[Fix, Test] (clientfb): heal a texture's record from its first set_texture_params - a texture born while the family was not live (the context's default textures are constructed before the backend registers its consumer) has no record, so the application's first glTexParameter* on texture 0 was refused, silently latched away before m-1 and the retrace census's one residual after it went loud; the params path now publishes the create, the storage if the texture has any, and then the parameters, the respecify path's own self-heal shape

- VulkanRenderer::LookupVaoDrawMemo's legacy arm read MG_Config::Features.PipeHandleAbaControl
  a fourth time, bypassing the one-question accessor MagmaPipeAbaControlDefeatsIdentity that
  55d2af9b's own message says all the control's sites go through. It could never be false
  there: the control's early return fires ahead of BOTH arms, so a run that reaches the legacy
  compare has the knob off. The compare is unconditional again, with a note saying where the
  question is answered and to route through the accessor if that early return is ever narrowed
  - which is the trap the dead line was, since narrowing it would have silently restored D18's
  retired semantics at this one site.
- HandleRecycleScenario's vertex-array case still said "BOTH the VAO and the buffer are
  recycled here so that a key built out of raw addresses matches", which the body has
  contradicted since 55d2af9b: both buffers are created before the ABA window and neither is
  deleted inside it, deliberately, so that no buffer traffic moves VkBufferManager's
  slice-epoch counter inside the window. Only the VAO is recycled, and only its GL name - the
  heap block is not handed back, which is why the knob rather than the allocator constructs the
  AbaControl arms' collision.

.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:

.github/workflows/test.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:
     inputs:
       baseline_sha:
@@ -416,13 +420,20 @@ jobs:
             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)"
+          if ! nm --defined-only "${BUILD_DIR}/libMobileGL.so" | grep -q "MGPipeVerifyInputs"; then
+            echo "::error::libMobileGL.so defines no MGPipeVerifyInputs: -DMOBILEGL_PIPE_VERIFY=ON did not take, and every lane that consumes this artifact would run the comparator-free library and pass having compared nothing"
+            exit 1
+          fi
+          # The per-verb entry point, under EITHER of its two names. P2 renames
+          # MGPipeFillForVerb to MGPipeValidateForVerb (the body becomes the tracker's walk and
+          # the fill is one of its five steps), so this check has to accept both or it goes red on
+          # the rename for a reason that has nothing to do with what it tests. What it tests is
+          # unchanged: that the library HAS a per-verb entry point compiled in.
+          if ! nm --defined-only "${BUILD_DIR}/libMobileGL.so" | grep -qE "MGPipeValidateForVerb|MGPipeFillForVerb"; then
+            echo "::error::libMobileGL.so defines neither MGPipeValidateForVerb nor MGPipeFillForVerb: there is no per-verb entry point in this artifact, so nothing fills the block the comparator compares"
+            exit 1
+          fi
+          echo "libMobileGL.so defines MGPipeVerifyInputs and a per-verb entry point (${defined} defined symbols)"
 
       - name: Show ccache stats
         if: always()
@@ -538,6 +549,91 @@ jobs:
             ctest --output-on-failure -L integration-verify --no-tests=error
           fi
 
+      # The push-only unit tests, on the verify runtime.
+      #
+      # WHY HERE AND NOT IN `test`. The entries themselves are registered in EVERY build - they
+      # have to be, or `ctest -N` would stop matching name-for-name between the pull and the push
+      # build (gate G2). What is push-only is what they assert about: MGPipeRenderStateSpans.cpp
+      # and PipeApply.cpp are appended to SOURCE_FILES inside the `if (MOBILEGL_PIPE_PUSH)` block,
+      # which is exactly how the pull build stays symbol-identical, so in a pull build each case
+      # opens with `#if !MOBILEGL_PIPE_PUSH GTEST_SKIP() << "push not compiled in"`. The `test`
+      # job therefore runs G6's chunk-table walk and G10's residual assertions as a column of
+      # skips: CI executes the NAMES and never one of the assertions. This job unpacks a build
+      # that compiled them, so it is the first place in CI where they actually run.
+      #
+      # This artifact already carries them - the packaging step above tars
+      # ${BUILD_DIR}/MobileGL/MG_Test whole - so the whole cost is the run, which is ~14 s for
+      # ~1490 entries. --no-tests=error, because a packaging change that stopped shipping the
+      # unit binaries would otherwise report a green run of nothing.
+      - name: Unit tests on the verify runtime (G6, G10)
+        working-directory: build-verify
+        run: ctest --output-on-failure -L unit --no-tests=error -j "$(nproc)"
+
+      # The always-on P2 and P3a negative controls (G8, G10, G12), which are labelled
+      # integration-gpu and not integration-verify - they are about the handle key, the subsystem
+      # switch and the map-persistent counter, not about the comparator - so the lane above does
+      # not reach them. They are run HERE because this is the only CI job that unpacks a
+      # MOBILEGL_PIPE_PUSH build: every one of them reads a counter out of the library's summary
+      # line, and both the counters and their brackets are #if MOBILEGL_PIPE_PUSH, so in the pull
+      # `integration` job the entries exist (gate G2 requires the same names in both builds) