCurrentContextFlags() : 0;
@@ -1511,15 +1749,15 @@ namespace MobileGL::MG_Impl::GLImpl {
         case GL_LINE_WIDTH:
             *params = static_cast<GLint>(MG_State::pGLContext->GetLineWidth());
             return;
-        case GL_LAYER_PROVOKING_VERTEX:
-            *params = GL_LAST_VERTEX_CONVENTION;
-            return;
         case GL_LOGIC_OP_MODE:
             *params = static_cast<GLint>(MG_Util::ConvertLogicOperationToGLEnum(MG_State::pGLContext->GetLogicOp()));
             return;
         case GL_MAX_COMBINED_ATOMIC_COUNTERS:
             *params = kFrontendMaxCombinedAtomicCounters;
             return;
+        case GL_MAX_COMBINED_ATOMIC_COUNTER_BUFFERS:
+            *params = kFrontendMaxCombinedAtomicCounterBuffers;
+            return;
         case GL_MAX_COMBINED_UNIFORM_BLOCKS:
             *params = ClampUniformBlockCount(kFrontendMaxCombinedUniformBlocks);
             return;
@@ -1535,8 +1773,11 @@ namespace MobileGL::MG_Impl::GLImpl {
         case GL_MAX_FRAGMENT_ATOMIC_COUNTERS:
             *params = kFrontendMaxFragmentAtomicCounters;
             return;
+        case GL_MAX_FRAGMENT_ATOMIC_COUNTER_BUFFERS:
+            *params = kFrontendMaxFragmentAtomicCounterBuffers;
+            return;
         case GL_MAX_FRAGMENT_SHADER_STORAGE_BLOCKS:
-            *params = ClampStorageBlockCount(16); // TODO
+            *params = StageStorageBlockCount(&MG_Backend::DynamicBackendParameters::MaxFragmentShaderStorageBlocks);
             return;
         case GL_MAX_FRAGMENT_INPUT_COMPONENTS:
             *params = kFrontendMaxFragmentInputComponents;
@@ -1562,7 +1803,7 @@ namespace MobileGL::MG_Impl::GLImpl {
             *params = kFrontendMaxGeometryAtomicCounterBuffers;
             return;
         case GL_MAX_GEOMETRY_SHADER_STORAGE_BLOCKS:
-            *params = ClampStorageBlockCount(16); // TODO
+            *params = StageStorageBlockCount(&MG_Backend::DynamicBackendParameters::MaxGeometryShaderStorageBlocks);
             return;
         case GL_MAX_GEOMETRY_INPUT_COMPONENTS:
             *params = kFrontendMaxGeometryInputComponents;
@@ -1590,6 +1831,9 @@ namespace MobileGL::MG_Impl::GLImpl {
         case GL_MAX_GEOMETRY_UNIFORM_COMPONENTS:
             *params = kFrontendMaxGeometryUniformComponents;
             return;
+        case GL_MAX_GEOMETRY_SHADER_INVOCATIONS:
+            *params = kFrontendMaxGeometryShaderInvocations;
+            return;
         case GL_MAX_IMAGE_SAMPLES:
             *params = 0; // multisampled image load/store is not exposed by the DirectGLES frontend
             return;
@@ -1597,7 +1841,11 @@ namespace MobileGL::MG_Impl::GLImpl {
             *params = MG_State::pGLContext->IsCapabilityEnabled(CapabilityInput::Multisample) ? GL_TRUE : GL_FALSE;
             return;
         case GL_MIN_MAP_BUFFER_ALIGNMENT:
-            *params = 64; // TODO
+            // The same constant the map paths align to (MG_State/GLState/BufferState/
+            // PipeResource.h), never a literal: this number is a PROMISE about the pointers
+            // glMapBuffer and glMapBufferRange return, and the two used to be unrelated - the
+            // query said 64 while the pointers came out of a std::vector aligned to 16.
+            *params = static_cast<GLint>(MG_State::GLState::MIN_MAP_BUFFER_ALIGNMENT);
             return;
         case GL_MAX_LABEL_LENGTH:
             *params = 256; // TODO
@@ -1633,16 +1881,71 @@ namespace MobileGL::MG_Impl::GLImpl {
             *params = 0;
             return;
         case GL_MAX_TESS_CONTROL_SHADER_STORAGE_BLOCKS:
-            *params = ClampStorageBlockCount(16); // TODO
+            *params = StageStorageBlockCount(&MG_Backend::DynamicBackendParameters::MaxTessControlShaderStorageBlocks);
             return;
         case GL_MAX_TESS_EVALUATION_SHADER_STORAGE_BLOCKS:
-            *params = ClampStorageBlockCount(16); // TODO
+            *params =
+                StageStorageBlockCount(&MG_Backend::DynamicBackendParameters::MaxTessEvaluationShaderStorageBlocks);
+            return;
+        // The tessellation per-stage resource limits. Every one of these is ALSO a GLSL built-in
+        // constant that BuildTBuiltInResource expands, and the two must report the same number
+        // (KHR-GL45.limits.max_tess_* compares them directly) - which is why the values come from
+        // the shared block in MG_Util/ShaderTranspiler/Types.h rather than from literals here.
+        // They were the whole per-stage tess family: the table had been filled in only where the
+        // honest answer was zero (the atomic counters, the image uniforms) or where a driver
+        // query existed (GL_MAX_PATCH_VERTICES, GL_MAX_TESS_GEN_LEVEL), so every pname whose
+        // answer is a real resource count fell through to GL_INVALID_ENUM.
+        case GL_MAX_TESS_CONTROL_INPUT_COMPONENTS:
+            *params = static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_TESS_CONTROL_INPUT_COMPONENTS);
+            return;
+        case GL_MAX_TESS_CONTROL_OUTPUT_COMPONENTS:
+            *params = static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_TESS_CONTROL_OUTPUT_COMPONENTS);
+            return;
+        case GL_MAX_TESS_CONTROL_TOTAL_OUTPUT_COMPONENTS:
+            *params = static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_TESS_CONTROL_TOTAL_OUTPUT_COMPONENTS);
+            return;
+        case GL_MAX_TESS_CONTROL_TEXTURE_IMAGE_UNITS:
+            *params = static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_TESS_CONTROL_TEXTURE_IMAGE_UNITS);
+            return;
+        case GL_MAX_TESS_CONTROL_UNIFORM_COMPONENTS:
+            *params = static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_TESS_CONTROL_UNIFORM_COMPONENTS);
+            return;
+        case GL_MAX_TESS_EVALUATION_INPUT_COMPONENTS:
+            *params = static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_TESS_EVALUATION_INPUT_COMPONENTS);
+            return;
+        case GL_MAX_TESS_EVALUATION_OUTPUT_COMPONENTS:
+            *params = static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_TESS_EVALUATION_OUTPUT_COMPONENTS);
+            return;
+        case GL_MAX_TESS_EVALUATION_TEXTURE_IMAGE_UNITS:
+            *params = static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_TESS_EVALUATION_TEXTURE_IMAGE_UNITS);
+            return;
+        case GL_MAX_TESS_EVALUATION_UNIFORM_COMPONENTS:
+            *params = static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_TESS_EVALUATION_UNIFORM_COMPONENTS);
+            return;
+        case GL_MAX_TESS_PATCH_COMPONENTS:
+            *params = static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_TESS_PATCH_COMPONENTS);
+            return;
+        // Routed through the same clamp as every other per-stage block count so the
+        // MAX_UNIFORM_BUFFER_BINDINGS >= MAX_COMBINED_UNIFORM_BLOCKS >= per-stage ordering of
+        // GL 4.6 table 23.64 cannot be broken by the two families moving independently.
+        case GL_MAX_TESS_CONTROL_UNIFORM_BLOCKS:
+            *params = ClampUniformBlockCount(kFrontendMaxTessControlUniformBlocks);
+            return;
+        case GL_MAX_TESS_EVALUATION_UNIFORM_BLOCKS:
+            *params = ClampUniformBlockCount(kFrontendMaxTessEvaluationUniformBlocks);
+            return;
+        case GL_MAX_SUBROUTINES:
+            *params = kFrontendMaxSubroutines;
+            return;
+        case GL_MAX_SUBROUTINE_UNIFORM_LOCATIONS:
+            *params = kFrontendMaxSubroutineUniformLocations;
             return;
         case GL_MAX_TEXTURE_LOD_BIAS:
             *params = 15; // TODO
             return;
         case GL_MAX_UNIFORM_LOCATIONS:
-            *params = 1024 * 4; // TODO
+            // The same constant the link's location allocator enforces - see ProgramObject.
+            *params = MG_State::GLState::ProgramObject::MAX_UNIFORM_LOCATIONS;
             return;
         case GL_MAX_VARYING_COMPONENTS:
             *params = kFrontendMaxVaryingComponents;
@@ -1662,7 +1965,7 @@ namespace MobileGL::MG_Impl::GLImpl {
                           : MG_Backend::DynamicBackendParameters{}.MaxVertexImageUniforms;
             return;
         case GL_MAX_VERTEX_SHADER_STORAGE_BLOCKS:
-            *params = ClampStorageBlockCount(16); // TODO
+            *params = StageStorageBlockCount(&MG_Backend::DynamicBackendParameters::MaxVertexShaderStorageBlocks);
             return;
         case GL_MAX_VERTEX_UNIFORM_COMPONENTS:
             *params = kFrontendMaxVertexUniformComponents;
@@ -1682,8 +1985,21 @@ namespace MobileGL::MG_Impl::GLImpl {
         case GL_NUM_PROGRAM_BINARY_FORMATS:
             *params = 0;
             return;
+        // GL_ARB_spirv_extensions / GL 4.6 core 22.2. An implementation that advertises no
+        // SPIR-V extension answers zero here, and glGetStringi(GL_SPIR_V_EXTENSIONS, i) is then
+        // never legally called - MobileGL runs the module through its own translation pipeline
+        // and relies on no SPIR-V extension to do it, so zero is the true answer rather than a
+        // placeholder.
+        case GL_NUM_SPIR_V_EXTENSIONS:
+            *params = 0;
+            return;
+        // GL_ARB_gl_spirv, core since 4.6: exactly one shader binary format, and the pair has to
+        // agree - an application sizes its GL_SHADER_BINARY_FORMATS array from the count.
         case GL_NUM_SHADER_BINARY_FORMATS:
-            *params = 0; // ShaderBinary entrypoints are stubbed
+            *params = 1;
+            return;
+        case GL_SHADER_BINARY_FORMATS:
+            *params = static_cast<GLint>(GL_SHADER_BINARY_FORMAT_SPIR_V);
             return;
         case GL_PACK_ALIGNMENT:
             *params = MG_State::pGLContext->GetPixelStoreParam(PixelStoreParam::PackAlignment);
@@ -1742,6 +2058,11 @@ namespace MobileGL::MG_Impl::GLImpl {
         case GL_PRIMITIVE_RESTART_INDEX:
             *params = static_cast<GLint>(MG_State::pGLContext->GetPrimitiveRestartIndex());
             return;
+        case GL_POLYGON_OFFSET_CLAMP:
+            // Float state (see GetFloatv); rounded to nearest for the integer query per GL 4.6
+            // core 22.1's float-to-integer rule.
+            *params = static_cast<GLint>(std::lround(MG_State::pGLContext->GetPolygonOffsetClamp()));
+            return;
         case GL_PROGRAM_BINARY_FORMATS:
             *params = 0; // program-binary entrypoints are stubbed
             return;
@@ -1827,6 +2148,13 @@ namespace MobileGL::MG_Impl::GLImpl {
         case GL_SAMPLE_MASK:
             *params = MG_State::pGLContext->IsCapabilityEnabled(CapabilityInput::SampleMask) ? GL_TRUE : GL_FALSE;
             return;
+        case GL_SAMPLE_SHADING:
+            *params = MG_State::pGLContext->IsCapabilityEnabled(CapabilityInput::SampleShading) ? GL_TRUE : GL_FALSE;
+            return;
+        case GL_MIN_SAMPLE_SHADING_VALUE:
+            // GL 4.6 core 22.2: a floating-point value queried as an integer rounds to nearest.
+            *params = static_cast<GLint>(std::lround(MG_State::pGLContext->GetMinSampleShadingValue()));
+            return;
         case GL_SAMPLE_MASK_VALUE:
             *params = static_cast<GLint>(MG_State::pGLContext->GetSampleMaskValue());
             return;
@@ -1938,6 +2266,7 @@ namespace MobileGL::MG_Impl::GLImpl {
             Int64 timestamp = 0;
             if (!MG_Config::Features.DisableTimerQuery) {
                 if (const auto getGpuTimestampNs = MG_Backend::gBackendFunctionsTable.GL.GetGpuTimestampNs) {
+                    MGP_FILL(GetGpuTimestampNs);
                     timestamp = getGpuTimestampNs();
                 }
             }
@@ -1972,6 +2301,24 @@ namespace MobileGL::MG_Impl::GLImpl {
         case GL_UNIFORM_BUFFER_START:
             RecordIndexedOnlyGetterError(__func__, pname);
             return;
+        // glBindBufferBase/Range set the GENERIC binding point too (GL 4.6 core 6.1.1), and this
+        // is the one indexed-buffer family whose non-indexed query was never answered - so it
+        // fell through to INVALID_ENUM and left the caller's variable holding whatever was in its
+        // stack slot. _START/_SIZE stay indexed-only, exactly like their uniform-buffer siblings.
+        case GL_ATOMIC_COUNTER_BUFFER_BINDING:
+            if (const auto& obj =
+                    MG_State::pGLContext->GetBufferBindingSlot(BufferTarget::AtomicCounter).GetBoundObject()) {
+                *params = static_cast<GLint>(obj->GetExternalIndex());
+            } else {
+                *params = 0;
+            }
+            return;
+        case GL_ATOMIC_COUNTER_BUFFER_START:
+            RecordIndexedOnlyGetterError(__func__, pname);
+            return;
+        case GL_ATOMIC_COUNTER_BUFFER_SIZE:
+            RecordIndexedOnlyGetterError(__func__, pname);
+            return;
         case GL_UNPACK_ALIGNMENT:
             *params = MG_State::pGLContext->GetPixelStoreParam(PixelStoreParam::UnpackAlignment);
             return;
@@ -2026,11 +2373,13 @@ namespace MobileGL::MG_Impl::GLImpl {
             params[3] = vp.w();
             return;
         }
-        case GL_VIEWPORT_INDEX_PROVOKING_VERTEX:
-            *params = GL_LAST_VERTEX_CONVENTION;
-            return;
         case GL_MAX_ELEMENT_INDEX:
-            *params = 1024 * 1024; // TODO
+            // 64-bit state (see GetInteger64v); the 32-bit query saturates, per the GL
+            // state-query conversion rules - the same shape GL_MAX_SHADER_STORAGE_BLOCK_SIZE
+            // uses. The real answer is 2^32-1 because both backends draw with GL_UNSIGNED_INT
+            // indices and neither bounds an index value; the old `1024 * 1024` was a placeholder
+            // that no draw path ever consulted.
+            *params = INT32_MAX;
             return;
         case GL_CONTEXT_PROFILE_MASK:
             // Reports the requested context profile (EGL defaults 3.x contexts to core);
@@ -2086,8 +2435,12 @@ namespace MobileGL::MG_Impl::GLImpl {
             *params = dynamicParameters.MaxComputeTextureImageUnits;
             break;
         case GL_MAX_COMBINED_COMPUTE_UNIFORM_COMPONENTS:
+            // The CLAMPED block count, i.e. exactly what GL_MAX_COMPUTE_UNIFORM_BLOCKS answers.
+            // GL 4.6 table 23.64 defines this as the components reachable through the blocks a
+            // stage may declare, so deriving it from the raw backend number described 256 blocks
+            // an application is only ever allowed 84 of.
             *params = GetMaxCombinedUniformComponents(kFrontendMaxComputeUniformComponents,
-                                                      dynamicParameters.MaxComputeUniformBlocks,
+                                                      ClampUniformBlockCount(dynamicParameters.MaxComputeUniformBlocks),
                                                       dynamicParameters.MaxUniformBlockSize);
             break;
         case GL_MAX_COMPUTE_WORK_GROUP_INVOCATIONS:
@@ -2116,17 +2469,31 @@ namespace MobileGL::MG_Impl::GLImpl {
         case GL_MAX_CLIP_DISTANCES:
             *params = dynamicParameters.MaxClipDistances;
             break;
+        // Both were a hard-coded GL_LAST_VERTEX_CONVENTION, derived from nothing. GL 4.6 table
+        // 23.65 permits GL_UNDEFINED_VERTEX for either, and that is what the backends report
+        // wherever they do not actually pin a convention - claiming one is a statement about
+        // which vertex of a primitive supplies gl_Layer / gl_ViewportIndex, and DirectGLES
+        // rasterizes only viewport 0 on a driver without GL_OES_viewport_array while
+        // DirectVulkan picks its provoking mode per pipeline. KHR-GLxx.viewport_array.query
+        // accepts all four values, and .provoking_vertex - which failed on both devices, in
+        // OPPOSITE directions - stops verifying as soon as either answer is undefined.
+        case GL_LAYER_PROVOKING_VERTEX:
+            *params = static_cast<GLint>(dynamicParameters.LayerProvokingVertex);
+            break;
+        case GL_VIEWPORT_INDEX_PROVOKING_VERTEX:
+            *params = static_cast<GLint>(dynamicParameters.ViewportIndexProvokingVertex);
+            break;
         case GL_MAX_COLOR_TEXTURE_SAMPLES:
-            *params = dynamicParameters.MaxColorTextureSamples;
+            *params = GetAdvertisedColorTextureMaxSamples();
             break;
         case GL_MAX_COMBINED_FRAGMENT_UNIFORM_COMPONENTS:
             *params = GetMaxCombinedUniformComponents(kFrontendMaxFragmentUniformComponents,
-                                                      kFrontendMaxFragmentUniformBlocks,
+                                                      ClampUniformBlockCount(kFrontendMaxFragmentUniformBlocks),
                                                       dynamicParameters.MaxUniformBlockSize);
             break;
         case GL_MAX_COMBINED_GEOMETRY_UNIFORM_COMPONENTS:
             *params = GetMaxCombinedUniformComponents(kFrontendMaxGeometryUniformComponents,
-                                                      kFrontendMaxGeometryUniformBlocks,
+                                                      ClampUniformBlockCount(kFrontendMaxGeometryUniformBlocks),
                                                       dynamicParameters.MaxUniformBlockSize);
             break;
         case GL_MAX_GEOMETRY_OUTPUT_VERTICES:
@@ -2140,14 +2507,14 @@ namespace MobileGL::MG_Impl::GLImpl {
             break;
         case GL_MAX_COMBINED_VERTEX_UNIFORM_COMPONENTS:
             *params = GetMaxCombinedUniformComponents(kFrontendMaxVertexUniformComponents,
-                                                      kFrontendMaxVertexUniformBlocks,
+                                                      ClampUniformBlockCount(kFrontendMaxVertexUniformBlocks),
                                                       dynamicParameters.MaxUniformBlockSize);
             break;
         case GL_MAX_CUBE_MAP_TEXTURE_SIZE:
             *params = dynamicParameters.MaxCubeMapTextureSize;
             break;
         case GL_MAX_DEPTH_TEXTURE_SAMPLES:
-            *params = dynamicParameters.MaxDepthTextureSamples;
+            *params = GetAdvertisedDepthTextureMaxSamples();
             break;
         case GL_MAX_FRAMEBUFFER_WIDTH:
             *params = dynamicParameters.MaxFramebufferWidth;
@@ -2174,7 +2541,7 @@ namespace MobileGL::MG_Impl::GLImpl {
             *params = dynamicParameters.MaxComputeImageUniforms;
             break;
         case GL_MAX_INTEGER_SAMPLES:
-            *params = dynamicParameters.MaxIntegerSamples;
+            *params = GetAdvertisedIntegerMaxSamples();
             break;
         case GL_MAX_RENDERBUFFER_SIZE:
             *params = dynamicParameters.MaxRenderbufferSize;
@@ -2185,12 +2552,56 @@ namespace MobileGL::MG_Impl::GLImpl {
         case GL_PATCH_VERTICES:
             *params = static_cast<GLint>(MG_State::pGLContext->GetPatchVertices());
             break;
+        // Float state, so glGetIntegerv rounds it (GL 4.6 core 2.2.2) - the exact values come back
+        // through glGetFloatv. Answered here so glGetBooleanv, which delegates to this getter for
+        // everything its own switch does not handle, does not report INVALID_ENUM for them.
+        case GL_PATCH_DEFAULT_OUTER_LEVEL: {
+            const FloatVec4& outer = MG_State::pGLContext->GetPatchDefaultOuterLevel();
+            for (Uint i = 0; i < 4; ++i) params[i] = static_cast<GLint>(std::lround(outer[i]));
+            break;
+        }
+        case GL_PATCH_DEFAULT_INNER_LEVEL: {
+            const FloatVec2& inner = MG_State::pGLContext->GetPatchDefaultInnerLevel();
+            for (Uint i = 0; i < 2; ++i) params[i] = static_cast<GLint>(std::lround(inner[i]));
+            break;
+        }
+        // GL 4.6 core table 23.66: whether the primitive-restart index terminates a patch.
+        // GL_FALSE is a legal answer and the true one - neither backend cuts a patch short, and
+        // the DirectVulkan draw path relies on this staying false (it resolves primitive restart
+        // to "never" for a PATCH_LIST topology on the strength of it).
+        case GL_PRIMITIVE_RESTART_FOR_PATCHES_SUPPORTED:
+            *params = GL_FALSE;
+            break;
         case GL_MAX_PATCH_VERTICES:
             *params = dynamicParameters.MaxPatchVertices;
             break;
         case GL_MAX_TESS_GEN_LEVEL:
             *params = dynamicParameters.MaxTessGenLevel;
             break;
+        // Same helper, and so the same arithmetic, as every other GL_MAX_COMBINED_*_UNIFORM_
+        // COMPONENTS: default-block components + blocks * (block size / 4). It reproduces the
+        // conformance suite's own formula exactly, so the two cannot drift.
+        case GL_MAX_COMBINED_TESS_CONTROL_UNIFORM_COMPONENTS:
+            *params = GetMaxCombinedUniformComponents(
+                static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_TESS_CONTROL_UNIFORM_COMPONENTS),
+                ClampUniformBlockCount(kFrontendMaxTessControlUniformBlocks), dynamicParameters.MaxUniformBlockSize);
+            break;
+        case GL_MAX_COMBINED_TESS_EVALUATION_UNIFORM_COMPONENTS:
+            *params = GetMaxCombinedUniformComponents(
+                static_cast<GLint>(MG_Util::ShaderTranspiler::MAX_TESS_EVALUATION_UNIFORM_COMPONENTS),
+                ClampUniformBlockCount(kFrontendMaxTessEvaluationUniformBlocks), dynamicParameters.MaxUniformBlockSize);
+            break;
+        // ARB_cull_distance. Backend-derived exactly like GL_MAX_CLIP_DISTANCES beside it, and
+        // for a stronger reason: a cull distance discards the whole primitive, so advertising
+        // eight the rasterizer cannot serve turns every culling draw into a silent no-op. Zero is
+        // the honest answer on a host with no cull-distance route, and the conformance suite then
+        // skips the functional cases instead of failing them deep inside a pixel comparison.
+        case GL_MAX_CULL_DISTANCES:
+            *params = dynamicParameters.MaxCullDistances;
+            break;
+        case GL_MAX_COMBINED_CLIP_AND_CULL_DISTANCES:
+            *params = dynamicParameters.MaxCombinedClipAndCullDistances;
+            break;
         case GL_MIN_PROGRAM_TEXTURE_GATHER_OFFSET:
             *params = dynamicParameters.MinProgramTextureGatherOffset;
             break;
@@ -2207,18 +2618,19 @@ namespace MobileGL::MG_Impl::GLImpl {
                                                           static_cast<Uint64>(INT32_MAX)));
             break;
         case GL_MAX_ATOMIC_COUNTER_BUFFER_BINDINGS:
+            // NOT the frontend's binding-point array size: GetIndexedBufferQueryPointCount
+            // clamps this family to the range a lowered counter block can actually be served
+            // from, which is the same number glslang compiles a layout(binding = N) atomic_uint
+            // against and the same one glBindBufferBase validates an index against.
             *params = static_cast<GLint>(GetIndexedBufferQueryPointCount(BufferTarget::AtomicCounter));
             break;
         case GL_MAX_ATOMIC_COUNTER_BUFFER_SIZE:
             // The conformance suite splits this evenly across every advertised binding point and
             // binds all of them in one glBindBuffersRange
-            // (KHR-GL44.multi_bind.functional_bind_buffers_range), so the pair has to divide:
-            // 32 bytes over 36 binding points is a zero-sized range, which BindBufferRange
-            // rejects with INVALID_VALUE before it binds anything. Floor the advertised size at
-            // one counter per binding point.
-            *params = std::max<GLint>(
-                kFrontendMaxAtomicCounterBufferSize,
-                static_cast<GLint>(GetIndexedBufferQueryPointCount(BufferTarget::AtomicCounter) * sizeof(GLuint)));
+            // (KHR-GL44.multi_bind.functional_bind_buffers_range), so the pair has to divide -
+            // a zero-sized range is INVALID_VALUE before BindBufferRange binds anything. The
+            // shared constant is 16384 over 8 binding points, which divides.
+            *params = kFrontendMaxAtomicCounterBufferSize;
             break;
         case GL_MAX_TEXTURE_BUFFER_SIZE:
             *params = dynamicParameters.MaxTextureBufferSize;
@@ -2240,7 +2652,25 @@ namespace MobileGL::MG_Impl::GLImpl {
             *params = kFrontendMaxTransformFeedbackSeparateAttribs;
             break;
         case GL_MAX_VERTEX_STREAMS:
-            *params = 1;
+            // ONE, which is under the GL 4.5 core table 23.62 minimum of four and is a known,
+            // deliberate non-conformance. It was briefly raised to 4 on the theory that streams
+            // 1..3 could exist and be permanently empty; measuring that decision refuted it.
+            // Raising the limit un-gates two CTS cases per package across KHR-GL40..GL46 -
+            // transform_feedback.draw_xfb_stream_test (which stops being skipped) and
+            // transform_feedback3.multiple_streams (which stops reporting NotSupported) - and
+            // both then fail, because nothing in the shader pipeline supports layout(stream = N),
+            // EmitStreamVertex or EndStreamPrimitive, and because the query state machine tracks
+            // one active query per TARGET rather than per (target, stream). That is 14 new
+            // failures against 2 gained limits passes, and a 4 nothing can back is the
+            // advertised-caps lie with the sign flipped.
+            //
+            // The real fix is the feature, not the number: per-stream capture needs
+            // layout(stream = N) through the transpiler plus per-(target, stream) query slots,
+            // which DirectVulkan could back with VK_EXT_transform_feedback's geometryStreams and
+            // DirectGLES cannot back at all (ES has no vertex streams). Until that lands, one is
+            // the honest count and every stream-addressing entry point bounds itself by THIS
+            // query, so raising it later moves them all together.
+            *params = kFrontendMaxVertexStreams;
             break;
         case GL_TRANSFORM_FEEDBACK_ACTIVE:
             *params = MG_State::pGLContext->IsTransformFeedbackActive() ? 1 : 0;
@@ -2257,15 +2687,36 @@ namespace MobileGL::MG_Impl::GLImpl {
         case GL_MAX_TEXTURE_SIZE:
             *params = dynamicParameters.MaxTextureSize;
             break;
-        case GL_MAX_UNIFORM_BUFFER_BINDINGS:
+        case GL_MAX_UNIFORM_BUFFER_BINDINGS: {
             // Never advertise more bindings than the state layer's indexed-binding array can track
             // (BufferState::BufferBindingPointCount): glBindBufferBase rejects indices past that
             // capacity, and the GL CTS per-case state reset calls glBindBufferBase on every
-            // advertised index and expects no error. The floor equals the GL 3.3 core minimum
-            // (36), so the clamp never under-advertises.
+            // advertised index and expects no error. The floor is the GL 4.5 core minimum, and
+            // the array was widened to exactly it, so the two coincide by construction.
+            //
+            // WHY THE BACKEND'S OWN COUNT IS NOT THE CEILING HERE, unlike the shader-storage
+            // family. A GL uniform binding point is where an APPLICATION parks a buffer; it is
+            // not a driver binding point. Neither backend forwards it as one on the draw path:
+            // DirectGLES rebinds the blocks a program declares onto COMPACTED ES points
+            // (BindCurrentProgramWithResources maps block i to ES point i+1) and DirectVulkan
+            // resolves each block to a descriptor. So what the host driver's count bounds is how
+            // many blocks ONE PROGRAM may use, not how many points an application may bind.
+            //
+            // That per-program number is NOT GL_MAX_COMBINED_UNIFORM_BLOCKS (84, the six-stage
+            // sum): no single program can reach it. A graphics program is bounded by the five
+            // graphics stages' per-stage counts, 14 each, so 70 blocks plus the global UBO at ES
+            // point 0 = 71 - inside the ES 3.2 minimum of 72. A compute program is bounded by
+            // GL_MAX_COMPUTE_UNIFORM_BLOCKS, which on DirectGLES is the ES driver's own count
+            // (GL-scale, ~14) and on DirectVulkan is served from descriptors with no ES binding
+            // points involved. Raising any per-stage graphics count past 14 is what would break
+            // this, so that is the edit to check against the ES ceiling - not this one.
+            static_assert(static_cast<GLint>(MG_State::GLState::BufferBindingPointCount) >=
+                              kFrontendMinUniformBufferBindings,
+                          "the indexed-binding array must be able to hold every advertised uniform binding point");
             *params = std::clamp(dynamicParameters.MaxUniformBufferBindings, kFrontendMinUniformBufferBindings,
                                  static_cast<GLint>(MG_State::GLState::BufferBindingPointCount));
             break;
+        }
         case GL_MAX_UNIFORM_BLOCK_SIZE:
             *params = dynamicParameters.MaxUniformBlockSize;
             break;
@@ -2303,7 +2754,12 @@ namespace MobileGL::MG_Impl::GLImpl {
             *params = static_cast<GLint>(dynamicParameters.PointSizeGranularity);
             break;
         case GL_SHADER_STORAGE_BUFFER_OFFSET_ALIGNMENT:
-            *params = static_cast<GLint>(dynamicParameters.UniformBufferOffsetAlignment);
+            // The STORAGE alignment, which is its own limit - this used to answer with the
+            // uniform one. They differ on real hardware (Adreno 830: 32 uniform, 64 storage), and
+            // under-reporting it is silent: ValidateBindBufferRange accepts the offset, the ES
+            // driver accepts it too without raising an error, and the shader's writes then land
+            // at an address the application never bound.
+            *params = static_cast<GLint>(dynamicParameters.ShaderStorageBufferOffsetAlignment);
             break;
         case GL_SMOOTH_LINE_WIDTH_RANGE:
             params[0] = static_cast<GLint>(dynamicParameters.SmoothLineWidthRangeMin);
@@ -2340,7 +2796,7 @@ namespace MobileGL::MG_Impl::GLImpl {
                                                         : dynamicParameters.MaxDrawBuffers;
             break;
         case GL_MAX_SAMPLES:
-            *params = std::max(dynamicParameters.MaxSamples, kFrontendMaxSamples);
+            *params = GetAdvertisedMaxSamples();
             break;
         case GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT:
             // Float state (see GetFloatv); rounded to nearest for the integer query per GL 3.3 6.1.2.

MobileGL/MG_Impl/GLImpl/Getter/GL_Getter.h

@@ -24,4 +24,25 @@ namespace MobileGL::MG_Impl::GLImpl {
     void GetInteger64i_v(GLenum target, GLuint index, GLint64* data);
     GLenum GetError();
     GLenum GetGraphicsResetStatus();
+    // The GL_MAX_SAMPLES value MobileGL advertises, i.e. the driver's value floored to the GL
+    // core minimum of 4. This is the RENDERBUFFER ceiling; the three per-category texture
+    // ceilings below have a minimum of one and are reported as probed.
+    GLint GetAdvertisedMaxSamples();
+    // Exactly what GL_MAX_COLOR_TEXTURE_SAMPLES / GL_MAX_DEPTH_TEXTURE_SAMPLES /
+    // GL_MAX_INTEGER_SAMPLES report: the probed backend limit floored at the GL 4.6 core minimum
+    // of ONE (table 23.53). Exported so the frontend's storage validation enforces exactly what
+    // the query promised - it used to floor both at 4 and then let the backend quietly
+    // under-allocate whatever the driver could not actually provide.
+    GLint GetAdvertisedColorTextureMaxSamples();
+    GLint GetAdvertisedDepthTextureMaxSamples();
+    GLint GetAdvertisedIntegerMaxSamples();
+    // What glGetIntegerv(GL_SAMPLES) answers for the CURRENT draw framebuffer: the largest sample
+    // count over its attachments, and 0 for a single-sample or default framebuffer (GL 4.6 core
+    // 9.2.3 / 22.2 - GL_SAMPLE_BUFFERS is 1 exactly when this is non-zero).
+    //
+    // Shared rather than duplicated because two callers need the identical number and disagreeing
+    // would be a silent bug: the query itself, and the draw path's write of the reserved
+    // gl_NumSamples stand-in - a shader comparing gl_NumSamples against glGetIntegerv(GL_SAMPLES)
+    // is exactly what the sample_variables CTS does.
+    GLint ResolveDrawFramebufferSampleCount();
 } // namespace MobileGL::MG_Impl::GLImpl

MobileGL/MG_Impl/GLImpl/Program/GL_Program.h

@@ -13,6 +13,12 @@ namespace MobileGL::MG_Impl::GLImpl {
     void AttachShader(GLuint program, GLuint shader);
     void BindAttribLocation(GLuint program, GLuint index, const GLchar* name);
     void CompileShader(GLuint shader);
+    // GL_ARB_gl_spirv, core since 4.6. The pair is a two-step operation: glShaderBinary attaches
+    // the module to one or more shader objects, glSpecializeShader names its entry point and
+    // supplies its specialization constants and is what actually compiles them.
+    void ShaderBinary(GLsizei count, const GLuint* shaders, GLenum binaryformat, const void* binary, GLsizei length);
+    void SpecializeShader(GLuint shader, const GLchar* pEntryPoint, GLuint numSpecializationConstants,
+                          const GLuint* pConstantIndex, const GLuint* pConstantValue);
     GLuint CreateProgram(void);
     GLuint CreateShader(GLenum type);
     void DeleteProgram(GLuint program);
@@ -140,6 +146,7 @@ namespace MobileGL::MG_Impl::GLImpl {
                               const GLenum* props, GLsizei bufSize, GLsizei* length, GLint* params);
     GLint GetProgramResourceLocation(GLuint program, GLenum programInterface, const GLchar* name);
     GLint GetProgramResourceLocationIndex(GLuint program, GLenum programInterface, const GLchar* name);
+    void GetActiveAtomicCounterBufferiv(GLuint program, GLuint bufferIndex, GLenum pname, GLint* params);
     void ShaderStorageBlockBinding(GLuint program, GLuint storageBlockIndex, GLuint storageBlockBinding);
     void Uniform1d(GLint location, GLdouble v0);
     void Uniform1dv(GLint location, GLsizei count, const GLdouble* value);

MobileGL/MG_Impl/GLImpl/Program/GL_ProgramPipeline.cpp

@@ -192,6 +192,15 @@ namespace MobileGL::MG_Impl::GLImpl {
                                     std::format("Program {} has not been linked successfully.", program));
                 return;
             }
+            // GL 4.6 core 7.4: "INVALID_OPERATION is generated if program was not linked with its
+            // PROGRAM_SEPARABLE status set". The LATCHED flag is the one that decides - a program
+            // whose live flag was cleared after a separable link is still a legal stage, and a
+            // program whose live flag was set after a non-separable link is not.
+            if (!programObject->GetLinkedSeparable()) {
+                RecordPipelineError(ErrorCode::InvalidOperation, __func__,
+                                    std::format("Program {} was not linked as a separable program.", program));
+                return;
+            }
         }
 
         const GLbitfield selected = stages == GL_ALL_SHADER_BITS ? kAllStageBits : stages;

MobileGL/MG_Impl/GLImpl/Program/ProgramInterface.cpp

@@ -19,7 +19,7 @@ namespace MobileGL::MG_Impl::GLImpl::ProgramInterface {
         // "<getAtomicCounterBlockName()>_<binding>" (ParseContextBase.cpp), one per GL
         // atomic-counter binding point. That block IS the GL_ATOMIC_COUNTER_BUFFER resource
         // and its trailing number IS GL_BUFFER_BINDING; its members stay GL_UNIFORMs.
-        constexpr const char* kAtomicCounterBlockPrefix = "gl_AtomicCounterBlock";
+        constexpr const char* kAtomicCounterBlockPrefix = MG_Util::ShaderTranspiler::ATOMIC_COUNTER_BLOCK_PREFIX;
 
         enum class BlockKind {
             Uniform,       // a real GL uniform block
@@ -81,19 +81,18 @@ namespace MobileGL::MG_Impl::GLImpl::ProgramInterface {
         // The enumerated spelling of an array resource is "name[0]". glslang already applies
         // that to uniforms and buffer variables (EShReflectionBasicArraySuffix), but never to
         // stage inputs/outputs, so those get it here.
-        String WithArraySuffix(const String& name, const glslang::TType* type) {
-            if (type == nullptr || !type->isArray() || EndsWithZeroSubscript(name)) return name;
+        String WithArraySuffix(const String& name, const ProgramObject::TypeFacts& type) {
+            if (!type.isArray || EndsWithZeroSubscript(name)) return name;
             return name + "[0]";
         }
 
         // GL_ARRAY_SIZE: element count for a sized array, 0 for a runtime-sized one
         // (a shader storage block's unsized trailing member), 1 for a non-array.
-        GLint ArraySizeOf(const glslang::TType* type, GLint reflectedSize) {
-            if (type != nullptr && type->isArray()) {
-                if (!type->isSizedArray()) return 0;
-                return type->getOuterArraySize();
-            }
-            return reflectedSize < 1 ? 1 : reflectedSize;
+        // `record.arraySize` is already the sized-array/reflected-size resolution; the only
+        // extra rule here is GL's 0 for a runtime-sized array.
+        GLint ArraySizeOf(const ProgramObject::ResourceReflection& record) {
+            if (record.type.isArray && !record.type.isSizedArray) return 0;
+            return record.arraySize;
         }
 
         // Two spellings name the same resource when they are equal, or differ only by the
@@ -174,22 +173,21 @@ namespace MobileGL::MG_Impl::GLImpl::ProgramInterface {
             return static_cast<GLint>(element);
         }
 
-        BlockKind ClassifyBlock(const glslang::TObjectReflection& block) {
+        BlockKind ClassifyBlock(const ProgramObject::BlockReflection& block) {
             if (std::strstr(block.name.c_str(), MG_Util::ShaderTranspiler::GLOBAL_UBO_NAME) != nullptr) {
                 return BlockKind::GlobalUbo;
             }
             if (IsAtomicCounterBlockName(block.name)) return BlockKind::AtomicCounter;
-            const glslang::TType* type = block.getType();
-            if (type != nullptr && type->getQualifier().storage == glslang::EvqBuffer) return BlockKind::Storage;
+            if (block.type.isBuffer) return BlockKind::Storage;
             return BlockKind::Uniform;
         }
 
         // std140/std430 column stride, the same vec4-rounded rule ProgramObject applies to
         // uniform matrices. 0 for a non-matrix.
-        GLint MatrixStrideOf(const glslang::TType* type) {
-            if (type == nullptr || !type->isMatrix()) return 0;
-            const bool rowMajor = type->getQualifier().layoutMatrix == glslang::ElmRowMajor;
-            const int strideVectorComponents = rowMajor ? type->getMatrixCols() : type->getMatrixRows();
+        GLint MatrixStrideOf(const ProgramObject::TypeFacts& type) {
+            if (!type.isMatrix) return 0;
+            const bool rowMajor = type.layoutMatrix == static_cast<Int>(glslang::ElmRowMajor);
+            const int strideVectorComponents = rowMajor ? type.matrixCols : type.matrixRows;
             constexpr int scalarSize = 4;
             const int vectorAlignment = (strideVectorComponents <= 1)    ? scalarSize
                                         : (strideVectorComponents == 2) ? 2 * scalarSize
@@ -197,9 +195,9 @@ namespace MobileGL::MG_Impl::GLImpl::ProgramInterface {
             return (vectorAlignment + 15) & ~15;
         }
 
-        GLint IsRowMajorOf(const glslang::TType* type) {
-            if (type == nullptr || !type->isMatrix()) return 0;
-            return type->getQualifier().layoutMatrix == glslang::ElmRowMajor ? 1 : 0;
+        GLint IsRowMajorOf(const ProgramObject::TypeFacts& type) {
+            if (!type.isMatrix) return 0;
+            return type.layoutMatrix == static_cast<Int>(glslang::ElmRowMajor) ? 1 : 0;
         }
 
         GLint MappedLocation(Int rawLocation) {
@@ -227,12 +225,12 @@ namespace MobileGL::MG_Impl::GLImpl::ProgramInterface {
         // Note the union is used even when it is empty: an array element nobody dereferenced has
         // no member bits and is genuinely referenced by nobody, which is the whole point - falling
         // back to the block's own mask there would restore the over-approximation.
-        Vector<Uint32> BuildBlockStagesFromMembers(const glslang::TProgram& reflection, Int blockCount) {
-            auto& mutableReflection = const_cast<glslang::TProgram&>(reflection);
+        Vector<Uint32> BuildBlockStagesFromMembers(const ProgramObject::LinkArtifacts& reflection,
+                                                    Int blockCount) {
             Vector<Uint32> stagesByBlock(static_cast<SizeT>(blockCount < 0 ? 0 : blockCount), 0u);
-            const Int uniformCount = mutableReflection.getNumUniformVariables();
+            const Int uniformCount = static_cast<Int>(reflection.uniformReflection.size());
             for (Int index = 0; index < uniformCount; ++index) {
-                const auto& uniform = mutableReflection.getUniform(index);
+                const auto& uniform = reflection.uniformReflection[index];
                 const Int owner = uniform.index;
                 if (owner < 0 || owner >= blockCount) continue;
                 stagesByBlock[static_cast<SizeT>(owner)] |= static_cast<Uint32>(uniform.stages);
@@ -250,7 +248,7 @@ namespace MobileGL::MG_Impl::GLImpl::ProgramInterface {
         // ss[1] and requires both to report the fragment stage, which only glslang's own
         // (deliberately over-approximating) block mask gets right. Storage and atomic-counter
         // blocks therefore keep that mask untouched.
-        Uint32 UniformBlockStages(const glslang::TObjectReflection& block, const Vector<Uint32>& stagesFromMembers,
+        Uint32 UniformBlockStages(const ProgramObject::BlockReflection& block, const Vector<Uint32>& stagesFromMembers,
                                   Int tIndex) {
             String arrayBase;
             Uint element = 0;
@@ -264,15 +262,15 @@ namespace MobileGL::MG_Impl::GLImpl::ProgramInterface {
             return stagesFromMembers[static_cast<SizeT>(tIndex)];
         }
 
-        void BuildBlocks(ProgramObject& program, const glslang::TProgram& reflection, Model& model,
+        void BuildBlocks(ProgramObject& program, const ProgramObject::LinkArtifacts& reflection, Model& model,
                          Vector<BlockKind>& blockKind, Vector<Int>& blockInterfaceIndex) {
-            const Int blockCount = const_cast<glslang::TProgram&>(reflection).getNumUniformBlocks();
+            const Int blockCount = static_cast<Int>(reflection.blockReflection.size());
             blockKind.assign(blockCount, BlockKind::Uniform);
             blockInterfaceIndex.assign(blockCount, -1);
             const Vector<Uint32> stagesFromMembers = BuildBlockStagesFromMembers(reflection, blockCount);
 
             for (Int tIndex = 0; tIndex < blockCount; ++tIndex) {
-                const auto& block = const_cast<glslang::TProgram&>(reflection).getUniformBlock(tIndex);
+                const auto& block = reflection.blockReflection[tIndex];
                 const BlockKind kind = ClassifyBlock(block);
                 blockKind[tIndex] = kind;
                 if (kind == BlockKind::AtomicCounter) {
@@ -293,7 +291,7 @@ namespace MobileGL::MG_Impl::GLImpl::ProgramInterface {
                     // glShaderStorageBlockBinding wins over the declaration (GL 4.6 §7.6.2 -
                     // exactly the same rule GL_UNIFORM_BLOCK follows through
                     // GetUniformBlockBinding below).
-                    const GLint declared = block.getBinding();
+                    const GLint declared = block.binding;
                     resource.bufferBinding = declared < 0 ? 0 : declared + BlockArrayElement(block.name);
                     const Int rebound = program.GetShaderStorageBlockBindingOverride(block.name);
                     if (rebound >= 0) resource.bufferBinding = static_cast<GLint>(rebound);
@@ -307,38 +305,53 @@ namespace MobileGL::MG_Impl::GLImpl::ProgramInterface {
             // GL_UNIFORM_BLOCK keeps the index space glUniformBlockBinding and
             // glGetActiveUniformBlockiv already use, so an index handed out here is usable
             // with them (which is exactly what the CTS does).
-            const Int glBlockCount = program.GetActiveUniformBlocksCount();
+            const Int glBlockCount = program.GetGlUniformBlockCount();
             for (Int glIndex = 0; glIndex < glBlockCount; ++glIndex) {
+                // The block-space index the block-keyed accessors want; the two spaces differ
+                // whenever the program also has a storage or atomic counter block, which
+                // glslang files under the same reflection list (no EShReflectionSeparateBuffers).
+                const Int blockIndex = program.BlockIndexFromGlUniformBlock(static_cast<Uint>(glIndex));
                 Resource resource;
-                resource.name = program.GetUniformBlockName(glIndex);
-                resource.bufferBinding = static_cast<GLint>(program.GetUniformBlockBinding(glIndex));
-                resource.bufferDataSize = static_cast<GLint>(program.GetUBOSizeAt(glIndex));
-                const Int tIndex = program.TProgramBlockIndex(static_cast<Uint>(glIndex));
+                resource.name = program.GetUniformBlockName(static_cast<Uint>(blockIndex));
+                resource.bufferBinding = static_cast<GLint>(program.GetUniformBlockBinding(static_cast<Uint>(blockIndex)));
+                resource.bufferDataSize = static_cast<GLint>(program.GetUBOSizeAt(static_cast<Uint>(blockIndex)));
+                const Int tIndex = program.TProgramBlockIndex(static_cast<Uint>(blockIndex));
                 if (tIndex >= 0 && tIndex < blockCount) {
-                    resource.stages = UniformBlockStages(const_cast<glslang::TProgram&>(reflection).getUniformBlock(tIndex),
+                    resource.stages = UniformBlockStages(reflection.blockReflection[tIndex],
                                                       stagesFromMembers, tIndex);
                 }
                 model.uniformBlocks.push_back(Move(resource));
             }
         }
 
-        void BuildUniformsAndBufferVariables(ProgramObject& program, const glslang::TProgram& reflection, Model& model,
+        void BuildUniformsAndBufferVariables(ProgramObject& program,
+                                             const ProgramObject::LinkArtifacts& reflection, Model& model,
                                              const Vector<BlockKind>& blockKind,
                                              const Vector<Int>& blockInterfaceIndex) {
-            const Uint uniformCount = program.GetUniformCount();
-            for (Uint glIndex = 0; glIndex < uniformCount; ++glIndex) {
-                const Int tIndex = program.TProgramUniformIndex(glIndex);
-                const auto& refl = const_cast<glslang::TProgram&>(reflection).getUniform(tIndex);
-                const glslang::TType* type = refl.getType();
+            // Walks the TPROGRAM uniform space, not the GL one. A buffer variable is not a GL
+            // uniform (GL 4.6 core 7.3.1) and DoReflection therefore keeps it out of the GL
+            // active-uniform index space - but GL_BUFFER_VARIABLE still has to enumerate it, and
+            // this is the only place that does. GL uniforms keep their GL index as their
+            // GL_UNIFORM resource index: the GL space is a subsequence of this one, so pushing
+            // the GL-visible entries in this order preserves the correspondence.
+            const Int tUniformCount = static_cast<Int>(reflection.uniformReflection.size());
+            for (Int tIndex = 0; tIndex < tUniformCount; ++tIndex) {
+                const auto& refl = ProgramObject::UniformAtIn(reflection, tIndex);
+                const auto& type = refl.type;
                 const Int owner = refl.index;
                 const BlockKind kind = (owner >= 0 && owner < static_cast<Int>(blockKind.size()))
                                            ? blockKind[owner]
                                            : BlockKind::GlobalUbo;
+                const Int glIndex = program.GlUniformIndexFromTProgram(tIndex);
+                // Everything except a buffer variable is enumerated through the GL space, so a
+                // uniform the relaxed parse swept out of it (a declared-but-dead default-block
+                // one) stays out of GL_UNIFORM too.
+                if (kind != BlockKind::Storage && glIndex < 0) continue;
 
                 Resource resource;
                 resource.name = refl.name;
                 resource.type = static_cast<GLenum>(refl.glDefineType);
-                resource.arraySize = ArraySizeOf(type, refl.size);
+                resource.arraySize = ArraySizeOf(refl);
                 resource.stages = static_cast<Uint32>(refl.stages);
 
                 if (kind == BlockKind::Storage) {
@@ -366,11 +379,12 @@ namespace MobileGL::MG_Impl::GLImpl::ProgramInterface {
                     resource.atomicCounterBufferIndex = blockInterfaceIndex[owner];
                     resource.location = -1;
                 } else {
-                    resource.blockIndex = program.GetActiveUniformBlockIndex(glIndex);
-                    resource.offset = program.GetActiveUniformOffset(glIndex);
-                    resource.arrayStride = program.GetActiveUniformArrayStride(glIndex);
-                    resource.matrixStride = program.GetActiveUniformMatrixStride(glIndex);
-                    resource.isRowMajor = program.GetActiveUniformIsRowMajor(glIndex);
+                    const Uint glUniformIndex = static_cast<Uint>(glIndex);
+                    resource.blockIndex = program.GetActiveUniformBlockIndex(glUniformIndex);
+                    resource.offset = program.GetActiveUniformOffset(glUniformIndex);
+                    resource.arrayStride = program.GetActiveUniformArrayStride(glUniformIndex);
+                    resource.matrixStride = program.GetActiveUniformMatrixStride(glUniformIndex);
+                    resource.isRowMajor = program.GetActiveUniformIsRowMajor(glUniformIndex);
                     // A member of a named uniform block has no location, whatever the
                     // frontend's own location table says (it hands one out to every uniform
                     // so glUniform* can address block members through the global UBO).
@@ -389,12 +403,16 @@ namespace MobileGL::MG_Impl::GLImpl::ProgramInterface {
                         static_cast<GLuint>(i));
                 }
             }
-            for (SizeT blockIndex = 0; blockIndex < model.uniformBlocks.size(); ++blockIndex) {
+            for (SizeT glBlockIndex = 0; glBlockIndex < model.uniformBlocks.size(); ++glBlockIndex) {
                 // Members of an arrayed block are reflected once, against instance [0].
-                const Int owner = static_cast<Int>(program.GetUniformBlockMemberOwnerIndex(static_cast<Uint>(blockIndex)));
+                // GetUniformBlockMemberOwnerIndex takes and answers BLOCK indices, while
+                // Resource::blockIndex is a GL_UNIFORM_BLOCK index, so translate both ways.
+                const Int blockIndex = program.BlockIndexFromGlUniformBlock(static_cast<Uint>(glBlockIndex));
+                const Int owner = program.GlUniformBlockIndexFromBlock(
+                    static_cast<Int>(program.GetUniformBlockMemberOwnerIndex(static_cast<Uint>(blockIndex))));
                 for (SizeT i = 0; i < model.uniforms.size(); ++i) {
                     if (model.uniforms[i].blockIndex == owner) {
-                        model.uniformBlocks[blockIndex].activeVariables.push_back(static_cast<GLuint>(i));
+                        model.uniformBlocks[glBlockIndex].activeVariables.push_back(static_cast<GLuint>(i));
                     }
                 }
             }
@@ -414,17 +432,13 @@ namespace MobileGL::MG_Impl::GLImpl::ProgramInterface {
         // program that redeclares `out gl_PerVertex { vec4 gl_Position; }` still carries
         // gl_PointSize and gl_ClipDistance through the block-unwrapping reflection, and they
         // are not part of its output interface.
-        Bool IsHiddenBlockMember(const glslang::TType* type) {
-            return type != nullptr && type->getBasicType() == glslang::EbtVoid;
-        }
+        Bool IsHiddenBlockMember(const ProgramObject::TypeFacts& type) { return type.isVoid; }
 
-        void BuildStageIO(ProgramObject& program, const glslang::TProgram& reflection, Model& model) {
-            auto& mutableReflection = const_cast<glslang::TProgram&>(reflection);
-
-            const Int inputCount = mutableReflection.getNumPipeInputs();
+        void BuildStageIO(ProgramObject& program, const ProgramObject::LinkArtifacts& reflection, Model& model) {
+            const Int inputCount = static_cast<Int>(reflection.pipeInputReflection.size());
             for (Int index = 0; index < inputCount; ++index) {
-                const auto& refl = mutableReflection.getPipeInput(index);
-                const glslang::TType* type = refl.getType();
+                const auto& refl = reflection.pipeInputReflection[index];
+                const auto& type = refl.type;
                 if (IsHiddenBlockMember(type)) continue;
                 Resource resource;
                 // The Vulkan-semantics parse reflects the vertex builtins under their SPIR-V
@@ -432,10 +446,10 @@ namespace MobileGL::MG_Impl::GLImpl::ProgramInterface {
                 const String& glName = ProgramObject::NormalizeBuiltinPipeInputName(refl.name);
                 resource.name = WithArraySuffix(glName, type);
                 resource.type = static_cast<GLenum>(refl.glDefineType);
-                resource.arraySize = ArraySizeOf(type, refl.size);
+                resource.arraySize = ArraySizeOf(refl);
                 resource.location = program.GetAttributeLocation(refl.name);
-                if (resource.location < 0) resource.location = MappedLocation(static_cast<Int>(refl.layoutLocation()));
-                resource.isPerPatch = (type != nullptr && type->getQualifier().patch) ? 1 : 0;
+                if (resource.location < 0) resource.location = MappedLocation(refl.location);
+                resource.isPerPatch = type.isPatch ? 1 : 0;
                 resource.stages = static_cast<Uint32>(refl.stages);
                 model.programInputs.push_back(Move(resource));
             }
@@ -447,16 +461,16 @@ namespace MobileGL::MG_Impl::GLImpl::ProgramInterface {
             // carries its own layout(location=N)), and a location then manufactures a color
             // index of 0 where GL requires -1
             // (KHR-GL43.program_interface_query.separate-programs-tess-control).
-            const Bool lastStageIsFragment = mutableReflection.getIntermediate(EShLangFragment) != nullptr;
-            const Int outputCount = mutableReflection.getNumPipeOutputs();
+            const Bool lastStageIsFragment = reflection.lastStageIsFragment;
+            const Int outputCount = static_cast<Int>(reflection.pipeOutputReflection.size());
             for (Int index = 0; index < outputCount; ++index) {
-                const auto& refl = mutableReflection.getPipeOutput(index);
-                const glslang::TType* type = refl.getType();
+                const auto& refl = reflection.pipeOutputReflection[index];
+                const auto& type = refl.type;
                 if (IsHiddenBlockMember(type)) continue;
                 Resource resource;
                 resource.name = WithArraySuffix(refl.name, type);
                 resource.type = static_cast<GLenum>(refl.glDefineType);
-                resource.arraySize = ArraySizeOf(type, refl.size);
+                resource.arraySize = ArraySizeOf(refl);
                 resource.location = MappedLocation(program.GetFragmentDataLocation(refl.name.c_str()));
                 if (resource.location < 0 || !lastStageIsFragment) {
                     // A built-in output (gl_FragDepth, gl_SampleMask) has no location, and a
@@ -467,11 +481,11 @@ namespace MobileGL::MG_Impl::GLImpl::ProgramInterface {
                     resource.locationIndex = program.GetFragmentDataIndex(refl.name.c_str());
                     // glBindFragDataLocationIndexed wins; otherwise the shader's
                     // layout(index = N), which the frag-data maps never saw.
-                    if (resource.locationIndex == 0 && type != nullptr && type->getQualifier().hasIndex()) {
-                        resource.locationIndex = static_cast<GLint>(type->getQualifier().layoutIndex);
+                    if (resource.locationIndex == 0 && type.hasIndex) {
+                        resource.locationIndex = static_cast<GLint>(type.layoutIndex);
                     }
                 }
-                resource.isPerPatch = (type != nullptr && type->getQualifier().patch) ? 1 : 0;
+                resource.isPerPatch = type.isPatch ? 1 : 0;
                 resource.stages = static_cast<Uint32>(refl.stages);
                 model.programOutputs.push_back(Move(resource));
             }
@@ -511,15 +525,14 @@ namespace MobileGL::MG_Impl::GLImpl::ProgramInterface {
         Model BuildModel(ProgramObject& program) {
             Model model;
             if (!program.GetLinkStatus()) return model;
-            const glslang::TProgram* reflection = program.GetReflection();
-            if (reflection == nullptr) return model;
+            const ProgramObject::LinkArtifacts& reflection = program.GetLinkReflection();
             model.valid = true;
 
             Vector<BlockKind> blockKind;
             Vector<Int> blockInterfaceIndex;
-            BuildBlocks(program, *reflection, model, blockKind, blockInterfaceIndex);
-            BuildUniformsAndBufferVariables(program, *reflection, model, blockKind, blockInterfaceIndex);
-            BuildStageIO(program, *reflection, model);
+            BuildBlocks(program, reflection, model, blockKind, blockInterfaceIndex);
+            BuildUniformsAndBufferVariables(program, reflection, model, blockKind, blockInterfaceIndex);
+            BuildStageIO(program, reflection, model);
             BuildXfb(program, model);
             return model;
         }

MobileGL/MG_Impl/GLImpl/Query/GL_Query.cpp

@@ -12,6 +12,7 @@
 #include <MG_Backend/BackendObjects.h>
 #include <MG_State/GLState/Core.h>
 #include <MG_State/GLState/ErrorState/ErrorInfo.h>
+#include <MG_Impl/Pipe/PipeFill.h>
 
 namespace MobileGL::MG_Impl::GLImpl {
     namespace {
@@ -31,8 +32,15 @@ namespace MobileGL::MG_Impl::GLImpl {
             Bool ended = false;
             Bool resultCached = false;
             Uint64 cachedResult = 0;
-            // Transform feedback primitive counter at BeginQuery time.
+            // The transform feedback primitive counter matching this query's target, at
+            // BeginQuery time.
             Uint64 counterSnapshot = 0;
+            // Capture-draw counters at BeginQuery time: how many capture draws the CPU
+            // accounting had reproduced exactly, and how many of those it could not (a
+            // geometry stage amplifies). Their deltas decide whether the CPU result may
+            // stand in for the backend's.
+            Uint64 accountedCaptureDrawSnapshot = 0;
+            Uint64 geometryCaptureDrawSnapshot = 0;
         };
 
         // Query calls may arrive from any thread (launchers migrate the context
@@ -52,6 +60,62 @@ namespace MobileGL::MG_Impl::GLImpl {
         GLuint g_activePrimitivesGeneratedQueryId = 0;
         // Id of the query active on GL_SAMPLES_PASSED (0 = none).
         GLuint g_activeSamplesPassedQueryId = 0;
+        // Ids of the queries active on the GL_ARB_pipeline_statistics_query targets, one slot per
+        // target (0 = none). A map rather than a field per target: the eleven behave identically
+        // and none of them has any state beyond "which object is counting".
+        UnorderedMap<GLenum, GLuint> g_activePipelineStatisticsQueryIds;
+
+        // Whether MobileGL puts GL_ARB_tessellation_shader in its extension string. Read from the
+        // ADVERTISED list rather than from a capability bit for the same reason
+        // BackendSupportsTextureViews does (GL_Texture.cpp): it makes "MobileGL claims tessellation
+        // support" and "the tessellation-conditional API surface is open" the same fact by
+        // construction, so the day a backend starts advertising the string the surface below opens
+        // with it and no second edit is owed.
+        Bool AdvertisesTessellationShaderExtension() {
+            const auto& activeBackendObject = MG_Backend::pActiveBackendObject;
+            if (!activeBackendObject) return false;
+            const auto& extensions = activeBackendObject->GetRendererInfo().RendererGLInfo.Extensions;
+            return std::find(extensions.begin(), extensions.end(), E_GL_ARB_tessellation_shader) != extensions.end();
+        }
+
+        // The eleven pipeline-statistics counters (GL 4.6 core table 4.3 / ARB_pipeline_statistics_query).
+        // A 4.6 core context ACCEPTS the nine unconditional ones at glBeginQuery - there is no query
+        // by which an application could learn otherwise before calling. MobileGL instruments none of
+        // them, and says so the way GL 4.6 core 4.2.1 provides for: GL_QUERY_COUNTER_BITS answers
+        // zero for these targets, which is the spec's own signal that the counter is unsupported and
+        // its results indeterminate. That is an honest zero, not an advertised capability - the
+        // alternative, GL_INVALID_ENUM on a core entry point, is both non-conformant AND less
+        // informative.
+        //
+        // The two TESSELLATION targets are the exception, because ARB_pipeline_statistics_query
+        // makes them conditional on tessellation support rather than unconditional, and the only
+        // thing an application (or the conformance suite) can read to decide whether an
+        // implementation has it is the GL_ARB_tessellation_shader string. MobileGL does not emit it
+        // today, so these two answer GL_INVALID_ENUM: an API surface that accepts a
+        // tessellation-conditional token while withholding the string that announces the condition
+        // is self-contradictory, and it is the contradiction the suite catches
+        // (KHR-GL46.pipeline_statistics_query_tests_ARB.api_coverage_unsupported_calls, whose
+        // support probe is gl4cPipelineStatisticsQueryTests.cpp:1166-1176). The gate is the
+        // advertisement itself, not a hardcoded "no", so this is one switch and not two.
+        Bool IsPipelineStatisticsQueryTarget(GLenum target) {
+            switch (target) {
+            case GL_VERTICES_SUBMITTED:
+            case GL_PRIMITIVES_SUBMITTED:
+            case GL_VERTEX_SHADER_INVOCATIONS:
+            case GL_GEOMETRY_SHADER_INVOCATIONS:
+            case GL_GEOMETRY_SHADER_PRIMITIVES_EMITTED:
+            case GL_FRAGMENT_SHADER_INVOCATIONS:
+            case GL_COMPUTE_SHADER_INVOCATIONS:
+            case GL_CLIPPING_INPUT_PRIMITIVES:
+            case GL_CLIPPING_OUTPUT_PRIMITIVES:
+                return true;
+            case GL_TESS_CONTROL_SHADER_PATCHES:
+            case GL_TESS_EVALUATION_SHADER_INVOCATIONS:
+                return AdvertisesTessellationShaderExtension();
+            default:
+                return false;
+            }
+        }
 
         Bool TimerQueryDisabled() {
             return MG_Config::Features.DisableTimerQuery;
@@ -101,6 +165,7 @@ namespace MobileGL::MG_Impl::GLImpl {
         void ResetQueryObjectLocked(QueryObject* queryObject) {
             if (queryObject->backendHandle) {
                 if (const auto deleteBackendQuery = MG_Backend::gBackendFunctionsTable.GL.DeleteBackendQuery) {
+                    MGP_FILL(DeleteBackendQuery);
                     deleteBackendQuery(queryObject->backendHandle);
                 }
                 queryObject->backendHandle = nullptr;
@@ -115,6 +180,7 @@ namespace MobileGL::MG_Impl::GLImpl {
         void EndTimeElapsedQueryLocked(QueryObject* queryObject) {
             const auto endTimeElapsedQuery = MG_Backend::gBackendFunctionsTable.GL.EndTimeElapsedQuery;
             if (endTimeElapsedQuery && queryObject->backendHandle) {
+                MGP_FILL(EndTimeElapsedQuery);
                 endTimeElapsedQuery(queryObject->backendHandle);
             }
             queryObject->active = false;
@@ -122,6 +188,46 @@ namespace MobileGL::MG_Impl::GLImpl {
             g_activeTimeElapsedQueryId = 0;
         }
 
+        // The CPU accounting counter a transform feedback query target reads: what the capture
+        // buffers took for GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN, and everything the capture
+        // stage assembled - a paused span included - for GL_PRIMITIVES_GENERATED. One counter
+        // for both targets would report the clamped written count as the generated one.
+        Uint64 TransformFeedbackCounterForTarget(GLenum target) {
+            return target == GL_PRIMITIVES_GENERATED
+                       ? MG_State::pGLContext->GetTransformFeedbackGeneratedCounter()
+                       : MG_State::pGLContext->GetTransformFeedbackPrimitiveCounter();
+        }
+
+        // The span's CPU accounting delta. Saturating: a snapshot left above its counter (a
+        // context switch between Begin and End, a counter that never moved) would otherwise
+        // wrap to 2^64-1, which GetQueryObjectuiv hands the app as 4294967295.
+        Uint64 TransformFeedbackCpuResult(const QueryObject* queryObject) {
+            const Uint64 counter = TransformFeedbackCounterForTarget(queryObject->target);
+            return counter > queryObject->counterSnapshot ? counter - queryObject->counterSnapshot : 0;
+        }
+
+        // Whether this ended span's result should come from the CPU accounting rather than from
+        // the backend query it also ran. Three conditions, all necessary:
+        //   * the backend asked for it (DirectGLES, whose ES driver counter is the unreliable
+        //     one; DirectVulkan never sets the bit and so is untouched by any of this);
+        //   * the target is PRIMITIVES_WRITTEN. GL_PRIMITIVES_GENERATED counts primitives
+        //     whether or not a capture is active, and the accounting only ever sees capture
+        //     draws, so the backend's counter is the more complete answer there;
+        //   * the span was fully accounted: at least one capture draw reached the accounting
+        //     (the instanced, indirect and multi-draw entry points do not call it at all, so a
+        //     span made of those is invisible to it) and none of them amplified through a
+        //     geometry stage, which the CPU cannot model.
+        Bool PrefersCpuTransformFeedbackResult(const QueryObject* queryObject) {
+            if (!MG_Backend::gBackendFunctionsTable.GL.PrefersCpuXfbPrimitiveAccounting) return false;
+            if (queryObject->target != GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN) return false;
+            if (MG_State::pGLContext->GetTransformFeedbackGeometryCaptureDraws() !=
+                queryObject->geometryCaptureDrawSnapshot) {
+                return false;
+            }
+            return MG_State::pGLContext->GetTransformFeedbackAccountedCaptureDraws() !=
+                   queryObject->accountedCaptureDrawSnapshot;
+        }
+
         // Shared GetQueryObject* implementation. Returns false when an error
         // was recorded and no value should be written back. `outValueProduced`, when given,
         // additionally distinguishes "succeeded with a value" from "succeeded but the result is not
@@ -154,6 +260,7 @@ namespace MobileGL::MG_Impl::GLImpl {
                 }
                 Uint64 result = 0;
                 const auto getQueryResult64 = MG_Backend::gBackendFunctionsTable.GL.GetQueryResult64;
+                MGP_FILL(GetQueryResult64);
                 if (queryObject->backendHandle && getQueryResult64 &&
                     !getQueryResult64(queryObject->backendHandle, /*wait=*/false, &result)) {
                     // Not ready. The whole point of the no-wait form is that the caller's
@@ -168,6 +275,7 @@ namespace MobileGL::MG_Impl::GLImpl {
                 }
                 if (queryObject->backendHandle) {
                     if (const auto deleteBackendQuery = MG_Backend::gBackendFunctionsTable.GL.DeleteBackendQuery) {
+                        MGP_FILL(DeleteBackendQuery);
                         deleteBackendQuery(queryObject->backendHandle);
                     }
                     queryObject->backendHandle = nullptr;
@@ -183,6 +291,7 @@ namespace MobileGL::MG_Impl::GLImpl {
                     return true;
                 }
                 const auto isQueryResultAvailable = MG_Backend::gBackendFunctionsTable.GL.IsQueryResultAvailable;
+                MGP_FILL(IsQueryResultAvailable);
                 outValue = (!isQueryResultAvailable || isQueryResultAvailable(queryObject->backendHandle)) ? 1 : 0;
                 return true;
             }
@@ -194,6 +303,7 @@ namespace MobileGL::MG_Impl::GLImpl {
                 Uint64 result = 0;
                 if (queryObject->backendHandle) {
                     const auto getQueryResult64 = MG_Backend::gBackendFunctionsTable.GL.GetQueryResult64;
+                    MGP_FILL(GetQueryResult64);
                     if (getQueryResult64 &&
                         !getQueryResult64(queryObject->backendHandle, /*wait=*/true, &result)) {
                         // The backend could not produce the result YET (e.g. a
@@ -214,6 +324,7 @@ namespace MobileGL::MG_Impl::GLImpl {
                     // query degrades to a zero result); the backend handle is
                     // consumed and the value cached for later reads.
                     if (const auto deleteBackendQuery = MG_Backend::gBackendFunctionsTable.GL.DeleteBackendQuery) {
+                        MGP_FILL(DeleteBackendQuery);
                         deleteBackendQuery(queryObject->backendHandle);
                     }
                     queryObject->backendHandle = nullptr;
@@ -319,10 +430,14 @@ namespace MobileGL::MG_Impl::GLImpl {
                     queryObject->target == GL_ANY_SAMPLES_PASSED_CONSERVATIVE) {
                     if (const auto endOcclusionQuery = MG_Backend::gBackendFunctionsTable.GL.EndOcclusionQuery;
                         endOcclusionQuery && queryObject->backendHandle) {
+                        MGP_FILL(EndOcclusionQuery);
                         endOcclusionQuery(queryObject->backendHandle);
                     }
                     queryObject->active = false;
                     g_activeSamplesPassedQueryId = 0;
+                } else if (IsPipelineStatisticsQueryTarget(queryObject->target)) {
+                    queryObject->active = false;
+                    g_activePipelineStatisticsQueryIds[queryObject->target] = 0;
                 } else if (queryObject->target == GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN ||
                            queryObject->target == GL_PRIMITIVES_GENERATED) {
                     queryObject->active = false;
@@ -335,6 +450,7 @@ namespace MobileGL::MG_Impl::GLImpl {
             }
             if (queryObject->backendHandle) {
                 if (const auto deleteBackendQuery = MG_Backend::gBackendFunctionsTable.GL.DeleteBackendQuery) {
+                    MGP_FILL(DeleteBackendQuery);
                     deleteBackendQuery(queryObject->backendHandle);
                 }
                 queryObject->backendHandle = nullptr;
@@ -363,7 +479,9 @@ namespace MobileGL::MG_Impl::GLImpl {
             (target == GL_SAMPLES_PASSED || target == GL_ANY_SAMPLES_PASSED ||
              target == GL_ANY_SAMPLES_PASSED_CONSERVATIVE) &&
             MG_Backend::gBackendFunctionsTable.GL.BeginOcclusionQuery != nullptr;
-        if (target != GL_TIME_ELAPSED && !isTransformFeedbackQuery && !isOcclusionQuery) {
+        const Bool isPipelineStatisticsQuery = IsPipelineStatisticsQueryTarget(target);
+        if (target != GL_TIME_ELAPSED && !isTransformFeedbackQuery && !isOcclusionQuery &&
+            !isPipelineStatisticsQuery) {
             // GL_TIMESTAMP is not a valid BeginQuery target; the occlusion targets
             // need backend support.
             RecordQueryError(ErrorCode::InvalidEnum, __FUNCTION__, "Query target is not supported.");
@@ -379,10 +497,12 @@ namespace MobileGL::MG_Impl::GLImpl {
             RecordQueryError(ErrorCode::InvalidOperation, __FUNCTION__, "Query object does not exist.");
             return;
         }
-        GLuint& activeQueryId = isTransformFeedbackQuery
-            ? (target == GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN ? g_activePrimitivesWrittenQueryId
-                                                                  : g_activePrimitivesGeneratedQueryId)
-            : (isOcclusionQuery ? g_activeSamplesPassedQueryId : g_activeTimeElapsedQueryId);
+        GLuint& activeQueryId = isPipelineStatisticsQuery
+            ? g_activePipelineStatisticsQueryIds[target]
+            : (isTransformFeedbackQuery
+                   ? (target == GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN ? g_activePrimitivesWrittenQueryId
+                                                                         : g_activePrimitivesGeneratedQueryId)
+                   : (isOcclusionQuery ? g_activeSamplesPassedQueryId : g_activeTimeElapsedQueryId));
         if (activeQueryId != 0) {
             RecordQueryError(ErrorCode::InvalidOperation, __FUNCTION__,
                              "A query is already active on this target.");
@@ -401,17 +521,28 @@ namespace MobileGL::MG_Impl::GLImpl {
         ResetQueryObjectLocked(queryObject); // discard any previous result
         queryObject->target = target;
         queryObject->active = true;
-        if (isTransformFeedbackQuery) {
+        if (isPipelineStatisticsQuery) {
+            // Nothing to start: the counter is uninstrumented and GL_QUERY_COUNTER_BITS says so.
+            // The object still becomes a real, target-latched query so every other rule about it
+            // (re-use with another target, double-begin, EndQuery pairing) keeps holding.
+        } else if (isTransformFeedbackQuery) {
             // Prefer real GPU transform-feedback queries (exact with geometry shaders);
             // the CPU accounting delta stays as the fallback when the backend lacks them.
             const auto beginXfbPrimitivesQuery = MG_Backend::gBackendFunctionsTable.GL.BeginXfbPrimitivesQuery;
+            MGP_FILL(BeginXfbPrimitivesQuery);
             queryObject->backendHandle =
                 beginXfbPrimitivesQuery ? beginXfbPrimitivesQuery(target == GL_PRIMITIVES_GENERATED) : nullptr;
-            queryObject->counterSnapshot = MG_State::pGLContext->GetTransformFeedbackPrimitiveCounter();
+            queryObject->counterSnapshot = TransformFeedbackCounterForTarget(target);
+            queryObject->accountedCaptureDrawSnapshot =
+                MG_State::pGLContext->GetTransformFeedbackAccountedCaptureDraws();
+            queryObject->geometryCaptureDrawSnapshot =
+                MG_State::pGLContext->GetTransformFeedbackGeometryCaptureDraws();
         } else if (isOcclusionQuery) {
+            MGP_FILL(BeginOcclusionQuery);
             queryObject->backendHandle = MG_Backend::gBackendFunctionsTable.GL.BeginOcclusionQuery();
         } else {
             const auto beginTimeElapsedQuery = MG_Backend::gBackendFunctionsTable.GL.BeginTimeElapsedQuery;
+            MGP_FILL(BeginTimeElapsedQuery);
             queryObject->backendHandle =
                 (!TimerQueryDisabled() && beginTimeElapsedQuery) ? beginTimeElapsedQuery() : nullptr;
         }
@@ -425,15 +556,19 @@ namespace MobileGL::MG_Impl::GLImpl {
             (target == GL_SAMPLES_PASSED || target == GL_ANY_SAMPLES_PASSED ||
              target == GL_ANY_SAMPLES_PASSED_CONSERVATIVE) &&
             MG_Backend::gBackendFunctionsTable.GL.BeginOcclusionQuery != nullptr;
-        if (target != GL_TIME_ELAPSED && !isTransformFeedbackQuery && !isOcclusionQuery) {
+        const Bool isPipelineStatisticsQuery = IsPipelineStatisticsQueryTarget(target);
+        if (target != GL_TIME_ELAPSED && !isTransformFeedbackQuery && !isOcclusionQuery &&
+            !isPipelineStatisticsQuery) {
             RecordQueryError(ErrorCode::InvalidEnum, __FUNCTION__, "Query target is not supported.");
             return;
         }
         const std::lock_guard<std::mutex> lock(g_queryObjectsMutex);
-        GLuint& activeQueryId = isTransformFeedbackQuery
-            ? (target == GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN ? g_activePrimitivesWrittenQueryId
-                                                                  : g_activePrimitivesGeneratedQueryId)
-            : (isOcclusionQuery ? g_activeSamplesPassedQueryId : g_activeTimeElapsedQueryId);
+        GLuint& activeQueryId = isPipelineStatisticsQuery
+            ? g_activePipelineStatisticsQueryIds[target]
+            : (isTransformFeedbackQuery
+                   ? (target == GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN ? g_activePrimitivesWrittenQueryId
+                                                                         : g_activePrimitivesGeneratedQueryId)
+                   : (isOcclusionQuery ? g_activeSamplesPassedQueryId : g_activeTimeElapsedQueryId));
         if (activeQueryId == 0) {
             RecordQueryError(ErrorCode::InvalidOperation, __FUNCTION__, "No query is active on this target.");
             return;
@@ -443,17 +578,39 @@ namespace MobileGL::MG_Impl::GLImpl {
             activeQueryId = 0; // should not happen; keep state consistent
             return;
         }
+        if (isPipelineStatisticsQuery) {
+            // The result is a definite zero rather than an unread backend handle, so a later
+            // GetQueryObject* answers immediately and never waits on something that was never
+            // started. GL_QUERY_COUNTER_BITS = 0 is what marks that zero indeterminate.
+            queryObject->cachedResult = 0;
+            queryObject->resultCached = true;
+            queryObject->active = false;
+            queryObject->ended = true;
+            activeQueryId = 0;
+            return;
+        }
         if (isTransformFeedbackQuery) {
             if (queryObject->backendHandle) {
                 if (const auto endXfbPrimitivesQuery = MG_Backend::gBackendFunctionsTable.GL.EndXfbPrimitivesQuery) {
+                    MGP_FILL(EndXfbPrimitivesQuery);
                     endXfbPrimitivesQuery(queryObject->backendHandle);
                 }
-                // Result comes from the GPU query at read time.
-            } else {
-                queryObject->cachedResult =
-                    MG_State::pGLContext->GetTransformFeedbackPrimitiveCounter() - queryObject->counterSnapshot;
+            }
+            // A backend query that is not going to be read is released here, not left to be
+            // collected later: the span is over, the driver object has nothing left to say.
+            // Ending it first is what makes that legal.
+            if (!queryObject->backendHandle || PrefersCpuTransformFeedbackResult(queryObject)) {
+                if (queryObject->backendHandle) {
+                    if (const auto deleteBackendQuery = MG_Backend::gBackendFunctionsTable.GL.DeleteBackendQuery) {
+                        MGP_FILL(DeleteBackendQuery);
+                        deleteBackendQuery(queryObject->backendHandle);
+                    }
+                    queryObject->backendHandle = nullptr;
+                }
+                queryObject->cachedResult = TransformFeedbackCpuResult(queryObject);
                 queryObject->resultCached = true;
             }
+            // Otherwise the result comes from the GPU query at read time.
             queryObject->active = false;
             queryObject->ended = true;
             activeQueryId = 0;
@@ -462,6 +619,7 @@ namespace MobileGL::MG_Impl::GLImpl {
         if (isOcclusionQuery) {
             if (const auto endOcclusionQuery = MG_Backend::gBackendFunctionsTable.GL.EndOcclusionQuery;
                 endOcclusionQuery && queryObject->backendHandle) {
+                MGP_FILL(EndOcclusionQuery);
                 endOcclusionQuery(queryObject->backendHandle);
             }
             queryObject->active = false;
@@ -500,11 +658,81 @@ namespace MobileGL::MG_Impl::GLImpl {
         ResetQueryObjectLocked(queryObject); // discard any previous result
         queryObject->target = target;
         const auto queryCounterTimestamp = MG_Backend::gBackendFunctionsTable.GL.QueryCounterTimestamp;
+        MGP_FILL(QueryCounterTimestamp);
         queryObject->backendHandle =
             (!TimerQueryDisabled() && queryCounterTimestamp) ? queryCounterTimestamp() : nullptr;
         queryObject->ended = true;
     }
 
+    void BeginConditionalRender(GLuint id, GLenum mode) {
+        // GL 4.6 core 10.9's eight modes. The _INVERTED half flips the sense of the predicate;
+        // the BY_REGION half only narrows WHERE an implementation is permitted to discard, so
+        // treating it as its whole-framebuffer sibling is what an implementation without region
+        // granularity does. The _NO_WAIT half is a permission to render rather than stall, not an
+        // obligation - see the resolve below.
+        Bool inverted = false;
+        switch (mode) {
+        case GL_QUERY_WAIT:
+        case GL_QUERY_NO_WAIT:
+        case GL_QUERY_BY_REGION_WAIT:
+        case GL_QUERY_BY_REGION_NO_WAIT:
+            inverted = false;
+            break;
+        case GL_QUERY_WAIT_INVERTED:
+        case GL_QUERY_NO_WAIT_INVERTED:
+        case GL_QUERY_BY_REGION_WAIT_INVERTED:
+        case GL_QUERY_BY_REGION_NO_WAIT_INVERTED:
+            inverted = true;
+            break;
+        default:
+            RecordQueryError(ErrorCode::InvalidEnum, __FUNCTION__, "mode is not a conditional render mode.");
+            return;
+        }
+
+        if (MG_State::pGLContext->IsConditionalRenderActive()) {
+            RecordQueryError(ErrorCode::InvalidOperation, __FUNCTION__, "Conditional rendering is already active.");
+            return;
+        }
+
+        {
+            const std::lock_guard<std::mutex> lock(g_queryObjectsMutex);
+            const auto* queryObject = FindQueryObjectLocked(id);
+            // A generated NAME is not yet a query object; it becomes one at its first use with a
+            // target (the same rule glIsQuery answers by).
+            if (!queryObject || (!queryObject->created && queryObject->target == 0)) {
+                RecordQueryError(ErrorCode::InvalidValue, __FUNCTION__, "id is not the name of a query object.");
+                return;
+            }
+            if (queryObject->active) {
+                RecordQueryError(ErrorCode::InvalidOperation, __FUNCTION__, "The query object is still active.");
+                return;
+            }
+            if (queryObject->target != GL_SAMPLES_PASSED && queryObject->target != GL_ANY_SAMPLES_PASSED &&
+                queryObject->target != GL_ANY_SAMPLES_PASSED_CONSERVATIVE) {
+                RecordQueryError(ErrorCode::InvalidOperation, __FUNCTION__,
+                                 "Conditional rendering requires an occlusion query object.");
+                return;
+            }
+        }
+
+        // Resolved ONCE, here, and by WAITING even for the _NO_WAIT modes: the spec lets those
+        // render instead of stalling, so always waiting is conforming and is the only choice that
+        // gives the whole block one deterministic verdict. Reading it per command instead would
+        // let a result that lands mid-block change the answer half way through.
+        Uint64 samplesPassed = 0;
+        if (!GetQueryObjectValue(id, GL_QUERY_RESULT, __FUNCTION__, samplesPassed)) return;
+        const Bool passed = samplesPassed != 0;
+        MG_State::pGLContext->BeginConditionalRender(id, mode, inverted ? passed : !passed);
+    }
+
+    void EndConditionalRender() {
+        if (!MG_State::pGLContext->IsConditionalRenderActive()) {
+            RecordQueryError(ErrorCode::InvalidOperation, __FUNCTION__, "Conditional rendering is not active.");
+            return;
+        }
+        MG_State::pGLContext->EndConditionalRender();
+    }
+
     void GetQueryiv(GLenum target, GLenum pname, GLint* params) {
         if (!params) {
             return;
@@ -528,7 +756,12 @@ namespace MobileGL::MG_Impl::GLImpl {
                 *params = static_cast<GLint>(g_activePrimitivesGeneratedQueryId);
                 break;
             default:
-                *params = 0;
+                if (IsPipelineStatisticsQueryTarget(target)) {
+                    const auto it = g_activePipelineStatisticsQueryIds.find(target);
+                    *params = it != g_activePipelineStatisticsQueryIds.end() ? static_cast<GLint>(it->second) : 0;
+                } else {
+                    *params = 0;
+                }
                 break;
             }
             return;
@@ -539,6 +772,14 @@ namespace MobileGL::MG_Impl::GLImpl {
             // entry points / timestamp valid bits at call time, not at table
             // init), and the MOBILEGL_DISABLE_TIMERQUERY kill switch always
             // wins.
+            if (IsPipelineStatisticsQueryTarget(target)) {
+                // Zero: GL 4.6 core 4.2.1's way of saying the counter is not implemented and its
+                // results are indeterminate. The conformance suite reads exactly this and skips
+                // the functional half of each such target, which is the outcome an uninstrumented
+                // counter should produce.
+                *params = 0;
+                return;
+            }
             if (target == GL_SAMPLES_PASSED || target == GL_ANY_SAMPLES_PASSED ||
                 target == GL_ANY_SAMPLES_PASSED_CONSERVATIVE) {
                 const Bool occlusionSupported = MG_Backend::gBackendFunctionsTable.GL.BeginOcclusionQuery != nullptr;
@@ -547,6 +788,7 @@ namespace MobileGL::MG_Impl::GLImpl {
             }
             const Bool timerTarget = target == GL_TIME_ELAPSED || target == GL_TIMESTAMP;
             const auto isTimerQuerySupported = MG_Backend::gBackendFunctionsTable.GL.IsTimerQuerySupported;
+            MGP_FILL(IsTimerQuerySupported);
             const Bool supported =
                 timerTarget && !TimerQueryDisabled() && isTimerQuerySupported && isTimerQuerySupported();
             *params = supported ? 64 : 0;
@@ -612,14 +854,24 @@ namespace MobileGL::MG_Impl::GLImpl {
     }
 
     namespace {
+        Bool IsPerVertexStreamQueryTarget(GLenum target) {
+            return target == GL_PRIMITIVES_GENERATED || target == GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN;
+        }
+
         // The indexed query entry points differ from the plain ones only in the vertex
         // stream they address (GL 4.6 core 4.2.1): index must be below GL_MAX_VERTEX_STREAMS
-        // for the two transform feedback targets and zero for every other target. With a
-        // single vertex stream both bounds are 1, so a valid call is always index 0 and
-        // forwards to the unindexed implementation.
+        // for the two transform feedback targets and zero for every other target. MobileGL
+        // implements ONE vertex stream, so both bounds are 1 and a valid call is always index 0 -
+        // which is what makes the three forwards below equivalent to the unindexed entry points.
+        //
+        // THAT EQUIVALENCE IS THE WHOLE JUSTIFICATION, and it is read out of the getter rather
+        // than assumed: the moment GL_MAX_VERTEX_STREAMS answers more than one, index 1..3 starts
+        // reaching EndQueryIndexed and GetQueryIndexediv, which resolve the active query from
+        // per-TARGET globals and would end - or report - a query begun on a different stream.
+        // Raising that limit therefore means giving each active query a stream index and
+        // comparing it here, not just changing the number.
         Bool ValidateQueryStreamIndex(const char* function, GLenum target, GLuint index) {
-            const Bool perStreamTarget =
-                target == GL_PRIMITIVES_GENERATED || target == GL_TRANSFORM_FEEDBACK_PRIMITIVES_WRITTEN;
+            const Bool perStreamTarget = IsPerVertexStreamQueryTarget(target);
             GLint maxVertexStreams = 1;
             if (perStreamTarget) {
                 GetIntegerv(GL_MAX_VERTEX_STREAMS, &maxVertexStreams);
@@ -632,6 +884,7 @@ namespace MobileGL::MG_Impl::GLImpl {
                                              : "index must be zero for this query target.");
             return false;
         }
+
     } // namespace
 
     void BeginQueryIndexed(GLenum target, GLuint index, GLuint id) {
@@ -648,4 +901,40 @@ namespace MobileGL::MG_Impl::GLImpl {
         if (!ValidateQueryStreamIndex(__FUNCTION__, target, index)) return;
         GetQueryiv(target, pname, params);
     }
+
+    void DestroyAllQueryObjects() {
+        // Detach the registry under the lock, release outside it - same discipline
+        // (and the same accepted teardown race) as DestroyAllSyncObjects. Without
+        // this drain, every query the app left undeleted survived full library
+        // teardown in the process-global registry: the objects and their backend
+        // wrappers leaked across Destroy/Initialize cycles, stale ids kept
+        // answering IsQuery == GL_TRUE in the re-initialized library, and a later
+        // glDeleteQueries could hand the OLD backend's handle to a DIFFERENT
+        // backend's DeleteBackendQuery, which casts it to the wrong wrapper type.
+        UnorderedMap<GLuint, QueryObject*> orphans;
+        {
+            const std::lock_guard<std::mutex> lock(g_queryObjectsMutex);
+            orphans.swap(g_liveQueryObjects);
+            g_activeTimeElapsedQueryId = 0;
+            g_activePrimitivesWrittenQueryId = 0;
+            g_activePrimitivesGeneratedQueryId = 0;
+            g_activeSamplesPassedQueryId = 0;
+        }
+        if (orphans.empty()) {
+            return;
+        }
+        // Backend handles must be released by the backend that created them, so
+        // this runs while the function table is still populated. Both backends'
+        // DeleteBackendQuery are generation-guarded, so a handle whose renderer
+        // or ES context is already gone frees only the wrapper.
+        const auto deleteBackendQuery = MG_Backend::gBackendFunctionsTable.GL.DeleteBackendQuery;
+        for (const auto& [_, queryObject] : orphans) {
+            if (deleteBackendQuery && queryObject->backendHandle) {
+                MGP_FILL(DeleteBackendQuery);
+                deleteBackendQuery(queryObject->backendHandle);
+            }
+            delete queryObject;
+        }
+        MGLOG_D("DestroyAllQueryObjects: reclaimed %zu query object(s) the app left undeleted", orphans.size());
+    }
 } // namespace MobileGL::MG_Impl::GLImpl

MobileGL/MG_Impl/GLImpl/Query/GL_Query.h

@@ -29,4 +29,18 @@ namespace MobileGL::MG_Impl::GLImpl {
     void GetQueryBufferObjecti64v(GLuint id, GLuint buffer, GLenum pname, GLintptr offset);
     void GetQueryBufferObjectui64v(GLuint id, GLuint buffer, GLenum pname, GLintptr offset);
     void QueryCounter(GLuint id, GLenum target);
+    // Conditional rendering (GL 4.6 core 10.9). Implemented here rather than beside the drawing
+    // entry points because the predicate is a QUERY OBJECT's result, and the object registry -
+    // with the lock that guards it - lives in this file.
+    void BeginConditionalRender(GLuint id, GLenum mode);
+    void EndConditionalRender();
+    // Destroys every still-registered query object exactly as DeleteQueries would.
+    // GL requires queries to die with their context; called only from full library
+    // teardown (DestroyImpl), where no context survives on any thread, so the
+    // process-global registry can be drained wholesale. Must run while the backend
+    // function table is still populated: each backend handle has to be released by
+    // the backend that created it, never by a later re-initialized one (whose
+    // DeleteBackendQuery would cast the wrapper to the wrong backend's type).
+    // Same contract as DestroyAllSyncObjects.
+    void DestroyAllQueryObjects();
 } // namespace MobileGL::MG_Impl::GLImpl

MobileGL/MG_Impl/GLImpl/RenderState/GL_RenderState.cpp

@@ -328,10 +328,50 @@ namespace MobileGL::MG_Impl::GLImpl {
         MG_State::pGLContext->SetSampleCoverage(std::clamp(static_cast<Float>(value), 0.0f, 1.0f), invert == GL_TRUE);
     }
 
+    // ARB_sample_shading / GL 4.6 core 14.3.1: "value is clamped to [0, 1] when specified", so
+    // there is no error to raise - a caller that asks for 2.0 gets 1.0 and GL_MIN_SAMPLE_SHADING_-
+    // VALUE reads back 1.0. Was a logging no-op while ARB_sample_shading was advertised, which
+    // let an application enable GL_SAMPLE_SHADING and then quietly get the driver's default rate.
+    void MinSampleShading_State(GLfloat value) {
+        MG_State::pGLContext->SetMinSampleShadingValue(std::clamp(static_cast<Float>(value), 0.0f, 1.0f));
+    }
+
     void PolygonOffset_State(GLfloat factor, GLfloat units) {
         MG_State::pGLContext->SetPolygonOffset(static_cast<Float>(factor), static_cast<Float>(units));
     }
 
+    void PolygonOffsetClamp_State(GLfloat factor, GLfloat units, GLfloat clamp) {
+        // GL 4.6 core 14.6.5 / GL_EXT_polygon_offset_clamp. No error cases: any three floats are
+        // legal, and clamp = 0 is exactly glPolygonOffset. Whether the backend can APPLY the clamp
+        // is a separate question (see the DirectGLES/DirectVulkan forwarding); the state is
+        // recorded either way, because GL_POLYGON_OFFSET_CLAMP has to read back what was written.
+        MG_State::pGLContext->SetPolygonOffsetClamped(static_cast<Float>(factor), static_cast<Float>(units),
+                                                      static_cast<Float>(clamp));
+    }
+
+    void ClipControl_State(GLenum origin, GLenum depth) {
+        // GL 4.5 core 13.5: both arguments are strict enums, and either being wrong is
+        // GL_INVALID_ENUM with the state left untouched.
+        if (origin != GL_LOWER_LEFT && origin != GL_UPPER_LEFT) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidEnum,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "glClipControl origin must be GL_LOWER_LEFT or GL_UPPER_LEFT; got " +
+                                                 MG_Util::ConvertGLEnumToString(origin) + "."));
+            return;
+        }
+        if (depth != GL_NEGATIVE_ONE_TO_ONE && depth != GL_ZERO_TO_ONE) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidEnum,
+                MakeUnique<GenericErrorInfo>(
+                    "MG_Impl/GLImpl", __func__,
+                    "glClipControl depth must be GL_NEGATIVE_ONE_TO_ONE or GL_ZERO_TO_ONE; got " +
+                        MG_Util::ConvertGLEnumToString(depth) + "."));
+            return;
+        }
+        MG_State::pGLContext->SetClipControl(origin, depth);
+    }
+
     void PolygonMode_State(GLenum face, GLenum mode) {
         // GL 3.3 core: separate front/back polygon modes were removed in 3.1, so the only legal
         // face is GL_FRONT_AND_BACK. GL_FRONT / GL_BACK must be rejected (some desktop drivers
@@ -1013,10 +1053,22 @@ namespace MobileGL::MG_Impl::GLImpl {
         SampleCoverage_State(value, invert);
     }
 
+    void MinSampleShading(GLfloat value) {
+        MinSampleShading_State(value);
+    }
+
     void PolygonOffset(GLfloat factor, GLfloat units) {
         PolygonOffset_State(factor, units);
     }
 
+    void PolygonOffsetClamp(GLfloat factor, GLfloat units, GLfloat clamp) {
+        PolygonOffsetClamp_State(factor, units, clamp);
+    }
+
+    void ClipControl(GLenum origin, GLenum depth) {
+        ClipControl_State(origin, depth);
+    }
+
     void PolygonMode(GLenum face, GLenum mode) {
         PolygonMode_State(face, mode);
     }

MobileGL/MG_Impl/GLImpl/RenderState/GL_RenderState.h

@@ -38,7 +38,10 @@ namespace MobileGL::MG_Impl::GLImpl {
     void StencilFunc(GLenum func, GLint ref, GLuint mask);
     void Scissor(GLint x, GLint y, GLsizei width, GLsizei height);
     void SampleCoverage(GLfloat value, GLboolean invert);
+    void MinSampleShading(GLfloat value);
     void PolygonOffset(GLfloat factor, GLfloat units);
+    void PolygonOffsetClamp(GLfloat factor, GLfloat units, GLfloat clamp);
+    void ClipControl(GLenum origin, GLenum depth);
     void PolygonMode(GLenum face, GLenum mode);
     void PointSize(GLfloat size);
     void PointParameterf(GLenum pname, GLfloat param);

MobileGL/MG_Impl/GLImpl/Sampler/GL_Sampler.cpp

@@ -13,6 +13,7 @@
 #include <MG_State/GLState/Core.h>
 #include <MG_Util/Converters/GLToMG/TextureEnumConverter.h>
 #include <MG_Util/Converters/MGToGL/TextureEnumConverter.h>
+#include <MG_Util/Math/FixedPointConversion.h>
 
 namespace MobileGL::MG_Impl::GLImpl {
     namespace {
@@ -22,6 +23,50 @@ namespace MobileGL::MG_Impl::GLImpl {
             return static_cast<Float>(*(const GLint*)param);
         }
 
+        // GL_TEXTURE_BORDER_COLOR is the only sampler parameter with more than one component, and it
+        // is also the only one whose meaning depends on WHICH entry point wrote it. Everything else
+        // reads exactly one component and does not care.
+        Bool IsVectorOnlySamplerPname(GLenum pname) {
+            return pname == GL_TEXTURE_BORDER_COLOR;
+        }
+
+        // A state query returns the value CONVERTED to the type the caller asked for (GL 4.6 core
+        // 2.2.2 / 6.1), never the other type's bits. These two are the sampler side of the numeric
+        // casts GetTexParameterfv_State/GetTexParameteriv_State already do on the texture side; the
+        // sampler path funnels all three spellings through one void* function, which is precisely how
+        // it came to write a fixed type regardless of the caller.
+        //
+        // Truncation rather than rounding for the float -> integer direction, matching the texture
+        // twin (GetTexParameteriv_State's static_cast<GLint> on MIN_LOD/MAX_LOD/LOD_BIAS): the two
+        // spellings of the same state disagreeing is the bug being fixed here, and a texture and a
+        // sampler queried the same way must answer the same number.
+        void StoreSamplerScalar(void* params, Bool isFloat, Bool isUnsignedInteger, Float value) {
+            if (isFloat) {
+                *(GLfloat*)params = value;
+                return;
+            }
+            // Via GLint in both integer spellings: a direct float -> GLuint cast of a negative value
+            // (GL_TEXTURE_MIN_LOD defaults to -1000) is undefined behaviour, while the two-step
+            // conversion is the well-defined modular one, and it is what the texture-side
+            // GetTexParameterIuiv fallback does.
+            const GLint asInt = static_cast<GLint>(value);
+            if (isUnsignedInteger) {
+                *(GLuint*)params = static_cast<GLuint>(asInt);
+            } else {
+                *(GLint*)params = asInt;
+            }
+        }
+
+        void StoreSamplerEnum(void* params, Bool isFloat, Bool isUnsignedInteger, GLenum value) {
+            if (isFloat) {
+                *(GLfloat*)params = static_cast<GLfloat>(value);
+            } else if (isUnsignedInteger) {
+                *(GLuint*)params = value;
+            } else {
+                *(GLint*)params = static_cast<GLint>(value);
+            }
+        }
+
         Bool ValidateSamplerParameterValue(GLenum pname, const void* param, Bool isFloat, Bool isUnsignedInteger) {
             if (param == nullptr) return false;
 
@@ -56,8 +101,15 @@ namespace MobileGL::MG_Impl::GLImpl {
         }
     } // namespace
 
+    // `isIntegerCommand` distinguishes the "I" spellings (glSamplerParameterIiv / Iuiv) from the
+    // plain ones. It only matters for GL_TEXTURE_BORDER_COLOR, and there it decides everything:
+    // GL 4.6 core 8.10 says the I forms store the components unmodified with an integer internal
+    // type, while glSamplerParameteriv converts them to floating point with equation 2.2. Routing
+    // both to the same setter - which is what this file used to do - meant glSamplerParameteriv
+    // stored raw integers (so a border of 255 became float 255.0 instead of the spec's ~1.19e-7)
+    // and glSamplerParameterIiv lost the fact that it was ever an integer at all.
     void SetSamplerParam_State(GLuint sampler, GLenum pname, const void* param, bool isFloat,
-                               bool isUnsignedInteger) {
+                               bool isUnsignedInteger, bool isIntegerCommand) {
         if (param == nullptr) return;
         if (!SamplerImpl::ValidateSamplerName(sampler)) return;
 
@@ -112,6 +164,13 @@ namespace MobileGL::MG_Impl::GLImpl {
             if (isFloat) {
                 const auto* values = (const GLfloat*)param;
                 samplerObj->SetBorderColor(FloatVec4(values[0], values[1], values[2], values[3]));
+            } else if (!isIntegerCommand) {
+                // glSamplerParameteriv: GL 4.6 core equation 2.2 into the FLOAT border colour.
+                const auto* values = (const GLint*)param;
+                samplerObj->SetBorderColor(FloatVec4(MG_Util::SignedNormalizedInt32ToFloat(values[0]),
+                                                     MG_Util::SignedNormalizedInt32ToFloat(values[1]),
+                                                     MG_Util::SignedNormalizedInt32ToFloat(values[2]),
+                                                     MG_Util::SignedNormalizedInt32ToFloat(values[3])));
             } else if (isUnsignedInteger) {
                 const auto* values = (const GLuint*)param;
                 samplerObj->SetBorderColorUI(UintVec4(values[0], values[1], values[2], values[3]));
@@ -128,7 +187,7 @@ namespace MobileGL::MG_Impl::GLImpl {
     }
 
     void GetSamplerParam_State(GLuint sampler, GLenum pname, void* params, bool isFloat,
-                               bool isUnsignedInteger) {
+                               bool isUnsignedInteger, bool isIntegerCommand) {
         if (params == nullptr) return;
         if (!SamplerImpl::ValidateSamplerName(sampler)) return;
 
@@ -141,47 +200,56 @@ namespace MobileGL::MG_Impl::GLImpl {
         if (!SamplerImpl::ValidateSamplerObject(sampler)) return;
 
         using namespace MG_Util;
+        // Every scalar pname goes through StoreSamplerScalar/StoreSamplerEnum so the CALLER'S form
+        // decides the destination type. Writing a fixed type regardless - which is what these case
+        // labels used to do - hands back the other type's bit pattern rather than a converted value:
+        // glGetSamplerParameterfv(GL_TEXTURE_WRAP_S) deposited the integer 10497 into a GLfloat and
+        // the caller read 1.47e-41, and glGetSamplerParameteriv(GL_TEXTURE_MIN_LOD) deposited the
+        // IEEE bits of -1000.0f and the caller read -998637568. Sixteen (pname, entry-point) pairs
+        // were broken this way; only MAX_ANISOTROPY_EXT and BORDER_COLOR branched correctly, which is
+        // how the same bug class was already found and fixed once for a single pname.
         switch (pname) {
         case GL_TEXTURE_WRAP_S:
-            *(GLuint*)params = MG_Util::ConvertSamplerWrapModeToGLEnum(samplerObj->GetWrapS());
+            StoreSamplerEnum(params, isFloat, isUnsignedInteger,
+                             MG_Util::ConvertSamplerWrapModeToGLEnum(samplerObj->GetWrapS()));
             break;
         case GL_TEXTURE_WRAP_T:
-            *(GLuint*)params = MG_Util::ConvertSamplerWrapModeToGLEnum(samplerObj->GetWrapT());
+            StoreSamplerEnum(params, isFloat, isUnsignedInteger,
+                             MG_Util::ConvertSamplerWrapModeToGLEnum(samplerObj->GetWrapT()));
             break;
         case GL_TEXTURE_WRAP_R:
-            *(GLuint*)params = MG_Util::ConvertSamplerWrapModeToGLEnum(samplerObj->GetWrapR());
+            StoreSamplerEnum(params, isFloat, isUnsignedInteger,
+                             MG_Util::ConvertSamplerWrapModeToGLEnum(samplerObj->GetWrapR()));
             break;
         case GL_TEXTURE_MIN_FILTER:
-            *(GLuint*)params =
-                MG_Util::ConvertSamplerFilterModeToGLEnum(samplerObj->GetMinFilter(), samplerObj->GetMipmapMode());
+            StoreSamplerEnum(params, isFloat, isUnsignedInteger,
+                             MG_Util::ConvertSamplerFilterModeToGLEnum(samplerObj->GetMinFilter(),
+                                                                       samplerObj->GetMipmapMode()));
             break;
         case GL_TEXTURE_MAG_FILTER:
-            *(GLuint*)params =
-                MG_Util::ConvertSamplerFilterModeToGLEnum(samplerObj->GetMagFilter(), SamplerMipmapMode::None);
+            StoreSamplerEnum(params, isFloat, isUnsignedInteger,
+                             MG_Util::ConvertSamplerFilterModeToGLEnum(samplerObj->GetMagFilter(),
+                                                                       SamplerMipmapMode::None));
             break;
         case GL_TEXTURE_MIN_LOD:
-            *(GLfloat*)params = samplerObj->GetMinLod();
+            StoreSamplerScalar(params, isFloat, isUnsignedInteger, samplerObj->GetMinLod());
             break;
         case GL_TEXTURE_MAX_LOD:
-            *(GLfloat*)params = samplerObj->GetMaxLod();
+            StoreSamplerScalar(params, isFloat, isUnsignedInteger, samplerObj->GetMaxLod());
             break;
         case GL_TEXTURE_LOD_BIAS:
-            *(GLfloat*)params = samplerObj->GetLodBias();
+            StoreSamplerScalar(params, isFloat, isUnsignedInteger, samplerObj->GetLodBias());
             break;
         case GL_TEXTURE_MAX_ANISOTROPY_EXT:
-            if (isFloat) {
-                *(GLfloat*)params = samplerObj->GetMaxAnisotropy();
-            } else if (isUnsignedInteger) {
-                *(GLuint*)params = static_cast<GLuint>(samplerObj->GetMaxAnisotropy());
-            } else {
-                *(GLint*)params = static_cast<GLint>(samplerObj->GetMaxAnisotropy());
-            }
+            StoreSamplerScalar(params, isFloat, isUnsignedInteger, samplerObj->GetMaxAnisotropy());
             break;
         case GL_TEXTURE_COMPARE_MODE:
-            *(GLuint*)params = MG_Util::ConvertSamplerCompareModeToGLEnum(samplerObj->GetCompareMode());
+            StoreSamplerEnum(params, isFloat, isUnsignedInteger,
+                             MG_Util::ConvertSamplerCompareModeToGLEnum(samplerObj->GetCompareMode()));
             break;
         case GL_TEXTURE_COMPARE_FUNC:
-            *(GLuint*)params = MG_Util::ConvertSamplerCompareFuncToGLEnum(samplerObj->GetSamplerCompareFunc());
+            StoreSamplerEnum(params, isFloat, isUnsignedInteger,
+                             MG_Util::ConvertSamplerCompareFuncToGLEnum(samplerObj->GetSamplerCompareFunc()));
             break;
         case GL_TEXTURE_BORDER_COLOR: {
             if (isFloat) {
@@ -191,6 +259,16 @@ namespace MobileGL::MG_Impl::GLImpl {
                 out[1] = color.y();
                 out[2] = color.z();
                 out[3] = color.w();
+            } else if (!isIntegerCommand) {
+                // glGetSamplerParameteriv: the inverse of the write side, GL 4.6 core equation 2.3.
+                // Exactly inverse, so a {0,1,2,4} written with glSamplerParameteriv reads back as
+                // {0,1,2,4}; a bare truncating cast answered {0,0,0,0}.
+                const auto& color = samplerObj->GetBorderColor();
+                auto* out = (GLint*)params;
+                out[0] = MG_Util::FloatToSignedNormalizedInt32(color.x());
+                out[1] = MG_Util::FloatToSignedNormalizedInt32(color.y());
+                out[2] = MG_Util::FloatToSignedNormalizedInt32(color.z());
+                out[3] = MG_Util::FloatToSignedNormalizedInt32(color.w());
             } else if (isUnsignedInteger) {
                 const auto& color = samplerObj->GetBorderColorUI();
                 auto* out = (GLuint*)params;
@@ -293,16 +371,10 @@ namespace MobileGL::MG_Impl::GLImpl {
         if (sampler == 0) {
             textureUnit.SetSamplerObject(nullptr);
         } else {
-            // GL 3.3 core 3.8.2: BindSampler on a name GenSamplers never returned - or one already
-            // deleted - is INVALID_OPERATION. SamplerParameter* raises INVALID_VALUE for the same
-            // name, which is why this cannot go through the shared SamplerImpl validator.
-            if (!MG_State::pGLContext->ValidateSamplerName(sampler)) {
-                MG_State::pGLContext->RecordError(
-                    ErrorCode::InvalidOperation,
-                    MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", "BindSampler_State",
-                                                 std::format("Invalid sampler name {}", sampler)));
-                return;
-            }
+            // GL 4.6 core 8.2: BindSampler on a name GenSamplers never returned - or one already
+            // deleted - is INVALID_OPERATION, and so is every other sampler entry point on such a
+            // name, so the shared validator answers for all of them.
+            if (!SamplerImpl::ValidateSamplerName(sampler)) return;
             Bool doesSamplerObjectCreated = MG_State::pGLContext->ValidateSamplerObject(sampler);
             if (!doesSamplerObjectCreated) {
                 MG_State::pGLContext->CreateSamplerObject(sampler);
@@ -356,30 +428,50 @@ namespace MobileGL::MG_Impl::GLImpl {
 
     /* @INSERTION_POINT:FUNCTION_IMPLEMENTATION@ */
     void GetSamplerParameteriv(GLuint sampler, GLenum pname, GLint* params) {
-        GetSamplerParam_State(sampler, pname, params, false, false);
+        GetSamplerParam_State(sampler, pname, params, false, false, false);
     }
 
     void SamplerParameterIuiv(GLuint sampler, GLenum pname, const GLuint* param) {
-        SetSamplerParam_State(sampler, pname, param, false, true);
+        SetSamplerParam_State(sampler, pname, param, false, true, true);
     }
 
     void SamplerParameterIiv(GLuint sampler, GLenum pname, const GLint* param) {
-        SetSamplerParam_State(sampler, pname, param, false, false);
+        SetSamplerParam_State(sampler, pname, param, false, false, true);
     }
 
     void SamplerParameteriv(GLuint sampler, GLenum pname, const GLint* param) {
-        SetSamplerParam_State(sampler, pname, param, false, false);
+        SetSamplerParam_State(sampler, pname, param, false, false, false);
     }
 
     void SamplerParameterfv(GLuint sampler, GLenum pname, const GLfloat* param) {
-        SetSamplerParam_State(sampler, pname, param, true, false);
+        SetSamplerParam_State(sampler, pname, param, true, false, false);
     }
 
+    // GL 4.6 core 8.10: the scalar spellings take "the value of pname", so a pname with more than one
+    // component is INVALID_ENUM here rather than something to read four components of. Guarding at
+    // the entry point rather than downstream is also what stops the vector path reading twelve bytes
+    // past the caller's single stack scalar - taking the address of a by-value argument and handing
+    // it to a four-component reader is what these used to do. The texture-side twins already answer
+    // INVALID_ENUM for GL_TEXTURE_BORDER_COLOR (TexParameteri/f name it as unsupported outright).
     void SamplerParameteri(GLuint sampler, GLenum pname, GLint param) {
+        if (IsVectorOnlySamplerPname(pname)) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidEnum,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", "SamplerParameteri",
+                                             "pname has more than one component and needs a vector form."));
+            return;
+        }
         SamplerParameteriv(sampler, pname, &param);
     }
 
     void SamplerParameterf(GLuint sampler, GLenum pname, GLfloat param) {
+        if (IsVectorOnlySamplerPname(pname)) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidEnum,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", "SamplerParameterf",
+                                             "pname has more than one component and needs a vector form."));
+            return;
+        }
         SamplerParameterfv(sampler, pname, &param);
     }
 
@@ -388,15 +480,15 @@ namespace MobileGL::MG_Impl::GLImpl {
     }
 
     void GetSamplerParameterIuiv(GLuint sampler, GLenum pname, GLuint* params) {
-        GetSamplerParam_State(sampler, pname, params, false, true);
+        GetSamplerParam_State(sampler, pname, params, false, true, true);
     }
 
     void GetSamplerParameterIiv(GLuint sampler, GLenum pname, GLint* params) {
-        GetSamplerParam_State(sampler, pname, params, false, false);
+        GetSamplerParam_State(sampler, pname, params, false, false, true);
     }
 
     void GetSamplerParameterfv(GLuint sampler, GLenum pname, GLfloat* params) {
-        GetSamplerParam_State(sampler, pname, params, true, false);
+        GetSamplerParam_State(sampler, pname, params, true, false, false);
     }
 
     void GenSamplers(GLsizei count, GLuint* samplers) {

MobileGL/MG_Impl/GLImpl/Sampler/Validators.cpp

@@ -12,11 +12,17 @@
 #include <MG_Util/Converters/GLToMG/TextureEnumConverter.h>
 
 namespace MobileGL::MG_Impl::GLImpl::SamplerImpl {
+    // GL 4.6 core 8.2: "An INVALID_OPERATION error is generated if sampler is not the name of a
+    // sampler object previously returned from a call to GenSamplers." That class is shared by every
+    // sampler entry point - BindSampler, SamplerParameter*, GetSamplerParameter* - so this one gate
+    // answers for all of them. It used to report INVALID_VALUE (the GL 3.3 wording), which forced
+    // BindSampler to carry a bespoke duplicate of the same check just to get the class right.
     Bool ValidateSamplerName(GLuint sampler) {
         if (!MG_State::pGLContext->ValidateSamplerName(sampler)) {
             MG_State::pGLContext->RecordError(
-                ErrorCode::InvalidValue, MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", "ValidateSamplerName",
-                                                                      std::format("Invalid sampler name {}", sampler)));
+                ErrorCode::InvalidOperation,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", "ValidateSamplerName",
+                                             std::format("Invalid sampler name {}", sampler)));
             return false;
         }
         return true;

MobileGL/MG_Impl/GLImpl/Sync/GL_Sync.cpp

@@ -8,6 +8,8 @@
 
 #include "GL_Sync.h"
 #include <MG_Backend/BackendObjects.h>
+#include <MG_State/GLState/Core.h>
+#include <MG_Impl/Pipe/PipeFill.h>
 
 namespace MobileGL::MG_Impl::GLImpl {
     namespace {
@@ -35,10 +37,27 @@ namespace MobileGL::MG_Impl::GLImpl {
     } // namespace
 
     GLsync FenceSync(GLenum condition, GLbitfield flags) {
+        // GL 4.6 core 4.1.2: GL_SYNC_GPU_COMMANDS_COMPLETE is the only condition and the only
+        // legal flags value is zero; both violations return 0 rather than a handle. A caller that
+        // then hands the 0 back to glDeleteSync hits the glDeleteSync(0) no-op below.
+        if (condition != GL_SYNC_GPU_COMMANDS_COMPLETE) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidEnum,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "condition must be GL_SYNC_GPU_COMMANDS_COMPLETE."));
+            return nullptr;
+        }
+        if (flags != 0) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "flags must be zero."));
+            return nullptr;
+        }
         auto* syncObject = new SyncObject;
         syncObject->condition = condition;
         syncObject->flags = flags;
         if (const auto backendFenceSync = MG_Backend::gBackendFunctionsTable.GL.FenceSync) {
+            MGP_FILL(FenceSync);
             syncObject->backendHandle = backendFenceSync();
         }
         const GLsync handle = reinterpret_cast<GLsync>(syncObject);
@@ -52,24 +71,58 @@ namespace MobileGL::MG_Impl::GLImpl {
     }
 
     GLenum ClientWaitSync(GLsync sync, GLbitfield flags, GLuint64 timeout) {
+        // GL 4.6 core 4.1.1: GL_SYNC_FLUSH_COMMANDS_BIT is the only bit this call accepts, and
+        // any other bit is INVALID_VALUE. Silently ignoring the stray bits used to make a caller
+        // that passed, say, GL_SYNC_GPU_COMMANDS_COMPLETE by mistake think it had asked for a
+        // flush it never got.
+        if ((flags & ~static_cast<GLbitfield>(GL_SYNC_FLUSH_COMMANDS_BIT)) != 0) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "flags must be zero or GL_SYNC_FLUSH_COMMANDS_BIT."));
+            return GL_WAIT_FAILED;
+        }
         const auto* syncObject = FindSyncObject(sync);
         if (!syncObject) {
+            // The spec pairs the GL_WAIT_FAILED return with a recorded INVALID_VALUE; returning
+            // the enum alone left glGetError() clean and the failure indistinguishable from a
+            // genuine wait failure on a live sync.
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "sync is not the name of a sync object."));
             return GL_WAIT_FAILED;
         }
         const auto backendClientWaitSync = MG_Backend::gBackendFunctionsTable.GL.ClientWaitSync;
         if (!backendClientWaitSync || !syncObject->backendHandle) {
             return GL_ALREADY_SIGNALED; // legacy always-signaled fallback
         }
+        MGP_FILL(ClientWaitSync);
         return backendClientWaitSync(syncObject->backendHandle, flags, timeout);
     }
 
     void WaitSync(GLsync sync, GLbitfield flags, GLuint64 timeout) {
+        // GL 4.6 core 4.1.2: the server-side wait takes no flags and no finite timeout - both
+        // arguments exist only to be forward-compatible, and anything else is INVALID_VALUE.
+        // Neither backend ever honored a nonzero timeout (DirectGLES hard-codes
+        // 0/GL_TIMEOUT_IGNORED, DirectVulkan's queue ordering makes the wait implicit), so
+        // rejecting the call loses no wait that used to happen.
+        if (flags != 0 || timeout != GL_TIMEOUT_IGNORED) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "flags must be zero and timeout must be GL_TIMEOUT_IGNORED."));
+            return;
+        }
         const auto* syncObject = FindSyncObject(sync);
         if (!syncObject) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "sync is not the name of a sync object."));
             return;
         }
         const auto backendWaitSync = MG_Backend::gBackendFunctionsTable.GL.WaitSync;
         if (backendWaitSync && syncObject->backendHandle) {
+            MGP_FILL(WaitSync);
             backendWaitSync(syncObject->backendHandle, flags, timeout);
         }
     }
@@ -90,14 +143,29 @@ namespace MobileGL::MG_Impl::GLImpl {
         }
         const auto backendDeleteSync = MG_Backend::gBackendFunctionsTable.GL.DeleteSync;
         if (backendDeleteSync && syncObject->backendHandle) {
+            MGP_FILL(DeleteSync);
             backendDeleteSync(syncObject->backendHandle);
         }
         delete syncObject;
     }
 
     void GetSynciv(GLsync sync, GLenum pname, GLsizei bufSize, GLsizei* length, GLint* values) {
+        // GL 4.6 core 4.1: a negative bufSize is INVALID_VALUE, an unnamed sync is INVALID_VALUE
+        // and an unrecognised pname is INVALID_ENUM. All three used to leave glGetError() clean
+        // and write a plausible-looking zero, which is the one failure mode a caller cannot tell
+        // apart from a real answer - GL_SYNC_STATUS legitimately answers GL_UNSIGNALED (0x9118),
+        // but a mistyped pname answered a bare 0 that no query ever returns.
+        if (bufSize < 0) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "bufSize must not be negative."));
+            return;
+        }
         const auto* syncObject = FindSyncObject(sync);
         if (!syncObject) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__, "sync is not the name of a sync object."));
             if (length) {
                 *length = 0;
             }
@@ -111,6 +179,7 @@ namespace MobileGL::MG_Impl::GLImpl {
             break;
         case GL_SYNC_STATUS: {
             const auto backendGetSyncStatus = MG_Backend::gBackendFunctionsTable.GL.GetSyncStatus;
+            MGP_FILL(GetSyncStatus);
             const Bool signaled = !backendGetSyncStatus || !syncObject->backendHandle ||
                                   backendGetSyncStatus(syncObject->backendHandle);
             value = signaled ? GL_SIGNALED : GL_UNSIGNALED;
@@ -123,7 +192,15 @@ namespace MobileGL::MG_Impl::GLImpl {
             value = static_cast<GLint>(syncObject->flags);
             break;
         default:
-            break;
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidEnum,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", __func__,
+                                             "pname must be GL_OBJECT_TYPE, GL_SYNC_STATUS, GL_SYNC_CONDITION or "
+                                             "GL_SYNC_FLAGS."));
+            if (length) {
+                *length = 0;
+            }
+            return;
         }
 
         if (length) {
@@ -156,6 +233,7 @@ namespace MobileGL::MG_Impl::GLImpl {
         const auto backendDeleteSync = MG_Backend::gBackendFunctionsTable.GL.DeleteSync;
         for (const auto& [_, syncObject] : orphans) {
             if (backendDeleteSync && syncObject->backendHandle) {
+                MGP_FILL(DeleteSync);
                 backendDeleteSync(syncObject->backendHandle);
             }
             delete syncObject;

MobileGL/MG_Impl/GLImpl/Texture/GL_Texture.h

@@ -8,9 +8,24 @@
 
 #pragma once
 #include <Includes.h>
+#include <MG_State/GLState/TextureState/TextureObject.h>
 
 namespace MobileGL::MG_Impl::GLImpl {
     /* @INSERTION_POINT:FUNCTION_DECLARATION@ */
+    // Answers a texture-image query straight out of the CPU shadow, into client memory or a bound
+    // PIXEL_PACK_BUFFER. This is the whole of glGetTexImage on a build with no backend readback, and
+    // it is also the sound fallback for a backend that has no GPU image to read: with no image,
+    // nothing GPU-side can ever have written the texture, so the shadow IS its content.
+    //
+    // It answers a NARROWER contract than glGetTexImage's, and refuses what it cannot do rather than
+    // answering wrongly. The copy is verbatim: it performs no format or type conversion, and it packs
+    // rows tightly, honouring only GL_PACK_SWAP_BYTES and the bitmap GL_PACK_LSB_FIRST path. A
+    // request whose (format, type) texel size differs from the texture's own, or a pixel-store state
+    // that adds row padding / a row-length override / a skip offset, is rejected with
+    // GL_INVALID_OPERATION (see ValidateShadowReadbackLayout, which spells out why each is unsafe).
+    void CopyTextureImageToClientOrPBO_State(const SharedPtr<MG_State::GLState::ITextureObject>& textureObject,
+                                             TextureUploadTarget textureUploadTarget, GLint level, GLenum format,
+                                             GLenum type, GLsizei bufSize, void* pixels, const char* caller);
     // The sized internal formats a buffer texture accepts (GL 4.6 core table 8.16). The buffer
     // clears take the same list, so it is shared rather than written out twice.
     Bool IsBufferTextureInternalFormat(GLenum internalformat);
@@ -37,8 +52,13 @@ namespace MobileGL::MG_Impl::GLImpl {
                            GLenum format, GLenum type, const void* pixels);
     void TextureSubImage3D(GLuint texture, GLint level, GLint xoffset, GLint yoffset, GLint zoffset, GLsizei width,
                            GLsizei height, GLsizei depth, GLenum format, GLenum type, const void* pixels);
+    void CompressedTextureSubImage1D(GLuint texture, GLint level, GLint xoffset, GLsizei width, GLenum format,
+                                     GLsizei imageSize, const void* data);
     void CompressedTextureSubImage2D(GLuint texture, GLint level, GLint xoffset, GLint yoffset, GLsizei width,
                                      GLsizei height, GLenum format, GLsizei imageSize, const void* data);
+    void CompressedTextureSubImage3D(GLuint texture, GLint level, GLint xoffset, GLint yoffset, GLint zoffset,
+                                     GLsizei width, GLsizei height, GLsizei depth, GLenum format, GLsizei imageSize,
+                                     const void* data);
     void TextureParameterf(GLuint texture, GLenum pname, GLfloat param);
     void TextureParameterfv(GLuint texture, GLenum pname, const GLfloat* params);
     void TextureParameteri(GLuint texture, GLenum pname, GLint param);
@@ -60,6 +80,8 @@ namespace MobileGL::MG_Impl::GLImpl {
     void GetTextureParameteriv(GLuint texture, GLenum pname, GLint* params);
     void GetTextureLevelParameterfv(GLuint texture, GLint level, GLenum pname, GLfloat* params);
     void GetTextureLevelParameteriv(GLuint texture, GLint level, GLenum pname, GLint* params);
+    void TextureView(GLuint texture, GLenum target, GLuint origtexture, GLenum internalformat, GLuint minlevel,
+                     GLuint numlevels, GLuint minlayer, GLuint numlayers);
     void TexStorage1D(GLenum target, GLsizei levels, GLenum internalformat, GLsizei width);
     void TexStorage2D(GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height);
     void TexStorage3D(GLenum target, GLsizei levels, GLenum internalformat, GLsizei width, GLsizei height,

MobileGL/MG_Impl/GLImpl/Texture/Validators.cpp

@@ -103,6 +103,28 @@ namespace MobileGL::MG_Impl::GLImpl::TextureImpl {
         return true;
     }
 
+    Bool ValidateCubeMapArrayShape(TextureUploadTarget target, GLsizei width, GLsizei height, GLsizei depth,
+                                   const char* caller) {
+        if (target != TextureUploadTarget::CubeMapArray && target != TextureUploadTarget::ProxyCubeMapArray) {
+            return true;
+        }
+        if (width != height) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", caller,
+                                             "Cube map array levels must be square (width == height)"));
+            return false;
+        }
+        if (depth % 6 != 0) {
+            MG_State::pGLContext->RecordError(
+                ErrorCode::InvalidValue,
+                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", caller,
+                                             "Cube map array depth must be a multiple of six"));
+            return false;
+        }
+        return true;
+    }
+
     Bool ValidateTextureSizeWithTextureUploadTarget(TextureUploadTarget target, GLsizei width, GLsizei height) {
         if (target == TextureUploadTarget::CubeMapPositiveX || target == TextureUploadTarget::CubeMapNegativeX ||
             target == TextureUploadTarget::CubeMapPositiveY || target == TextureUploadTarget::CubeMapNegativeY ||
@@ -313,9 +335,13 @@ namespace MobileGL::MG_Impl::GLImpl::TextureImpl {
             return false;
         }
 
-        // TexImage in core 3.3 has no stencil-only upload path (that arrived with GL 4.4).
-        if (format == TextureInputFormat::StencilIndex) {
-            return recordInvalidOperation("STENCIL_INDEX is not a valid texture upload format");
+        // The stencil-only transfer path arrived with GL 4.4 / ARB_texture_stencil8, and only ever
+        // pairs with stencil-only storage: against a depth, depth-stencil or colour internal format
+        // STENCIL_INDEX keeps the pre-4.4 answer (GL CTS packed_pixels feeds exactly that pairing
+        // and expects INVALID_OPERATION).
+        if (format == TextureInputFormat::StencilIndex &&
+            internalFormat != TextureInternalFormat::StencilIndex8) {
+            return recordInvalidOperation("STENCIL_INDEX requires a stencil-only internal format");
         }
 
         if (IsDepthLikeInputFormat(format) != IsDepthLikeInternalFormat(internalFormat)) {
@@ -619,4 +645,144 @@ namespace MobileGL::MG_Impl::GLImpl::TextureImpl {
         }
         return true;
     }
+
+    // GL 4.6 core table 8.21 ("Compatible internal formats for TextureView"), transcribed whole.
+    // Written against the raw GLenum rather than TextureInternalFormat on purpose: MobileGL's own
+    // enum collapses every compressed format onto uncompressed storage and drops formats it
+    // cannot carry, so classifying the converted value would silently widen the compatibility
+    // rule - GL_COMPRESSED_RG_RGTC2 and GL_RGBA8 would end up in the same class.
+    TextureViewClass GetTextureViewClass(GLenum internalformat) {
+        switch (internalformat) {
+        case GL_RGBA32F:
+        case GL_RGBA32UI:
+        case GL_RGBA32I:
+            return TextureViewClass::Bits128;
+        case GL_RGB32F:
+        case GL_RGB32UI:
+        case GL_RGB32I:
+            return TextureViewClass::Bits96;
+        case GL_RGBA16F:
+        case GL_RG32F:
+        case GL_RGBA16UI:
+        case GL_RG32UI:
+        case GL_RGBA16I:
+        case GL_RG32I:
+        case GL_RGBA16:
+        case GL_RGBA16_SNORM:
+            return TextureViewClass::Bits64;
+        case GL_RGB16:
+        case GL_RGB16_SNORM:
+        case GL_RGB16F:
+        case GL_RGB16UI:
+        case GL_RGB16I:
+            return TextureViewClass::Bits48;
+        case GL_RG16F:
+        case GL_R11F_G11F_B10F:
+        case GL_R32F:
+        case GL_RGB10_A2UI:
+        case GL_RGBA8UI:
+        case GL_RG16UI:
+        case GL_R32UI:
+        case GL_RGBA8I:
+        case GL_RG16I:
+        case GL_R32I:
+        case GL_RGB10_A2:
+        case GL_RGBA8:
+        case GL_RG16:
+        case GL_RGBA8_SNORM:
+        case GL_RG16_SNORM:
+        case GL_SRGB8_ALPHA8:
+        case GL_RGB9_E5:
+            return TextureViewClass::Bits32;
+        case GL_RGB8:
+        case GL_RGB8_SNORM:
+        case GL_SRGB8:
+        case GL_RGB8UI:
+        case GL_RGB8I:
+            return TextureViewClass::Bits24;
+        case GL_R16F:
+        case GL_RG8UI:
+        case GL_R16UI:
+        case GL_RG8I:
+        case GL_R16I:
+        case GL_RG8:
+        case GL_R16:
+        case GL_RG8_SNORM:
+        case GL_R16_SNORM:
+            return TextureViewClass::Bits16;
+        case GL_R8UI:
+        case GL_R8I:
+        case GL_R8:
+        case GL_R8_SNORM:
+            return TextureViewClass::Bits8;
+        case GL_COMPRESSED_RED_RGTC1:
+        case GL_COMPRESSED_SIGNED_RED_RGTC1:
+            return TextureViewClass::Rgtc1Red;
+        case GL_COMPRESSED_RG_RGTC2:
+        case GL_COMPRESSED_SIGNED_RG_RGTC2:
+            return TextureViewClass::Rgtc2Rg;
+        case GL_COMPRESSED_RGBA_BPTC_UNORM:
+        case GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM:
+            return TextureViewClass::BptcUnorm;
+        case GL_COMPRESSED_RGB_BPTC_SIGNED_FLOAT:
+        case GL_COMPRESSED_RGB_BPTC_UNSIGNED_FLOAT:
+            return TextureViewClass::BptcFloat;
+        default:
+            // Every depth/stencil format, every S3TC/ETC/ASTC format and every unsized format
+            // reaches here. The caller must then demand an EXACT format match.
+            return TextureViewClass::None;
+        }
+    }
+
+    // GL 4.6 core table 8.20 ("Legal texture targets for TextureView").
+    Bool IsLegalTextureViewTargetPair(TextureTarget origTarget, TextureTarget viewTarget) {
+        switch (origTarget) {
+        case TextureTarget::Texture1D:
+            return viewTarget == TextureTarget::Texture1D || viewTarget == TextureTarget::Texture1DArray;
+        case TextureTarget::Texture2D:
+            return viewTarget == TextureTarget::Texture2D || viewTarget == TextureTarget::Texture2DArray;
+        case TextureTarget::Texture3D:
+            return viewTarget == TextureTarget::Texture3D;
+        case TextureTarget::TextureCubeMap:
+            return viewTarget == TextureTarget::TextureCubeMap || viewTarget == TextureTarget::Texture2D ||
+                   viewTarget == TextureTarget::Texture2DArray || viewTarget == TextureTarget::TextureCubeMapArray;
+        case TextureTarget::TextureRectangle:
+            return viewTarget == TextureTarget::TextureRectangle;
+        case TextureTarget::Texture1DArray:
+            return viewTarget == TextureTarget::Texture1DArray || viewTarget == TextureTarget::Texture1D;
+        case TextureTarget::Texture2DArray:
+            return viewTarget == TextureTarget::Texture2DArray || viewTarget == TextureTarget::Texture2D ||
+                   viewTarget == TextureTarget::TextureCubeMap || viewTarget == TextureTarget::TextureCubeMapArray;
+        case TextureTarget::TextureCubeMapArray:
+            return viewTarget == TextureTarget::TextureCubeMapArray || viewTarget == TextureTarget::Texture2DArray ||
+                   viewTarget == TextureTarget::Texture2D || viewTarget == TextureTarget::TextureCubeMap;
+        case TextureTarget::Texture2DMultisample:
+        case TextureTarget::Texture2DMultisampleArray:
+            return viewTarget == TextureTarget::Texture2DMultisample ||
+                   viewTarget == TextureTarget::Texture2DMultisampleArray;
+        case TextureTarget::TextureBuffer:
+            // The table lists no legal target for a buffer texture: its storage is a buffer
+            // object, and there is nothing to make a view of.
+            return false;
+        default:
+            return false;
+        }
+    }
+
+    Uint RequiredTextureViewLayerCount(TextureTarget viewTarget) {
+        switch (viewTarget) {
+        case TextureTarget::TextureCubeMap:
+            return 6;
+        case TextureTarget::Texture1D:
+        case TextureTarget::Texture2D:
+        case TextureTarget::Texture3D:
+        case TextureTarget::TextureRectangle:
+        case TextureTarget::Texture2DMultisample:
+            return 1;
+        default:
+            // 1D/2D array, cube-map array, 2D multisample array: any count (the cube-map array's
+            // "multiple of 6" is checked by the caller).
+            return 0;
+        }
+    }
 } // namespace MobileGL::MG_Impl::GLImpl::TextureImpl

MobileGL/MG_Impl/GLImpl/Texture/Validators.h

@@ -20,6 +20,13 @@ namespace MobileGL::MG_Impl::GLImpl::TextureImpl {
     Bool ValidateTexturePixelDataType(TexturePixelDataType texturePixelDataType);
     Bool ValidateTextureLevelNumber(Int level);
     Bool ValidateTextureSizeWithTextureUploadTarget(TextureUploadTarget target, GLsizei width, GLsizei height);
+    // The two shape rules a cube-map-array level owes (GL 4.6 core 8.5): its faces are square, and
+    // its depth counts whole cubes. Both are GL_INVALID_VALUE. This used to be spelled inline in
+    // glTexStorage3D only, which is why glTexImage3D let both violations through - every entry
+    // point that DEFINES a cube-array level calls this now, so the two cannot drift again. A
+    // non-cube-array upload target answers true untouched.
+    Bool ValidateCubeMapArrayShape(TextureUploadTarget target, GLsizei width, GLsizei height, GLsizei depth,
+                                   const char* caller);
     Bool ValidateTextureSizeRange(Int width, Int height, Int depth);
     Bool ValidateTextureInternalFormat(TextureInternalFormat format);
     Bool ValidateTextureBorderNumber(Int border);
@@ -79,4 +86,33 @@ namespace MobileGL::MG_Impl::GLImpl::TextureImpl {
     // GL 4.6 SS 8.6 subset rule for glCopyTexImage*: the read buffer must supply every component
     // the requested internalformat asks for, but may supply more.
     Bool ValidateCopyTexImageBaseFormatSubset(TextureInternalFormat destFormat, TextureInternalFormat srcFormat);
+
+    // ---- glTextureView (ARB_texture_view / GL 4.6 core 8.18) ----
+    // Table 8.21's view classes. `None` is not a class - it means the format has NO entry in the
+    // table, which the spec turns into a much stricter rule than "same class": such a format can
+    // only ever be viewed as ITSELF. Every depth, stencil and depth/stencil format lands here,
+    // which is why the Better Clouds D24S8 view must name GL_DEPTH24_STENCIL8 exactly.
+    enum class TextureViewClass {
+        None = 0,
+        Bits128,
+        Bits96,
+        Bits64,
+        Bits48,
+        Bits32,
+        Bits24,
+        Bits16,
+        Bits8,
+        Rgtc1Red,
+        Rgtc2Rg,
+        BptcUnorm,
+        BptcFloat,
+    };
+    TextureViewClass GetTextureViewClass(GLenum internalformat);
+    // Table 8.20: which <target> values glTextureView accepts for a given origtexture target.
+    Bool IsLegalTextureViewTargetPair(TextureTarget origTarget, TextureTarget viewTarget);
+    // Table 8.20 again, read the other way: how many layers <target> requires. Returns 0 for the
+    // targets whose layer count is unconstrained (the array targets), 6 for GL_TEXTURE_CUBE_MAP,
+    // and 1 for every single-layer target. GL_TEXTURE_CUBE_MAP_ARRAY is special-cased by the
+    // caller because its constraint is "a multiple of 6", not an exact count.
+    Uint RequiredTextureViewLayerCount(TextureTarget viewTarget);
 } // namespace MobileGL::MG_Impl::GLImpl::TextureImpl

MobileGL/MG_Impl/GLImpl/VertexArray/GL_VertexArray.cpp

@@ -514,10 +514,17 @@ namespace MobileGL::MG_Impl::GLImpl {
     // recorded DataType is always Float64 - what IsLong adds is that this is the *unconverted* form,
     // as opposed to VertexAttribFormat(GL_DOUBLE), which asks for a float conversion.
     //
-    // Whether the backend can feed it is detected, not assumed: DirectVulkan needs shaderFloat64,
-    // and DirectGLES can never have it at all. A backend without it declines here, loudly - GL error
-    // plus a log line naming the reason - rather than accepting state no draw could honour and
-    // rendering garbage. The matching startup POST row is in MG_Util/SelfTest/DriverPost.cpp.
+    // Whether the backend can FEED it at full precision is detected, not assumed: DirectVulkan
+    // needs shaderFloat64, and DirectGLES can never have it at all. What that costs is PRECISION,
+    // not the call and no longer the array: GL 4.6 core 10.3.2 defines no error for a well-formed
+    // glVertexAttribLFormat, and a GL 4.3 context has 64-bit attributes in core, so declining the
+    // call would be non-conformant and would make the four pure state queries
+    // (VERTEX_ATTRIB_ARRAY_SIZE / _TYPE / _LONG / _RELATIVE_OFFSET) unanswerable
+    // (KHR-GL43.vertex_attrib_binding.basic-state1/3). The format is therefore RECORDED here and
+    // the array is NARROWED to float32 at draw, matching the fp64 demotion every shader already
+    // gets (DemoteFloat64Pass) - loudly, once, naming the cost. The matching startup POST row is in
+    // MG_Util/SelfTest/DriverPost.cpp; the draw-side narrowing is DirectGLES/Managers.cpp and, on
+    // DirectVulkan, VertexInputStateFactory's Float64 case.
     static void VertexAttribLFormatSeparate_State(const SharedPtr<MG_State::GLState::VertexArrayObject>& vao,
                                                   GLuint attribindex, GLint size, GLenum type,
                                                   GLuint relativeoffset) {
@@ -528,14 +535,11 @@ namespace MobileGL::MG_Impl::GLImpl {
         if (!MG_Backend::pActiveBackendObject ||
             !MG_Backend::pActiveBackendObject->GetDynamicParameters().SupportsFloat64VertexAttributes) {
             MGLOG_W_ONCE("VertexAttribLFormat: attribute %u asked for a 64-bit (GL_DOUBLE) format, but this "
-                    "backend has no double-precision vertex attribute support - see the "
-                    "\"64-bit vertex attributes\" / \"shaderFloat64\" POST row for what that costs",
+                    "backend has no double-precision vertex attribute support - the format is recorded "
+                    "and queryable, and the array is FETCHED AT FLOAT32 PRECISION at draw (the same "
+                    "narrowing the shader's dvec inputs already get); see the \"64-bit vertex "
+                    "attributes\" / \"shaderFloat64\" POST row for what that costs",
                     attribindex);
-            MG_State::pGLContext->RecordError(
-                ErrorCode::InvalidOperation,
-                MakeUnique<GenericErrorInfo>("MG_Impl/GLImpl", "VertexAttribLFormat",
-                                             "64-bit vertex attributes are not supported by this backend."));
-            return;
         }
 
         vao->SetAttributeFormatSeparate(attribindex, size, MG_Util::ConvertGLEnumToDataType(type),

MobileGL/MG_Impl/GLImpl/VertexArray/Validators.cpp

@@ -12,15 +12,15 @@
 #include <MG_State/GLState/ErrorState/Error.h>
 #include <MG_Util/Converters/MGToGL/DataTypeConverter.h>
 #include <MG_Util/Converters/MGToStr/DataTypeConverter.h>
+#include <MG_Util/ShaderTranspiler/CompileEnv.h>
 
 namespace MobileGL::MG_Impl::GLImpl::VertexArrayImpl {
     Uint GetMaxVertexAttribs() {
-        constexpr Uint capacity = static_cast<Uint>(MG_State::GLState::VertexArrayObject::MAX_VERTEX_ATTRIBS);
-        if (!MG_Backend::pActiveBackendObject) return capacity;
-
-        const Int backendLimit = MG_Backend::pActiveBackendObject->GetDynamicParameters().MaxVertexAttribs;
-        if (backendLimit <= 0) return capacity;
-        return std::min(static_cast<Uint>(backendLimit), capacity);
+        // Shared with reflection's limit and with gl_MaxVertexAttribs; see ResolveMaxVertexAttribs.
+        const Bool hasBackend = MG_Backend::pActiveBackendObject != nullptr;
+        const Int backendLimit =
+            hasBackend ? MG_Backend::pActiveBackendObject->GetDynamicParameters().MaxVertexAttribs : 0;
+        return static_cast<Uint>(MG_Util::ShaderTranspiler::ResolveMaxVertexAttribs(hasBackend, backendLimit));
     }
 
     Uint GetMaxVertexAttribBindings() {

MobileGL/MG_Impl/Pipe/PipeFill.cpp

@@ -0,0 +1,614 @@
+// MobileGL - MobileGL/MG_Impl/Pipe/PipeFill.cpp
+// Copyright (c) 2025-2026 MobileGL-Dev
+// Licensed under the GNU Lesser General Public License v3.0:
+//   https://www.gnu.org/licenses/gpl-3.0.txt
+//   https://www.gnu.org/licenses/lgpl-3.0.txt
+// SPDX-License-Identifier: LGPL-3.0-only
+// End of Source File Header
+
+// The client side of the PipeInputs block (ARCHITECTURE.md 9.2 phase A): the only place in
+// the push arm that reads MG_State::pGLContext. Holds the per-verb filler, the F-class
+// forwarders, IsLive, the MOBILEGL_PIPE_POISON_OMIT knob and - in a verify build - the
+// second arm (SnapshotFromGLContext), the entry compare, the compare-at-read hook and the
+// MOBILEGL_PIPE_VERIFY_CORRUPT / _FATAL knobs. Compiled only under MOBILEGL_PIPE_PUSH
+// (CMakeLists.txt appends it to SOURCE_FILES there).
+#include <MG_State/GLState/Core.h>
+#include <MG_State/GLState/BufferState/BufferState.h>
+#include <MG_Backend/MGPipe/PipeInputs.h>
+#include <MG_Impl/Pipe/PipeFill.h>
+#include <MG_Pipe/PipeMutation.h>
+#include <Config.h>
+
+#include <atomic>
+#include <cstdlib>
+#include <cstring>
+
+namespace MobileGL::MG_Pipe {
+    using GLContext = MG_State::GLState::GLContext;
+
+    // The one door into PipeInputs' storage on the client side. A struct rather than a
+    // list of friend functions so the header names exactly one friend.
+    struct MGPipeFillAccess {
+        // Copies ONE field's storage out of the live context by calling the GLContext
+        // accessor of the same name (P1 brief D4: no derivation logic is re-implemented
+        // here, which is what keeps the copy semantically identical by construction).
+        // A forwarded field has no storage and copies nothing.
+        static void CopyField(PipeInputs& dst, GLContext& ctx, MGPipeInputField field) {
+            using F = MGPipeInputField;
+            using MG_State::GLState::BufferBindPointTargets;
+            using MG_State::GLState::GlobalBufferTargets;
+            switch (field) {
+            case F::GetActiveTextureUnit:
+                dst.m_activeTextureUnit = ctx.GetActiveTextureUnit();
+                break;
+            case F::GetBlendColor:
+                dst.m_blendColor = ctx.GetBlendColor();
+                break;
+            case F::GetBlendEquationIndexed:
+                for (Uint i = 0; i < kMGMaxDrawBuffers; ++i) {
+                    ctx.GetBlendEquationIndexed(i, dst.m_blendEquation[i][0], dst.m_blendEquation[i][1]);
+                }
+                break;
+            case F::GetBlendFuncIndexed:
+                for (Uint i = 0; i < kMGMaxDrawBuffers; ++i) {
+                    ctx.GetBlendFuncIndexed(i, dst.m_blendFunc[i][0], dst.m_blendFunc[i][1], dst.m_blendFunc[i][2],
+                                            dst.m_blendFunc[i][3]);
+                }
+                break;
+            case F::GetBoundTransformFeedbackName:
+                dst.m_boundTransformFeedbackName = ctx.GetBoundTransformFeedbackName();
+                break;
+            case F::GetBoundVertexArray:
+                dst.m_boundVertexArray = ctx.GetBoundVertexArray();
+                break;
+            case F::GetBufferBindingSlot:
+                // Every global target has a slot; Index stays null - GLContext resolves it
+                // through the bound VAO's element-buffer slot (Core.cpp), a derivation no
+                // FillPoints.def row can copy - and a read of it is the poison Fatal in the
+                // accessor. No backend reads it today (every slot read is DrawIndirect,
+                // DispatchIndirect, Parameter or PixelPack).
+                for (const auto target : GlobalBufferTargets) {
+                    dst.m_bufferBindingSlot[static_cast<SizeT>(target)] = &ctx.GetBufferBindingSlot(target);
+                }
+                break;
+            case F::GetBufferBindingPoint:
+                // The live storage is Array<Array<BindingSlotRange1D, BufferBindingPointCount>, N>
+                // (BufferState.h), so the address of point 0 is the base of that target's row.
+                for (const auto target : BufferBindPointTargets) {
+                    dst.m_bufferBindingPointBase[static_cast<SizeT>(target)] = &ctx.GetBufferBindingPoint(target, 0);
+                }
+                break;
+            case F::GetTouchedBufferBindingPointCount:
+                for (const auto target : BufferBindPointTargets) {
+                    dst.m_touchedBindingPointCount[static_cast<SizeT>(target)] =
+                        ctx.GetTouchedBufferBindingPointCount(target);
+                }
+                break;
+            case F::GetClampReadColor:
+                dst.m_clampReadColor = ctx.GetClampReadColor();
+                break;
+            case F::GetClearColor:
+                dst.m_clearColor = ctx.GetClearColor();
+                break;
+            case F::GetClearDepth:
+                dst.m_clearDepth = ctx.GetClearDepth();
+                break;
+            case F::GetClearStencil:
+                dst.m_clearStencil = ctx.GetClearStencil();
+                break;
+            case F::GetColorMaskIndexed:
+                for (Uint i = 0; i < kMGMaxDrawBuffers; ++i) {
+                    dst.m_colorMask[i] = ctx.GetColorMaskIndexed(i);
+                }
+                break;
+            case F::GetCullFaceMode:
+                dst.m_cullFaceMode = ctx.GetCullFaceMode();
+                break;
+            case F::GetCurrentVertexAttribute:
+                for (Uint i = 0; i < PipeInputs::kMaxVertexAttribs; ++i) {
+                    dst.m_currentVertexAttribute[i] = ctx.GetCurrentVertexAttribute(i);
+                }
+                break;
+            case F::GetDepthFunc:
+                dst.m_depthFunc = ctx.GetDepthFunc();
+                break;
+            case F::GetDepthMask:
+                dst.m_depthMask = ctx.GetDepthMask();
+                break;
+            case F::GetDepthRangeIndexed:
+                for (Uint i = 0; i < PipeInputs::kMaxViewports; ++i) {
+                    dst.m_depthRange[i] = ctx.GetDepthRangeIndexed(i);
+                }
+                break;
+            case F::GetFramebufferBindingSlot:
+                for (SizeT i = 0; i < PipeInputs::kFramebufferTargetCount; ++i) {
+                    dst.m_framebufferBindingSlot[i] =
+                        &ctx.GetFramebufferBindingSlot(static_cast<PipeInputs::FramebufferTarget>(i));
+                }
+                break;
+            case F::GetImageTextureBinding:
+                // Array<ImageTextureBinding, MAX_TEXTURE_IMAGE_UNITS> (TextureState.h): unit 0's
+                // address is the base.
+                dst.m_imageTextureBindingBase = &ctx.GetImageTextureBinding(0);
+                break;
+            case F::GetLineWidth:
+                dst.m_lineWidth = ctx.GetLineWidth();
+                break;
+            case F::GetLogicOp:
+                dst.m_logicOp = ctx.GetLogicOp();
+                break;
+            case F::GetMaxTouchedTextureUnit:
+                dst.m_maxTouchedTextureUnit = ctx.GetMaxTouchedTextureUnit();
+                break;
+            case F::GetMinSampleShadingValue:
+                dst.m_minSampleShadingValue = ctx.GetMinSampleShadingValue();
+                break;
+            case F::GetPatchDefaultInnerLevel:
+                dst.m_patchDefaultInnerLevel = ctx.GetPatchDefaultInnerLevel();
+                break;
+            case F::GetPatchDefaultOuterLevel:
+                dst.m_patchDefaultOuterLevel = ctx.GetPatchDefaultOuterLevel();
+                break;
+            case F::GetPatchVertices:
+                dst.m_patchVertices = ctx.GetPatchVertices();
+                break;
+            case F::GetPipelineStateVersion:
+                dst.m_pipelineStateVersion = ctx.GetPipelineStateVersion();
+                break;
+            case F::GetPixelStoreParameters:
+                dst.m_pixelStore[0] = ctx.GetPixelStoreParameters(false);
+                dst.m_pixelStore[1] = ctx.GetPixelStoreParameters(true);
+                break;
+            case F::GetPolygonModeFront:
+                dst.m_polygonModeFront = ctx.GetPolygonModeFront();
+                break;
+            case F::GetPolygonOffsetFactor:
+                dst.m_polygonOffsetFactor = ctx.GetPolygonOffsetFactor();
+                break;
+            case F::GetPolygonOffsetUnits:
+                dst.m_polygonOffsetUnits = ctx.GetPolygonOffsetUnits();
+                break;
+            case F::GetPrimitiveRestartIndex:
+                dst.m_primitiveRestartIndex = ctx.GetPrimitiveRestartIndex();
+                break;
+            case F::GetProgramForDispatch:
+                dst.m_programForDispatch = ctx.GetProgramForDispatch();
+                break;
+            case F::GetProgramForDraw:
+                dst.m_programForDraw = ctx.GetProgramForDraw();
+                break;
+            case F::GetProvokingVertexMode:
+                dst.m_provokingVertexMode = ctx.GetProvokingVertexMode();
+                break;
+            case F::GetRenderStateParameters:
+                dst.m_renderState = ctx.GetRenderStateParameters();
+                break;
+            case F::GetRenderStateParametersVersion:
+                dst.m_renderStateParametersVersion = ctx.GetRenderStateParametersVersion();
+                break;
+            case F::GetSamplingResolutionGeneration:
+                dst.m_samplingResolutionGeneration = ctx.GetSamplingResolutionGeneration();
+                break;
+            case F::GetScissorBox:
+                dst.m_scissorBox = ctx.GetScissorBox();
+                break;
+            case F::GetStencilState:
+                dst.m_stencil[0] = ctx.GetStencilState(StencilFace::Front);
+                dst.m_stencil[1] = ctx.GetStencilState(StencilFace::Back);
+                break;
+            case F::GetTextureBindGeneration:
+                dst.m_textureBindGeneration = ctx.GetTextureBindGeneration();
+                break;
+            case F::GetTextureContextId:
+                dst.m_textureContextId = ctx.GetTextureContextId();
+                break;
+            case F::GetTextureUnitObject:
+                // Array<TextureUnit, MAX_TEXTURE_IMAGE_UNITS> (TextureState.h): unit 0 is the base.
+                dst.m_textureUnitBase = &ctx.GetTextureUnitObject(0);
+                break;
+            case F::GetTransformFeedbackCapturedVertices:
+                dst.m_transformFeedbackCapturedVertices = ctx.GetTransformFeedbackCapturedVertices();
+                break;
+            case F::GetTransformFeedbackGeneration:
+                dst.m_transformFeedbackGeneration = ctx.GetTransformFeedbackGeneration();
+                break;
+            case F::GetTransformFeedbackPausedPrimitiveCounter:
+                dst.m_transformFeedbackPausedPrimitiveCounter = ctx.GetTransformFeedbackPausedPrimitiveCounter();
+                break;
+            case F::GetTransformFeedbackProgram:
+                dst.m_transformFeedbackProgram = ctx.GetTransformFeedbackProgram();
+                break;
+            case F::GetViewport:
+                dst.m_viewport = ctx.GetViewport();
+                break;
+            case F::GetViewportIndexed:
+                for (Uint i = 0; i < PipeInputs::kMaxViewports; ++i) {
+                    dst.m_viewportIndexed[i] = ctx.GetViewportIndexed(i);
+                }
+                break;
+            case F::IsCapabilityEnabled:
+                // Every capability, FramebufferSrgb included: it copies today's constant false
+                // (MEASUREMENTS.md), so no value changes.
+                for (SizeT i = 0; i < PipeInputs::kCapabilityCount; ++i) {
+                    dst.m_capability[i] = ctx.IsCapabilityEnabled(static_cast<CapabilityInput>(i));
+                }
+                break;
+            case F::IsCapabilityEnabledIndexed:
+                // The only two indexed capabilities GLContext keeps (RenderState).
+                for (Uint i = 0; i < kMGMaxDrawBuffers; ++i) {
+                    dst.m_capabilityIndexed.Blend[i] = ctx.IsCapabilityEnabledIndexed(CapabilityInput::Blend, i);
+                }
+                for (Uint i = 0; i < PipeInputs::kMaxViewports; ++i) {
+                    dst.m_capabilityIndexed.ScissorTest[i] =
+                        ctx.IsCapabilityEnabledIndexed(CapabilityInput::ScissorTest, i);
+                }
+                break;
+            case F::IsTransformFeedbackActive:
+                dst.m_transformFeedbackActive = ctx.IsTransformFeedbackActive();
+                break;
+            case F::IsTransformFeedbackPaused:
+                dst.m_transformFeedbackPaused = ctx.IsTransformFeedbackPaused();
+                break;
+            case F::GetBoundTransformFeedbackLifetimeId:
+                dst.m_boundTransformFeedbackLifetimeId = ctx.GetBoundTransformFeedbackLifetimeId();
+                break;
+            // The seven forwarded fields: nothing to copy.
+            case F::GetBufferBindingPointCount:
+            case F::GetProgramObject:
+            case F::GetTextureObject:
+            case F::HasOpenTransformFeedbackSpan:
+            case F::InvalidateCompileEnv:
+            case F::ValidateProgramName:
+            case F::RecordError:
+            case F::kFieldCount:
+                break;
+            }
+        }
+
+        static void SetIdentity(PipeInputs& inputs, GLContext* ctx) {
+            inputs.m_live = ctx != nullptr;
+            inputs.m_contextIdentity = ctx;
+        }
+        static void SetVerb(PipeInputs& inputs, MGPipeVerb verb) { inputs.m_currentVerb = verb; }
+#if MOBILEGL_PIPE_POISON
+        static MGPipeFilledState& Filled(PipeInputs& inputs) { return inputs.m_filled; }
+#endif
+    };
+
+    namespace {
+        GLContext* LiveContext() { return MG_State::pGLContext.get(); }
+
+        template <class T>
+        const SharedPtr<T>& NullShared() {
+            static const SharedPtr<T> null;
+            return null;
+        }
+
+        [[noreturn]] void BadKnob(const char* knob, const char* value, const char* why) {
+            MGLOG_F("MGPipe: Fatal{PipeVerifyBadKnob, \"%s=%s\": %s}", knob, value, why);
+            std::abort();
+        }
+
+        // ---- MOBILEGL_PIPE_POISON_OMIT (negative control B, P1 brief D6) ----
+        // The filler skips the STAMP (never the value) of one (verb, field) pair: an omission
+        // indistinguishable from a forgotten FillPoints.def row, so that verb's read of the
+        // field is Fatal{UnmigratedPipeInput, "Field@Verb"} and no other verb is affected.
+        struct PoisonOmission {
+            Bool Armed = false;
+            MGPipeVerb Verb = MGPipeVerb::kVerbCount;
+            MGPipeInputField Field = MGPipeInputField::kFieldCount;
+        };
+        PoisonOmission g_omission;
+        Bool g_omissionKnobParsed = false;
+        String g_omissionKnobValue; // the value the last parse saw
+
+        // Parsed on the first fill and again only when the value changes. A lane loads
+        // Features once, before any fill, so that is one parse per process there; a forked
+        // test child that sets Features after its parent already filled gets its own parse,
+        // which is what puts the parser and its Fatal{PipeVerifyBadKnob} under a unit test.
+        // An empty value never clears an omission a test armed through MGPipeSetPoisonOmission.
+        void ParsePoisonOmissionKnob() {
+            const String& knob = MG_Config::Features.PipePoisonOmit;
+            if (g_omissionKnobParsed && knob == g_omissionKnobValue) return;
+            g_omissionKnobParsed = true;
+            g_omissionKnobValue = knob;
+            if (knob.empty()) return;
+            const auto colon = knob.find(':');
+            if (colon == String::npos || colon == 0 || colon + 1 >= knob.size()) {
+                BadKnob("MOBILEGL_PIPE_POISON_OMIT", knob.c_str(), "expected <Verb>:<FieldName>");
+            }
+            const String verbName = knob.substr(0, colon);
+            const String fieldName = knob.substr(colon + 1);
+            const auto verb = MGPipeFindVerb(verbName.c_str());
+            if (!verb) BadKnob("MOBILEGL_PIPE_POISON_OMIT", knob.c_str(), "no such verb in kMGPipeVerbNames");
+            const auto field = MGPipeFindInputField(fieldName.c_str());
+            if (!field) BadKnob("MOBILEGL_PIPE_POISON_OMIT", knob.c_str(), "no such field in kMGPipeInputFieldNames");
+            MGPipeSetPoisonOmission(verbName.c_str(), fieldName.c_str());
+        }
+
+        [[maybe_unused]] Bool IsOmitted(MGPipeVerb verb, MGPipeInputField field) {
+            return g_omission.Armed && g_omission.Verb == verb && g_omission.Field == field;
+        }
+
+#if MOBILEGL_PIPE_VERIFY
+        // ---- the MOBILEGL_PIPE_VERIFY comparator (P1 brief D8) ----
+        // Two mechanisms, both active only when Features.PipeVerify is set: the ENTRY compare
+        // once per verb (the pushed block against a second snapshot of the live context,
+        // taken at the same instant - tautological until P2 gives the first arm a real
+        // filler, and kept falsifiable by MOBILEGL_PIPE_VERIFY_CORRUPT), and the
+        // COMPARE-AT-READ in every accessor (the stored value against a fresh read of the
+        // live context at the moment the backend reads it - the arm that is real in P1: it
+        // catches a value that changed between the verb boundary and the read).
+        PipeInputs g_snapshot{};    // the second arm
+        PipeInputs g_readScratch{}; // where the compare-at-read re-read lands
+
+        // The read hook arms at the first fill (ArmVerify below), so it cannot see a read
+        // made before that. That window is covered by the poison instead: MGP_INPUT_CHECK
+        // precedes MGP_INPUT_VERIFY_READ in every accessor and a stamp of 0 is never fresh,
+        // so such a read is Fatal{UnmigratedPipeInput, "<Field>@<none>"} before the hook
+        // could matter - which holds only while a verify build always carries the poison.
+        static_assert(MOBILEGL_PIPE_POISON, "the compare-at-read hook relies on the poison for reads before the first fill");
+
+        struct VerifyState {
+            Bool Parsed = false;
+            Bool Enabled = false;
+            Bool Fatal = true;
+            Bool InHook = false; // a re-read that re-enters an accessor is not re-verified
+            Optional<MGPipeInputField> Corrupt;
+            String CorruptKnob; // the MOBILEGL_PIPE_VERIFY_CORRUPT value the last arm saw
+            std::atomic<Uint64> Divergences{0};
+            ~VerifyState() {
+                const Uint64 count = Divergences.load(std::memory_order_relaxed);
+                if (count != 0) {
+                    MGLOG_E("MGPipe: verify summary - %llu divergence(s) survived MOBILEGL_PIPE_VERIFY_FATAL=0",
+                            static_cast<unsigned long long>(count));
+                }
+            }
+        };
+        VerifyState g_verify;
+
+        // Armed on the first fill and re-armed when any of the three verify knobs'
+        // Features value (PipeVerify, PipeVerifyFatal, PipeVerifyCorrupt) differs from what
+        // the last arm latched (the same reason as ParsePoisonOmissionKnob: one arm per lane
+        // process, a fresh arm for a forked test child that turns a knob after its parent
+        // filled). Cost: two Bool compares and one String compare per fill, verify builds only.
+        void ArmVerify() {
+            const auto& features = MG_Config::Features;
+            if (g_verify.Parsed && g_verify.Enabled == features.PipeVerify && g_verify.Fatal == features.PipeVerifyFatal &&
+                g_verify.CorruptKnob == features.PipeVerifyCorrupt) {
+                return;
+            }
+            g_verify.Parsed = true;
+            g_verify.Enabled = features.PipeVerify;
+            g_verify.Fatal = features.PipeVerifyFatal;
+            g_verify.CorruptKnob = features.PipeVerifyCorrupt;
+            g_verify.Corrupt = Optional<MGPipeInputField>{};
+            if (!g_verify.Enabled) return;
+            const String& corrupt = g_verify.CorruptKnob;
+            if (!corrupt.empty()) {
+                const auto field = MGPipeFindInputField(corrupt.c_str());
+                if (!field) {
+                    BadKnob("MOBILEGL_PIPE_VERIFY_CORRUPT", corrupt.c_str(), "no such field in kMGPipeInputFieldNames");
+                }
+                g_verify.Corrupt = field;
+            }
+            // The lanes grep for this line: a verify run whose log lacks it never armed.
+            MGLOG_I("MGPipe: verify armed - %u fields, %u verbs, fatal=%d", static_cast<unsigned>(kMGPipeInputFieldCount),
+                    static_cast<unsigned>(kMGPipeVerbCount), g_verify.Fatal ? 1 : 0);
+            if (g_verify.Corrupt) {
+                MGLOG_I("MGPipe: verify corruption armed - %s", kMGPipeInputFieldNames[static_cast<SizeT>(*g_verify.Corrupt)]);
+            }
+        }
+
+        void ReportDivergence(MGPipeInputField field, const char* where) {
+            const Uint64 serial = MGPipeFillAccess::Filled(gPipeInputs).CurrentVerbSerial;
+            MGLOG_F("MGPipe: Fatal{PipeVerifyDiffer, \"%s@%s\", verb=%llu, where=%s}",
+                    kMGPipeInputFieldNames[static_cast<SizeT>(field)], MGPipeVerbName(gPipeInputs.CurrentVerb()),
+                    static_cast<unsigned long long>(serial), where);
+            if (g_verify.Fatal) std::abort();
+            g_verify.Divergences.fetch_add(1, std::memory_order_relaxed);
+        }
+
+        void EntryCompare(PipeInputs& inputs, const MGPipeFieldMask& mask) {
+            if (!g_verify.Enabled) return;
+            SnapshotFromGLContext(g_snapshot, mask);
+            // Negative control A: perturb the SNAPSHOT arm, so a green run goes red naming the
+            // field. A field outside this verb's mask is not compared and stays untouched.
+            if (g_verify.Corrupt && MGPipeFieldMaskHas(mask, *g_verify.Corrupt)) {
+                MGPipeApplyVerifyCorruption(g_snapshot, *g_verify.Corrupt);
+            }
+            MGPipeInputField differing = MGPipeInputField::kFieldCount;
+            if (!MGPipeVerifyInputs(inputs, g_snapshot, mask, &differing)) ReportDivergence(differing, "entry");
+        }
+#endif // MOBILEGL_PIPE_VERIFY
+    } // namespace
+
+#if MOBILEGL_PIPE_VERIFY
+    void SnapshotFromGLContext(PipeInputs& snapshot, const MGPipeFieldMask& mask) {
+        auto* ctx = LiveContext();
+        MGPipeFillAccess::SetIdentity(snapshot, ctx);
+        MGPipeFillAccess::SetVerb(snapshot, gPipeInputs.CurrentVerb());
+        if (ctx == nullptr) return;
+        for (SizeT i = 0; i < kMGPipeInputFieldCount; ++i) {
+            const auto field = static_cast<MGPipeInputField>(i);
+            if (!MGPipeFieldMaskHas(mask, field) || kMGPipeInputFieldSticky[i]) continue;
+            MGPipeFillAccess::CopyField(snapshot, *ctx, field);
+        }
+    }
+
+    void MGPipeVerifyReadHook(const PipeInputs& self, MGPipeInputField field, Uint index0, Uint index1) {
+        if (&self != &gPipeInputs || !g_verify.Enabled || g_verify.InHook) return;
+        const auto index = static_cast<SizeT>(field);
+        if (kMGPipeInputFieldSticky[index]) return;
+        auto* ctx = LiveContext();
+        if (ctx == nullptr) return;
+        // The whole field is re-read and compared - a superset of "the same indices", so a
+        // divergence in an index the backend did not ask for is still a divergence between
+        // the boundary value and the live value. The indices only decorate the report. The
+        // cost is per backend read (GetRenderStateParameters re-copies and compares the whole
+        // struct; GetProgramForDraw re-joins the pending link), inside the verify budget and
+        // to be kept in mind when reading the verify lane's wall time.
+        // InHook: the re-read calls the same GLContext accessor the filler calls, and
+        // GetProgramForDraw's join can re-enter a backend and with it another gPipeInputs
+        // accessor; that inner read is a plain load rather than a second hook, so the hook
+        // never recurses (and never reports the inner read against a half-copied scratch).
+        g_verify.InHook = true;
+        MGPipeFillAccess::CopyField(g_readScratch, *ctx, field);
+        const Bool equal = MGPipeInputsFieldEqual(field, self, g_readScratch);
+        g_verify.InHook = false;
+        if (equal) return;
+        MGLOG_E("MGPipe: verify read of %s (index %u, %u) differs from the live context", kMGPipeInputFieldNames[index],
+                index0, index1);
+        ReportDivergence(field, "read");
+    }
+#endif // MOBILEGL_PIPE_VERIFY
+
+    // ---- push on mutation (P1 lane finding F2) ----
+    // A backend that writes a frontend object inside its own verb moves a value the verb
+    // boundary already copied: Magma's ResolveSamplerDescriptor synthesises a fallback
+    // texture for an unbound sampler and its AllocateStorage/SetInternalFormat bump the
+    // context's sampling-resolution generation, so every read of that field after the
+    // fallback differs from the live context (the two SampledSetStaleness / six
+    // UnboundImageDescriptor entries the verify lane aborted on). The frontend mutator
+    // spells MGP_NOTE_MUTATION(Field) at the point of the move and lands here.
+    //
+    // Only the value is refreshed. The stamp is deliberately left alone: a field whose stamp
+    // this verb withheld (negative control B) must stay stale, and a field the verb never
+    // filled must stay Fatal{UnmigratedPipeInput} on the next read rather than be healed by
+    // an unrelated frontend write.
+    void MGPipeNoteFrontendMutation(MGPipeInputField field) {
+        PipeInputs& inputs = gPipeInputs;
+        auto* ctx = LiveContext();
+        if (ctx == nullptr) return;
+        const auto verb = inputs.CurrentVerb();
+        if (verb == MGPipeVerb::kVerbCount) return; // nothing has filled the block yet
+        const auto index = static_cast<SizeT>(field);
+        if (kMGPipeInputFieldSticky[index]) return; // forwarded: no storage to refresh
+        const MGPipeFieldMask& mask =
+            kMGPipeClassFieldMask[static_cast<SizeT>(kMGPipeVerbClass[static_cast<SizeT>(verb)])];
+        if (!MGPipeFieldMaskHas(mask, field)) return; // this verb never pushed it
+        MGPipeFillAccess::CopyField(inputs, *ctx, field);
+    }
+
+    void MGPipeSetPoisonOmission(const char* verb, const char* field) {
+        if (verb == nullptr || field == nullptr) {
+            g_omission = PoisonOmission{};
+            return;
+        }
+        const auto v = MGPipeFindVerb(verb);
+        const auto f = MGPipeFindInputField(field);
+        if (!v || !f) BadKnob("MOBILEGL_PIPE_POISON_OMIT", verb, "unknown verb or field");
+        g_omission.Armed = true;
+        g_omission.Verb = *v;
+        g_omission.Field = *f;
+#if MOBILEGL_PIPE_POISON
+        MGLOG_I("MGPipe: poison omission armed - %s@%s", field, verb);
+#else
+        MGLOG_W_ONCE("MGPipe: poison omission %s@%s requested but the poison is not compiled in "
+                     "(MOBILEGL_PIPE_POISON=0): no stamp exists to omit",
+                     field, verb);
+#endif
+    }
+
+    // ---- liveness ----
+    Bool PipeInputs::IsLive() const { return LiveContext() != nullptr; }
+
+    // ---- the seven F-class forwarders ----
+    SizeT PipeInputs::GetBufferBindingPointCount(BufferTarget target) const {
+        const auto* ctx = LiveContext();
+        return ctx != nullptr ? ctx->GetBufferBindingPointCount(target) : 0;
+    }
+
+    const SharedPtr<PipeInputs::ProgramObject>& PipeInputs::GetProgramObject(Uint index) {
+        auto* ctx = LiveContext();
+        return ctx != nullptr ? ctx->GetProgramObject(index) : NullShared<ProgramObject>();
+    }
+
+    const SharedPtr<PipeInputs::ITextureObject>& PipeInputs::GetTextureObject(Uint index) {
+        auto* ctx = LiveContext();
+        return ctx != nullptr ? ctx->GetTextureObject(index) : NullShared<ITextureObject>();
+    }
+
+    Bool PipeInputs::HasOpenTransformFeedbackSpan(Uint64 lifetimeId) const {
+        const auto* ctx = LiveContext();
+        return ctx != nullptr && ctx->HasOpenTransformFeedbackSpan(lifetimeId);
+    }
+
+    void PipeInputs::InvalidateCompileEnv() {
+        if (auto* ctx = LiveContext()) ctx->InvalidateCompileEnv();
+    }
+
+    Bool PipeInputs::ValidateProgramName(Uint index) const {
+        const auto* ctx = LiveContext();
+        return ctx != nullptr && ctx->ValidateProgramName(index);
+    }
+
+    void PipeInputs::RecordError(ErrorCode code, UniquePtr<ErrorInfo> info) {
+        auto* ctx = LiveContext();
+        if (ctx == nullptr) {
+            MGLOG_E_ONCE("PipeInputs::RecordError: no live context, dropping error %d", static_cast<int>(code));
+            return;
+        }
+        ctx->RecordError(code, Move(info));
+    }
+
+    void MGPipeLeaveVerb() {
+        PipeInputs& inputs = gPipeInputs;
+#if MOBILEGL_PIPE_POISON
+        // Same bump the next fill would make, without a verb to fill from: no field is
+        // stamped, so every stamp this verb made falls behind the serial.
+        ++MGPipeFillAccess::Filled(inputs).CurrentVerbSerial;
+#endif
+        MGPipeFillAccess::SetVerb(inputs, MGPipeVerb::kVerbCount);
+    }
+
+    // ---- the filler ----
+    void MGPipeFillForVerb(MGPipeVerb verb) {
+        PipeInputs& inputs = gPipeInputs;
+        ParsePoisonOmissionKnob();
+#if MOBILEGL_PIPE_VERIFY
+        ArmVerify();
+#else
+        // The runtime knob without the compiled comparator is a no-op that would look green;
+        // this warning is what a lane's arming assertion turns into red.
+        if (MG_Config::Features.PipeVerify) {
+            MGLOG_W_ONCE("MGPipe: MOBILEGL_PIPE_VERIFY=1 requested but the comparator is not compiled in "
+                         "(configure with -DMOBILEGL_PIPE_VERIFY=ON)");
+        }
+#endif
+#if MOBILEGL_PIPE_POISON
+        MGPipeFilledState& filled = MGPipeFillAccess::Filled(inputs);
+        // Starts at 1: FilledGen == 0 is "never filled", and MGPipeInputFieldIsFresh refuses
+        // it on both branches, so a read before this first bump is
+        // Fatal{UnmigratedPipeInput, "<Field>@<none>"} rather than default storage.
+        ++filled.CurrentVerbSerial;
+#endif
+        MGPipeFillAccess::SetVerb(inputs, verb);
+        auto* ctx = LiveContext();
+        MGPipeFillAccess::SetIdentity(inputs, ctx);
+        if (ctx == nullptr) return;
+        const MGPipeFieldMask& mask = kMGPipeClassFieldMask[static_cast<SizeT>(kMGPipeVerbClass[static_cast<SizeT>(verb)])];
+        for (SizeT i = 0; i < kMGPipeInputFieldCount; ++i) {
+            const auto field = static_cast<MGPipeInputField>(i);
+            if (!MGPipeFieldMaskHas(mask, field)) continue;
+#if MOBILEGL_PIPE_POISON
+            if (kMGPipeInputFieldSticky[i]) {
+                // Stamped once by the first fill that sees a live context; fresh through the
+                // Sticky -> FilledGen != 0 branch of MGPipeInputFieldIsFresh from then on.
+                if (filled.FilledGen[i] == 0) filled.FilledGen[i] = 1;
+                continue;
+            }
+#else
+            if (kMGPipeInputFieldSticky[i]) continue;
+#endif
+            MGPipeFillAccess::CopyField(inputs, *ctx, field);
+#if MOBILEGL_PIPE_POISON
+            // The value is copied either way; only the stamp is withheld for the omitted pair.
+            if (!IsOmitted(verb, field)) filled.FilledGen[i] = filled.CurrentVerbSerial;
+#endif
+        }
+#if MOBILEGL_PIPE_VERIFY
+        EntryCompare(inputs, mask);
+#endif
+    }
+} // namespace MobileGL::MG_Pipe

MobileGL/MG_Impl/Pipe/PipeFill.h

@@ -0,0 +1,55 @@
+// MobileGL - MobileGL/MG_Impl/Pipe/PipeFill.h
+// Copyright (c) 2025-2026 MobileGL-Dev
+// Licensed under the GNU Lesser General Public License v3.0:
+//   https://www.gnu.org/licenses/gpl-3.0.txt
+//   https://www.gnu.org/licenses/lgpl-3.0.txt
+// SPDX-License-Identifier: LGPL-3.0-only
+// End of Source File Header
+
+#pragma once
+// The fill point (ARCHITECTURE.md 9.2, P1 brief D7). MG_Impl spells MGP_FILL(Verb); as the
+// statement immediately before every call through gBackendFunctionsTable.GL - after every
+// early return the call is behind, inside the loop body for a call made in a loop - so the
+// frontend fills the PipeInputs block for exactly the verbs that reach a backend. In the
+// pull build the macro is ((void)0) and the pull build is byte-identical to a tree without
+// it.
+#if MOBILEGL_PIPE_PUSH
+#include <MG_Pipe/MGPipe.h>
+namespace MobileGL::MG_Pipe {
+    struct PipeInputs;
+
+    // PipeFill.cpp. Bumps the per-verb serial, records the verb and the context identity,
+    // and copies every field in the verb class's may-read mask (kMGPipeClassFieldMask) out
+    // of the live GLContext, stamping each with the new serial. In a verify build it then
+    // runs the entry compare against a second snapshot (P1 brief D8).
+    void MGPipeFillForVerb(MGPipeVerb verb);
+
+    // Ends the verb in flight without starting another: bumps the serial, so every field the
+    // verb stamped goes stale, and puts the current verb back to "none", so a read made after
+    // it aborts as Fatal{UnmigratedPipeInput, "<Field>@<none>"} - which is what such a read
+    // is - instead of naming whichever verb happened to be filled last. Nothing in the GL
+    // entry points calls this: a real verb is always followed by the next verb's fill. It
+    // exists for a caller that drives a backend helper directly and wants its declaration to
+    // stop where it says it stops (MG_Test/ScopedPipeVerb.h).
+    void MGPipeLeaveVerb();
+
+    // PipeFill.cpp. Negative control B (P1 brief D6): the filler withholds the STAMP - never
+    // the value - of `field` at `verb`, so that verb's read of it is
+    // Fatal{UnmigratedPipeInput, "Field@Verb"} while every other verb is unaffected. The
+    // MOBILEGL_PIPE_POISON_OMIT knob ("<Verb>:<FieldName>") calls this once, on the first
+    // fill; tests call it directly. Both null clears the omission. An unknown name is
+    // Fatal{PipeVerifyBadKnob}.
+    void MGPipeSetPoisonOmission(const char* verb, const char* field);
+
+#if MOBILEGL_PIPE_VERIFY
+    // PipeFill.cpp. The second arm of the comparator (P1 brief D8, ARCHITECTURE.md 13.2-2):
+    // fills `snapshot` from the live GLContext the old way, for every field in `mask`. This
+    // is the branch that survives P13, which is why it is its own function rather than the
+    // filler's loop.
+    void SnapshotFromGLContext(PipeInputs& snapshot, const MGPipeFieldMask& mask);
+#endif
+} // namespace MobileGL::MG_Pipe
+#define MGP_FILL(Verb) ::MobileGL::MG_Pipe::MGPipeFillForVerb(::MobileGL::MG_Pipe::MGPipeVerb::Verb)
+#else
+#define MGP_FILL(Verb) ((void)0)
+#endif

MobileGL/MG_IntegrationTest/CMakeLists.txt

@@ -24,9 +24,14 @@ set(CMAKE_CXX_STANDARD_REQUIRED ON)
 
 set(MGL_ITEST_ROOT ${CMAKE_CURRENT_LIST_DIR}/../..)
 
-# Only meaningful where MobileGL_s exists (i.e. not Android).
-if (NOT TARGET MobileGL_s)
-    message(STATUS "MobileGL_s is not available; skipping the integration test module")
+# Desktop links the static implementation directly. Android runs the same
+# executable from adb shell and links the shipping shared library instead.
+if (ANDROID)
+    set(MGL_ITEST_MOBILEGL_TARGET MobileGL)
+elseif (TARGET MobileGL_s)
+    set(MGL_ITEST_MOBILEGL_TARGET MobileGL_s)
+else()
+    message(STATUS "No MobileGL library target is available; skipping the integration test module")
     return()
 endif()
 
@@ -46,6 +51,7 @@ endif()
 add_executable(MobileGLIntegrationTest
     Main.cpp
     Harness/HeadlessGL.cpp
+    Harness/BackendCapsPeek.cpp
     Scenarios/OrientationScenario.cpp
     Scenarios/CrossFrameBufferScenario.cpp
     Scenarios/ResidentIndexScenario.cpp
@@ -53,12 +59,18 @@ add_executable(MobileGLIntegrationTest
     Scenarios/DrawParametersScenario.cpp
     Scenarios/AsyncCompileScenario.cpp
     Scenarios/XfbAfterClipDistanceScenario.cpp
+    Scenarios/UnwrittenPositionOutputScenario.cpp
+    Scenarios/SampleMaskScopeScenario.cpp
+    Scenarios/SampledSetStalenessScenario.cpp
     Scenarios/ThreeChannelAttachmentScenario.cpp
+    Scenarios/SnormAttachmentScenario.cpp
     Scenarios/PipelineFailureScenario.cpp
     Scenarios/AdvertisedLimitsScenario.cpp
     Scenarios/PixelStoreSweepScenario.cpp
+    Scenarios/PrimitiveRestartScenario.cpp
     Scenarios/FragCoordOriginScenario.cpp
     Scenarios/ClearThenReadPixelsScenario.cpp
+    Scenarios/SampleVariablesScenario.cpp
     Scenarios/DepthStencilReadbackScenario.cpp
     Scenarios/DepthStencilReadbackMatrixScenario.cpp
     Scenarios/DepthStencilReadbackAttachmentShapeScenario.cpp
@@ -68,20 +80,55 @@ add_executable(MobileGLIntegrationTest
     Scenarios/DoublePrecisionScenario.cpp
     Scenarios/UniformInitializerScenario.cpp
     Scenarios/SwizzleAccessRoutineScenario.cpp
+    Scenarios/IterationRPFirstReductionScenario.cpp
+    Scenarios/IterationRPProgram203Scenario.cpp
+    Scenarios/IterationRPScratchFixScenario.cpp
     Scenarios/ProgramPipelineScenario.cpp
     Scenarios/ImageLoadStoreSsoScenario.cpp
     Scenarios/ImageTargetKindScenario.cpp
     Scenarios/ImageFormatQualifierScenario.cpp
+    Scenarios/NonCoreImageFormatScenario.cpp
+    Scenarios/ImageSizeAfterRespecScenario.cpp
     Scenarios/SsboDeclarationFormScenario.cpp
     Scenarios/Glsl420DeclarationScenario.cpp
+    Scenarios/IoBlockNameCollisionScenario.cpp
+    Scenarios/UnlocatedIoBlockScenario.cpp
+    Scenarios/TessellationDrawModeScenario.cpp
+    Scenarios/GeometryDrawModeScenario.cpp
+    Scenarios/PostLinkAttachScenario.cpp
+    Scenarios/FormatlessImageBakeScenario.cpp
     Scenarios/FragmentOutputArrayIndexScenario.cpp
     Scenarios/BufferTextureScenario.cpp
     Scenarios/VertexAttribBindingScenario.cpp
     Scenarios/XfbCaptureBufferReuseScenario.cpp
+    Scenarios/XfbPrimitiveQueryScenario.cpp
+    Scenarios/PrimitivesGeneratedNoXfbScenario.cpp
+    Scenarios/XfbRepeatedCaptureScenario.cpp
+    Scenarios/TessellationXfbCaptureScenario.cpp
+    Scenarios/PointSizeDemotionScenario.cpp
     Scenarios/VertexArrayEnableDisableScenario.cpp
     Scenarios/CopyImageLevelRangeScenario.cpp
     Scenarios/CopyImageLayeredScenario.cpp
+    Scenarios/CopyImagePacked16Scenario.cpp
+    Scenarios/TextureViewScenario.cpp
+    Scenarios/PackedWordReadbackScenario.cpp
     Scenarios/LayeredAttachmentBarrierScenario.cpp
+    Scenarios/LayeredAttachmentShapeScenario.cpp
+    Scenarios/LayeredTextureReadbackScenario.cpp
+    Scenarios/AtomicCounterScenario.cpp
+    Scenarios/LargeArenaAdoptionScenario.cpp
+    Scenarios/SsboArrayDynamicIndexScenario.cpp
+    Scenarios/StorageBufferRegrowScenario.cpp
+    Scenarios/SpirvShaderBinaryScenario.cpp
+    Scenarios/RelinkStageSetScenario.cpp
+    Scenarios/GuiBatchScenario.cpp
+    Scenarios/UnboundImageDescriptorScenario.cpp
+    Scenarios/IntegerBorderColorScenario.cpp
+    Scenarios/ClearTexImageUndefinedLevelZeroScenario.cpp
+    Scenarios/RenderbufferBlendFormatScenario.cpp
+    Scenarios/DualSourceBlendScenario.cpp
+    Scenarios/PipeVerifyArmingScenario.cpp
+    Scenarios/PoisonOmissionScenario.cpp
 )
 
 target_include_directories(MobileGLIntegrationTest PRIVATE
@@ -92,9 +139,20 @@ target_include_directories(MobileGLIntegrationTest PRIVATE
 # gtest, not gtest_main: Main.cpp installs the harness banner itself.
 target_link_libraries(MobileGLIntegrationTest PRIVATE
     GTest::gtest
-    MobileGL_s
+    ${MGL_ITEST_MOBILEGL_TARGET}
 )
 
+if (ANDROID)
+    find_library(MGL_ITEST_ANDROID_LIBRARY android REQUIRED)
+    find_library(MGL_ITEST_LOG_LIBRARY log REQUIRED)
+    find_library(MGL_ITEST_MEDIANDK_LIBRARY mediandk REQUIRED)
+    target_link_libraries(MobileGLIntegrationTest PRIVATE
+        ${MGL_ITEST_ANDROID_LIBRARY}
+        ${MGL_ITEST_LOG_LIBRARY}
+        ${MGL_ITEST_MEDIANDK_LIBRARY}
+    )
+endif()
+
 if (MSVC)
     # Same reason as MG_Test/Backend/DirectVulkan: the GLES headers declare gl*
     # as dllimport on Windows, so the in-library GL entry-point definitions only
@@ -103,6 +161,10 @@ if (MSVC)
 endif()
 target_compile_definitions(MobileGLIntegrationTest PRIVATE -DNOMINMAX)
 
+if (ANDROID)
+    return()
+endif()
+
 # --- ctest wiring --------------------------------------------------------
 # A bare libEGL on a glvnd box resolves to whatever vendor comes first, which is
 # usually Mesa/llvmpipe - a software rasteriser silently replacing the GPU under
@@ -223,6 +285,19 @@ endif()
 set(MGL_ITEST_VULKAN_ENV ${MGL_ITEST_COMMON_ENV})
 if (MOBILEGL_ITEST_VK_ICD)
     list(APPEND MGL_ITEST_VULKAN_ENV "VK_ICD_FILENAMES=${MOBILEGL_ITEST_VK_ICD}")
+    # The three iterationRP repairs are tri-state quirks that default to device
+    # auto-detection, and lavapipe is not on any auto list - so on lavapipe the
+    # iterationRP scenarios run unrepaired and Program 203 misses its golden
+    # output. CI's integration-gpu job exports these three by hand; pinning them
+    # to the ICD instead means a local `ctest -L integration-gpu` measures the
+    # same thing the gate does, with no environment to remember.
+    if (MOBILEGL_ITEST_VK_ICD MATCHES "lvp_icd|lavapipe")
+        message(STATUS "Integration tests: lavapipe ICD - forcing the iterationRP repairs on")
+        list(APPEND MGL_ITEST_VULKAN_ENV
+            "MOBILEGL_MAGMA_FIX_ITERATIONRP_SUBGROUP_SCRATCH=1"
+            "MOBILEGL_MAGMA_DERIVE_NUM_SUBGROUPS=1"
+            "MOBILEGL_MAGMA_ITERATIONRP_FIX_BARRIER=1")
+    endif()
 endif()
 
 # The ENVIRONMENT test property is itself a `;`-list, and gtest_discover_tests
@@ -252,6 +327,70 @@ mgl_itest_join_environment(MGL_ITEST_VULKAN_ASYNC_ENVIRONMENT
 mgl_itest_join_environment(MGL_ITEST_GLES_FORCED_DS_ENVIRONMENT
     "MOBILEGL_BACKEND_TYPE=DirectGLES" "MOBILEGL_ESPRYT_FORCE_DS_READBACK_EMULATION=1" ${MGL_ITEST_COMMON_ENV})
 
+# The shader-compiler configurations AsyncCompileScenario needs, and the one
+# ViewportArrayScenario's negative control needs.
+#
+# These used to be poked into MG_Config::Features from inside the test bodies. They
+# cannot be any more - on Android this module links the SHIPPING libMobileGL.so, which
+# exports nothing internal - and they should not have been anyway: half of what each of
+# them decides is latched before the first GL call (the compile pool and its threads;
+# the advertised extension list, which a backend builds once from the configuration in
+# force at its first use), so an in-process write could only ever have moved the other
+# half. Every one of them is a whole-process property, and a whole-process property is
+# spelled with an environment variable and a ctest entry of its own.
+#
+# Note the shape of every list here: it APPENDS to MGL_ITEST_COMMON_ENV /
+# MGL_ITEST_VULKAN_ENV rather than standing alone. A ctest ENVIRONMENT property REPLACES
+# the job environment rather than adding to it, so an entry that lists only its mode
+# variable would silently lose the EGL vendor and Vulkan ICD pinning and run against
+# whatever the loader found first.
+mgl_itest_join_environment(MGL_ITEST_GLES_ASYNC_ON_ENVIRONMENT
+    "MOBILEGL_BACKEND_TYPE=DirectGLES" "MOBILEGL_ASYNC_SHADER_COMPILE=1" ${MGL_ITEST_COMMON_ENV})
+mgl_itest_join_environment(MGL_ITEST_GLES_ASYNC_OFF_ENVIRONMENT
+    "MOBILEGL_BACKEND_TYPE=DirectGLES" "MOBILEGL_ASYNC_SHADER_COMPILE=0" ${MGL_ITEST_COMMON_ENV})
+mgl_itest_join_environment(MGL_ITEST_VULKAN_ASYNC_ON_ENVIRONMENT
+    "MOBILEGL_BACKEND_TYPE=DirectVulkan" "MOBILEGL_ASYNC_SHADER_COMPILE=1" ${MGL_ITEST_VULKAN_ENV})
+mgl_itest_join_environment(MGL_ITEST_VULKAN_ASYNC_OFF_ENVIRONMENT
+    "MOBILEGL_BACKEND_TYPE=DirectVulkan" "MOBILEGL_ASYNC_SHADER_COMPILE=0" ${MGL_ITEST_VULKAN_ENV})
+mgl_itest_join_environment(MGL_ITEST_GLES_OPTIMISTIC_ENVIRONMENT
+    "MOBILEGL_BACKEND_TYPE=DirectGLES" "MOBILEGL_ASYNC_SHADER_COMPILE=1"
+    "MOBILEGL_ASYNC_OPTIMISTIC_SHADER_STATUS=1" ${MGL_ITEST_COMMON_ENV})
+mgl_itest_join_environment(MGL_ITEST_VULKAN_OPTIMISTIC_ENVIRONMENT
+    "MOBILEGL_BACKEND_TYPE=DirectVulkan" "MOBILEGL_ASYNC_SHADER_COMPILE=1"
+    "MOBILEGL_ASYNC_OPTIMISTIC_SHADER_STATUS=1" ${MGL_ITEST_VULKAN_ENV})
+mgl_itest_join_environment(MGL_ITEST_GLES_NO_VIEWPORT_EMULATION_ENVIRONMENT
+    "MOBILEGL_BACKEND_TYPE=DirectGLES" "MOBILEGL_ESPRYT_FORCE_VIEWPORT_ARRAY_EMULATION=0" ${MGL_ITEST_COMMON_ENV})
+# MOBILEGL_LOG_FILE_PATH alongside the pin, because the arming assertion needs somewhere to
+# read the library's own report from. The strip's arming signal is a latched MGLOG_I and there
+# is no other way for a test process to learn that it fired - MG_Config is not reachable from
+# this module on Android, where it links the shipping library. The path is per-lane so nothing
+# else appends to it, and the case only trusts the bytes written after it started.
+mgl_itest_join_environment(MGL_ITEST_GLES_UNLOCATED_IO_BLOCKS_ENVIRONMENT
+    "MOBILEGL_BACKEND_TYPE=DirectGLES" "MOBILEGL_ESPRYT_UNLOCATED_IO_BLOCKS=1"
+    "MOBILEGL_LOG_FILE_PATH=${CMAKE_CURRENT_BINARY_DIR}/unlocated-io-blocks.log"
+    ${MGL_ITEST_COMMON_ENV})
+mgl_itest_join_environment(MGL_ITEST_GLES_WIDENED_PACKED16_ENVIRONMENT
+    "MOBILEGL_BACKEND_TYPE=DirectGLES" "MOBILEGL_ESPRYT_WIDEN_PACKED16_STORAGE=1" ${MGL_ITEST_COMMON_ENV})
+# Same shape as the UnlocatedIoBlocks entry: the log path is where the reroute's latched
+# MGLOG_I lands, and the arming case only trusts the bytes written after it started.
+mgl_itest_join_environment(MGL_ITEST_VULKAN_PRIMGEN_REROUTE_ENVIRONMENT
+    "MOBILEGL_BACKEND_TYPE=DirectVulkan" "MOBILEGL_MAGMA_PRIMGEN_QUERY_REROUTE=1"
+    "MOBILEGL_LOG_FILE_PATH=${CMAKE_CURRENT_BINARY_DIR}/primgen-query-reroute.log"
+    ${MGL_ITEST_VULKAN_ENV})
+# The point-size demotion pinned on, per backend, with a per-lane log file for the arming
+# assertion - the same MOBILEGL_LOG_FILE_PATH reasoning as the UnlocatedIoBlocks lane above.
+# Two lanes because the demotion runs in the SHARED phase-B chain and each backend then
+# consumes it differently (Espryt respells the driver-side capture request, Magma binds the
+# SPIR-V Xfb decorations to the carrier).
+mgl_itest_join_environment(MGL_ITEST_GLES_POINT_SIZE_DEMOTION_ENVIRONMENT
+    "MOBILEGL_BACKEND_TYPE=DirectGLES" "MOBILEGL_POINT_SIZE_DEMOTION=1"
+    "MOBILEGL_LOG_FILE_PATH=${CMAKE_CURRENT_BINARY_DIR}/point-size-demotion-gles.log"
+    ${MGL_ITEST_COMMON_ENV})
+mgl_itest_join_environment(MGL_ITEST_VULKAN_POINT_SIZE_DEMOTION_ENVIRONMENT
+    "MOBILEGL_BACKEND_TYPE=DirectVulkan" "MOBILEGL_POINT_SIZE_DEMOTION=1"
+    "MOBILEGL_LOG_FILE_PATH=${CMAKE_CURRENT_BINARY_DIR}/point-size-demotion-vulkan.log"
+    ${MGL_ITEST_VULKAN_ENV})
+
 # TIMEOUT on every entry: a GPU test that wedges must fail the run, not hang it.
 set(MGL_ITEST_TIMEOUT 120)
 
@@ -316,3 +455,359 @@ gtest_discover_tests(MobileGLIntegrationTest
         TIMEOUT ${MGL_ITEST_TIMEOUT}
         ENVIRONMENT "${MGL_ITEST_GLES_FORCED_DS_ENVIRONMENT}"
 )
+
+# UnlocatedIoBlockScenario with the interface-block location strip PINNED ON, for the same
+# reason the depth/stencil entry above pins its emulation: without it this scenario is
+# UNFALSIFIABLE on the machines this suite runs on. llvmpipe carries a located interface block
+# correctly, so the driver POST that arms the strip on Mali answers "healthy" here and the
+# emulation never runs - the ambient registration would be exercising the un-stripped path
+# twice and calling it coverage. With the variable set, the blocks really are emitted with no
+# location and the assertion is about the spelling the device gets.
+gtest_discover_tests(MobileGLIntegrationTest
+    TEST_PREFIX "DirectGLES.UnlocatedIoBlocks."
+    TEST_FILTER "UnlocatedIoBlockScenario.*"
+    DISCOVERY_TIMEOUT 30
+    PROPERTIES
+        LABELS integration-gpu
+        TIMEOUT ${MGL_ITEST_TIMEOUT}
+        ENVIRONMENT "${MGL_ITEST_GLES_UNLOCATED_IO_BLOCKS_ENVIRONMENT}"
+)
+
+# AsyncCompileScenario, with asynchronous compilation PINNED ON per backend.
+#
+# Not a duplicate of what the two ambient registrations already run: they run whatever
+# MobileGL's built-in default happens to be, and the day that default flips they would
+# stop covering the asynchronous path without anything going red. These entries are the
+# ones that keep the asynchronous half tested no matter what ships. They are also the
+# only place ExtensionStringMatchesTheConfiguration can assert that the extension IS
+# advertised - the case derives its expectation from this variable and nothing else, and
+# skips where it is unset, precisely so that it is not asserting the implementation
+# against itself.
+gtest_discover_tests(MobileGLIntegrationTest
+    TEST_PREFIX "DirectGLES.AsyncOn."
+    TEST_FILTER "AsyncCompileScenario.*"
+    DISCOVERY_TIMEOUT 30
+    PROPERTIES
+        LABELS integration-gpu
+        TIMEOUT ${MGL_ITEST_TIMEOUT}
+        ENVIRONMENT "${MGL_ITEST_GLES_ASYNC_ON_ENVIRONMENT}"
+)
+gtest_discover_tests(MobileGLIntegrationTest
+    TEST_PREFIX "DirectVulkan.AsyncOn."
+    TEST_FILTER "AsyncCompileScenario.*"
+    DISCOVERY_TIMEOUT 30
+    PROPERTIES
+        LABELS integration-gpu
+        TIMEOUT ${MGL_ITEST_TIMEOUT}
+        ENVIRONMENT "${MGL_ITEST_VULKAN_ASYNC_ON_ENVIRONMENT}"
+)
+
+# The other side of the same switch: asynchronous compilation OFF, so
+# GL_KHR_parallel_shader_compile must be WITHDRAWN from both spellings of the extension
+# list and GL_MAX_SHADER_COMPILER_THREADS_KHR must read 0. Only that one case is
+# registered here because it is the only one that has anything to say in this
+# configuration - the other four exist to observe worker-built artifacts, and there are
+# none - so registering the whole scenario would buy four guaranteed skips per backend.
+# Together with the AsyncOn. entries above, one ctest run still covers both flag states,
+# which is what the in-process forcing used to be for.
+gtest_discover_tests(MobileGLIntegrationTest
+    TEST_PREFIX "DirectGLES.AsyncOff."
+    TEST_FILTER "AsyncCompileScenario.ExtensionStringMatchesTheConfiguration"
+    DISCOVERY_TIMEOUT 30
+    PROPERTIES
+        LABELS integration-gpu
+        TIMEOUT ${MGL_ITEST_TIMEOUT}
+        ENVIRONMENT "${MGL_ITEST_GLES_ASYNC_OFF_ENVIRONMENT}"
+)
+gtest_discover_tests(MobileGLIntegrationTest
+    TEST_PREFIX "DirectVulkan.AsyncOff."
+    TEST_FILTER "AsyncCompileScenario.ExtensionStringMatchesTheConfiguration"
+    DISCOVERY_TIMEOUT 30
+    PROPERTIES
+        LABELS integration-gpu
+        TIMEOUT ${MGL_ITEST_TIMEOUT}
+        ENVIRONMENT "${MGL_ITEST_VULKAN_ASYNC_OFF_ENVIRONMENT}"
+)
+
+# The optimistic-status quirk's end-to-end shape. Its own entries and not part of the
+# AsyncOn. ones because the quirk is not neutral for the rest of the scenario: with it in
+# force glGetShaderiv(GL_COMPILE_STATUS) deliberately answers without joining, which is
+# exactly what CompletionStatusPollingThenForcedJoin asserts must NOT happen. Off by
+# default and never advertised, so - unlike asynchronous compilation, which announces
+# itself through the extension string - the variable is the only thing that can tell the
+# case it is in force.
+gtest_discover_tests(MobileGLIntegrationTest
+    TEST_PREFIX "DirectGLES.OptimisticShaderStatus."
+    TEST_FILTER "AsyncCompileScenario.IrisShapedTwoPhaseBatchRendersCorrectly"
+    DISCOVERY_TIMEOUT 30
+    PROPERTIES
+        LABELS integration-gpu
+        TIMEOUT ${MGL_ITEST_TIMEOUT}
+        ENVIRONMENT "${MGL_ITEST_GLES_OPTIMISTIC_ENVIRONMENT}"
+)
+gtest_discover_tests(MobileGLIntegrationTest
+    TEST_PREFIX "DirectVulkan.OptimisticShaderStatus."
+    TEST_FILTER "AsyncCompileScenario.IrisShapedTwoPhaseBatchRendersCorrectly"
+    DISCOVERY_TIMEOUT 30
+    PROPERTIES
+        LABELS integration-gpu
+        TIMEOUT ${MGL_ITEST_TIMEOUT}
+        ENVIRONMENT "${MGL_ITEST_VULKAN_OPTIMISTIC_ENVIRONMENT}"
+)
+
+# The negative control for the DirectGLES gl_ViewportIndex emulation, in a process that
+# has it switched off. One case, because it is the only one the switch may touch: with
+# the emulation off the three positive cases in the same fixture describe behaviour the
+# backend does not have, so a whole-scenario registration would be three guaranteed reds.
+# DirectGLES only - the flag steers nothing on DirectVulkan, which routes natively.
+gtest_discover_tests(MobileGLIntegrationTest
+    TEST_PREFIX "DirectGLES.NoViewportArrayEmulation."
+    TEST_FILTER "ViewportArrayScenario.WithoutTheEmulationEveryIndexCollapsesOntoViewportZero"
+    DISCOVERY_TIMEOUT 30
+    PROPERTIES
+        LABELS integration-gpu
+        TIMEOUT ${MGL_ITEST_TIMEOUT}
+        ENVIRONMENT "${MGL_ITEST_GLES_NO_VIEWPORT_EMULATION_ENVIRONMENT}"
+)
+
+# PrimitivesGeneratedNoXfbScenario again, with the GL_PRIMITIVES_GENERATED statistics
+# reroute PINNED ON. The ambient DirectVulkan registration runs the same cases under the
+# bring-up probe's Auto verdict, so between the two entries both accounting paths answer
+# the same GL questions and must produce the same numbers - the "two pools must agree"
+# gate this machine can hold that the affected device cannot. The pinned entry is also
+# the only one whose arming case runs: it asserts the renderer's latched MGLOG_I, so a
+# silently-disarmed reroute (an inverted override mapping, a lost gate) fails here
+# instead of leaving every equality case vacuously green. DirectVulkan only - the flag
+# steers nothing on DirectGLES.
+gtest_discover_tests(MobileGLIntegrationTest
+    TEST_PREFIX "DirectVulkan.PrimGenReroute."
+    TEST_FILTER "PrimitivesGeneratedNoXfbScenario.*"
+    DISCOVERY_TIMEOUT 30
+    PROPERTIES
+        LABELS integration-gpu
+        TIMEOUT ${MGL_ITEST_TIMEOUT}
+        ENVIRONMENT "${MGL_ITEST_VULKAN_PRIMGEN_REROUTE_ENVIRONMENT}"
+)
+
+# The packed16 copy scenarios again, with the 8-bit storage widening PINNED ON. The ambient
+# registrations above cover the narrow storage - on every CI driver the widening's POST
+# probe finds no field-order mirror, so Auto keeps the native 16-bit path - which means the
+# storage every AFFECTED device will actually run would otherwise execute nowhere at all:
+# no CI driver has the Mali bug that arms it. This lane is what proves the widened storage
+# is client-invisible (same packed words in and out on every leg the 18 failing CTS bodies
+# used, the renderbuffer one included). DirectGLES only - the flag steers nothing on
+# DirectVulkan, which has always stored these formats widened.
+gtest_discover_tests(MobileGLIntegrationTest
+    TEST_PREFIX "DirectGLES.WidenedPacked16."
+    TEST_FILTER "CopyImagePacked16Scenario.*"
+    DISCOVERY_TIMEOUT 30
+    PROPERTIES
+        LABELS integration-gpu
+        TIMEOUT ${MGL_ITEST_TIMEOUT}
+        ENVIRONMENT "${MGL_ITEST_GLES_WIDENED_PACKED16_ENVIRONMENT}"
+)
+
+# PointSizeDemotionScenario with the demotion PINNED ON, per backend, for the reason every
+# pinned lane above exists: llvmpipe and lavapipe both HOST gl_PointSize in tessellation and
+# geometry stages, so the ambient registrations run these captures through the built-in and
+# the demotion - the path every affected Mali device actually takes - would execute nowhere.
+# The ambient runs stay the negative control: same scenario, same CPU-computed bytes, native
+# path. Both backends, because the demotion is shared phase-B work with two different
+# consumers (the ESSL capture respelling vs the SPIR-V Xfb carrier binding).
+gtest_discover_tests(MobileGLIntegrationTest
+    TEST_PREFIX "DirectGLES.PointSizeDemotion."
+    TEST_FILTER "PointSizeDemotionScenario.*"
+    DISCOVERY_TIMEOUT 30
+    PROPERTIES
+        LABELS integration-gpu
+        TIMEOUT ${MGL_ITEST_TIMEOUT}
+        ENVIRONMENT "${MGL_ITEST_GLES_POINT_SIZE_DEMOTION_ENVIRONMENT}"
+)
+
+gtest_discover_tests(MobileGLIntegrationTest
+    TEST_PREFIX "DirectVulkan.PointSizeDemotion."
+    TEST_FILTER "PointSizeDemotionScenario.*"
+    DISCOVERY_TIMEOUT 30
+    PROPERTIES
+        LABELS integration-gpu
+        TIMEOUT ${MGL_ITEST_TIMEOUT}
+        ENVIRONMENT "${MGL_ITEST_VULKAN_POINT_SIZE_DEMOTION_ENVIRONMENT}"
+)
+
+# --- the third CI mode: MOBILEGL_PIPE_VERIFY -----------------------------------
+#
+# ARCHITECTURE.md 13.2-(2) asks for a THIRD build mode next to pull and push: two state models in
+# one address space, compared field by field at every verb boundary and again at every accessor
+# read, 5-10x slower and never shipped. These entries are that mode's lane. They exist only when
+# the library was configured with -DMOBILEGL_PIPE_VERIFY=ON, which is deliberate and is half of
+# what makes the lane falsifiable: `ctest -L integration-verify --no-tests=error` in a build that
+# forgot the option matches NO tests and fails, instead of reporting a green run of nothing.
+#
+# The other half is PipeVerifyArmingScenario.Armed, which asserts the library's own arming line -
+# because MOBILEGL_PIPE_VERIFY=1 in the environment of a library that never compiled the
+# comparator in is a silent no-op that looks exactly like a clean pass.
+#
+# Three things about the ENVIRONMENT properties below, each of which has already gone wrong once
+# in this file:
+#   * every list APPENDS ${MGL_ITEST_COMMON_ENV} / ${MGL_ITEST_VULKAN_ENV}. A ctest ENVIRONMENT
+#     entry overrides the job environment for the names it lists, so an entry that named only its
+#     own knobs would lose the EGL vendor and Vulkan ICD pinning and run against whichever driver
+#     the loader found first.
+#   * the ambient Verify. entries name NEITHER MOBILEGL_PIPE_VERIFY_CORRUPT NOR
+#     MOBILEGL_PIPE_POISON_OMIT. That is what lets CI's two always-on negative-control steps
+#     export those knobs in the JOB environment and have them reach the test processes; a
+#     property entry of the same name would silently win and the controls would prove nothing.
+#   * MOBILEGL_LOG_FILE_PATH is per lane, and "per lane" is the exact limit of what it proves. It
+#     is the only channel a test process has for reading the library's own report (MG_Config is not
+#     reachable from this module), but the log is opened fopen(path, "w"), so every process in a
+#     lane TRUNCATES it: after an ambient lane of 400-odd entries the file holds the LAST process
+#     and nothing else. Reading it is therefore only sound in a filtered, one-entry lane - which is
+#     why the arming case has a lane and a log of its own below, and why neither this file nor CI
+#     may read the ambient logs as evidence about the entries that ran before the last one. The
+#     ambient path is kept for post-mortems (and to keep library chatter out of ctest's capture).
+if (MOBILEGL_PIPE_VERIFY)
+    # 900s, not the ambient 120: the comparator re-reads every field of the fill mask at the verb
+    # boundary and again at every accessor read, which the design budgets at 5-10x.
+    set(MGL_ITEST_VERIFY_TIMEOUT 900)
+
+    mgl_itest_join_environment(MGL_ITEST_GLES_VERIFY_ENVIRONMENT
+        "MOBILEGL_BACKEND_TYPE=DirectGLES" "MOBILEGL_PIPE_VERIFY=1"
+        "MOBILEGL_LOG_FILE_PATH=${CMAKE_CURRENT_BINARY_DIR}/pipe-verify-DirectGLES.log"
+        ${MGL_ITEST_COMMON_ENV})
+    mgl_itest_join_environment(MGL_ITEST_VULKAN_VERIFY_ENVIRONMENT
+        "MOBILEGL_BACKEND_TYPE=DirectVulkan" "MOBILEGL_PIPE_VERIFY=1"
+        "MOBILEGL_LOG_FILE_PATH=${CMAKE_CURRENT_BINARY_DIR}/pipe-verify-DirectVulkan.log"
+        ${MGL_ITEST_VULKAN_ENV})
+
+    # The arming assertion's own lane, one case per backend, with a log path nothing else writes to.
+    #
+    # PipeVerifyArmingScenario.Armed reads the library's log, and the log is a per-LANE resource: it
+    # is opened fopen(path, "w"), so every process in a lane truncates it. In the ambient Verify.
+    # lane that is 400-odd processes on one path, run `-j 4` in CI, and a whole-file read there
+    # races a neighbour's bring-up. Every other log-reading scenario in this file (UnlocatedIoBlocks,
+    # the primgen reroute, the point-size demotion) is registered exactly like this for the same
+    # reason. MGITEST_PIPE_ARMING_LANE is a harness marker - the library never reads it - and it is
+    # what makes the case skip in the ambient lane instead of racing there.
+    mgl_itest_join_environment(MGL_ITEST_GLES_VERIFY_ARMING_ENVIRONMENT
+        "MOBILEGL_BACKEND_TYPE=DirectGLES" "MOBILEGL_PIPE_VERIFY=1" "MGITEST_PIPE_ARMING_LANE=1"
+        "MOBILEGL_LOG_FILE_PATH=${CMAKE_CURRENT_BINARY_DIR}/pipe-verify-arming-DirectGLES.log"
+        ${MGL_ITEST_COMMON_ENV})
+    mgl_itest_join_environment(MGL_ITEST_VULKAN_VERIFY_ARMING_ENVIRONMENT
+        "MOBILEGL_BACKEND_TYPE=DirectVulkan" "MOBILEGL_PIPE_VERIFY=1" "MGITEST_PIPE_ARMING_LANE=1"
+        "MOBILEGL_LOG_FILE_PATH=${CMAKE_CURRENT_BINARY_DIR}/pipe-verify-arming-DirectVulkan.log"
+        ${MGL_ITEST_VULKAN_ENV})
+
+    # Negative control A (G4). MOBILEGL_PIPE_VERIFY_FATAL=0 so the process SURVIVES its own
+    # divergence and the case can read the report back out of the log; the CI step that exports
+    # the same corruption against the ambient lane, where FATAL keeps its default of 1, asserts
+    # the other half - that a divergence aborts and reds the entry.
+    mgl_itest_join_environment(MGL_ITEST_GLES_VERIFY_CORRUPT_ENVIRONMENT
+        "MOBILEGL_BACKEND_TYPE=DirectGLES" "MOBILEGL_PIPE_VERIFY=1"
+        "MOBILEGL_PIPE_VERIFY_CORRUPT=GetRenderStateParameters" "MOBILEGL_PIPE_VERIFY_FATAL=0"
+        "MOBILEGL_LOG_FILE_PATH=${CMAKE_CURRENT_BINARY_DIR}/pipe-verify-corrupt-DirectGLES.log"
+        ${MGL_ITEST_COMMON_ENV})
+    mgl_itest_join_environment(MGL_ITEST_VULKAN_VERIFY_CORRUPT_ENVIRONMENT
+        "MOBILEGL_BACKEND_TYPE=DirectVulkan" "MOBILEGL_PIPE_VERIFY=1"
+        "MOBILEGL_PIPE_VERIFY_CORRUPT=GetRenderStateParameters" "MOBILEGL_PIPE_VERIFY_FATAL=0"
+        "MOBILEGL_LOG_FILE_PATH=${CMAKE_CURRENT_BINARY_DIR}/pipe-verify-corrupt-DirectVulkan.log"
+        ${MGL_ITEST_VULKAN_ENV})
+
+    # Negative control B (G5). The omission skips the STAMP of one field for one verb while still
+    # copying its value, which is indistinguishable from a fill row nobody wrote; the scenario
+    # forks, so the resulting std::abort() is a datum in waitpid() rather than a dead lane.
+    mgl_itest_join_environment(MGL_ITEST_GLES_POISON_OMIT_ENVIRONMENT
+        "MOBILEGL_BACKEND_TYPE=DirectGLES" "MOBILEGL_PIPE_VERIFY=1"
+        "MOBILEGL_PIPE_POISON_OMIT=GenerateMipmap:GetActiveTextureUnit"
+        "MOBILEGL_LOG_FILE_PATH=${CMAKE_CURRENT_BINARY_DIR}/pipe-poison-omit-DirectGLES.log"
+        ${MGL_ITEST_COMMON_ENV})
+    mgl_itest_join_environment(MGL_ITEST_VULKAN_POISON_OMIT_ENVIRONMENT
+        "MOBILEGL_BACKEND_TYPE=DirectVulkan" "MOBILEGL_PIPE_VERIFY=1"
+        "MOBILEGL_PIPE_POISON_OMIT=GenerateMipmap:GetActiveTextureUnit"
+        "MOBILEGL_LOG_FILE_PATH=${CMAKE_CURRENT_BINARY_DIR}/pipe-poison-omit-DirectVulkan.log"
+        ${MGL_ITEST_VULKAN_ENV})
+
+    # The whole suite again, per backend, with the comparator armed. Same scenarios, same
+    # assertions, but every backend read of frontend state is now checked against a snapshot taken
+    # from the live context at the verb boundary - which is what "the 742 integration entries
+    # prove push equals pull" means. Labelled integration-gpu as well so a verify build's
+    # `ctest -L integration-gpu` still describes the whole registration set.
+    gtest_discover_tests(MobileGLIntegrationTest
+        TEST_PREFIX "DirectGLES.Verify."
+        DISCOVERY_TIMEOUT 30
+        PROPERTIES
+            LABELS "integration-gpu\;integration-verify"
+            TIMEOUT ${MGL_ITEST_VERIFY_TIMEOUT}
+            ENVIRONMENT "${MGL_ITEST_GLES_VERIFY_ENVIRONMENT}"
+    )
+    gtest_discover_tests(MobileGLIntegrationTest
+        TEST_PREFIX "DirectVulkan.Verify."
+        DISCOVERY_TIMEOUT 30
+        PROPERTIES
+            LABELS "integration-gpu\;integration-verify"
+            TIMEOUT ${MGL_ITEST_VERIFY_TIMEOUT}
+            ENVIRONMENT "${MGL_ITEST_VULKAN_VERIFY_ENVIRONMENT}"
+    )
+
+    # The arming assertion, one entry per backend. This is the entry that fails a lane whose library
+    # never armed: it runs the same library and the same MOBILEGL_PIPE_VERIFY=1 as the ambient
+    # entries above, but unlike them it cannot be green against a library with no comparator
+    # compiled in. Its log is its own, so `-j 4` cannot make it flake.
+    gtest_discover_tests(MobileGLIntegrationTest
+        TEST_PREFIX "DirectGLES.VerifyArming."
+        TEST_FILTER "PipeVerifyArmingScenario.Armed"
+        DISCOVERY_TIMEOUT 30
+        PROPERTIES
+            LABELS "integration-gpu\;integration-verify"
+            TIMEOUT ${MGL_ITEST_VERIFY_TIMEOUT}
+            ENVIRONMENT "${MGL_ITEST_GLES_VERIFY_ARMING_ENVIRONMENT}"
+    )
+    gtest_discover_tests(MobileGLIntegrationTest
+        TEST_PREFIX "DirectVulkan.VerifyArming."
+        TEST_FILTER "PipeVerifyArmingScenario.Armed"
+        DISCOVERY_TIMEOUT 30
+        PROPERTIES
+            LABELS "integration-gpu\;integration-verify"
+            TIMEOUT ${MGL_ITEST_VERIFY_TIMEOUT}
+            ENVIRONMENT "${MGL_ITEST_VULKAN_VERIFY_ARMING_ENVIRONMENT}"
+    )
+
+    # One case each: the knobs are process-wide, so a corrupted or poisoned process cannot also be
+    # running the ambient assertions. These four entries are the ones that assert the RED - they
+    # pass when the comparator and the poison report, and go red when either stops.
+    gtest_discover_tests(MobileGLIntegrationTest
+        TEST_PREFIX "DirectGLES.VerifyCorrupted."
+        TEST_FILTER "PipeVerifyArmingScenario.CorruptedFieldIsReported"
+        DISCOVERY_TIMEOUT 30
+        PROPERTIES
+            LABELS "integration-gpu\;integration-verify"
+            TIMEOUT ${MGL_ITEST_VERIFY_TIMEOUT}
+            ENVIRONMENT "${MGL_ITEST_GLES_VERIFY_CORRUPT_ENVIRONMENT}"
+    )
+    gtest_discover_tests(MobileGLIntegrationTest
+        TEST_PREFIX "DirectVulkan.VerifyCorrupted."
+        TEST_FILTER "PipeVerifyArmingScenario.CorruptedFieldIsReported"
+        DISCOVERY_TIMEOUT 30
+        PROPERTIES
+            LABELS "integration-gpu\;integration-verify"
+            TIMEOUT ${MGL_ITEST_VERIFY_TIMEOUT}
+            ENVIRONMENT "${MGL_ITEST_VULKAN_VERIFY_CORRUPT_ENVIRONMENT}"
+    )
+    gtest_discover_tests(MobileGLIntegrationTest
+        TEST_PREFIX "DirectGLES.PoisonOmitted."
+        TEST_FILTER "PoisonOmissionScenario.OmittedFieldAbortsOnThatVerb"
+        DISCOVERY_TIMEOUT 30
+        PROPERTIES
+            LABELS "integration-gpu\;integration-verify"
+            TIMEOUT ${MGL_ITEST_VERIFY_TIMEOUT}
+            ENVIRONMENT "${MGL_ITEST_GLES_POISON_OMIT_ENVIRONMENT}"
+    )
+    gtest_discover_tests(MobileGLIntegrationTest
+        TEST_PREFIX "DirectVulkan.PoisonOmitted."
+        TEST_FILTER "PoisonOmissionScenario.OmittedFieldAbortsOnThatVerb"
+        DISCOVERY_TIMEOUT 30
+        PROPERTIES
+            LABELS "integration-gpu\;integration-verify"
+            TIMEOUT ${MGL_ITEST_VERIFY_TIMEOUT}
+            ENVIRONMENT "${MGL_ITEST_VULKAN_POISON_OMIT_ENVIRONMENT}"
+    )
+endif()

MobileGL/MG_IntegrationTest/Harness/BackendCapsPeek.cpp

@@ -0,0 +1,42 @@
+// MobileGL - MobileGL/MG_IntegrationTest/Harness/BackendCapsPeek.cpp
+// Copyright (c) 2025-2026 MobileGL-Dev
+// Licensed under the GNU Lesser General Public License v3.0:
+//   https://www.gnu.org/licenses/gpl-3.0.txt
+//   https://www.gnu.org/licenses/lgpl-3.0.txt
+// SPDX-License-Identifier: LGPL-3.0-only
+// End of Source File Header
+
+#include "BackendCapsPeek.h"
+
+#if !defined(__ANDROID__)
+#include <MG_Backend/BackendObject.h>
+
+namespace MobileGL::MG_Backend {
+    // Declared in MG_Backend/BackendObjects.h, which also pulls in both backends' headers
+    // and, through them, their loaders; the reference alone is all that is needed here.
+    extern UniquePtr<BackendObject>& pActiveBackendObject;
+} // namespace MobileGL::MG_Backend
+#endif
+
+namespace MGITest {
+
+    bool PeekComputeWorkGroupCaps(int outCount[3], int outSize[3]) {
+#if defined(__ANDROID__)
+        (void)outCount;
+        (void)outSize;
+        return false;
+#else
+        const auto& backend = MobileGL::MG_Backend::pActiveBackendObject;
+        if (!backend) {
+            return false;
+        }
+        const MobileGL::MG_Backend::DynamicBackendParameters& caps = backend->GetDynamicParameters();
+        for (int axis = 0; axis < 3; ++axis) {
+            outCount[axis] = caps.MaxComputeWorkGroupCount[axis];
+            outSize[axis] = caps.MaxComputeWorkGroupSize[axis];
+        }
+        return true;
+#endif
+    }
+
+} // namespace MGITest

MobileGL/MG_IntegrationTest/Harness/BackendCapsPeek.h

@@ -0,0 +1,29 @@
+// MobileGL - MobileGL/MG_IntegrationTest/Harness/BackendCapsPeek.h
+// Copyright (c) 2025-2026 MobileGL-Dev
+// Licensed under the GNU Lesser General Public License v3.0:
+//   https://www.gnu.org/licenses/gpl-3.0.txt
+//   https://www.gnu.org/licenses/lgpl-3.0.txt
+// SPDX-License-Identifier: LGPL-3.0-only
+// End of Source File Header
+//
+// The one place this module looks past the GL API into the active backend's caps block.
+//
+// It exists for exactly one assertion: that the six per-axis compute limits the MGPipe
+// caps block carries (DynamicBackendParameters::MaxComputeWorkGroupCount/Size, plan B
+// section 4.4.1) are the same numbers glGetIntegeri_v answers today, since P0.5 retires
+// the getter in favour of the caps. A separate translation unit, because the scenario
+// sources include the GL headers with prototypes and MobileGL's umbrella header is not
+// meant to meet them in one file.
+
+#pragma once
+
+namespace MGITest {
+
+    // Copies the active backend's MaxComputeWorkGroupCount / MaxComputeWorkGroupSize into the
+    // two arrays and returns true. Returns false, touching nothing, where the caps block is
+    // out of reach: on Android this module links the SHIPPING libMobileGL.so, built
+    // -fvisibility=hidden, so no internal symbol resolves; on desktop it links MobileGL_s and
+    // the read is direct.
+    bool PeekComputeWorkGroupCaps(int outCount[3], int outSize[3]);
+
+} // namespace MGITest

MobileGL/MG_IntegrationTest/Harness/HeadlessGL.cpp

@@ -15,6 +15,16 @@
 #include <ostream>
 #include <sstream>
 
+#if defined(_WIN32)
+#define WIN32_LEAN_AND_MEAN
+#include <windows.h>
+#elif defined(__ANDROID__)
+#include <android/hardware_buffer.h>
+#include <android/native_window.h>
+#include <media/NdkImage.h>
+#include <media/NdkImageReader.h>
+#endif
+
 // MobileGL's own headers, in the order MobileGL/Includes.h uses them: GL/gl.h
 // first, then glcorearb.h for the 3.x+ entry points. This binary links
 // MobileGL_s, so every gl*/egl* below binds to MobileGL's implementation, not
@@ -32,7 +42,7 @@
 // the only construction that is actually predictive here: MobileGL ABORTS
 // (MOBILEGL_ASSERT -> SIGTRAP) rather than returning an error on an unusable
 // platform, so nothing the parent can call in-process is allowed to be wrong.
-#if !defined(_WIN32) && !defined(__APPLE__) && __has_include(<sys/wait.h>)
+#if !defined(_WIN32) && !defined(__APPLE__) && !defined(__ANDROID__) && __has_include(<sys/wait.h>)
 #define MGITEST_HAVE_FORK_PREFLIGHT 1
 #include <csignal>
 #include <ctime>
@@ -53,6 +63,83 @@ namespace MGITest {
         constexpr int kSurfaceWidth = 128;
         constexpr int kSurfaceHeight = 96;
 
+#if defined(_WIN32)
+        HWND g_testWindow = nullptr;
+
+        HWND CreateTestWindow() {
+            static const wchar_t* const kClassName = L"MobileGLIntegrationTestWindow";
+            static bool registered = false;
+            if (!registered) {
+                WNDCLASSW windowClass{};
+                windowClass.lpfnWndProc = DefWindowProcW;
+                windowClass.hInstance = GetModuleHandleW(nullptr);
+                windowClass.lpszClassName = kClassName;
+                if (RegisterClassW(&windowClass) == 0 && GetLastError() != ERROR_CLASS_ALREADY_EXISTS) {
+                    return nullptr;
+                }
+                registered = true;
+            }
+            return CreateWindowExW(0, kClassName, L"MobileGL Integration Test", WS_OVERLAPPEDWINDOW,
+                                   CW_USEDEFAULT, CW_USEDEFAULT, kSurfaceWidth, kSurfaceHeight, nullptr, nullptr,
+                                   GetModuleHandleW(nullptr), nullptr);
+        }
+#elif defined(__ANDROID__)
+        AImageReader* g_imageReader = nullptr;
+        ANativeWindow* g_imageReaderWindow = nullptr;
+
+        void DrainImageReader(void*, AImageReader* reader) {
+            AImage* image = nullptr;
+            if (AImageReader_acquireNextImage(reader, &image) == AMEDIA_OK && image != nullptr) {
+                AImage_delete(image);
+            }
+        }
+
+        bool CreateImageReaderWindow() {
+            if (g_imageReaderWindow != nullptr) return true;
+            constexpr int kMaxImages = 4;
+            const media_status_t status = AImageReader_newWithUsage(
+                kSurfaceWidth, kSurfaceHeight, AIMAGE_FORMAT_RGBA_8888,
+                AHARDWAREBUFFER_USAGE_GPU_SAMPLED_IMAGE | AHARDWAREBUFFER_USAGE_GPU_COLOR_OUTPUT,
+                kMaxImages, &g_imageReader);
+            if (status != AMEDIA_OK || g_imageReader == nullptr) return false;
+
+            AImageReader_ImageListener listener = {nullptr, DrainImageReader};
+            AImageReader_setImageListener(g_imageReader, &listener);
+            if (AImageReader_getWindow(g_imageReader, &g_imageReaderWindow) != AMEDIA_OK ||
+                g_imageReaderWindow == nullptr) {
+                AImageReader_setImageListener(g_imageReader, nullptr);
+                AImageReader_delete(g_imageReader);
+                g_imageReader = nullptr;
+                return false;
+            }
+            ANativeWindow_acquire(g_imageReaderWindow);
+            return true;
+        }
+
+        void DestroyImageReaderWindow() {
+            if (g_imageReaderWindow != nullptr) {
+                ANativeWindow_release(g_imageReaderWindow);
+                g_imageReaderWindow = nullptr;
+            }
+            if (g_imageReader != nullptr) {
+                AImageReader_setImageListener(g_imageReader, nullptr);
+                AImageReader_delete(g_imageReader);
+                g_imageReader = nullptr;
+            }
+        }
+#endif
+
+        bool UseWindowSurface() {
+#if defined(_WIN32)
+            const char* value = std::getenv("MOBILEGL_ITEST_WINDOW_SURFACE");
+            return value != nullptr && value[0] != '\0' && std::strcmp(value, "0") != 0;
+#elif defined(__ANDROID__)
+            return true;
+#else
+            return false;
+#endif
+        }
+
         std::string EnvOr(const char* name, const char* fallback) {
             const char* value = std::getenv(name);
             return (value != nullptr && value[0] != '\0') ? std::string(value) : std::string(fallback);
@@ -87,10 +174,10 @@ namespace MGITest {
         // callers). surfaceless is the platform with no window-system dependency at
         // all; the surface this file then creates is still a pbuffer, which every
         // platform supports and which the amendment to this rule requires as the
-        // fallback shape. DISPLAY/WAYLAND_DISPLAY are cleared as well so that a
+        // fallback shape on desktop. Android instead supplies an AImageReader
+        // ANativeWindow. DISPLAY/WAYLAND_DISPLAY are cleared as well so that a
         // driver that consults them directly cannot reintroduce the dependency
-        // behind EGL's back. Desktop-only file: MG_IntegrationTest never builds
-        // for Android, so no device path is affected.
+        // behind EGL's back.
         void EnsureHeadlessPlatform() {
 #if defined(__linux__) && !defined(__ANDROID__)
             static bool done = false;
@@ -134,8 +221,9 @@ namespace MGITest {
                 return 3;
             }
 
+            const bool useWindowSurface = UseWindowSurface();
             const EGLint configAttribs[] = {EGL_SURFACE_TYPE,
-                                            EGL_PBUFFER_BIT,
+                                            useWindowSurface ? EGL_WINDOW_BIT : EGL_PBUFFER_BIT,
                                             EGL_RED_SIZE,
                                             8,
                                             EGL_GREEN_SIZE,
@@ -152,7 +240,9 @@ namespace MGITest {
             EGLConfig config = nullptr;
             EGLint configCount = 0;
             if (eglChooseConfig(display, configAttribs, &config, 1, &configCount) != EGL_TRUE || configCount < 1) {
-                outReason = WithEglError("eglChooseConfig found no pbuffer-capable RGBA8/D24 config");
+                outReason = WithEglError(useWindowSurface
+                                             ? "eglChooseConfig found no window-capable RGBA8/D24 config"
+                                             : "eglChooseConfig found no pbuffer-capable RGBA8/D24 config");
                 return 4;
             }
 
@@ -166,10 +256,32 @@ namespace MGITest {
                 return 5;
             }
 
-            const EGLint pbufferAttribs[] = {EGL_WIDTH, kSurfaceWidth, EGL_HEIGHT, kSurfaceHeight, EGL_NONE};
-            EGLSurface surface = eglCreatePbufferSurface(display, config, pbufferAttribs);
+            EGLSurface surface = EGL_NO_SURFACE;
+            if (useWindowSurface) {
+#if defined(_WIN32)
+                if (g_testWindow == nullptr) g_testWindow = CreateTestWindow();
+                if (g_testWindow == nullptr) {
+                    outReason = "failed to create the Windows integration-test window";
+                    return 6;
+                }
+                surface = eglCreateWindowSurface(display, config, g_testWindow, nullptr);
+#elif defined(__ANDROID__)
+                if (!CreateImageReaderWindow()) {
+                    outReason = "failed to create the Android AImageReader integration-test window";
+                    return 6;
+                }
+                surface = eglCreateWindowSurface(display, config, g_imageReaderWindow, nullptr);
+#endif
+            } else {
+                const EGLint pbufferAttribs[] = {EGL_WIDTH, kSurfaceWidth, EGL_HEIGHT, kSurfaceHeight, EGL_NONE};
+                surface = eglCreatePbufferSurface(display, config, pbufferAttribs);
+            }
             if (surface == EGL_NO_SURFACE) {
-                outReason = WithEglError("eglCreatePbufferSurface failed");
+#if defined(__ANDROID__)
+                DestroyImageReaderWindow();
+#endif
+                outReason = WithEglError(useWindowSurface ? "eglCreateWindowSurface failed"
+                                                         : "eglCreatePbufferSurface failed");
                 return 6;
             }
             // The step that brings the whole backend up (DirectVulkan creates its
@@ -440,7 +552,15 @@ namespace MGITest {
         // before the pre-flight forks - the child must measure the same platform
         // the parent will use.
         EnsureHeadlessPlatform();
-        m_backendName = EnvOr("MOBILEGL_BACKEND_TYPE", "<unset>");
+        // The backend that is actually about to come up, which is what every
+        // `BackendName() == "DirectGLES"` gate in the scenarios means by the question.
+        // MG_ConfigLoader::InitBackendType defaults an unset MOBILEGL_BACKEND_TYPE to
+        // DirectGLES, so the same default belongs here; this used to report the literal
+        // "<unset>" instead. Under ctest the variable is always set by the ENVIRONMENT
+        // property, which is why that never showed - but run straight from a device
+        // shell, where nothing sets it, DirectGLES came up and every case gated on the
+        // NAME DirectGLES skipped as though it had not.
+        m_backendName = EnvOr("MOBILEGL_BACKEND_TYPE", "DirectGLES");
         m_usable = BringUp();
     }
 
@@ -491,6 +611,14 @@ namespace MGITest {
         if (m_context != nullptr) eglDestroyContext(display, static_cast<EGLContext>(m_context));
         if (m_surface != nullptr) eglDestroySurface(display, static_cast<EGLSurface>(m_surface));
         eglTerminate(display);
+#if defined(_WIN32)
+        if (g_testWindow != nullptr) {
+            DestroyWindow(g_testWindow);
+            g_testWindow = nullptr;
+        }
+#elif defined(__ANDROID__)
+        DestroyImageReaderWindow();
+#endif
         m_context = nullptr;
         m_surface = nullptr;
         m_display = nullptr;

MobileGL/MG_IntegrationTest/Harness/HeadlessGL.h

@@ -14,11 +14,11 @@
 // inspects backend state - both bugs this module pins were invisible to
 // state-level assertions and visible only in pixels.
 //
-// Headless by construction, following MG_Benchmark/Driver/DriverBench.c: an EGL
-// context on a PBUFFER surface. No window, no window manager, no human. Unlike
-// DriverBench the scenarios do draw to the DEFAULT framebuffer (that is where
-// the Y-flip lives) and do call eglSwapBuffers (that is the frame boundary the
-// cross-frame scenarios need to be real).
+// Headless by construction: desktop uses an EGL pbuffer and Android uses an
+// AImageReader-backed ANativeWindow that needs no Activity. No window manager,
+// no human. Unlike DriverBench the scenarios do draw to the DEFAULT framebuffer
+// (that is where the Y-flip lives) and do call eglSwapBuffers (that is the frame
+// boundary the cross-frame scenarios need to be real).
 //
 // One process is one backend: MOBILEGL_BACKEND_TYPE is latched at
 // initialization, so the CMake wiring runs this binary once per backend rather

MobileGL/MG_IntegrationTest/Harness/ScenarioFixture.h

@@ -21,12 +21,49 @@
 
 #pragma once
 
+#include <cctype>
+#include <cstdlib>
+#include <string>
+
 #include <gtest/gtest.h>
 
 #include "HeadlessGL.h"
 
 namespace MGITest {
 
+    // How a MOBILEGL_* quirk variable reads in THIS process's environment.
+    //
+    // A scenario that needs a non-default configuration takes it from here and skips
+    // when the process it was launched into is not in that configuration, rather than
+    // writing MG_Config::Features itself. Two reasons, and the second one decides it:
+    //
+    //   - the feature table is an internal symbol. On Android this module links against
+    //     the SHIPPING libMobileGL.so - deliberately, so the on-device run validates the
+    //     real artifact - and that library is built -fvisibility=hidden, so nothing
+    //     internal is reachable from here at all.
+    //   - a quirk poked in-process is already too late for everything latched at
+    //     initialization: the compile pool and its threads, and the backend's advertised
+    //     extension list, which is built once from the configuration in force at first
+    //     use. The process-wide variable is the only spelling that covers the whole
+    //     configuration instead of the half of it that is still mutable afterwards.
+    //
+    // The reading rule is MG_ConfigLoader's, character for character (ConfigLoader.cpp,
+    // QueryEnvQuirkOverride / IsTruthyValue): unset is Auto - device auto-detection or a
+    // built-in default, i.e. a value only the implementation knows - a truthy value is
+    // On, and anything else that IS set ("0", "false", "") is Off.
+    enum class AmbientQuirk { Auto, On, Off };
+
+    inline AmbientQuirk AmbientQuirkFromEnvironment(const char* name) {
+        const char* value = std::getenv(name);
+        if (value == nullptr) return AmbientQuirk::Auto;
+        std::string lowered(value);
+        for (char& c : lowered) {
+            c = static_cast<char>(std::tolower(static_cast<unsigned char>(c)));
+        }
+        if (lowered.empty() || lowered == "0" || lowered == "false") return AmbientQuirk::Off;
+        return AmbientQuirk::On;
+    }
+
     class ScenarioTest : public ::testing::Test {
     protected:
         void SetUp() override {

MobileGL/MG_IntegrationTest/Main.cpp

@@ -31,8 +31,14 @@ namespace {
                 // silently bound to a workstation's window system is a different
                 // run from CI's and must be visible as one in the log.
                 const char* eglPlatform = std::getenv("EGL_PLATFORM");
-                std::fprintf(stderr, "  renderer: %s\n  surface:  %dx%d pbuffer (headless, EGL_PLATFORM=%s)\n",
+#if defined(__ANDROID__)
+                constexpr const char* surfaceKind = "AImageReader window";
+#else
+                constexpr const char* surfaceKind = "pbuffer";
+#endif
+                std::fprintf(stderr, "  renderer: %s\n  surface:  %dx%d %s (headless, EGL_PLATFORM=%s)\n",
                              gl.RendererString().c_str(), gl.Width(), gl.Height(),
+                             surfaceKind,
                              eglPlatform != nullptr ? eglPlatform : "<unset>");
             } else if (MGITest::RequireGpu()) {
                 std::fprintf(stderr,

MobileGL/MG_IntegrationTest/Scenarios/AdvertisedLimitsScenario.cpp

@@ -26,9 +26,11 @@
 // quantities, so an entry that only fails on DirectVulkan is a translation bug and one that
 // fails on both is a table bug.
 
+#include <algorithm>
 #include <string>
 #include <vector>
 
+#include "../Harness/BackendCapsPeek.h"
 #include "../Harness/HeadlessGL.h"
 #include "../Harness/ScenarioFixture.h"
 
@@ -56,7 +58,13 @@ namespace MGITest {
 
         const std::vector<LimitBound>& BufferLimitTable() {
             static const std::vector<LimitBound> table = {
-                {GL_MAX_UNIFORM_BUFFER_BINDINGS, "GL_MAX_UNIFORM_BUFFER_BINDINGS", 36, 256},
+                // 84 is the GL 4.5 core table 23.64 minimum, and also the width of the state
+                // layer's indexed-binding array - the two were made to coincide when the array
+                // was widened from 36, which had made the clamp in GL_Getter degenerate.
+                {GL_MAX_UNIFORM_BUFFER_BINDINGS, "GL_MAX_UNIFORM_BUFFER_BINDINGS", 84, 256},
+                // 14 uniform blocks on each of the FIVE graphics stages. The sum used to count
+                // three, and the two tessellation stages were simply missing from it.
+                {GL_MAX_COMBINED_UNIFORM_BLOCKS, "GL_MAX_COMBINED_UNIFORM_BLOCKS", 70, 256},
                 {GL_MAX_COMPUTE_UNIFORM_BLOCKS, "GL_MAX_COMPUTE_UNIFORM_BLOCKS", 12, 256},
                 {GL_MAX_COMPUTE_SHADER_STORAGE_BLOCKS, "GL_MAX_COMPUTE_SHADER_STORAGE_BLOCKS", 8, 256},
                 {GL_MAX_COMBINED_SHADER_STORAGE_BLOCKS, "GL_MAX_COMBINED_SHADER_STORAGE_BLOCKS", 8, 256},
@@ -133,6 +141,49 @@ namespace MGITest {
                     << relation.blocksName << " = " << blocks << " exceeds " << relation.bindingsName << " = "
                     << bindings << "; a shader may declare more blocks than there are binding points to bind them to";
             }
+
+            // THE MIDDLE TERM, which the relation quoted above always had and this case never
+            // checked. It is the one that actually broke: widening the binding-point array to 84
+            // raised what every PER-STAGE count clamps to, while the combined value was a
+            // five-stage sum of 70 - so a device reporting descriptor-indexing-scale uniform
+            // buffers (Adreno: maxPerStageDescriptorUniformBuffers = 16777216) advertised 84
+            // compute uniform blocks inside a combined limit of 70. Per-stage <= combined is
+            // exactly the assertion that says so, and it costs one glGetIntegerv per row.
+            struct StageAgainstCombined {
+                GLenum stage;
+                const char* stageName;
+                GLenum combined;
+                const char* combinedName;
+            };
+            const StageAgainstCombined stageRelations[] = {
+                {GL_MAX_COMPUTE_UNIFORM_BLOCKS, "GL_MAX_COMPUTE_UNIFORM_BLOCKS", GL_MAX_COMBINED_UNIFORM_BLOCKS,
+                 "GL_MAX_COMBINED_UNIFORM_BLOCKS"},
+                {GL_MAX_VERTEX_UNIFORM_BLOCKS, "GL_MAX_VERTEX_UNIFORM_BLOCKS", GL_MAX_COMBINED_UNIFORM_BLOCKS,
+                 "GL_MAX_COMBINED_UNIFORM_BLOCKS"},
+                {GL_MAX_TESS_CONTROL_UNIFORM_BLOCKS, "GL_MAX_TESS_CONTROL_UNIFORM_BLOCKS",
+                 GL_MAX_COMBINED_UNIFORM_BLOCKS, "GL_MAX_COMBINED_UNIFORM_BLOCKS"},
+                {GL_MAX_TESS_EVALUATION_UNIFORM_BLOCKS, "GL_MAX_TESS_EVALUATION_UNIFORM_BLOCKS",
+                 GL_MAX_COMBINED_UNIFORM_BLOCKS, "GL_MAX_COMBINED_UNIFORM_BLOCKS"},
+                {GL_MAX_GEOMETRY_UNIFORM_BLOCKS, "GL_MAX_GEOMETRY_UNIFORM_BLOCKS", GL_MAX_COMBINED_UNIFORM_BLOCKS,
+                 "GL_MAX_COMBINED_UNIFORM_BLOCKS"},
+                {GL_MAX_FRAGMENT_UNIFORM_BLOCKS, "GL_MAX_FRAGMENT_UNIFORM_BLOCKS", GL_MAX_COMBINED_UNIFORM_BLOCKS,
+                 "GL_MAX_COMBINED_UNIFORM_BLOCKS"},
+                {GL_MAX_COMPUTE_SHADER_STORAGE_BLOCKS, "GL_MAX_COMPUTE_SHADER_STORAGE_BLOCKS",
+                 GL_MAX_COMBINED_SHADER_STORAGE_BLOCKS, "GL_MAX_COMBINED_SHADER_STORAGE_BLOCKS"},
+                {GL_MAX_FRAGMENT_SHADER_STORAGE_BLOCKS, "GL_MAX_FRAGMENT_SHADER_STORAGE_BLOCKS",
+                 GL_MAX_COMBINED_SHADER_STORAGE_BLOCKS, "GL_MAX_COMBINED_SHADER_STORAGE_BLOCKS"},
+            };
+            for (const StageAgainstCombined& relation : stageRelations) {
+                GLint stage = -1;
+                GLint combined = -1;
+                glGetIntegerv(relation.stage, &stage);
+                glGetIntegerv(relation.combined, &combined);
+                ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR)) << relation.stageName;
+                EXPECT_LE(stage, combined)
+                    << relation.stageName << " = " << stage << " exceeds " << relation.combinedName << " = "
+                    << combined << "; GL 4.6 table 23.64 orders MAX_*_BUFFER_BINDINGS >= MAX_COMBINED_*_BLOCKS >= "
+                       "every per-stage count, and a single-stage program may use its whole per-stage allowance";
+            }
         }
 
         // KHR-GL44.multi_bind.functional_bind_buffers_range sizes each of an indexed target's
@@ -199,6 +250,80 @@ namespace MGITest {
                    "derived component limits are computed in";
         }
 
+        // The GL 4.5 core minimums that had no case in the getter at all, or that were still
+        // carrying an ES/GL3.3-tier number. Every one of these answered GL_INVALID_ENUM or a
+        // too-small value against a context advertising 4.6, and each is the FIRST call its
+        // conformance case makes - so the case died before it could measure anything.
+        //
+        // The cull pair is deliberately absent: zero is a legal answer there (a backend with no
+        // cull-distance route MUST report it), so it is checked for answerability only, below.
+        TEST_F(AdvertisedLimitsScenario, EveryGL45CoreMinimumIsMet) {
+            const std::vector<LimitBound> table = {
+                {GL_MAX_VARYING_VECTORS, "GL_MAX_VARYING_VECTORS", 15, 256},
+                {GL_MAX_VERTEX_UNIFORM_VECTORS, "GL_MAX_VERTEX_UNIFORM_VECTORS", 256, 1 << 20},
+                {GL_MAX_VARYING_COMPONENTS, "GL_MAX_VARYING_COMPONENTS", 60, 1 << 20},
+                // GL_MAX_VERTEX_STREAMS is deliberately absent. GL 4.5 requires 4 and MobileGL
+                // answers 1, which is a KNOWN non-conformance rather than an oversight: raising
+                // the number un-gates two transform-feedback CTS cases per package across
+                // KHR-GL40..GL46 that then fail, because no part of the shader pipeline supports
+                // layout(stream = N). See the GL_MAX_VERTEX_STREAMS case in GL_Getter.cpp. Adding
+                // a row here would pin a number the implementation cannot back.
+                {GL_MAX_GEOMETRY_SHADER_INVOCATIONS, "GL_MAX_GEOMETRY_SHADER_INVOCATIONS", 32, 256},
+                {GL_MAX_SUBROUTINES, "GL_MAX_SUBROUTINES", 256, 1 << 20},
+                {GL_MAX_SUBROUTINE_UNIFORM_LOCATIONS, "GL_MAX_SUBROUTINE_UNIFORM_LOCATIONS", 1024, 1 << 20},
+                {GL_MAX_TESS_CONTROL_INPUT_COMPONENTS, "GL_MAX_TESS_CONTROL_INPUT_COMPONENTS", 128, 1 << 16},
+                {GL_MAX_TESS_CONTROL_OUTPUT_COMPONENTS, "GL_MAX_TESS_CONTROL_OUTPUT_COMPONENTS", 128, 1 << 16},
+                {GL_MAX_TESS_CONTROL_TOTAL_OUTPUT_COMPONENTS, "GL_MAX_TESS_CONTROL_TOTAL_OUTPUT_COMPONENTS", 4096,
+                 1 << 20},
+                {GL_MAX_TESS_CONTROL_TEXTURE_IMAGE_UNITS, "GL_MAX_TESS_CONTROL_TEXTURE_IMAGE_UNITS", 16, 256},
+                {GL_MAX_TESS_CONTROL_UNIFORM_COMPONENTS, "GL_MAX_TESS_CONTROL_UNIFORM_COMPONENTS", 1024, 1 << 20},
+                {GL_MAX_TESS_CONTROL_UNIFORM_BLOCKS, "GL_MAX_TESS_CONTROL_UNIFORM_BLOCKS", 14, 256},
+                {GL_MAX_TESS_EVALUATION_INPUT_COMPONENTS, "GL_MAX_TESS_EVALUATION_INPUT_COMPONENTS", 128, 1 << 16},
+                {GL_MAX_TESS_EVALUATION_OUTPUT_COMPONENTS, "GL_MAX_TESS_EVALUATION_OUTPUT_COMPONENTS", 128, 1 << 16},
+                {GL_MAX_TESS_EVALUATION_TEXTURE_IMAGE_UNITS, "GL_MAX_TESS_EVALUATION_TEXTURE_IMAGE_UNITS", 16, 256},
+                {GL_MAX_TESS_EVALUATION_UNIFORM_COMPONENTS, "GL_MAX_TESS_EVALUATION_UNIFORM_COMPONENTS", 1024,
+                 1 << 20},
+                {GL_MAX_TESS_EVALUATION_UNIFORM_BLOCKS, "GL_MAX_TESS_EVALUATION_UNIFORM_BLOCKS", 14, 256},
+                {GL_MAX_TESS_PATCH_COMPONENTS, "GL_MAX_TESS_PATCH_COMPONENTS", 120, 1 << 16},
+                {GL_MAX_COMBINED_TESS_CONTROL_UNIFORM_COMPONENTS, "GL_MAX_COMBINED_TESS_CONTROL_UNIFORM_COMPONENTS",
+                 58368, 1 << 30},
+                {GL_MAX_COMBINED_TESS_EVALUATION_UNIFORM_COMPONENTS,
+                 "GL_MAX_COMBINED_TESS_EVALUATION_UNIFORM_COMPONENTS", 58368, 1 << 30},
+            };
+            for (const LimitBound& bound : table) {
+                GLint value = -424242;
+                glGetIntegerv(bound.pname, &value);
+                const unsigned int error = FirstGLError();
+                EXPECT_EQ(error, GLenum(GL_NO_ERROR)) << bound.name << " is not answerable: " << GLErrorName(error);
+                if (error != GL_NO_ERROR) continue;
+                EXPECT_GE(value, bound.minimum) << bound.name << " = " << value << " is below the GL 4.5 minimum "
+                                                << bound.minimum;
+                EXPECT_LE(value, bound.ceiling) << bound.name << " = " << value << " exceeds the ceiling "
+                                                << bound.ceiling;
+            }
+
+            // ARB_cull_distance's pair. Zero is honest on a backend with no cull-distance route,
+            // so only answerability and the combined-limit ordering are checked here.
+            GLint cull = -1;
+            GLint clip = -1;
+            GLint combined = -1;
+            glGetIntegerv(GL_MAX_CULL_DISTANCES, &cull);
+            glGetIntegerv(GL_MAX_CLIP_DISTANCES, &clip);
+            glGetIntegerv(GL_MAX_COMBINED_CLIP_AND_CULL_DISTANCES, &combined);
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR)) << "the ARB_cull_distance queries must not error";
+            EXPECT_GE(cull, 0);
+            EXPECT_GE(combined, cull) << "GL 4.6 core 11.1.3.10: the combined limit is at least the cull one";
+            EXPECT_GE(combined, clip) << "GL 4.6 core 11.1.3.10: the combined limit is at least the clip one";
+
+            // GL_MAX_ELEMENT_INDEX is 64-bit state: the required 2^32-1 does not fit a GLint, so
+            // the wide query must answer it and the narrow one must saturate rather than wrap.
+            GLint64 elementIndex = -1;
+            glGetInteger64v(GL_MAX_ELEMENT_INDEX, &elementIndex);
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+            EXPECT_GE(elementIndex, static_cast<GLint64>(4294967295LL))
+                << "GL 4.5 core table 23.55 sets the GL_MAX_ELEMENT_INDEX minimum at 2^32-1";
+        }
+
         // ARB_viewport_array's own limits. They are advertised from three different places -
         // GL_MAX_VIEWPORTS from the frontend's indexed state width, the bounds range and the
         // subpixel bits from the backend caps table - and each backend fills that table from a
@@ -255,5 +380,250 @@ namespace MGITest {
             EXPECT_GE(viewportDims[1], maxRenderbufferSize);
         }
 
+
+        // THE INDEXED AND PER-PROGRAM QUERIES THAT NAME FRONTEND STATE, pinned on both lanes.
+        //
+        // Both backends used to carry their own arms for GL_SHADER_STORAGE_BUFFER_* and
+        // GL_IMAGE_BINDING_* inside GLFunctionsTable::GetIntegeri_v, and their own
+        // GetInteger64i_v / GetProgramiv table entries. None of it was reachable: GL_Getter and
+        // GL_Program answer every one of these pnames from the frontend's own state and return
+        // before the table is consulted. The duplicates did not even agree - the backend arms
+        // clamped a bound range to the buffer's current storage, which GL 4.6 core tables
+        // 23.4/23.5 do not permit - so the code was one refactor away from becoming the answer.
+        // These cases pin what the frontend actually reports, so a future move of any of it back
+        // behind the interface has to keep saying the same thing.
+        TEST_F(AdvertisedLimitsScenario, IndexedBufferBindingsAreReportedVerbatimOnBothWidths) {
+            GLuint buffer = 0;
+            glGenBuffers(1, &buffer);
+            glBindBuffer(GL_SHADER_STORAGE_BUFFER, buffer);
+            glBufferData(GL_SHADER_STORAGE_BUFFER, 1024, nullptr, GL_DYNAMIC_DRAW);
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+
+            // A range that is NOT the whole buffer, so a clamp to the store would be visible.
+            glBindBufferRange(GL_SHADER_STORAGE_BUFFER, 1, buffer, 256, 512);
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+
+            GLint binding32 = -1;
+            GLint start32 = -1;
+            GLint size32 = -1;
+            glGetIntegeri_v(GL_SHADER_STORAGE_BUFFER_BINDING, 1, &binding32);
+            glGetIntegeri_v(GL_SHADER_STORAGE_BUFFER_START, 1, &start32);
+            glGetIntegeri_v(GL_SHADER_STORAGE_BUFFER_SIZE, 1, &size32);
+            EXPECT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+            EXPECT_EQ(binding32, static_cast<GLint>(buffer));
+            EXPECT_EQ(start32, 256);
+            EXPECT_EQ(size32, 512);
+
+            // The 64-bit width has to agree pname for pname. It has no backend entry of its own
+            // and derives everything from the 32-bit answer above plus its own buffer arm.
+            GLint64 binding64 = -1;
+            GLint64 start64 = -1;
+            GLint64 size64 = -1;
+            glGetInteger64i_v(GL_SHADER_STORAGE_BUFFER_BINDING, 1, &binding64);
+            glGetInteger64i_v(GL_SHADER_STORAGE_BUFFER_START, 1, &start64);
+            glGetInteger64i_v(GL_SHADER_STORAGE_BUFFER_SIZE, 1, &size64);
+            EXPECT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+            EXPECT_EQ(binding64, static_cast<GLint64>(buffer));
+            EXPECT_EQ(start64, static_cast<GLint64>(256));
+            EXPECT_EQ(size64, static_cast<GLint64>(512));
+
+            // An unbound index answers zero rather than erroring or leaking the driver's answer.
+            GLint unbound = -1;
+            glGetIntegeri_v(GL_SHADER_STORAGE_BUFFER_BINDING, 0, &unbound);
+            EXPECT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+            EXPECT_EQ(unbound, 0);
+
+            // THE ARM THAT SEPARATES VERBATIM FROM CLAMPED. GL 4.6 core tables 23.4/23.5 report
+            // the size glBindBufferRange was ASKED for; it does not follow the buffer, so
+            // shrinking the store underneath the binding must not move it. A clamp to the
+            // current storage - which is exactly what both backends' deleted arms did - answers
+            // 128 here, and answers 0 for the bind-then-allocate shape
+            // KHR-GL43.shader_storage_buffer_object.basic-binding uses.
+            glBufferData(GL_SHADER_STORAGE_BUFFER, 128, nullptr, GL_DYNAMIC_DRAW);
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+            GLint startAfterShrink = -1;
+            GLint sizeAfterShrink = -1;
+            GLint64 sizeAfterShrink64 = -1;
+            glGetIntegeri_v(GL_SHADER_STORAGE_BUFFER_START, 1, &startAfterShrink);
+            glGetIntegeri_v(GL_SHADER_STORAGE_BUFFER_SIZE, 1, &sizeAfterShrink);
+            glGetInteger64i_v(GL_SHADER_STORAGE_BUFFER_SIZE, 1, &sizeAfterShrink64);
+            EXPECT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+            EXPECT_EQ(startAfterShrink, 256)
+                << "the bound range's start followed the buffer through a re-specification";
+            EXPECT_EQ(sizeAfterShrink, 512)
+                << "the bound range's size was clamped to the buffer's current 128-byte storage; the range is "
+                   "state of the BINDING POINT and is reported verbatim";
+            EXPECT_EQ(sizeAfterShrink64, static_cast<GLint64>(512))
+                << "the 64-bit width disagreed with the 32-bit one about the same pname";
+
+            glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 1, 0);
+            glDeleteBuffers(1, &buffer);
+            (void)FirstGLError();
+        }
+
+        TEST_F(AdvertisedLimitsScenario, ImageUnitBindingsAreReportedFromTheFrontendState) {
+            GLint maxImageUnits = 0;
+            glGetIntegerv(GL_MAX_IMAGE_UNITS, &maxImageUnits);
+            (void)FirstGLError();
+            if (maxImageUnits < 2) GTEST_SKIP() << "no image units to bind on this lane";
+
+            GLuint texture = 0;
+            glGenTextures(1, &texture);
+            glBindTexture(GL_TEXTURE_2D, texture);
+            glTexStorage2D(GL_TEXTURE_2D, 2, GL_RGBA8, 8, 8);
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+
+            glBindImageTexture(1, texture, 1, GL_FALSE, 0, GL_READ_ONLY, GL_RGBA8);
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+
+            struct Expectation {
+                GLenum pname;
+                const char* name;
+                GLint expected;
+            };
+            const Expectation expectations[] = {
+                {GL_IMAGE_BINDING_NAME, "GL_IMAGE_BINDING_NAME", static_cast<GLint>(texture)},
+                {GL_IMAGE_BINDING_LEVEL, "GL_IMAGE_BINDING_LEVEL", 1},
+                {GL_IMAGE_BINDING_LAYERED, "GL_IMAGE_BINDING_LAYERED", GL_FALSE},
+                {GL_IMAGE_BINDING_LAYER, "GL_IMAGE_BINDING_LAYER", 0},
+                {GL_IMAGE_BINDING_ACCESS, "GL_IMAGE_BINDING_ACCESS", GL_READ_ONLY},
+                {GL_IMAGE_BINDING_FORMAT, "GL_IMAGE_BINDING_FORMAT", GL_RGBA8},
+            };
+            for (const Expectation& expectation : expectations) {
+                GLint value = -424242;
+                glGetIntegeri_v(expectation.pname, 1, &value);
+                EXPECT_EQ(FirstGLError(), GLenum(GL_NO_ERROR)) << expectation.name;
+                EXPECT_EQ(value, expectation.expected) << expectation.name;
+
+                // Same pname through the wide width - it must not fall through to a driver that
+                // knows nothing about MobileGL's image-unit state.
+                GLint64 wide = -424242;
+                glGetInteger64i_v(expectation.pname, 1, &wide);
+                EXPECT_EQ(FirstGLError(), GLenum(GL_NO_ERROR)) << expectation.name << " (64-bit)";
+                EXPECT_EQ(wide, static_cast<GLint64>(expectation.expected)) << expectation.name << " (64-bit)";
+            }
+
+            glBindImageTexture(1, 0, 0, GL_FALSE, 0, GL_READ_ONLY, GL_RGBA8);
+            glDeleteTextures(1, &texture);
+            (void)FirstGLError();
+        }
+
+        // glGetProgramiv(GL_COMPUTE_WORK_GROUP_SIZE) is a LINK ARTIFACT of the program the
+        // application wrote. DirectVulkan used to answer it from its own spirv-reflect cache and
+        // DirectGLES by forwarding to the driver's ESSL program - neither of which the
+        // application ever named - while GL_Program.cpp has always answered it from
+        // ProgramObject::GetComputeLocalSize. This pins the declared local size on both lanes.
+        TEST_F(AdvertisedLimitsScenario, ComputeLocalSizeComesFromTheLinkedProgram) {
+            static const char* kSource = R"(#version 430 core
+layout(local_size_x = 4, local_size_y = 3, local_size_z = 2) in;
+layout(std430, binding = 0) buffer Output { uint g_data[]; };
+void main() { g_data[gl_LocalInvocationIndex] = 1u; }
+)";
+            const GLuint shader = glCreateShader(GL_COMPUTE_SHADER);
+            glShaderSource(shader, 1, &kSource, nullptr);
+            glCompileShader(shader);
+            GLint compiled = 0;
+            glGetShaderiv(shader, GL_COMPILE_STATUS, &compiled);
+            if (compiled == GL_FALSE) {
+                char log[2048] = {};
+                glGetShaderInfoLog(shader, sizeof(log) - 1, nullptr, log);
+                glDeleteShader(shader);
+                (void)FirstGLError();
+                GTEST_SKIP() << "no compute shader support on this lane: " << log;
+            }
+            const GLuint program = glCreateProgram();
+            glAttachShader(program, shader);
+            glLinkProgram(program);
+            glDeleteShader(shader);
+            GLint linked = 0;
+            glGetProgramiv(program, GL_LINK_STATUS, &linked);
+            if (linked == GL_FALSE) {
+                char log[2048] = {};
+                glGetProgramInfoLog(program, sizeof(log) - 1, nullptr, log);
+                glDeleteProgram(program);
+                (void)FirstGLError();
+                GTEST_SKIP() << "the compute program did not link on this lane: " << log;
+            }
+
+            GLint localSize[3] = {-1, -1, -1};
+            glGetProgramiv(program, GL_COMPUTE_WORK_GROUP_SIZE, localSize);
+            EXPECT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+            EXPECT_EQ(localSize[0], 4);
+            EXPECT_EQ(localSize[1], 3);
+            EXPECT_EQ(localSize[2], 2);
+
+            // A program with no compute stage must answer INVALID_OPERATION, not a stale or
+            // defaulted (1, 1, 1) - the frontend's rule, and the one a backend that answers from
+            // its own reflection cache cannot express.
+            const GLuint empty = glCreateProgram();
+            GLint ignored[3] = {0, 0, 0};
+            glGetProgramiv(empty, GL_COMPUTE_WORK_GROUP_SIZE, ignored);
+            EXPECT_EQ(FirstGLError(), GLenum(GL_INVALID_OPERATION))
+                << "GL 4.6 core 7.13: the query is only defined for a linked program with a compute shader";
+
+            glDeleteProgram(empty);
+            glDeleteProgram(program);
+            (void)FirstGLError();
+        }
+
+        // THE SIX COMPUTE LIMITS THAT OUTLIVE THE GETTER. GL_MAX_COMPUTE_WORK_GROUP_COUNT and
+        // GL_MAX_COMPUTE_WORK_GROUP_SIZE, three axes each, are the only indexed pnames the
+        // DEVICE answers rather than the frontend (glGetIntegeri_v on Espryt, VkPhysicalDevice-
+        // Limits on Magma), and therefore the only ones that have to cross the MGPipe boundary
+        // once GetIntegeri_v is retired (plan B section 4.4.6 / P0.5). They ride in MGPCaps by
+        // inclusion, as DynamicBackendParameters::MaxComputeWorkGroupCount/Size, filled by both
+        // backends at capability init. This case pins that the caps copy and the live getter
+        // answer are one number - the getter floors the backend's raw answer at the GL 4.3
+        // minimum, so the comparison is against the floored caps value - and pins the
+        // GL-visible half on every lane: answerability, the floors, vector/indexed agreement
+        // and the index bound. On a lane where the caps block is out of reach (Android links
+        // the shipping .so) only the GL-visible half runs.
+        TEST_F(AdvertisedLimitsScenario, ComputeWorkGroupLimitsAreTheCapsBlocksAnswer) {
+            struct Axis {
+                GLenum pname;
+                const char* name;
+                GLint minimum[3]; // GL 4.3 core table 23.60
+            };
+            const Axis axes[] = {
+                {GL_MAX_COMPUTE_WORK_GROUP_COUNT, "GL_MAX_COMPUTE_WORK_GROUP_COUNT", {65535, 65535, 65535}},
+                {GL_MAX_COMPUTE_WORK_GROUP_SIZE, "GL_MAX_COMPUTE_WORK_GROUP_SIZE", {1024, 1024, 64}},
+            };
+            int capsCount[3] = {0, 0, 0};
+            int capsSize[3] = {0, 0, 0};
+            const bool capsVisible = PeekComputeWorkGroupCaps(capsCount, capsSize);
+
+            for (const Axis& axis : axes) {
+                GLint indexed[3] = {-1, -1, -1};
+                for (GLuint i = 0; i < 3; ++i) {
+                    glGetIntegeri_v(axis.pname, i, &indexed[i]);
+                    ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR)) << axis.name << "[" << i << "]";
+                    EXPECT_GE(indexed[i], axis.minimum[i])
+                        << axis.name << "[" << i << "] = " << indexed[i]
+                        << " is below the GL 4.3 core table 23.60 minimum " << axis.minimum[i];
+                }
+                GLint vector[3] = {-1, -1, -1};
+                glGetIntegerv(axis.pname, vector);
+                ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR)) << axis.name;
+                for (int i = 0; i < 3; ++i) {
+                    EXPECT_EQ(vector[i], indexed[i])
+                        << axis.name << "[" << i << "]: the vector query and the indexed query disagree";
+                }
+                GLint outOfRange = -424242;
+                glGetIntegeri_v(axis.pname, 3, &outOfRange);
+                EXPECT_EQ(FirstGLError(), GLenum(GL_INVALID_VALUE))
+                    << axis.name << "[3]: an index past the three axes is INVALID_VALUE (GL 4.6 core 22.1)";
+
+                if (!capsVisible) continue;
+                const int* capsAxis = axis.pname == GL_MAX_COMPUTE_WORK_GROUP_COUNT ? capsCount : capsSize;
+                for (int i = 0; i < 3; ++i) {
+                    EXPECT_EQ(std::max(capsAxis[i], axis.minimum[i]), indexed[i])
+                        << axis.name << "[" << i << "]: MGPCaps carries " << capsAxis[i]
+                        << " but glGetIntegeri_v answers " << indexed[i]
+                        << " - the caps block and the getter path must be one number, because P0.5 retires "
+                           "the getter in favour of the caps";
+                }
+            }
+        }
+
     } // namespace
 } // namespace MGITest

MobileGL/MG_IntegrationTest/Scenarios/AsyncCompileScenario.cpp

@@ -25,10 +25,12 @@
 //       be able to turn this into a red.
 //   (b) Forcing the join afterwards produces the right answer for every one of them:
 //       GL_COMPILE_STATUS true, an empty info log, and a program that links.
-//   (c) The extension string matches the configuration. This is the half a recorded
-//       trace can never cover - Iris and Sodium change their submission schedule the
-//       moment they see the string - so it is asserted against a real backend's real
-//       GL_EXTENSIONS, through both glGetString and glGetStringi.
+//   (c) The extension string matches the configuration - where "the configuration" is
+//       MOBILEGL_ASYNC_SHADER_COMPILE as this process inherited it, and NOT anything the
+//       implementation says about itself. This is the half a recorded trace can never
+//       cover - Iris and Sodium change their submission schedule the moment they see the
+//       string - so it is asserted against a real backend's real GL_EXTENSIONS, through
+//       both glGetString and glGetStringi.
 //   (d) glMaxShaderCompilerThreadsKHR(0) leaves nothing in flight: every subsequent
 //       GL_COMPLETION_STATUS_KHR reads GL_TRUE immediately, and compilation after it
 //       is synchronous. That is what the extension requires of a zero count.
@@ -40,6 +42,27 @@
 //
 // Backend selection is the module's usual one process, one backend (MOBILEGL_BACKEND_TYPE),
 // so this file runs twice per ctest invocation.
+//
+// COMPILATION MODE IS PER PROCESS TOO. Every case here needs a particular configuration of
+// MobileGL's shader compiler, and takes it from the ENVIRONMENT
+// (MOBILEGL_ASYNC_SHADER_COMPILE, MOBILEGL_ASYNC_OPTIMISTIC_SHADER_STATUS) rather than by
+// writing MG_Config::Features on the way past. Half of what those variables decide is
+// latched before the first GL call - the compile pool and its threads, and the advertised
+// extension list a backend builds once from the configuration in force at its first use -
+// so an in-process poke could only ever have moved the other half; and on Android it could
+// move nothing at all, because this module links against the shipping libMobileGL.so, which
+// exports no such symbol. A case whose process is not in the configuration it needs SKIPS
+// with that as its reason. CMakeLists.txt registers the extra ctest entries that put a
+// process into each configuration (AsyncOn., AsyncOff., OptimisticShaderStatus.), so one
+// ctest run still covers both sides of every switch. Run straight from a shell with nothing
+// set - the on-device shape - the ambient configuration runs and the rest skip cleanly.
+//
+// WITHIN one process, "compiled on a worker" versus "compiled on this thread" is switched
+// through glMaxShaderCompilerThreadsKHR, the extension's own entry point: a zero count joins
+// everything outstanding and compiles inline from then on, any nonzero count lifts that
+// again, and 0xFFFFFFFF asks for the implementation maximum (GL_Program.cpp,
+// MaxShaderCompilerThreadsKHR_State). Doing it through the public call rather than the
+// feature table means the switching is itself part of what these cases exercise.
 
 #include <string>
 #include <vector>
@@ -47,9 +70,6 @@
 #include "../Harness/HeadlessGL.h"
 #include "../Harness/ScenarioFixture.h"
 
-#include "Config.h"
-#include "MG_Util/Async/ShaderCompilePool.h"
-
 #ifdef GLAPI
 #undef GLAPI
 #endif
@@ -76,8 +96,6 @@ extern "C" void glMaxShaderCompilerThreadsKHR(GLuint count);
 namespace MGITest {
     namespace {
 
-        using MobileGL::MG_Config::QuirkOverride;
-
         // Same shape as the other scenarios: a two-attribute pass-through, so the only
         // thing that can differ between the two compilation modes is the compilation.
         constexpr const char* kVertexSource = R"(#version 330 core
@@ -139,50 +157,40 @@ void main() {
             return source;
         }
 
-        // MOBILEGL_ASYNC_SHADER_COMPILE decides the ambient mode; a scenario that wants
-        // the other one says so here and gets the ambient one back on scope exit. Forcing
-        // it in-process is what lets ONE ctest run compare the two modes against each
-        // other - the whole point of (e).
-        class AsyncModeScope {
-        public:
-            explicit AsyncModeScope(bool async) : m_saved(MobileGL::MG_Config::Features.AsyncShaderCompile) {
-                MobileGL::MG_Config::Features.AsyncShaderCompile =
-                    async ? QuirkOverride::ForceOn : QuirkOverride::ForceOff;
+        // Whether this context advertises GL_KHR_parallel_shader_compile, which is exactly
+        // "MobileGL is configured to compile asynchronously" as an application can see it:
+        // the backends gate the string on AsyncShaderCompileEnabled() and on nothing else
+        // (BackendObject_DirectGLES.cpp / BackendObject_DirectVulkan.cpp), and the string
+        // is the only way MobileGL ever tells anyone. A case that needs asynchronous
+        // compilation checks for it the way an application would, and skips without it.
+        //
+        // The INDEXED form, because that is the one a core-profile application reads.
+        bool HasParallelShaderCompile() {
+            GLint count = 0;
+            glGetIntegerv(GL_NUM_EXTENSIONS, &count);
+            for (GLint i = 0; i < count; ++i) {
+                const char* name = reinterpret_cast<const char*>(glGetStringi(GL_EXTENSIONS, GLuint(i)));
+                if (name != nullptr && std::string(name) == "GL_KHR_parallel_shader_compile") return true;
             }
-            ~AsyncModeScope() { MobileGL::MG_Config::Features.AsyncShaderCompile = m_saved; }
-            AsyncModeScope(const AsyncModeScope&) = delete;
-            AsyncModeScope& operator=(const AsyncModeScope&) = delete;
+            return false;
+        }
 
-        private:
-            const QuirkOverride m_saved;
-        };
-
-        // MOBILEGL_ASYNC_OPTIMISTIC_SHADER_STATUS, forced in-process for the same reason
-        // as AsyncModeScope: one ctest run asserts the quirk against the ambient default.
-        class OptimisticStatusScope {
-        public:
-            explicit OptimisticStatusScope(const QuirkOverride mode)
-                : m_saved(MobileGL::MG_Config::Features.AsyncOptimisticShaderStatus) {
-                MobileGL::MG_Config::Features.AsyncOptimisticShaderStatus = mode;
-            }
-            ~OptimisticStatusScope() { MobileGL::MG_Config::Features.AsyncOptimisticShaderStatus = m_saved; }
-            OptimisticStatusScope(const OptimisticStatusScope&) = delete;
-            OptimisticStatusScope& operator=(const OptimisticStatusScope&) = delete;
-
-        private:
-            const QuirkOverride m_saved;
-        };
-
-        // glMaxShaderCompilerThreadsKHR writes process-wide state; a scenario that calls
-        // it has to put the pool back or it changes how every scenario after it compiles.
+        // glMaxShaderCompilerThreadsKHR writes process-wide state; a scenario that calls it
+        // has to put the pool back or it changes how every scenario after it compiles.
+        //
+        // The restore is the extension's own "implementation maximum" spelling rather than a
+        // hand-rolled poke at the pool. glMaxShaderCompilerThreadsKHR(0xFFFFFFFF) is defined
+        // (GL_Program.cpp, MaxShaderCompilerThreadsKHR_State) as precisely the two steps this
+        // used to perform through internal entry points - concurrency := the pool's full
+        // thread count, then lift any suspension a zero count had armed - in the safer order,
+        // since it raises the budget before re-admitting work rather than after. Going through
+        // the public call also puts the restore path itself under test, and it is the only
+        // spelling available on Android, where this module links the shipping shared library
+        // and can reach nothing but the GL entry points.
         class CompilerThreadScope {
         public:
             CompilerThreadScope() = default;
-            ~CompilerThreadScope() {
-                MobileGL::MG_Util::Async::SetAsyncShaderCompileSuspended(false);
-                auto& pool = MobileGL::MG_Util::Async::ShaderCompilePool::Get();
-                pool.SetMaxConcurrency(pool.GetThreadCount());
-            }
+            ~CompilerThreadScope() { glMaxShaderCompilerThreadsKHR(0xFFFFFFFFu); }
             CompilerThreadScope(const CompilerThreadScope&) = delete;
             CompilerThreadScope& operator=(const CompilerThreadScope&) = delete;
         };
@@ -293,7 +301,12 @@ void main() {
         // interesting for shaders that (a) proved were genuinely still outstanding.
         TEST_F(AsyncCompileScenario, CompletionStatusPollingThenForcedJoin) {
             if (!Ready()) return;
-            const AsyncModeScope async(true);
+            if (!HasParallelShaderCompile()) {
+                GTEST_SKIP() << "this process is configured to compile inline "
+                                "(GL_KHR_parallel_shader_compile is not advertised), so no compile can be "
+                                "outstanding; the AsyncOn. ctest entries run this case with "
+                                "MOBILEGL_ASYNC_SHADER_COMPILE=1";
+            }
             const CompilerThreadScope threads;
             // One worker, so the queue behind it is what the poll observes.
             glMaxShaderCompilerThreadsKHR(1);
@@ -341,14 +354,33 @@ void main() {
         }
 
         // ---- (c) ------------------------------------------------------------------
-        // The extension string, read from a real backend that really brought a driver
-        // up. No mode forcing here: a backend builds its advertised list once, from the
-        // configuration in force at its first use, so the meaningful assertion is
-        // against the AMBIENT configuration - which is exactly what makes this case
-        // worth running in both of the suite's flag states.
+        // The extension string, read from a real backend that really brought a driver up.
+        //
+        // The expectation comes from the ENVIRONMENT, never from the implementation. This
+        // case used to derive it by calling AsyncShaderCompileEnabled() - which is the same
+        // function the backends gate the string on, so the two halves could only ever agree
+        // and the case would have passed however wrong both of them were. Asserting an
+        // implementation against itself pins nothing.
+        //
+        // MOBILEGL_ASYNC_SHADER_COMPILE is the whole input: the process inherited it before
+        // any GL call, a backend builds its advertised list once from the configuration in
+        // force at first use, and nothing in this process can move it afterwards. So reading
+        // the variable IS reading the configuration, independently. With the variable unset
+        // the configuration in force is MobileGL's built-in default, which only the
+        // implementation knows - there is nothing independent left to compare against, and
+        // this case says so rather than inventing an expectation. The AsyncOn. and AsyncOff.
+        // ctest entries pin the variable to each of its two values, so one ctest run still
+        // asserts both the advertised and the withdrawn side.
         TEST_F(AsyncCompileScenario, ExtensionStringMatchesTheConfiguration) {
             if (!Ready()) return;
-            const bool expected = MobileGL::MG_Util::Async::AsyncShaderCompileEnabled();
+            const AmbientQuirk configured = AmbientQuirkFromEnvironment("MOBILEGL_ASYNC_SHADER_COMPILE");
+            if (configured == AmbientQuirk::Auto) {
+                GTEST_SKIP() << "MOBILEGL_ASYNC_SHADER_COMPILE is unset, so the configuration in force is "
+                                "MobileGL's built-in default and the only way to learn it would be to ask "
+                                "the implementation this case exists to check; the AsyncOn. and AsyncOff. "
+                                "ctest entries run it with the variable pinned to each of its two values";
+            }
+            const bool expected = configured == AmbientQuirk::On;
 
             const char* extensions = reinterpret_cast<const char*>(glGetString(GL_EXTENSIONS));
             ASSERT_NE(extensions, nullptr);
@@ -385,7 +417,12 @@ void main() {
         // A zero count must leave nothing in flight and keep it that way.
         TEST_F(AsyncCompileScenario, ZeroCompilerThreadsSettlesEverythingImmediately) {
             if (!Ready()) return;
-            const AsyncModeScope async(true);
+            if (!HasParallelShaderCompile()) {
+                GTEST_SKIP() << "this process is configured to compile inline "
+                                "(GL_KHR_parallel_shader_compile is not advertised), so a zero count has "
+                                "nothing to settle; the AsyncOn. ctest entries run this case with "
+                                "MOBILEGL_ASYNC_SHADER_COMPILE=1";
+            }
             const CompilerThreadScope threads;
             glMaxShaderCompilerThreadsKHR(1);
 
@@ -417,12 +454,28 @@ void main() {
         // Compared through the DEFAULT framebuffer deliberately: that is where the
         // backend's orientation and present path live, so the comparison covers the
         // whole pipeline rather than the reflection tables alone.
+        //
+        // The two modes are selected through glMaxShaderCompilerThreadsKHR, the extension's
+        // own entry point, rather than through the feature table: a zero count joins
+        // everything outstanding and makes every later glCompileShader/glLinkProgram run its
+        // body on the calling thread, and 0xFFFFFFFF lifts that again with the pool at its
+        // full thread count (GL_Program.cpp, MaxShaderCompilerThreadsKHR_State; the compile
+        // and link paths both gate on AsyncShaderCompileActive(), which is what the zero
+        // count switches). So this is still one process comparing worker-built artifacts
+        // against inline-built ones - just asked for the way an application asks.
         TEST_F(AsyncCompileScenario, AsyncAndSyncProgramsRenderIdenticalFrames) {
             if (!Ready()) return;
+            if (!HasParallelShaderCompile()) {
+                GTEST_SKIP() << "this process is configured to compile inline "
+                                "(GL_KHR_parallel_shader_compile is not advertised), so both halves would "
+                                "be the same inline build and the comparison would be vacuous; the "
+                                "AsyncOn. ctest entries run this case with MOBILEGL_ASYNC_SHADER_COMPILE=1";
+            }
+            const CompilerThreadScope threads;
 
             Image asyncImage;
             {
-                const AsyncModeScope async(true);
+                glMaxShaderCompilerThreadsKHR(0xFFFFFFFFu);
                 const GLuint program = BuildProgram();
                 ASSERT_NE(program, 0u);
                 asyncImage = DrawFrameWith(program);
@@ -431,7 +484,7 @@ void main() {
 
             Image syncImage;
             {
-                const AsyncModeScope async(false);
+                glMaxShaderCompilerThreadsKHR(0);
                 const GLuint program = BuildProgram();
                 ASSERT_NE(program, 0u);
                 syncImage = DrawFrameWith(program);
@@ -456,11 +509,16 @@ void main() {
         // candidate) shows up here and not in the single-program case above.
         TEST_F(AsyncCompileScenario, ABatchOfAsyncProgramsAllRenderCorrectly) {
             if (!Ready()) return;
+            if (!HasParallelShaderCompile()) {
+                GTEST_SKIP() << "this process is configured to compile inline "
+                                "(GL_KHR_parallel_shader_compile is not advertised), so nothing would be "
+                                "built on a worker and there is no per-worker state to leak; the AsyncOn. "
+                                "ctest entries run this case with MOBILEGL_ASYNC_SHADER_COMPILE=1";
+            }
             constexpr int kPrograms = 12;
 
             std::vector<GLuint> programs;
             {
-                const AsyncModeScope async(true);
                 const CompilerThreadScope threads;
                 glMaxShaderCompilerThreadsKHR(1);
                 // Everything enqueued before anything is read: the only shape in which
@@ -489,6 +547,21 @@ void main() {
         // then mis-renders - shows up here as a wrong quadrant signature.
         TEST_F(AsyncCompileScenario, IrisShapedTwoPhaseBatchRendersCorrectly) {
             if (!Ready()) return;
+            // The quirk is off by default and never advertised, so unlike the cases above
+            // there is no GL observable that says whether it is in force - only the variable
+            // that put it there. It also has to be set BEFORE this process started for the
+            // shape to be the real one: the optimistic answer is latched per compile, and a
+            // quirk switched on mid-process would only cover the compiles after it.
+            if (AmbientQuirkFromEnvironment("MOBILEGL_ASYNC_OPTIMISTIC_SHADER_STATUS") != AmbientQuirk::On) {
+                GTEST_SKIP() << "this case is the optimistic-status quirk's end-to-end shape and needs it on "
+                                "for the whole process; the OptimisticShaderStatus. ctest entries run it with "
+                                "MOBILEGL_ASYNC_OPTIMISTIC_SHADER_STATUS=1";
+            }
+            if (!HasParallelShaderCompile()) {
+                GTEST_SKIP() << "the optimistic status only ever applies to a compile that is still in flight "
+                                "(OptimisticShaderStatusActive() requires AsyncShaderCompileActive()), and "
+                                "this process is configured to compile inline";
+            }
             constexpr int kPrograms = 12;
 
             // Distinct per program (so neither the source memo nor the adoption map turns
@@ -508,8 +581,6 @@ void main() {
 
             std::vector<GLuint> programs;
             {
-                const AsyncModeScope async(true);
-                const OptimisticStatusScope quirk(QuirkOverride::ForceOn);
                 const CompilerThreadScope threads;
                 glMaxShaderCompilerThreadsKHR(1);
 

MobileGL/MG_IntegrationTest/Scenarios/AtomicCounterScenario.cpp

@@ -0,0 +1,267 @@
+// MobileGL - MobileGL/MG_IntegrationTest/Scenarios/AtomicCounterScenario.cpp
+// Copyright (c) 2025-2026 MobileGL-Dev
+// Licensed under the GNU Lesser General Public License v3.0:
+//   https://www.gnu.org/licenses/gpl-3.0.txt
+//   https://www.gnu.org/licenses/lgpl-3.0.txt
+// SPDX-License-Identifier: LGPL-3.0-only
+// End of Source File Header
+//
+// Scenario - ATOMIC COUNTERS, END TO END.
+//
+// GL_ATOMIC_COUNTER_BUFFER does not exist in ES, and glslang does not hand one to a backend
+// either: its Vulkan-relaxed parse rewrites every atomic_uint into a uint member of a
+// synthesized gl_AtomicCounterBlock_<N> STORAGE block. Making counters work therefore means
+// closing two open ends that used to be missing entirely -
+//
+//   * the block's shader-storage binding, which the IO mapper picked at random and which had no
+//     relation to the GL binding point N the application bound its buffer to (and could alias an
+//     SSBO the application binds itself), is moved to a slot reserved at the top of the driver's
+//     range; and
+//   * the buffer bound at GL_ATOMIC_COUNTER_BUFFER point N, which nothing in the ES backend ever
+//     read, is re-issued as a shader-storage binding at that reserved slot.
+//
+// Neither end alone is observable: with only the first the shader increments a block nobody
+// bound a buffer to, with only the second the buffer lands where the shader does not look. The
+// only thing that proves both is the VALUE, so every assertion here reads the counter back.
+//
+// Compute rather than a draw on purpose: the invocation count is exactly what was dispatched,
+// while a fragment stage's is a property of the rasterizer (helper invocations, early depth).
+// Conformance cases behind this: KHR-GL42/GL43.shader_atomic_counters.basic-usage-cs,
+// .advanced-usage-multi-stage and .advanced-usage-draw-update-draw.
+
+#include <string>
+#include <vector>
+
+#include "../Harness/HeadlessGL.h"
+#include "../Harness/ScenarioFixture.h"
+
+#ifdef GLAPI
+#undef GLAPI
+#endif
+#define GL_GLEXT_PROTOTYPES
+#include <GL/gl.h>
+#include <GL/glcorearb.h>
+#undef GL_GLEXT_PROTOTYPES
+
+namespace MGITest {
+    namespace {
+
+        // Two counters share binding 0 at DIFFERENT offsets and a third sits alone on binding 1.
+        // The offsets are what separates "the buffer arrived" from "the buffer arrived and the
+        // block is laid out the way GL says": a lowering that packed the members in declaration
+        // order without honouring `offset` would still pass a single-counter check.
+        constexpr const char* kCounterComputeSource = R"(#version 430 core
+layout(local_size_x = 4) in;
+layout(binding = 0, offset = 0) uniform atomic_uint g_first;
+layout(binding = 0, offset = 4) uniform atomic_uint g_second;
+layout(binding = 1, offset = 0) uniform atomic_uint g_other;
+void main() {
+    atomicCounterIncrement(g_first);
+    atomicCounterIncrement(g_second);
+    atomicCounterIncrement(g_second);
+    atomicCounterIncrement(g_other);
+}
+)";
+
+        constexpr int kLocalSizeX = 4;
+        constexpr int kWorkGroups = 2;
+        constexpr unsigned int kInvocations = kLocalSizeX * kWorkGroups;
+
+        // Deliberately non-zero: the shader adds to whatever the application uploaded, so a seed
+        // that survives is also proof that the buffer's CPU-side contents reached the driver.
+        constexpr unsigned int kSeedFirst = 5;
+        constexpr unsigned int kSeedSecond = 100;
+        constexpr unsigned int kSeedOther = 7;
+
+        class AtomicCounterScenario : public ScenarioTest {
+        protected:
+            void SetUp() override {
+                ScenarioTest::SetUp();
+                if (!Ready()) return;
+                GLint counters = 0;
+                glGetIntegerv(GL_MAX_COMPUTE_ATOMIC_COUNTERS, &counters);
+                GLint buffers = 0;
+                glGetIntegerv(GL_MAX_COMPUTE_ATOMIC_COUNTER_BUFFERS, &buffers);
+                if (counters < 3 || buffers < 2) {
+                    GTEST_SKIP() << "GL_MAX_COMPUTE_ATOMIC_COUNTERS is " << counters
+                                 << " and GL_MAX_COMPUTE_ATOMIC_COUNTER_BUFFERS is " << buffers
+                                 << "; this needs 3 and 2";
+                }
+                m_program = CompileComputeProgram(kCounterComputeSource);
+                ASSERT_NE(m_program, 0u) << m_buildLog;
+            }
+
+            void TearDown() override {
+                if (!Ready()) return;
+                glUseProgram(0);
+                if (!m_buffers.empty()) glDeleteBuffers(static_cast<GLsizei>(m_buffers.size()), m_buffers.data());
+                if (m_program != 0) glDeleteProgram(m_program);
+                m_buffers.clear();
+                m_program = 0;
+            }
+
+            unsigned int CompileComputeProgram(const char* source) {
+                const GLuint shader = glCreateShader(GL_COMPUTE_SHADER);
+                glShaderSource(shader, 1, &source, nullptr);
+                glCompileShader(shader);
+                GLint compiled = 0;
+                glGetShaderiv(shader, GL_COMPILE_STATUS, &compiled);
+                if (compiled == GL_FALSE) {
+                    char log[2048] = {};
+                    glGetShaderInfoLog(shader, sizeof(log) - 1, nullptr, log);
+                    m_buildLog = std::string("compute shader did not compile: ") + log;
+                    glDeleteShader(shader);
+                    return 0;
+                }
+                const GLuint program = glCreateProgram();
+                glAttachShader(program, shader);
+                glLinkProgram(program);
+                glDeleteShader(shader);
+                GLint linked = 0;
+                glGetProgramiv(program, GL_LINK_STATUS, &linked);
+                if (linked == GL_FALSE) {
+                    char log[2048] = {};
+                    glGetProgramInfoLog(program, sizeof(log) - 1, nullptr, log);
+                    m_buildLog = std::string("compute program did not link: ") + log;
+                    glDeleteProgram(program);
+                    return 0;
+                }
+                return program;
+            }
+
+            // A counter buffer of `count` uints, seeded and bound to atomic-counter point
+            // `binding`.
+            GLuint MakeCounterBuffer(GLuint binding, const std::vector<unsigned int>& seed) {
+                GLuint buffer = 0;
+                glGenBuffers(1, &buffer);
+                glBindBuffer(GL_ATOMIC_COUNTER_BUFFER, buffer);
+                glBufferData(GL_ATOMIC_COUNTER_BUFFER,
+                             static_cast<GLsizeiptr>(seed.size() * sizeof(unsigned int)), seed.data(),
+                             GL_DYNAMIC_DRAW);
+                glBindBufferBase(GL_ATOMIC_COUNTER_BUFFER, binding, buffer);
+                glBindBuffer(GL_ATOMIC_COUNTER_BUFFER, 0);
+                m_buffers.push_back(buffer);
+                return buffer;
+            }
+
+            std::vector<unsigned int> ReadCounters(GLuint buffer, int count) {
+                std::vector<unsigned int> values(static_cast<std::size_t>(count), 0xDEADBEEFu);
+                glBindBuffer(GL_ATOMIC_COUNTER_BUFFER, buffer);
+                glGetBufferSubData(GL_ATOMIC_COUNTER_BUFFER, 0,
+                                   static_cast<GLsizeiptr>(values.size() * sizeof(unsigned int)), values.data());
+                glBindBuffer(GL_ATOMIC_COUNTER_BUFFER, 0);
+                return values;
+            }
+
+            void Dispatch() {
+                glUseProgram(m_program);
+                glDispatchCompute(kWorkGroups, 1, 1);
+                glMemoryBarrier(GL_ATOMIC_COUNTER_BARRIER_BIT | GL_BUFFER_UPDATE_BARRIER_BIT);
+            }
+
+            unsigned int m_program = 0;
+            std::string m_buildLog;
+            std::vector<GLuint> m_buffers;
+        };
+
+    } // namespace
+
+    // The counter values a dispatch leaves behind, per binding point and per offset within one
+    // binding. Nothing in the ES backend used to touch BufferTarget::AtomicCounter at all, so
+    // before the wiring landed every one of these read back its seed unchanged.
+    TEST_F(AtomicCounterScenario, DispatchIncrementsTheBoundCounterBuffers) {
+        if (!Ready() || IsSkipped()) return;
+
+        const GLuint zero = MakeCounterBuffer(0, {kSeedFirst, kSeedSecond});
+        const GLuint one = MakeCounterBuffer(1, {kSeedOther});
+        ASSERT_EQ(FirstGLError(), 0u) << "binding the counter buffers raised a GL error";
+
+        Dispatch();
+        EXPECT_EQ(FirstGLError(), 0u) << "the dispatch raised a GL error";
+
+        const std::vector<unsigned int> zeroValues = ReadCounters(zero, 2);
+        const std::vector<unsigned int> oneValues = ReadCounters(one, 1);
+        EXPECT_EQ(FirstGLError(), 0u) << "reading the counters back raised a GL error";
+
+        EXPECT_EQ(zeroValues[0], kSeedFirst + kInvocations)
+            << "binding 0 offset 0 read back " << zeroValues[0] << "; " << kSeedFirst
+            << " means the shader's increments never reached the buffer the application bound";
+        EXPECT_EQ(zeroValues[1], kSeedSecond + 2 * kInvocations)
+            << "binding 0 offset 4 read back " << zeroValues[1] << "; the seed means the counter at a NON-ZERO "
+            << "offset was not carried through the lowering, even though offset 0 was";
+        EXPECT_EQ(oneValues[0], kSeedOther + kInvocations)
+            << "binding 1 read back " << oneValues[0] << "; a counter buffer past the first binding point "
+            << "resolves to a different reserved slot and is where an off-by-one shows up";
+    }
+
+    // A second dispatch continues from where the first left off, and a re-seed between them is
+    // visible to the shader. Both halves of the buffer's traffic have to work, in both
+    // directions: the increments are only observable through the readback path, and the re-seed
+    // is only observable if the upload reaches the driver AFTER the buffer has been GPU-written.
+    TEST_F(AtomicCounterScenario, CountersAccumulateAcrossDispatchesAndFollowAReseed) {
+        if (!Ready() || IsSkipped()) return;
+
+        const GLuint zero = MakeCounterBuffer(0, {0u, 0u});
+        MakeCounterBuffer(1, {0u});
+        ASSERT_EQ(FirstGLError(), 0u);
+
+        Dispatch();
+        Dispatch();
+        std::vector<unsigned int> values = ReadCounters(zero, 2);
+        EXPECT_EQ(FirstGLError(), 0u);
+        EXPECT_EQ(values[0], 2 * kInvocations) << "two dispatches did not accumulate";
+        EXPECT_EQ(values[1], 4 * kInvocations) << "two dispatches did not accumulate at offset 4";
+
+        const unsigned int reseed[2] = {1000u, 2000u};
+        glBindBuffer(GL_ATOMIC_COUNTER_BUFFER, zero);
+        glBufferSubData(GL_ATOMIC_COUNTER_BUFFER, 0, sizeof(reseed), reseed);
+        glBindBuffer(GL_ATOMIC_COUNTER_BUFFER, 0);
+        ASSERT_EQ(FirstGLError(), 0u) << "re-seeding the counter buffer raised a GL error";
+
+        Dispatch();
+        values = ReadCounters(zero, 2);
+        EXPECT_EQ(FirstGLError(), 0u);
+        EXPECT_EQ(values[0], reseed[0] + kInvocations) << "the re-seeded value did not reach the shader";
+        EXPECT_EQ(values[1], reseed[1] + 2 * kInvocations) << "the re-seeded value at offset 4 did not reach the shader";
+    }
+
+    // A CPU glBufferSubData issued AFTER a dispatch, read back with NO further GPU work in
+    // between. Each backend has its own way to invert this pair, and both are pinned here.
+    // DirectGLES queues app SubData ranges for the draw-time staged-copy flush (the upload
+    // ring) instead of uploading in place, and readback of a GPU-written buffer overwrites
+    // the frontend shadow with the driver copy - so if the readback path forgets to flush the
+    // queued range first, the newer CPU write is REVERTED by the readback and offset 0 reads
+    // the dispatch's value instead of the reseed. DirectVulkan adopts the buffer into
+    // coherent GPU memory the moment the dispatch resolves its descriptor, so the SubData
+    // write lands in the very bytes the GPU reads - while the dispatch still sits recorded in
+    // the deferred frame command buffer. Unless the frontend retires that pending work before
+    // writing the adopted store (BufferObject::UploadSubData), the dispatch executes ON TOP
+    // of the reseed and offset 0 reads reseed + increments instead of the reseed. Offset 4
+    // pins the other direction for both: the upload must leave bytes outside its range - the
+    // dispatch's results - untouched.
+    TEST_F(AtomicCounterScenario, SubDataAfterDispatchSurvivesAnImmediateReadback) {
+        if (!Ready() || IsSkipped()) return;
+
+        const GLuint zero = MakeCounterBuffer(0, {0u, 0u});
+        MakeCounterBuffer(1, {0u});
+        ASSERT_EQ(FirstGLError(), 0u);
+
+        Dispatch();
+
+        const unsigned int reseed = 4242u;
+        glBindBuffer(GL_ATOMIC_COUNTER_BUFFER, zero);
+        glBufferSubData(GL_ATOMIC_COUNTER_BUFFER, 0, sizeof(reseed), &reseed);
+        glBindBuffer(GL_ATOMIC_COUNTER_BUFFER, 0);
+        ASSERT_EQ(FirstGLError(), 0u) << "re-seeding the counter buffer raised a GL error";
+
+        const std::vector<unsigned int> values = ReadCounters(zero, 2);
+        EXPECT_EQ(FirstGLError(), 0u);
+        EXPECT_EQ(values[0], reseed)
+            << "offset 0 read back " << values[0] << "; the dispatch's value (" << kInvocations
+            << ") means the readback ran before the queued SubData range was flushed and reverted it";
+        EXPECT_EQ(values[1], 2 * kInvocations)
+            << "offset 4 read back " << values[1] << "; the SubData flush must leave bytes outside its "
+            << "range untouched";
+    }
+
+} // namespace MGITest

MobileGL/MG_IntegrationTest/Scenarios/BufferTextureScenario.cpp

@@ -299,4 +299,99 @@ void main() {
         EXPECT_EQ(FirstGLError(), 0u);
     }
 
+    // glGetTexLevelParameter used to refuse EVERY pname on a buffer texture: WIDTH/HEIGHT/DEPTH
+    // fell out of a mipmap-only switch as GL_INVALID_OPERATION, and GL_TEXTURE_BUFFER_SIZE /
+    // GL_TEXTURE_BUFFER_OFFSET were not in the switch at all, so they came back GL_INVALID_ENUM.
+    // KHR-GL43.texture_buffer wraps both queries in GLU_EXPECT_NO_ERROR, so the error alone fails
+    // the case before any value is compared.
+    //
+    // The two halves report DIFFERENT units and only one of them is clamped, which is the thing
+    // easiest to get backwards: WIDTH is a TEXEL count clamped to GL_MAX_TEXTURE_BUFFER_SIZE,
+    // BUFFER_SIZE is the range in basic machine units exactly as it was given.
+    TEST_F(BufferTextureScenario, LevelQueriesDescribeTheAttachedBufferRange) {
+        if (!Ready()) return;
+        FirstGLError();
+
+        GLint offsetAlignment = 1;
+        glGetIntegerv(GL_TEXTURE_BUFFER_OFFSET_ALIGNMENT, &offsetAlignment);
+        if (offsetAlignment < 1) offsetAlignment = 1;
+        GLint maxTexels = 0;
+        glGetIntegerv(GL_MAX_TEXTURE_BUFFER_SIZE, &maxTexels);
+        ASSERT_EQ(FirstGLError(), 0u);
+        ASSERT_GT(maxTexels, 0) << "an OpenGL 4.x context may not advertise a zero buffer-texture limit";
+
+        constexpr GLint kTexelBytes = 4; // GL_RGBA8
+        const GLsizeiptr rangeOffset = static_cast<GLsizeiptr>(offsetAlignment);
+        const GLsizeiptr rangeBytes = 32 * kTexelBytes;
+        // Deliberately bigger than the range, so a getter that answered out of the BUFFER rather
+        // than out of the texture's window would be caught.
+        const GLsizeiptr bufferBytes = rangeOffset + rangeBytes + 16 * kTexelBytes;
+
+        const std::vector<GLubyte> zeros(static_cast<size_t>(bufferBytes), 0);
+        GLuint buffer = 0;
+        glGenBuffers(1, &buffer);
+        glBindBuffer(GL_TEXTURE_BUFFER, buffer);
+        glBufferData(GL_TEXTURE_BUFFER, bufferBytes, zeros.data(), GL_STATIC_DRAW);
+
+        GLuint texture = 0;
+        glGenTextures(1, &texture);
+        glBindTexture(GL_TEXTURE_BUFFER, texture);
+        glTexBufferRange(GL_TEXTURE_BUFFER, GL_RGBA8, buffer, rangeOffset, rangeBytes);
+        ASSERT_EQ(FirstGLError(), 0u) << "glTexBufferRange(GL_RGBA8) was refused";
+
+        const auto levelQuery = [](GLenum pname) {
+            GLint value = -1;
+            glGetTexLevelParameteriv(GL_TEXTURE_BUFFER, 0, pname, &value);
+            return value;
+        };
+        const auto levelQueryF = [](GLenum pname) {
+            GLfloat value = -1.0f;
+            glGetTexLevelParameterfv(GL_TEXTURE_BUFFER, 0, pname, &value);
+            return value;
+        };
+
+        EXPECT_EQ(levelQuery(GL_TEXTURE_WIDTH), static_cast<GLint>(rangeBytes / kTexelBytes))
+            << "GL_TEXTURE_WIDTH is a texel count over the attached RANGE";
+        EXPECT_EQ(levelQuery(GL_TEXTURE_HEIGHT), 1);
+        EXPECT_EQ(levelQuery(GL_TEXTURE_DEPTH), 1);
+        EXPECT_EQ(levelQuery(GL_TEXTURE_BUFFER_SIZE), static_cast<GLint>(rangeBytes))
+            << "GL_TEXTURE_BUFFER_SIZE reports basic machine units, not texels";
+        EXPECT_EQ(levelQuery(GL_TEXTURE_BUFFER_OFFSET), static_cast<GLint>(rangeOffset));
+        EXPECT_EQ(FirstGLError(), 0u) << "a buffer-texture level query raised an error";
+        EXPECT_LE(levelQuery(GL_TEXTURE_WIDTH), maxTexels)
+            << "GL_TEXTURE_WIDTH must stay clamped to GL_MAX_TEXTURE_BUFFER_SIZE";
+
+        // The float getter is a separate switch and has drifted from the integer one before.
+        EXPECT_FLOAT_EQ(levelQueryF(GL_TEXTURE_WIDTH), static_cast<GLfloat>(rangeBytes / kTexelBytes));
+        EXPECT_FLOAT_EQ(levelQueryF(GL_TEXTURE_HEIGHT), 1.0f);
+        EXPECT_FLOAT_EQ(levelQueryF(GL_TEXTURE_BUFFER_SIZE), static_cast<GLfloat>(rangeBytes));
+        EXPECT_EQ(FirstGLError(), 0u) << "the float form of a buffer-texture level query raised an error";
+
+        // The whole-buffer form follows the buffer's current size instead of freezing a window.
+        glTexBuffer(GL_TEXTURE_BUFFER, GL_RGBA8, buffer);
+        EXPECT_EQ(levelQuery(GL_TEXTURE_BUFFER_OFFSET), 0);
+        EXPECT_EQ(levelQuery(GL_TEXTURE_BUFFER_SIZE), static_cast<GLint>(bufferBytes));
+        EXPECT_EQ(levelQuery(GL_TEXTURE_WIDTH), static_cast<GLint>(bufferBytes / kTexelBytes));
+        EXPECT_EQ(FirstGLError(), 0u);
+
+        // Both buffer pnames belong to buffer textures alone; anything else is INVALID_OPERATION,
+        // the same shape GL_TEXTURE_COMPRESSED_IMAGE_SIZE uses for an uncompressed image.
+        GLuint plainTexture = 0;
+        glGenTextures(1, &plainTexture);
+        glBindTexture(GL_TEXTURE_2D, plainTexture);
+        glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA8, 4, 4, 0, GL_RGBA, GL_UNSIGNED_BYTE, nullptr);
+        EXPECT_EQ(FirstGLError(), 0u);
+        GLint unused = -1;
+        glGetTexLevelParameteriv(GL_TEXTURE_2D, 0, GL_TEXTURE_BUFFER_SIZE, &unused);
+        EXPECT_EQ(FirstGLError(), static_cast<unsigned int>(GL_INVALID_OPERATION));
+
+        glBindTexture(GL_TEXTURE_2D, 0);
+        glBindTexture(GL_TEXTURE_BUFFER, 0);
+        glBindBuffer(GL_TEXTURE_BUFFER, 0);
+        glDeleteTextures(1, &plainTexture);
+        glDeleteTextures(1, &texture);
+        glDeleteBuffers(1, &buffer);
+        EXPECT_EQ(FirstGLError(), 0u);
+    }
+
 } // namespace MGITest

MobileGL/MG_IntegrationTest/Scenarios/ClearTexImageUndefinedLevelZeroScenario.cpp

@@ -0,0 +1,250 @@
+// MobileGL - MobileGL/MG_IntegrationTest/Scenarios/ClearTexImageUndefinedLevelZeroScenario.cpp
+// Copyright (c) 2025-2026 MobileGL-Dev
+// Licensed under the GNU Lesser General Public License v3.0:
+//   https://www.gnu.org/licenses/gpl-3.0.txt
+//   https://www.gnu.org/licenses/lgpl-3.0.txt
+// SPDX-License-Identifier: LGPL-3.0-only
+// End of Source File Header
+//
+// Scenario - glClearTexImage ON A TEXTURE WHOSE GL LEVEL 0 WAS NEVER DEFINED.
+//
+// KHR-GL4[456].clear_tex_image.* builds exactly one shape: fillTexture() issues ONE
+// glTexImage2D(GL_TEXTURE_2D, m_texLevel, ...) - the only texImage2D in the whole format/level
+// family - sets GL_TEXTURE_MAX_LEVEL to that level, clears it and reads it back with
+// glGetTexImage(..., m_texLevel, ...). For m_texLevel > 0 the levels BELOW the defined one have no
+// storage at all, and the split in the conformance results was on that alone: every texLevel_0 body
+// passed on DirectVulkan and every texLevel != 0 body failed, across all four internal formats and
+// all three entry points.
+//
+// The frontend understands this shape - the clear is a pure CPU-shadow write, and
+// ValidateTextureImageQuery deliberately does not demand mip completeness for a readback. The
+// Vulkan backend did not: VkTextureManager takes storage mip 0 as the physical image extent, so a
+// texture with no level 0 got no VkImage, SyncTextureAndGetDescriptor answered nullptr, and
+// VulkanRenderer::GetTextureImage took a silent early return - leaving the caller's buffer exactly
+// as it found it. The conformance failures carried no <Text> at all, because nothing raised a GL
+// error: the destination was simply never written, so the test compared its own zero-initialized
+// buffer against the clear value.
+//
+// The fix this pins is the readback fallback: with NO VkImage, nothing GPU-side can ever have
+// written the texture, so the CPU shadow IS its content and is the correct answer. It is gated on
+// "no image exists at all" and not on "syncing was inconvenient - a blanket shadow answer would
+// return stale bytes for every render-to-texture result instead.
+//
+// NOT covered here, and deliberately: such a texture still has no VkImage, so it remains invisible
+// to SAMPLING and rendering on DirectVulkan. Backing the image from the lowest defined level is a
+// separate change (it moves every GL-level-to-subresource translation in the backend); this
+// scenario asserts the readback contract only, and the DirectGLES leg - which has always been able
+// to define a lone level N - is the built-in control for what the answer should be.
+
+#include <array>
+#include <cstdint>
+#include <vector>
+
+#include "../Harness/HeadlessGL.h"
+#include "../Harness/ScenarioFixture.h"
+
+#ifdef GLAPI
+#undef GLAPI
+#endif
+#define GL_GLEXT_PROTOTYPES
+#include <GL/gl.h>
+#include <GL/glcorearb.h>
+#undef GL_GLEXT_PROTOTYPES
+
+namespace MGITest {
+    namespace {
+
+        // The conformance family's own shape: a mid-chain level of a texture that has nothing else.
+        constexpr GLint kDefinedLevel = 3;
+        constexpr GLsizei kLevelExtent = 8;
+
+        struct Texel8 {
+            GLubyte r = 0, g = 0, b = 0, a = 0;
+            bool operator==(const Texel8& other) const {
+                return r == other.r && g == other.g && b == other.b && a == other.a;
+            }
+        };
+
+        std::ostream& operator<<(std::ostream& os, const Texel8& c) {
+            return os << "rgba(" << int(c.r) << "," << int(c.g) << "," << int(c.b) << "," << int(c.a) << ")";
+        }
+
+        // The conformance test's clear value is a single repeated component; 5 is what it uses, and
+        // it is deliberately neither 0 (an unwritten destination) nor 255 (a saturated one).
+        constexpr Texel8 kClearValue{5, 5, 5, 5};
+        constexpr Texel8 kInitialValue{200, 100, 50, 255};
+
+        class ClearTexImageUndefinedLevelZeroScenario : public ScenarioTest {
+        protected:
+            void SetUp() override {
+                ScenarioTest::SetUp();
+                if (!Ready()) return;
+                DrainErrors();
+            }
+
+            void TearDown() override {
+                if (!Ready()) return;
+                if (m_texture != 0) {
+                    glBindTexture(GL_TEXTURE_2D, 0);
+                    glDeleteTextures(1, &m_texture);
+                    m_texture = 0;
+                }
+                DrainErrors();
+            }
+
+            static void DrainErrors() {
+                for (int i = 0; i < 16 && glGetError() != GL_NO_ERROR; ++i) {
+                }
+            }
+
+            // One level and nothing else, through glTexImage2D - deliberately NOT glTexStorage2D,
+            // which would define the whole chain and could not express "level 0 does not exist".
+            void MakeTextureWithOnlyLevel(GLint level) {
+                if (m_texture != 0) glDeleteTextures(1, &m_texture);
+                glGenTextures(1, &m_texture);
+                glBindTexture(GL_TEXTURE_2D, m_texture);
+                const std::vector<Texel8> initial(static_cast<std::size_t>(kLevelExtent) * kLevelExtent, kInitialValue);
+                glTexImage2D(GL_TEXTURE_2D, level, GL_RGBA8, kLevelExtent, kLevelExtent, 0, GL_RGBA, GL_UNSIGNED_BYTE,
+                             initial.data());
+                // What the conformance case does: MAX_LEVEL names the one level that exists, and
+                // BASE_LEVEL is left at its default 0 - which is what makes level 0 undefined AND
+                // nominally the base level, the shape the backend could not express.
+                glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, level);
+                glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+                glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
+                ASSERT_EQ(FirstGLError(), 0u) << "texture setup with only level " << level;
+            }
+
+            std::vector<Texel8> ReadLevel(GLint level) {
+                std::vector<Texel8> pixels(static_cast<std::size_t>(kLevelExtent) * kLevelExtent, Texel8{0, 0, 0, 0});
+                glBindTexture(GL_TEXTURE_2D, m_texture);
+                glGetTexImage(GL_TEXTURE_2D, level, GL_RGBA, GL_UNSIGNED_BYTE, pixels.data());
+                EXPECT_EQ(FirstGLError(), 0u) << "glGetTexImage(level " << level << ") left a GL error behind";
+                return pixels;
+            }
+
+            void ExpectAllTexels(const char* what, const std::vector<Texel8>& pixels, Texel8 expected) {
+                std::size_t offenders = 0;
+                Texel8 firstBad{};
+                for (const Texel8& pixel : pixels) {
+                    if (pixel == expected) continue;
+                    if (offenders == 0) firstBad = pixel;
+                    ++offenders;
+                }
+                EXPECT_EQ(offenders, 0u) << what << ": got " << firstBad << " instead of " << expected << " ("
+                                         << offenders << " of " << pixels.size() << " texels wrong)";
+            }
+
+            // Level 0 defined, a GAP, then `level` defined. GL keeps the intervening levels at a zero
+            // extent, so the backend's mip walk stops at the gap and the VkImage ends up with FEWER
+            // mip levels than the GL level count - which is a different shape from "no image at all"
+            // and is why the readback has to bound the level against the IMAGE.
+            void MakeTextureWithAGapBefore(GLint level) {
+                if (m_texture != 0) glDeleteTextures(1, &m_texture);
+                glGenTextures(1, &m_texture);
+                glBindTexture(GL_TEXTURE_2D, m_texture);
+                const std::vector<Texel8> base(static_cast<std::size_t>(kLevelExtent) * kLevelExtent, kInitialValue);
+                glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA8, kLevelExtent, kLevelExtent, 0, GL_RGBA, GL_UNSIGNED_BYTE,
+                             base.data());
+                const std::vector<Texel8> gapped(static_cast<std::size_t>(kLevelExtent) * kLevelExtent, kInitialValue);
+                glTexImage2D(GL_TEXTURE_2D, level, GL_RGBA8, kLevelExtent, kLevelExtent, 0, GL_RGBA,
+                             GL_UNSIGNED_BYTE, gapped.data());
+                glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, level);
+                glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+                glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
+                ASSERT_EQ(FirstGLError(), 0u) << "texture setup with a gap before level " << level;
+            }
+
+            GLuint m_texture = 0;
+        };
+
+    } // namespace
+
+    // The regression. Before the fix glGetTexImage wrote nothing at all on DirectVulkan, so the
+    // caller's buffer kept whatever it already held - which is why the conformance failures showed
+    // the test's own zero-initialized memory and carried no GL error.
+    TEST_F(ClearTexImageUndefinedLevelZeroScenario, ClearAndReadBackALevelWhoseLowerLevelsDoNotExist) {
+        if (!Ready()) GTEST_SKIP();
+        MakeTextureWithOnlyLevel(kDefinedLevel);
+
+        // Pre-flight: the level reads back as what was uploaded. This is what makes the assertion
+        // after the clear falsifiable - without it, a readback that silently wrote nothing could not
+        // be told from one that wrote the right answer.
+        ExpectAllTexels("before the clear", ReadLevel(kDefinedLevel), kInitialValue);
+
+        glClearTexImage(m_texture, kDefinedLevel, GL_RGBA, GL_UNSIGNED_BYTE, &kClearValue);
+        EXPECT_EQ(FirstGLError(), 0u) << "glClearTexImage was rejected";
+
+        ExpectAllTexels("after the clear", ReadLevel(kDefinedLevel), kClearValue);
+        Gl().EndFrame();
+    }
+
+    // The same shape through glClearTexSubImage, which is a separate entry point in the conformance
+    // family and failed on exactly the same bodies.
+    TEST_F(ClearTexImageUndefinedLevelZeroScenario, ClearSubImageOfALevelWhoseLowerLevelsDoNotExist) {
+        if (!Ready()) GTEST_SKIP();
+        MakeTextureWithOnlyLevel(kDefinedLevel);
+
+        glClearTexSubImage(m_texture, kDefinedLevel, 0, 0, 0, kLevelExtent, kLevelExtent, 1, GL_RGBA,
+                           GL_UNSIGNED_BYTE, &kClearValue);
+        EXPECT_EQ(FirstGLError(), 0u) << "glClearTexSubImage was rejected";
+
+        ExpectAllTexels("after the sub-image clear", ReadLevel(kDefinedLevel), kClearValue);
+        Gl().EndFrame();
+    }
+
+    // The negative control: an ORDINARY texture, whose level 0 does exist, must keep answering from
+    // the GPU image rather than being diverted onto the shadow. A fallback that fired unconditionally
+    // would pass the two tests above and this one too - but it would also hand back stale bytes for
+    // anything the GPU had written, which is why the partial-clear check below matters: the readback
+    // has to see a region the backend cleared and a region it did not, in one image.
+    TEST_F(ClearTexImageUndefinedLevelZeroScenario, AnOrdinaryLevelZeroTextureStillReadsBackCorrectly) {
+        if (!Ready()) GTEST_SKIP();
+        MakeTextureWithOnlyLevel(0);
+
+        ExpectAllTexels("before the clear", ReadLevel(0), kInitialValue);
+
+        // Clear only the left half, so the answer is neither "all initial" nor "all cleared".
+        glClearTexSubImage(m_texture, 0, 0, 0, 0, kLevelExtent / 2, kLevelExtent, 1, GL_RGBA, GL_UNSIGNED_BYTE,
+                           &kClearValue);
+        EXPECT_EQ(FirstGLError(), 0u) << "glClearTexSubImage was rejected";
+
+        const std::vector<Texel8> pixels = ReadLevel(0);
+        ASSERT_EQ(pixels.size(), static_cast<std::size_t>(kLevelExtent) * kLevelExtent);
+        for (int y = 0; y < kLevelExtent; ++y) {
+            for (int x = 0; x < kLevelExtent; ++x) {
+                const Texel8 expected = x < kLevelExtent / 2 ? kClearValue : kInitialValue;
+                const Texel8 actual = pixels[static_cast<std::size_t>(y) * kLevelExtent + x];
+                ASSERT_EQ(actual, expected) << "at (" << x << "," << y << ")";
+            }
+        }
+        Gl().EndFrame();
+    }
+
+    // The adjacent shape the first fix did NOT cover: level 0 defined, a gap, then the level being
+    // read. This one DOES get a VkImage - just one with fewer mip levels than GL thinks the texture
+    // has - so the "no VkImage" test passes and the GL level was written straight into
+    // imageSubresource.mipLevel and into a VkImageMemoryBarrier's baseMipLevel. An out-of-range
+    // subresource is a promise the driver takes at face value; the glCopyImageSubData path two
+    // functions away grew the same guard after it SIGSEGV'd inside the Adreno driver.
+    //
+    // The level being read really does hold its own data (the shadow is its only copy, since nothing
+    // ever uploaded it), so the correct answer is the uploaded bytes - not a decline.
+    TEST_F(ClearTexImageUndefinedLevelZeroScenario, ReadBackALevelSeparatedFromLevelZeroByAGap) {
+        if (!Ready()) GTEST_SKIP();
+        MakeTextureWithAGapBefore(kDefinedLevel);
+
+        ExpectAllTexels("before the clear", ReadLevel(kDefinedLevel), kInitialValue);
+
+        glClearTexImage(m_texture, kDefinedLevel, GL_RGBA, GL_UNSIGNED_BYTE, &kClearValue);
+        EXPECT_EQ(FirstGLError(), 0u) << "glClearTexImage was rejected";
+
+        ExpectAllTexels("after the clear", ReadLevel(kDefinedLevel), kClearValue);
+
+        // Level 0 is backed by the real image and must still read back from it, so the level bound is
+        // about the level and not about the texture.
+        ExpectAllTexels("level 0 after clearing level 3", ReadLevel(0), kInitialValue);
+        Gl().EndFrame();
+    }
+
+} // namespace MGITest

MobileGL/MG_IntegrationTest/Scenarios/ClipDistanceScenario.cpp

@@ -151,6 +151,18 @@ void main() { fragColor = vec4(0.0, 1.0, 0.0, 1.0); }
                 glReadPixels(x, y, 1, 1, GL_RGBA, GL_UNSIGNED_BYTE, out);
             }
 
+            // GL_MAX_CLIP_DISTANCES is a real backend answer, not a constant: DirectGLES reports
+            // 0 on a driver without GL_EXT_clip_cull_distance, and DirectVulkan reports 0 without
+            // the shaderClipDistance device feature. On such a stack the shader above cannot
+            // compile - and MUST not, because declaring a clip distance the backend cannot host
+            // is exactly what used to link cleanly and then render nothing. Skip rather than
+            // fail: there is no clipping to assert about.
+            static bool BackendHostsTwoClipDistances() {
+                GLint maxClipDistances = 0;
+                glGetIntegerv(GL_MAX_CLIP_DISTANCES, &maxClipDistances);
+                return maxClipDistances >= 2;
+            }
+
             // Never assume the eight start disabled - see the header note about
             // XfbAfterClipDistanceScenario leaving one on for the rest of the process.
             static void DisableEveryClipDistance() {
@@ -229,6 +241,9 @@ void main() { fragColor = vec4(0.0, 1.0, 0.0, 1.0); }
     // The claim: an enabled clip distance removes the fragments where it is negative.
     TEST_F(ClipDistanceScenario, AnEnabledClipDistanceRemovesTheNegativeHalf) {
         if (!Ready()) return;
+        if (!BackendHostsTwoClipDistances()) {
+            GTEST_SKIP() << "this backend advertises no clip distances, so there is nothing to clip with";
+        }
         HeadlessGL& gl = Gl();
         const int width = gl.Width();
         const int height = gl.Height();
@@ -280,6 +295,9 @@ void main() { fragColor = vec4(0.0, 1.0, 0.0, 1.0); }
     // draw simply failed - would pass the case above.
     TEST_F(ClipDistanceScenario, ADisabledClipDistanceRemovesNothing) {
         if (!Ready()) return;
+        if (!BackendHostsTwoClipDistances()) {
+            GTEST_SKIP() << "this backend advertises no clip distances, so there is nothing to clip with";
+        }
         HeadlessGL& gl = Gl();
         const int width = gl.Width();
         const int height = gl.Height();
@@ -329,6 +347,9 @@ void main() { fragColor = vec4(0.0, 1.0, 0.0, 1.0); }
     // passes both cases above and fails this one.
     TEST_F(ClipDistanceScenario, TheEnablesAreIndependentPerDistance) {
         if (!Ready()) return;
+        if (!BackendHostsTwoClipDistances()) {
+            GTEST_SKIP() << "this backend advertises no clip distances, so there is nothing to clip with";
+        }
         HeadlessGL& gl = Gl();
         const int width = gl.Width();
         const int height = gl.Height();

MobileGL/MG_IntegrationTest/Scenarios/CopyImagePacked16Scenario.cpp

@@ -0,0 +1,375 @@
+// MobileGL - MobileGL/MG_IntegrationTest/Scenarios/CopyImagePacked16Scenario.cpp
+// Copyright (c) 2026 MobileGL-Dev
+// Licensed under the GNU Lesser General Public License v3.0:
+//   https://www.gnu.org/licenses/gpl-3.0.txt
+//   https://www.gnu.org/licenses/lgpl-3.0.txt
+// SPDX-License-Identifier: LGPL-3.0-only
+// End of Source File Header
+//
+// Scenario - glCopyImageSubData PRESERVES 16-BIT PACKED WORDS ACROSS AN ARRAY MIP LEVEL.
+//
+// The shape is lifted verbatim from the 18 Espryt bodies of KHR-GL4x.copy_image.functional
+// that survived every earlier wave: the three internal formats MobileGL can keep as 16-bit
+// packed ES storage - GL_RGB5 (stored GL_RGB565), GL_RGB5_A1, GL_RGBA4 - crossed with the
+// target pairs that put a GL_TEXTURE_2D_ARRAY's MIP LEVEL 1 on one side of the copy. On the
+// affected Mali the mirrored *_REV field order is a property of WHOLE ALLOCATIONS (shape-
+// and context-dependent; the failing 30x30x12 arrays carry it at every level, the small
+// arrays of the suite's passing iterations do not), and glCopyImageSubData - a raw
+// texel-block move - between a mirrored allocation and a plain one lands the fields
+// reversed: src word 0x0047 arrives as 0x8C20 (its 5_5_5_1 -> 1_5_5_5_REV re-encoding),
+// 0x0007 as 0x3800, byte-exact on every failing body. Uploads and readbacks of the same
+// image are clean (the driver decodes its own layout consistently), which is why only the
+// copy path ever crossed the two layouts and why the CTS's "source image was not modified"
+// checks always passed.
+//
+// The array is 30x30x12 with THREE levels and the flat endpoint is 7x7 with three levels
+// (7/3/1) because that is the allocation the failures pin - the CTS builds every functional
+// texture with FUNCTIONAL_TEST_N_LEVELS = 3 (makeTextureComplete(0, 2)) - and any deviation
+// from the measured shape might sit on the clean side of whatever allocation heuristic picks
+// the driver's layout.
+//
+// The repair under test is the packed16 storage widening
+// (PixelFormatNormalizeOptionBit::WidenPacked16Norm): where the POST probe
+// (SelfTest::CopyImageMirrorsPacked16FieldOrder) measures the mirror - or
+// MOBILEGL_ESPRYT_WIDEN_PACKED16_STORAGE forces it - the three formats are stored as
+// GL_RGB8/GL_RGBA8, leaving no 16-bit packed image for a copy to disagree about. The client
+// word still round-trips exactly: the canonical shadow is already UNorm8, and an n-bit field
+// encodes to UNorm8 and back losslessly for every n <= 8.
+//
+// This scenario runs in BOTH configurations, and both must hand back identical client words:
+//   * the ambient registrations take the narrow path on a clean driver (llvmpipe has no
+//     mirror, so Auto keeps the native 16-bit storage - the pre-existing behaviour stays
+//     covered);
+//   * the DirectGLES.WidenedPacked16. registration pins MOBILEGL_ESPRYT_WIDEN_PACKED16_STORAGE=1,
+//     which is the storage every affected device will actually run - without it the repair
+//     is unfalsifiable off-device, because no CI driver has the bug that arms it.
+// The Mali mirror itself CANNOT be reproduced here; only the on-device CTS run can show the
+// widening killing the 18 bodies. What this scenario pins is that the widened storage is
+// client-invisible: same words in, same words out, on every leg the failing bodies used.
+//
+// DirectVulkan is the control - Magma has always resolved these formats to RGBA8 - so a
+// failure on both backends means the scenario is wrong, and a failure on DirectGLES alone
+// means the widening (or the narrow path it replaces) is.
+
+#include <algorithm>
+#include <string>
+#include <vector>
+
+#include "../Harness/HeadlessGL.h"
+#include "../Harness/ScenarioFixture.h"
+
+#ifdef GLAPI
+#undef GLAPI
+#endif
+#define GL_GLEXT_PROTOTYPES
+#include <GL/gl.h>
+#include <GL/glcorearb.h>
+#undef GL_GLEXT_PROTOTYPES
+
+namespace MGITest {
+    namespace {
+
+        constexpr int kBaseSize = 30; // array level 0; level 1 is 15x15
+        constexpr int kLevel1Size = kBaseSize / 2;
+        constexpr int kLayers = 12;
+        constexpr int kFlatSize = 7; // the plain-2D / renderbuffer endpoint, level 0
+        // Copies cover the whole flat endpoint and land at (8, 8) inside the 15x15 level so
+        // that offsets are honoured, not just texel (0, 0): 8 + 7 == 15 reaches the far edge.
+        constexpr int kRegion = kFlatSize;
+        constexpr int kArrayOffset = 8;
+
+        struct PackedFormatCase {
+            GLenum internalFormat; // the spelling the CTS uses
+            GLenum transferFormat;
+            GLenum transferType;
+            const char* name;
+        };
+
+        // Per-texel varying words, every field inside its width, so a swapped field order (or
+        // a mis-addressed row) cannot cancel out the way a uniform fill would let it.
+        GLushort MakeWord(GLenum type, int i) {
+            switch (type) {
+            case GL_UNSIGNED_SHORT_5_6_5: {
+                const int r = i % 32, g = (i * 7 + 3) % 64, b = (i * 5 + 11) % 32;
+                return static_cast<GLushort>((r << 11) | (g << 5) | b);
+            }
+            case GL_UNSIGNED_SHORT_4_4_4_4: {
+                const int r = i % 16, g = (i * 3 + 1) % 16, b = (i * 7 + 5) % 16, a = (i * 5 + 2) % 16;
+                return static_cast<GLushort>((r << 12) | (g << 8) | (b << 4) | a);
+            }
+            case GL_UNSIGNED_SHORT_5_5_5_1: {
+                const int r = i % 32, g = (i * 7 + 3) % 32, b = (i * 3 + 11) % 32, a = i % 2;
+                return static_cast<GLushort>((r << 11) | (g << 6) | (b << 1) | a);
+            }
+            default:
+                return 0;
+            }
+        }
+
+        std::vector<GLushort> MakeWords(GLenum type, int count, int seed) {
+            std::vector<GLushort> words(static_cast<size_t>(count));
+            for (int i = 0; i < count; ++i) {
+                words[static_cast<size_t>(i)] = MakeWord(type, i + seed);
+            }
+            return words;
+        }
+
+        class CopyImagePacked16Scenario : public ScenarioTest {
+        protected:
+            void SetUp() override {
+                ScenarioTest::SetUp();
+                if (!Ready()) return;
+                // 16-bit rows are 2-byte aligned; the default 4-byte row alignment would pad
+                // every odd-width row of the 15x15 level and shear the comparisons.
+                glPixelStorei(GL_UNPACK_ALIGNMENT, 2);
+                glPixelStorei(GL_PACK_ALIGNMENT, 2);
+                if (!CopyImageSubDataUsable()) {
+                    GTEST_SKIP() << "glCopyImageSubData is unavailable on backend " << Gl().BackendName();
+                }
+            }
+
+            void TearDown() override {
+                if (!Ready()) return;
+                glPixelStorei(GL_UNPACK_ALIGNMENT, 4);
+                glPixelStorei(GL_PACK_ALIGNMENT, 4);
+                for (const GLuint texture : m_textures) {
+                    glDeleteTextures(1, &texture);
+                }
+                m_textures.clear();
+                if (m_renderbuffer != 0) {
+                    glDeleteRenderbuffers(1, &m_renderbuffer);
+                    m_renderbuffer = 0;
+                }
+                if (m_fbo != 0) {
+                    glBindFramebuffer(GL_FRAMEBUFFER, 0);
+                    glDeleteFramebuffers(1, &m_fbo);
+                    m_fbo = 0;
+                }
+            }
+
+            bool CopyImageSubDataUsable() {
+                GLuint probe[2] = {0, 0};
+                glGenTextures(2, probe);
+                for (const GLuint texture : probe) {
+                    glBindTexture(GL_TEXTURE_2D_ARRAY, texture);
+                    glTexStorage3D(GL_TEXTURE_2D_ARRAY, 1, GL_RGBA8, 1, 1, 1);
+                }
+                glBindTexture(GL_TEXTURE_2D_ARRAY, 0);
+                while (glGetError() != GL_NO_ERROR) {
+                }
+                glCopyImageSubData(probe[0], GL_TEXTURE_2D_ARRAY, 0, 0, 0, 0, probe[1], GL_TEXTURE_2D_ARRAY, 0, 0, 0,
+                                   0, 1, 1, 1);
+                const bool usable = glGetError() == GL_NO_ERROR;
+                glDeleteTextures(2, probe);
+                return usable;
+            }
+
+            // The CTS's own mutable shape: glTexImage3D per level, filter NEAREST, THREE levels
+            // (30/15/7) with the chain clamped to them. Level 2 carries its own fill so nothing
+            // below can pass by reading a level that was never written.
+            GLuint MakeArrayTexture(const PackedFormatCase& format, const std::vector<GLushort>& level0,
+                                    const std::vector<GLushort>& level1) {
+                GLuint texture = 0;
+                glGenTextures(1, &texture);
+                m_textures.push_back(texture);
+                glBindTexture(GL_TEXTURE_2D_ARRAY, texture);
+                glTexParameteri(GL_TEXTURE_2D_ARRAY, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+                glTexParameteri(GL_TEXTURE_2D_ARRAY, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
+                glTexParameteri(GL_TEXTURE_2D_ARRAY, GL_TEXTURE_MAX_LEVEL, 2);
+                glTexImage3D(GL_TEXTURE_2D_ARRAY, 0, static_cast<GLint>(format.internalFormat), kBaseSize, kBaseSize,
+                             kLayers, 0, format.transferFormat, format.transferType, level0.data());
+                glTexImage3D(GL_TEXTURE_2D_ARRAY, 1, static_cast<GLint>(format.internalFormat), kLevel1Size,
+                             kLevel1Size, kLayers, 0, format.transferFormat, format.transferType, level1.data());
+                const int level2Size = kLevel1Size / 2;
+                const auto level2 = MakeWords(format.transferType, level2Size * level2Size * kLayers, 211);
+                glTexImage3D(GL_TEXTURE_2D_ARRAY, 2, static_cast<GLint>(format.internalFormat), level2Size,
+                             level2Size, kLayers, 0, format.transferFormat, format.transferType, level2.data());
+                glBindTexture(GL_TEXTURE_2D_ARRAY, 0);
+                return texture;
+            }
+
+            // Three levels (7/3/1) like the CTS's plain endpoints; `texels` is level 0, the one
+            // every assertion reads.
+            GLuint MakeFlatTexture(const PackedFormatCase& format, const std::vector<GLushort>& texels) {
+                GLuint texture = 0;
+                glGenTextures(1, &texture);
+                m_textures.push_back(texture);
+                glBindTexture(GL_TEXTURE_2D, texture);
+                glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+                glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
+                glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, 2);
+                glTexImage2D(GL_TEXTURE_2D, 0, static_cast<GLint>(format.internalFormat), kFlatSize, kFlatSize, 0,
+                             format.transferFormat, format.transferType, texels.data());
+                for (int level = 1; level <= 2; ++level) {
+                    const int size = std::max(kFlatSize >> level, 1);
+                    const auto fill = MakeWords(format.transferType, size * size, 97 + level);
+                    glTexImage2D(GL_TEXTURE_2D, level, static_cast<GLint>(format.internalFormat), size, size, 0,
+                                 format.transferFormat, format.transferType, fill.data());
+                }
+                glBindTexture(GL_TEXTURE_2D, 0);
+                return texture;
+            }
+
+            std::vector<GLushort> ReadTexImage(GLenum target, GLuint texture, int level,
+                                               const PackedFormatCase& format, size_t texelCount) {
+                std::vector<GLushort> words(texelCount, 0);
+                glBindTexture(target, texture);
+                glGetTexImage(target, level, format.transferFormat, format.transferType, words.data());
+                glBindTexture(target, 0);
+                return words;
+            }
+
+            // Every word of `got` inside the kRegion-square at (x0, y0) of a width-wide layer-0
+            // image equals the corresponding source word, and every word outside it still holds
+            // `fill`'s. Failures name the texel and both words, which is what turns a field-order
+            // regression into a one-line diagnosis.
+            void ExpectRegion(const std::vector<GLushort>& got, int width, int x0, int y0,
+                              const std::vector<GLushort>& source, int sourceWidth, int sourceX0, int sourceY0,
+                              const std::vector<GLushort>& fill, const char* what) {
+                for (int y = 0; y < width; ++y) {
+                    for (int x = 0; x < width && static_cast<size_t>(y * width + x) < got.size(); ++x) {
+                        const bool inRegion =
+                            x >= x0 && x < x0 + kRegion && y >= y0 && y < y0 + kRegion;
+                        const GLushort actual = got[static_cast<size_t>(y * width + x)];
+                        const GLushort expected =
+                            inRegion ? source[static_cast<size_t>((sourceY0 + y - y0) * sourceWidth + sourceX0 +
+                                                                  (x - x0))]
+                                     : fill[static_cast<size_t>(y * width + x)];
+                        EXPECT_EQ(actual, expected)
+                            << what << ": texel (" << x << ", " << y << ")"
+                            << (inRegion ? " (copied)" : " (untouched)") << " holds 0x" << std::hex << actual
+                            << ", expected 0x" << expected;
+                        if (actual != expected) return; // one texel names the defect; 224 more would bury it
+                    }
+                }
+            }
+
+            std::vector<GLuint> m_textures;
+            GLuint m_renderbuffer = 0;
+            GLuint m_fbo = 0;
+        };
+
+        const PackedFormatCase kFormats[] = {
+            {GL_RGB5, GL_RGB, GL_UNSIGNED_SHORT_5_6_5, "rgb5"},
+            {GL_RGB5_A1, GL_RGBA, GL_UNSIGNED_SHORT_5_5_5_1, "rgb5_a1"},
+            {GL_RGBA4, GL_RGBA, GL_UNSIGNED_SHORT_4_4_4_4, "rgba4"},
+        };
+
+        // texture_2d (the ES image behind GL_TEXTURE_RECTANGLE too) -> the array's level 1:
+        // the array-as-destination direction of 12 of the 18 failing bodies.
+        TEST_F(CopyImagePacked16Scenario, FlatImageLandsInArrayMipLevelIntact) {
+            if (!Ready() || IsSkipped()) return;
+            for (const PackedFormatCase& format : kFormats) {
+                const auto level0 = MakeWords(format.transferType, kBaseSize * kBaseSize * kLayers, 1);
+                const auto level1 = MakeWords(format.transferType, kLevel1Size * kLevel1Size * kLayers, 7);
+                const auto flat = MakeWords(format.transferType, kFlatSize * kFlatSize, 131);
+                const GLuint array = MakeArrayTexture(format, level0, level1);
+                const GLuint source = MakeFlatTexture(format, flat);
+                ASSERT_EQ(glGetError(), static_cast<GLenum>(GL_NO_ERROR)) << format.name << ": setup failed";
+
+                glCopyImageSubData(source, GL_TEXTURE_2D, 0, 0, 0, 0, array, GL_TEXTURE_2D_ARRAY, 1, kArrayOffset,
+                                   kArrayOffset, 0, kRegion, kRegion, 1);
+                ASSERT_EQ(glGetError(), static_cast<GLenum>(GL_NO_ERROR))
+                    << format.name << ": glCopyImageSubData raised an error";
+
+                const auto got = ReadTexImage(GL_TEXTURE_2D_ARRAY, array, 1, format,
+                                              static_cast<size_t>(kLevel1Size) * kLevel1Size * kLayers);
+                ExpectRegion(got, kLevel1Size, kArrayOffset, kArrayOffset, flat, kFlatSize, 0, 0, level1,
+                             (std::string("2d->2d_array level 1, ") + format.name).c_str());
+                // The source must not have moved - the CTS asserts this before it ever looks at
+                // the destination, and it is what pins the corruption to the copy itself.
+                const auto sourceAfter =
+                    ReadTexImage(GL_TEXTURE_2D, source, 0, format, static_cast<size_t>(kFlatSize) * kFlatSize);
+                ExpectRegion(sourceAfter, kFlatSize, 0, 0, flat, kFlatSize, 0, 0, flat,
+                             (std::string("source after 2d->2d_array, ") + format.name).c_str());
+            }
+        }
+
+        // The array's level 1 -> texture_2d: the array-as-source direction of the other 6
+        // bodies (2d_array -> 3d and 2d_array -> rectangle both read the level-1 array).
+        TEST_F(CopyImagePacked16Scenario, ArrayMipLevelLandsInFlatImageIntact) {
+            if (!Ready() || IsSkipped()) return;
+            for (const PackedFormatCase& format : kFormats) {
+                const auto level0 = MakeWords(format.transferType, kBaseSize * kBaseSize * kLayers, 1);
+                const auto level1 = MakeWords(format.transferType, kLevel1Size * kLevel1Size * kLayers, 7);
+                const auto fill = MakeWords(format.transferType, kFlatSize * kFlatSize, 131);
+                const GLuint array = MakeArrayTexture(format, level0, level1);
+                const GLuint destination = MakeFlatTexture(format, fill);
+                ASSERT_EQ(glGetError(), static_cast<GLenum>(GL_NO_ERROR)) << format.name << ": setup failed";
+
+                glCopyImageSubData(array, GL_TEXTURE_2D_ARRAY, 1, kArrayOffset, kArrayOffset, 0, destination,
+                                   GL_TEXTURE_2D, 0, 0, 0, 0, kRegion, kRegion, 1);
+                ASSERT_EQ(glGetError(), static_cast<GLenum>(GL_NO_ERROR))
+                    << format.name << ": glCopyImageSubData raised an error";
+
+                const auto got = ReadTexImage(GL_TEXTURE_2D, destination, 0, format,
+                                              static_cast<size_t>(kFlatSize) * kFlatSize);
+                ExpectRegion(got, kFlatSize, 0, 0, level1, kLevel1Size, kArrayOffset, kArrayOffset, fill,
+                             (std::string("2d_array level 1 -> 2d, ") + format.name).c_str());
+            }
+        }
+
+        // renderbuffer -> the array's level 1: the leg the remaining 3 bodies use, and the one
+        // that requires the renderbuffer's ES storage to move together with the textures' -
+        // glCopyImageSubData needs both endpoints in the same driver format, so a widening that
+        // reached textures alone would break exactly here.
+        TEST_F(CopyImagePacked16Scenario, RenderbufferLandsInArrayMipLevelIntact) {
+            if (!Ready() || IsSkipped()) return;
+            for (const PackedFormatCase& format : kFormats) {
+                const auto level0 = MakeWords(format.transferType, kBaseSize * kBaseSize * kLayers, 1);
+                const auto level1 = MakeWords(format.transferType, kLevel1Size * kLevel1Size * kLayers, 7);
+                const GLuint array = MakeArrayTexture(format, level0, level1);
+
+                if (m_renderbuffer == 0) glGenRenderbuffers(1, &m_renderbuffer);
+                glBindRenderbuffer(GL_RENDERBUFFER, m_renderbuffer);
+                glRenderbufferStorage(GL_RENDERBUFFER, format.internalFormat, kFlatSize, kFlatSize);
+                if (m_fbo == 0) glGenFramebuffers(1, &m_fbo);
+                glBindFramebuffer(GL_FRAMEBUFFER, m_fbo);
+                glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_RENDERBUFFER, m_renderbuffer);
+                ASSERT_EQ(glCheckFramebufferStatus(GL_FRAMEBUFFER), static_cast<GLenum>(GL_FRAMEBUFFER_COMPLETE))
+                    << format.name << ": the renderbuffer is not attachable";
+                // Field values picked to encode exactly in the narrow fields AND in their
+                // UNorm8 expansions, so the expected word is the same whichever storage the
+                // configuration picked - which is the point of the whole scenario.
+                const int maxG = format.transferType == GL_UNSIGNED_SHORT_5_6_5 ? 63 : 31;
+                const int max = format.transferType == GL_UNSIGNED_SHORT_4_4_4_4 ? 15 : 31;
+                const int maxGreen = format.transferType == GL_UNSIGNED_SHORT_4_4_4_4 ? 15 : maxG;
+                const GLfloat clearColor[4] = {static_cast<GLfloat>(8 % (max + 1)) / max,
+                                               static_cast<GLfloat>(maxGreen / 2) / maxGreen,
+                                               static_cast<GLfloat>(max - 2) / max, 1.0f};
+                // The context is shared with every scenario in this process; a scissor left on
+                // would clip the clear and hand the copy undefined renderbuffer texels.
+                glDisable(GL_SCISSOR_TEST);
+                glClearBufferfv(GL_COLOR, 0, clearColor);
+                glBindFramebuffer(GL_FRAMEBUFFER, 0);
+                ASSERT_EQ(glGetError(), static_cast<GLenum>(GL_NO_ERROR)) << format.name << ": setup failed";
+
+                glCopyImageSubData(m_renderbuffer, GL_RENDERBUFFER, 0, 0, 0, 0, array, GL_TEXTURE_2D_ARRAY, 1,
+                                   kArrayOffset, kArrayOffset, 0, kRegion, kRegion, 1);
+                ASSERT_EQ(glGetError(), static_cast<GLenum>(GL_NO_ERROR))
+                    << format.name << ": glCopyImageSubData raised an error";
+
+                GLushort clearedWord = 0;
+                switch (format.transferType) {
+                case GL_UNSIGNED_SHORT_5_6_5:
+                    clearedWord = static_cast<GLushort>((8 << 11) | ((maxGreen / 2) << 5) | (max - 2));
+                    break;
+                case GL_UNSIGNED_SHORT_5_5_5_1:
+                    clearedWord = static_cast<GLushort>((8 << 11) | ((maxGreen / 2) << 6) | ((max - 2) << 1) | 1);
+                    break;
+                case GL_UNSIGNED_SHORT_4_4_4_4:
+                    clearedWord = static_cast<GLushort>((8 << 12) | ((maxGreen / 2) << 8) | ((max - 2) << 4) | 15);
+                    break;
+                default:
+                    break;
+                }
+                std::vector<GLushort> expectedRegion(static_cast<size_t>(kRegion) * kRegion, clearedWord);
+                const auto got = ReadTexImage(GL_TEXTURE_2D_ARRAY, array, 1, format,
+                                              static_cast<size_t>(kLevel1Size) * kLevel1Size * kLayers);
+                ExpectRegion(got, kLevel1Size, kArrayOffset, kArrayOffset, expectedRegion, kRegion, 0, 0, level1,
+                             (std::string("renderbuffer -> 2d_array level 1, ") + format.name).c_str());
+            }
+        }
+
+    } // namespace
+} // namespace MGITest

MobileGL/MG_IntegrationTest/Scenarios/DoublePrecisionScenario.cpp

@@ -6,27 +6,32 @@
 // SPDX-License-Identifier: LGPL-3.0-only
 // End of Source File Header
 //
-// Scenario - GLSL DOUBLES, RUN AT SINGLE PRECISION.
+// Scenario - GLSL DOUBLES, AT WHATEVER PRECISION THE BACKEND CAN GIVE.
 //
 // No mobile GPU has 64-bit floats. Adreno and Mali both report shaderFloat64 == VK_FALSE, so
-// Magma cannot build a module that declares the Float64 capability, and ESSL has no fp64 type
-// at all, so SPIRV-Cross refuses the module outright on Espryt ("FP64 not supported in ES
-// profile") and the program never reaches the driver. MobileGL therefore narrows every 64-bit
-// float in a shader to 32 bits (ShaderTranspiler::DemoteFloat64Pass) rather than declining the
-// shader: `double` compiles and runs everywhere, at float precision.
+// Magma cannot build a module that declares the Float64 capability there, and ESSL has no fp64
+// type at all, so SPIRV-Cross refuses the module outright on Espryt ("FP64 not supported in ES
+// profile") and the program never reaches the driver. On every such backend MobileGL narrows
+// every 64-bit float in a shader to 32 bits (ShaderTranspiler::DemoteFloat64Pass) rather than
+// declining the shader: `double` compiles and runs everywhere, at float precision. Where the
+// backend DOES consume 64-bit floats - lavapipe is the one that does - the narrowing is skipped
+// and the doubles reach the driver whole.
 //
-// The narrowing is only half a contract. The other half is the API side: the global UBO is
-// laid out by reflecting the DEMOTED module, so glUniform*d has to store a float where the
-// shader reads a float, glGetUniform*v has to read one back, and a dmat4's columns are now
-// std140-padded like any other matrix's. Every one of those is a byte offset that fails
-// silently - the uniform simply reads as something else - so the cases below set values
-// through the API and have the SHADER report what it saw.
+// Either way it is only half a contract. The other half is the API side: the global UBO is laid
+// out by reflecting whichever module was produced, so glUniform*d has to store the width the
+// shader reads, glGetUniform*v has to read that width back, and a matrix's columns are
+// std140-padded to a vec4 or a dvec4 to match. Every one of those is a byte offset that fails
+// silently - the uniform simply reads as something else - so the cases below set values through
+// the API and have the SHADER report what it saw.
 //
-// What is deliberately NOT asserted: that the values are exact to double precision. They are
-// not, and cannot be. Every expectation here is the float value of the double that was set,
-// which is the whole point.
+// WHY ALMOST EVERY EXPECTATION HERE IS A FLOAT VALUE, and why that is not an accident of the
+// demotion: the shader reports through a `float` SSBO, and every value chosen is exact in
+// float32, so the same number is correct in both regimes and the assertions test the LAYOUT
+// rather than the precision. Exactly one case (GetUniformdvReadsBackWhatWasStored) uses a value
+// that is not - 0.1 - and it names both answers explicitly.
 
 #include <cmath>
+#include <cstring>
 #include <string>
 #include <vector>
 
@@ -153,6 +158,151 @@ void main() {
             std::string m_buildLog;
         };
 
+        // A SHADER STORAGE BLOCK that holds doubles is the one place the narrowing is NOT free:
+        // demoting `double` to `float` also repacks the block, and the bytes the application
+        // wrote into the buffer do not move with it. Every member past the first double then
+        // reads and writes at the wrong offset, and the block is simply shorter than the one
+        // that was bound - the tail of it is never touched at all
+        // (KHR-GL43.shader_storage_buffer_object.basic-stdLayout-case3, whose output matched its
+        // input up to the first double's slot and was zero from there on).
+        //
+        // The block layout is fixed by GL 4.6 core 7.6.2.2 and is asserted here as literal byte
+        // offsets rather than queried, so this says what the SPEC requires and not what MobileGL
+        // happens to report. Both packings are covered because they differ in exactly the places
+        // that matter: std140 rounds an array's stride and a matrix's column stride up to 16,
+        // std430 does not, and only std430 packs the scalars tightly.
+        //
+        // Every value is exactly representable in binary32, so a correct implementation copies
+        // the block BYTE FOR BYTE even though it narrows each double on the way through.
+        constexpr const char* kBlockCopySource = R"(#version 430 core
+layout(local_size_x = 1) in;
+layout(std140, binding = 0) buffer In140 {
+    int    data0;
+    float  data1[3];
+    mat3x2 data2;
+    double data3;
+    double data4[2];
+    int    data5;
+    dvec3  data6;
+} g_in140;
+layout(std430, binding = 1) buffer In430 {
+    int    data0;
+    float  data1[3];
+    mat3x2 data2;
+    double data3;
+    double data4[2];
+    int    data5;
+    dvec3  data6;
+} g_in430;
+layout(std140, binding = 2) buffer Out140 {
+    int    data0;
+    float  data1[3];
+    mat3x2 data2;
+    double data3;
+    double data4[2];
+    int    data5;
+    dvec3  data6;
+} g_out140;
+layout(std430, binding = 3) buffer Out430 {
+    int    data0;
+    float  data1[3];
+    mat3x2 data2;
+    double data3;
+    double data4[2];
+    int    data5;
+    dvec3  data6;
+} g_out430;
+void main() {
+    g_out140.data0 = g_in140.data0;
+    for (int i = 0; i < 3; ++i) g_out140.data1[i] = g_in140.data1[i];
+    g_out140.data2 = g_in140.data2;
+    g_out140.data3 = g_in140.data3;
+    for (int i = 0; i < 2; ++i) g_out140.data4[i] = g_in140.data4[i];
+    g_out140.data5 = g_in140.data5;
+    g_out140.data6 = g_in140.data6;
+
+    g_out430.data0 = g_in430.data0;
+    for (int i = 0; i < 3; ++i) g_out430.data1[i] = g_in430.data1[i];
+    g_out430.data2 = g_in430.data2;
+    g_out430.data3 = g_in430.data3;
+    for (int i = 0; i < 2; ++i) g_out430.data4[i] = g_in430.data4[i];
+    g_out430.data5 = g_in430.data5;
+    g_out430.data6 = g_in430.data6;
+}
+)";
+
+        // GL 4.6 core 7.6.2.2 rule by rule, for the block above.
+        //   std140: an array's element stride and a matrix's column stride round up to 16, a
+        //           double aligns to 8 and a dvec3 to 32.
+        //   std430: the same without the rounding - so the scalars pack tightly and only the
+        //           dvec3's 32-byte alignment leaves a hole.
+        struct BlockLayout {
+            int data0;
+            int data1;
+            int data1Stride;
+            int data2;
+            int data2ColumnStride;
+            int data3;
+            int data4;
+            int data4Stride;
+            int data5;
+            int data6;
+            int size;
+        };
+        constexpr BlockLayout kStd140{0, 16, 16, 64, 16, 112, 128, 16, 160, 192, 216};
+        constexpr BlockLayout kStd430{0, 4, 4, 16, 8, 40, 48, 8, 64, 96, 120};
+
+        void PokeInt(std::vector<unsigned char>& bytes, int offset, int value) {
+            std::memcpy(&bytes[static_cast<std::size_t>(offset)], &value, sizeof(value));
+        }
+        void PokeFloat(std::vector<unsigned char>& bytes, int offset, float value) {
+            std::memcpy(&bytes[static_cast<std::size_t>(offset)], &value, sizeof(value));
+        }
+        void PokeDouble(std::vector<unsigned char>& bytes, int offset, double value) {
+            std::memcpy(&bytes[static_cast<std::size_t>(offset)], &value, sizeof(value));
+        }
+
+        // The block's contents, at the offsets the standard puts them. Padding stays zero, which
+        // is what makes a byte-for-byte comparison against the (zero-initialised) output buffer
+        // catch a member that landed somewhere it should not have.
+        std::vector<unsigned char> MakeBlockContents(const BlockLayout& layout) {
+            std::vector<unsigned char> bytes(static_cast<std::size_t>(layout.size), 0);
+            PokeInt(bytes, layout.data0, 1);
+            for (int i = 0; i < 3; ++i) {
+                PokeFloat(bytes, layout.data1 + i * layout.data1Stride, 2.0f + static_cast<float>(i));
+            }
+            // Column-major, two rows per column.
+            for (int column = 0; column < 3; ++column) {
+                for (int row = 0; row < 2; ++row) {
+                    PokeFloat(bytes, layout.data2 + column * layout.data2ColumnStride + row * 4,
+                              5.0f + static_cast<float>(column * 2 + row));
+                }
+            }
+            PokeDouble(bytes, layout.data3, 11.0);
+            for (int i = 0; i < 2; ++i) {
+                PokeDouble(bytes, layout.data4 + i * layout.data4Stride, 12.0 + static_cast<double>(i));
+            }
+            PokeInt(bytes, layout.data5, 14);
+            for (int i = 0; i < 3; ++i) {
+                PokeDouble(bytes, layout.data6 + i * 8, 15.0 + static_cast<double>(i));
+            }
+            return bytes;
+        }
+
+        // Names the first byte that differs, and which member owns it, so a failure is a
+        // diagnosis rather than "the buffer is wrong".
+        std::string DescribeOffset(const BlockLayout& layout, int offset) {
+            const std::pair<int, const char*> members[] = {
+                {layout.data0, "data0"}, {layout.data1, "data1"}, {layout.data2, "data2"},
+                {layout.data3, "data3"}, {layout.data4, "data4"}, {layout.data5, "data5"},
+                {layout.data6, "data6"}};
+            const char* owner = "(padding before data0)";
+            for (const auto& [start, name] : members) {
+                if (offset >= start) owner = name;
+            }
+            return std::string(owner);
+        }
+
         // Every double-typed uniform shape GLSL has, all thirteen of them, in one program - the
         // shape of KHR-GL43.compute_shader.fp64-case2. The scalar and the square matrices are
         // covered by the cases above; what only a set like this reaches is the NON-SQUARE
@@ -428,12 +578,24 @@ void main() {
             glUseProgram(0);
 
             // The readback has to undo exactly what the write did - the same std140 column
-            // padding, the same 4-byte components - or a dmat4 comes back with its columns
-            // shifted and nothing else in the API would say so.
+            // padding, the same component width - or a dmat4 comes back with its columns
+            // shifted and nothing else in the API would say so. Every value below except the
+            // scalar is exact in float32, so those expectations pin the LAYOUT and hold in
+            // either regime; the scalar is the one that also pins the PRECISION.
             GLdouble readScalar = 0.0;
             glGetUniformdv(m_program, scalar, &readScalar);
-            EXPECT_DOUBLE_EQ(readScalar, static_cast<double>(static_cast<float>(0.1)))
-                << "the value is what a float can hold, not the double that was passed in";
+            // 0.1 is not representable in float32, so what comes back names the regime: a
+            // backend without native fp64 narrowed it at the glUniform1d above (the module's own
+            // doubles were demoted, so its storage is 4 bytes per component), and one with it
+            // stored the double whole. Both are correct; asserting only the narrow answer would
+            // fail the moment fp64 stops being emulated, and asserting only the wide one would
+            // fail on every mobile device there is.
+            if (readScalar == 0.1) {
+                SUCCEED() << "this backend consumes 64-bit floats natively; the double survived whole";
+            } else {
+                EXPECT_DOUBLE_EQ(readScalar, static_cast<double>(static_cast<float>(0.1)))
+                    << "the value is what a float can hold, not the double that was passed in";
+            }
 
             GLdouble readVector[3] = {};
             glGetUniformdv(m_program, vector, readVector);
@@ -447,7 +609,8 @@ void main() {
                 EXPECT_DOUBLE_EQ(readMatrix[i], 100.0 + i) << "dmat4 component " << i;
             }
 
-            // The float query sees the same storage through the type it is actually stored as.
+            // The float query sees the same storage through a narrower type, and answers the
+            // same float either way: GL 4.6 core 7.6 converts on the way out.
             GLfloat readFloat = 0.0f;
             glGetUniformfv(m_program, scalar, &readFloat);
             EXPECT_FLOAT_EQ(readFloat, static_cast<float>(0.1));
@@ -697,24 +860,188 @@ void main() {
             EXPECT_EQ(glGetError(), static_cast<GLenum>(GL_NO_ERROR));
         }
 
-        TEST_F(DoublePrecisionScenario, A64BitVertexFormatIsDeclinedOnEveryBackend) {
+        TEST_F(DoublePrecisionScenario, A64BitVertexFormatIsRecordedAndItsArrayIsDroppedAtDraw) {
             if (!Ready()) return;
             // The demotion leaves no 64-bit shader input to feed, so there is nothing a 64-bit
             // vertex FETCH could be fetched into - on either backend, and no longer only on the
-            // ones whose device lacks shaderFloat64. Declined loudly rather than accepted and
-            // drawn as garbage; the matching POST row says the same thing at startup.
+            // ones whose device lacks shaderFloat64.
+            //
+            // What that costs is the ARRAY, not the CALL. GL 4.6 core 10.3.2 defines no error for
+            // a well-formed glVertexAttribLFormat and 64-bit attributes are core in the GL 4.3
+            // context MobileGL advertises, so refusing the call would be non-conformant and would
+            // leave four pure state queries unanswerable
+            // (KHR-GL43.vertex_attrib_binding.basic-state1/3). The format is therefore recorded and
+            // queryable; the enabled array is what gets dropped, and the attribute then reads its
+            // generic current value. The matching POST row says exactly that at startup.
             GLuint vao = 0;
             glGenVertexArrays(1, &vao);
             glBindVertexArray(vao);
             while (glGetError() != GL_NO_ERROR) {}
 
-            glVertexAttribLFormat(0, 3, GL_DOUBLE, 0);
-            EXPECT_EQ(glGetError(), static_cast<GLenum>(GL_INVALID_OPERATION));
+            glVertexAttribLFormat(1, 3, GL_DOUBLE, 8);
+            EXPECT_EQ(glGetError(), static_cast<GLenum>(GL_NO_ERROR))
+                << "glVertexAttribLFormat is a legal call in a GL 4.3 context";
+
+            GLint attribSize = 0;
+            GLint attribType = 0;
+            GLint attribIsLong = 0;
+            GLint attribRelativeOffset = 0;
+            glGetVertexAttribiv(1, GL_VERTEX_ATTRIB_ARRAY_SIZE, &attribSize);
+            glGetVertexAttribiv(1, GL_VERTEX_ATTRIB_ARRAY_TYPE, &attribType);
+            glGetVertexAttribiv(1, GL_VERTEX_ATTRIB_ARRAY_LONG, &attribIsLong);
+            glGetVertexAttribiv(1, GL_VERTEX_ATTRIB_RELATIVE_OFFSET, &attribRelativeOffset);
+            EXPECT_EQ(attribSize, 3);
+            EXPECT_EQ(attribType, static_cast<GLint>(GL_DOUBLE));
+            EXPECT_EQ(attribIsLong, GL_TRUE) << "GL_VERTEX_ATTRIB_ARRAY_LONG is what makes this the "
+                                                "unconverted form; without it the state is a lie";
+            EXPECT_EQ(attribRelativeOffset, 8);
+            EXPECT_EQ(glGetError(), static_cast<GLenum>(GL_NO_ERROR));
 
             glBindVertexArray(0);
             glDeleteVertexArrays(1, &vao);
             while (glGetError() != GL_NO_ERROR) {}
         }
 
+        // The consequence of recording the state rather than refusing the call: a 64-bit array can
+        // now be ENABLED in a VAO that a draw uses, which it never could before. That must not
+        // take the draw down. Leaving such an array enabled with no pointer behind it is exactly
+        // the documented Adreno null-deref (SIGSEGV inside the next glDraw*), so DirectGLES
+        // disables it before glVertexAttribPointer can ever see GL_DOUBLE, and DirectVulkan maps
+        // the format to VK_FORMAT_UNDEFINED so it never enters the pipeline's vertex input state.
+        //
+        // The shader deliberately does NOT read location 1: that keeps the two backends on the
+        // same path (DirectVulkan declines a draw whose SHADER reads an unsupported enabled array,
+        // by design and loudly, which is a different assertion from this one) and it is the shape
+        // the crash needed - an enabled array nothing set a pointer for.
+        TEST_F(DoublePrecisionScenario, AnEnabledLongArrayDoesNotBreakADrawThatIgnoresIt) {
+            if (!Ready()) return;
+
+            constexpr const char* kVs = R"(#version 430 core
+layout(location = 0) in vec2 aPos;
+void main() { gl_Position = vec4(aPos, 0.0, 1.0); }
+)";
+            constexpr const char* kFs = R"(#version 430 core
+out vec4 o_color;
+void main() { o_color = vec4(0.0, 1.0, 0.0, 1.0); }
+)";
+            std::string error;
+            const unsigned int program = CompileProgram(kVs, kFs, &error);
+            ASSERT_NE(program, 0u) << error;
+
+            ColorFbo target = MakeColorFbo(32, 32);
+            ASSERT_NE(target.fbo, 0u) << "could not create the render target";
+            BindFbo(target);
+
+            const float positions[8] = {-1.0f, -1.0f, 1.0f, -1.0f, -1.0f, 1.0f, 1.0f, 1.0f};
+            const double doubles[4] = {1.0, 2.0, 3.0, 4.0};
+
+            GLuint vao = 0;
+            GLuint positionBuffer = 0;
+            GLuint doubleBuffer = 0;
+            glGenVertexArrays(1, &vao);
+            glBindVertexArray(vao);
+            glGenBuffers(1, &positionBuffer);
+            glBindBuffer(GL_ARRAY_BUFFER, positionBuffer);
+            glBufferData(GL_ARRAY_BUFFER, sizeof(positions), positions, GL_STATIC_DRAW);
+            glGenBuffers(1, &doubleBuffer);
+            glBindBuffer(GL_ARRAY_BUFFER, doubleBuffer);
+            glBufferData(GL_ARRAY_BUFFER, sizeof(doubles), doubles, GL_STATIC_DRAW);
+            glBindBuffer(GL_ARRAY_BUFFER, 0);
+
+            glVertexAttribFormat(0, 2, GL_FLOAT, GL_FALSE, 0);
+            glVertexAttribBinding(0, 0);
+            glBindVertexBuffer(0, positionBuffer, 0, static_cast<GLsizei>(2 * sizeof(float)));
+            glEnableVertexAttribArray(0);
+
+            glVertexAttribLFormat(1, 1, GL_DOUBLE, 0);
+            glVertexAttribBinding(1, 1);
+            glBindVertexBuffer(1, doubleBuffer, 0, static_cast<GLsizei>(sizeof(double)));
+            glEnableVertexAttribArray(1);
+            EXPECT_EQ(FirstGLError(), 0u) << "setting up the 64-bit array was refused";
+
+            ClearTo(0.0f, 0.0f, 0.0f, 1.0f);
+            glUseProgram(program);
+            glDrawArrays(GL_TRIANGLE_STRIP, 0, 4);
+            EXPECT_EQ(FirstGLError(), 0u) << "a draw with an enabled 64-bit array must not raise an error";
+
+            const Image image = ReadPixels(target.width, target.height);
+            ASSERT_FALSE(image.Empty());
+            EXPECT_GT(image.At(target.width / 2, target.height / 2).g, 200)
+                << "the draw did not happen; the enabled 64-bit array must be dropped, not fatal";
+
+            glDisableVertexAttribArray(0);
+            glDisableVertexAttribArray(1);
+            glBindVertexArray(0);
+            glDeleteVertexArrays(1, &vao);
+            glDeleteBuffers(1, &positionBuffer);
+            glDeleteBuffers(1, &doubleBuffer);
+            BindDefaultFramebuffer();
+            DestroyColorFbo(target);
+            glUseProgram(0);
+            glDeleteProgram(program);
+            EXPECT_EQ(FirstGLError(), 0u);
+        }
+
+        TEST_F(DoublePrecisionScenario, AStorageBlockWithDoublesKeepsTheLayoutItWasBoundWith) {
+            if (!Ready()) return;
+
+            GLint blocks = 0;
+            glGetIntegerv(GL_MAX_COMPUTE_SHADER_STORAGE_BLOCKS, &blocks);
+            if (blocks < 4) {
+                GTEST_SKIP() << "GL_MAX_COMPUTE_SHADER_STORAGE_BLOCKS is " << blocks << "; this needs 4";
+            }
+
+            const unsigned int program = CompileComputeProgram(kBlockCopySource);
+            ASSERT_NE(program, 0u) << m_buildLog;
+
+            const std::vector<unsigned char> in140 = MakeBlockContents(kStd140);
+            const std::vector<unsigned char> in430 = MakeBlockContents(kStd430);
+            const std::vector<unsigned char> zero140(in140.size(), 0);
+            const std::vector<unsigned char> zero430(in430.size(), 0);
+
+            GLuint buffers[4] = {};
+            glGenBuffers(4, buffers);
+            const std::vector<unsigned char>* contents[4] = {&in140, &in430, &zero140, &zero430};
+            for (int i = 0; i < 4; ++i) {
+                glBindBufferBase(GL_SHADER_STORAGE_BUFFER, static_cast<GLuint>(i), buffers[i]);
+                glBufferData(GL_SHADER_STORAGE_BUFFER, static_cast<GLsizeiptr>(contents[i]->size()),
+                             contents[i]->data(), GL_DYNAMIC_COPY);
+            }
+            ASSERT_EQ(FirstGLError(), 0u);
+
+            glUseProgram(program);
+            glDispatchCompute(1, 1, 1);
+            glMemoryBarrier(GL_BUFFER_UPDATE_BARRIER_BIT);
+            EXPECT_EQ(FirstGLError(), 0u);
+
+            for (int pass = 0; pass < 2; ++pass) {
+                const BlockLayout& layout = pass == 0 ? kStd140 : kStd430;
+                const std::vector<unsigned char>& expected = pass == 0 ? in140 : in430;
+                const char* packing = pass == 0 ? "std140" : "std430";
+                std::vector<unsigned char> observed(expected.size(), 0xEE);
+                glBindBuffer(GL_SHADER_STORAGE_BUFFER, buffers[2 + pass]);
+                glGetBufferSubData(GL_SHADER_STORAGE_BUFFER, 0,
+                                   static_cast<GLsizeiptr>(observed.size()), observed.data());
+                int mismatches = 0;
+                int firstMismatch = -1;
+                for (std::size_t i = 0; i < expected.size(); ++i) {
+                    if (expected[i] == observed[i]) continue;
+                    ++mismatches;
+                    if (firstMismatch < 0) firstMismatch = static_cast<int>(i);
+                }
+                EXPECT_EQ(mismatches, 0)
+                    << packing << " block: " << mismatches << " of " << expected.size()
+                    << " bytes differ, first at byte " << firstMismatch << " (in "
+                    << DescribeOffset(layout, firstMismatch < 0 ? 0 : firstMismatch)
+                    << "); a block that was repacked around its doubles reads and writes every "
+                       "member after the first one at the wrong offset";
+            }
+
+            glUseProgram(0);
+            glDeleteProgram(program);
+            glDeleteBuffers(4, buffers);
+            EXPECT_EQ(FirstGLError(), 0u);
+        }
+
     } // namespace
 } // namespace MGITest

MobileGL/MG_IntegrationTest/Scenarios/DualSourceBlendScenario.cpp

@@ -0,0 +1,238 @@
+// MobileGL - MobileGL/MG_IntegrationTest/Scenarios/DualSourceBlendScenario.cpp
+// Copyright (c) 2025-2026 MobileGL-Dev
+// Licensed under the GNU Lesser General Public License v3.0:
+//   https://www.gnu.org/licenses/gpl-3.0.txt
+//   https://www.gnu.org/licenses/lgpl-3.0.txt
+// SPDX-License-Identifier: LGPL-3.0-only
+// End of Source File Header
+//
+// Scenario - A DUAL-SOURCE BLEND DRAW HAS TO SURVIVE ON EVERY DRIVER.
+//
+// GL_SRC1_COLOR / GL_ONE_MINUS_SRC1_COLOR / GL_SRC1_ALPHA / GL_ONE_MINUS_SRC1_ALPHA
+// (ARB_blend_func_extended, core since 3.3) need a backend capability that not every device has:
+// GL_EXT_blend_func_extended on the ES driver, or the dualSrcBlend device feature on Vulkan. When
+// the capability IS there both backends translate the factors properly, and that has always
+// worked. When it is NOT, both backends used to THROW_EXCEPTION at draw time - and
+// MG_Util/Types.h's THROW_EXCEPTION is a plain `throw`, with no catch anywhere in MG_Impl or
+// MG_Backend, so the exception unwound out through the C GL ABI and killed the process. An
+// application asking for a blend factor the device cannot do is a picture problem, never a reason
+// to take the process down.
+//
+// Both are now a DECLINE: the attachment is drawn with blending off and neutral One/Zero factors,
+// and the loss is logged once. So a dual-source draw has exactly two defined outcomes, and this
+// scenario pins that it lands on one of them and never on a crash:
+//
+//   capability present - src0 * src1 + dst * (1 - src1)
+//   capability absent  - src0, written straight through
+//
+// What each CI lane actually reaches: lavapipe has dualSrcBlend, so the DirectVulkan lane runs the
+// whole sequence and measures the blend. Mesa's GLES front end on llvmpipe has no
+// GL_EXT_blend_func_extended, so the ESSL stage carrying `layout(index = 1)` never compiles and the
+// program renders nothing - the DirectGLES lane therefore SKIPS on the capability probe in SetUp
+// rather than measuring a picture the driver never produced. The DECLINE arm itself - the path this
+// scenario exists for - is unit-tested against stubbed capabilities in
+// MG_Test/Framebuffer/FramebufferTest.cpp (DualSourceBlendIsDeclinedRatherThanThrownWhenTheExtensionIsMissing),
+// which is the only place it can be reached without a driver that lacks the extension.
+//
+// The Vulkan half has a second edge the last case covers: the dual-source VUIDs
+// (VUID-VkPipelineColorBlendAttachmentState-srcColorBlendFactor-00608 and its three siblings)
+// forbid a VK_BLEND_FACTOR_SRC1_* anywhere in VkPipelineColorBlendAttachmentState without the
+// feature, whatever blendEnable says - so leaving the factors in place while clearing the enable
+// would still be invalid pipeline state.
+
+#include <cstdint>
+#include <string>
+
+#include "../Harness/HeadlessGL.h"
+#include "../Harness/ScenarioFixture.h"
+
+#ifdef GLAPI
+#undef GLAPI
+#endif
+#define GL_GLEXT_PROTOTYPES
+#include <GL/gl.h>
+#include <GL/glcorearb.h>
+#undef GL_GLEXT_PROTOTYPES
+
+namespace MGITest {
+    namespace {
+
+        constexpr int kExtent = 16;
+
+        constexpr const char* kVertexSource = R"(#version 330 core
+void main()
+{
+    switch (gl_VertexID)
+    {
+      case 0: gl_Position = vec4(-1.0, 1.0, 0.0, 1.0); break;
+      case 1: gl_Position = vec4( 1.0, 1.0, 0.0, 1.0); break;
+      case 2: gl_Position = vec4(-1.0,-1.0, 0.0, 1.0); break;
+      case 3: gl_Position = vec4( 1.0,-1.0, 0.0, 1.0); break;
+    }
+}
+)";
+
+        // Two outputs on the SAME location, indices 0 and 1: the shader-side spelling of
+        // dual-source output (GLSL 3.30 4.4.2, the `index` layout qualifier). No
+        // glBindFragDataLocationIndexed needed, which keeps the program buildable through the
+        // harness's compile-and-link helper.
+        constexpr const char* kDualSourceFragmentSource = R"(#version 330 core
+uniform vec4 uSrc0;
+uniform vec4 uSrc1;
+layout(location = 0, index = 0) out vec4 fragColor0;
+layout(location = 0, index = 1) out vec4 fragColor1;
+void main()
+{
+    fragColor0 = uSrc0;
+    fragColor1 = uSrc1;
+}
+)";
+
+        class DualSourceBlendScenario : public ScenarioTest {
+        protected:
+            void SetUp() override {
+                ScenarioTest::SetUp();
+                if (!Ready()) return;
+                glGenVertexArrays(1, &m_vao);
+                glGenRenderbuffers(1, &m_renderbuffer);
+                glBindRenderbuffer(GL_RENDERBUFFER, m_renderbuffer);
+                glRenderbufferStorage(GL_RENDERBUFFER, GL_RGBA8, kExtent, kExtent);
+                glGenFramebuffers(1, &m_fbo);
+                glBindFramebuffer(GL_FRAMEBUFFER, m_fbo);
+                glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_RENDERBUFFER, m_renderbuffer);
+                ASSERT_EQ(glCheckFramebufferStatus(GL_FRAMEBUFFER), static_cast<GLenum>(GL_FRAMEBUFFER_COMPLETE));
+
+                std::string error;
+                m_program = CompileProgram(kVertexSource, kDualSourceFragmentSource, &error);
+                m_programError = error;
+                glViewport(0, 0, kExtent, kExtent);
+                glDisable(GL_SCISSOR_TEST);
+                glDisable(GL_DEPTH_TEST);
+
+                // Capability probe, not an assertion. A GL link that succeeded is not proof that
+                // the BACKEND can run the program: DirectGLES transpiles to ESSL lazily at first
+                // use, and GLSL ES has no `index` layout qualifier outside
+                // GL_EXT_blend_func_extended, so on a driver without it the stage never compiles
+                // and the draw renders nothing. One unblended white draw tells the two apart, and
+                // the cases skip rather than measure a picture the driver never produced.
+                if (m_program != 0) {
+                    glDisable(GL_BLEND);
+                    glBlendFunc(GL_ONE, GL_ZERO);
+                    Draw(/*src0=*/1.0f, /*src1=*/1.0f);
+                    glFinish();
+                    const Image probe = ReadPixels(kExtent, kExtent);
+                    m_programRenders =
+                        !probe.Empty() && static_cast<int>(probe.At(kExtent / 2, kExtent / 2).r) > 245;
+                }
+                for (int i = 0; i < 16 && glGetError() != GL_NO_ERROR; ++i) {
+                }
+            }
+
+            void TearDown() override {
+                if (!Ready()) return;
+                glDisable(GL_BLEND);
+                glBlendFunc(GL_ONE, GL_ZERO);
+                glBindFramebuffer(GL_FRAMEBUFFER, 0);
+                if (m_fbo != 0) glDeleteFramebuffers(1, &m_fbo);
+                if (m_renderbuffer != 0) glDeleteRenderbuffers(1, &m_renderbuffer);
+                if (m_program != 0) glDeleteProgram(m_program);
+                if (m_vao != 0) glDeleteVertexArrays(1, &m_vao);
+            }
+
+            void Draw(float src0, float src1) {
+                glUseProgram(m_program);
+                glUniform4f(glGetUniformLocation(m_program, "uSrc0"), src0, src0, src0, 1.0f);
+                glUniform4f(glGetUniformLocation(m_program, "uSrc1"), src1, src1, src1, 1.0f);
+                glBindVertexArray(m_vao);
+                glDrawArrays(GL_TRIANGLE_STRIP, 0, 4);
+                glBindVertexArray(0);
+                glUseProgram(0);
+            }
+
+            // Both cases share this gate: nothing below can be measured on a backend that cannot
+            // run a dual-source fragment program at all. Returns the skip reason, empty when the
+            // program runs - NOT a void helper that calls GTEST_SKIP itself, because GTEST_SKIP
+            // expands to a `return` and would leave only the HELPER, letting the case run its
+            // assertions anyway and report Failed instead of Skipped.
+            std::string WhyTheProgramCannotRun() const {
+                if (m_program == 0) {
+                    return "this driver cannot build a dual-source fragment shader: " + m_programError;
+                }
+                if (!m_programRenders) {
+                    return "this backend links a dual-source fragment program but renders nothing with it "
+                           "(GLSL ES has no `index` layout qualifier without GL_EXT_blend_func_extended)";
+                }
+                return {};
+            }
+
+            GLuint m_renderbuffer = 0;
+            GLuint m_fbo = 0;
+            GLuint m_vao = 0;
+            unsigned int m_program = 0;
+            bool m_programRenders = false;
+            std::string m_programError;
+        };
+
+    } // namespace
+
+    // The whole point of the scenario: this sequence used to be a process kill on any device
+    // without the capability, and it has to be a picture either way.
+    //
+    // dst is black, src0 is white and src1 is mid-grey, with SRC1_COLOR / ONE_MINUS_SRC1_COLOR.
+    //   blended  = 1.0 * 0.5 + 0.0 * 0.5 = 0.5  -> ~128
+    //   declined = 1.0                          ->  255
+    // Anything else means the factors were mistranslated rather than either honoured or declined.
+    TEST_F(DualSourceBlendScenario, DualSourceBlendDrawProducesOneOfTheTwoDefinedResults) {
+        if (!Ready()) GTEST_SKIP();
+        if (const std::string reason = WhyTheProgramCannotRun(); !reason.empty()) GTEST_SKIP() << reason;
+
+        glDisable(GL_BLEND);
+        glBlendFunc(GL_ONE, GL_ZERO);
+        Draw(/*src0=*/0.0f, /*src1=*/0.0f);
+
+        glEnable(GL_BLEND);
+        glBlendFunc(GL_SRC1_COLOR, GL_ONE_MINUS_SRC1_COLOR);
+        EXPECT_EQ(FirstGLError(), 0u) << "glBlendFunc must accept the GL_SRC1_* factors - they are core since 3.3";
+        Draw(/*src0=*/1.0f, /*src1=*/0.5f);
+        glFinish();
+        glDisable(GL_BLEND);
+        EXPECT_EQ(FirstGLError(), 0u) << "the dual-source draw left a GL error behind";
+
+        const Image image = ReadPixels(kExtent, kExtent);
+        ASSERT_FALSE(image.Empty());
+        const Rgba8 centre = image.At(kExtent / 2, kExtent / 2);
+        const int red = static_cast<int>(centre.r);
+        const bool blended = red > 100 && red < 160;
+        const bool declined = red > 245;
+        EXPECT_TRUE(blended || declined)
+            << "got " << centre << ", which is neither the dual-source blend (~128) nor the declined "
+            << "straight-through source (255) - the SRC1 factors were mistranslated";
+        Gl().EndFrame();
+    }
+
+    // The same factors with blending DISABLED. Nothing may blend, and on the Vulkan side nothing
+    // may reach VkPipelineColorBlendAttachmentState carrying a VK_BLEND_FACTOR_SRC1_* on a device
+    // without dualSrcBlend - the VUIDs bind to the struct, not to blendEnable. The picture is the
+    // source either way, so this case is really "no crash, no error, no surprise".
+    TEST_F(DualSourceBlendScenario, DualSourceFactorsWithBlendingDisabledJustWriteTheSource) {
+        if (!Ready()) GTEST_SKIP();
+        if (const std::string reason = WhyTheProgramCannotRun(); !reason.empty()) GTEST_SKIP() << reason;
+
+        glDisable(GL_BLEND);
+        glBlendFunc(GL_ONE, GL_ZERO);
+        Draw(/*src0=*/0.0f, /*src1=*/0.0f);
+
+        glBlendFunc(GL_SRC1_ALPHA, GL_ONE_MINUS_SRC1_ALPHA);
+        Draw(/*src0=*/1.0f, /*src1=*/0.25f);
+        glFinish();
+        EXPECT_EQ(FirstGLError(), 0u) << "a draw with SRC1 factors and blending off left a GL error behind";
+
+        const Image image = ReadPixels(kExtent, kExtent);
+        ASSERT_FALSE(image.Empty());
+        const Rgba8 centre = image.At(kExtent / 2, kExtent / 2);
+        EXPECT_GT(static_cast<int>(centre.r), 245)
+            << "got " << centre << ": blending is disabled, so the source has to be written straight through";
+        Gl().EndFrame();
+    }
+
+} // namespace MGITest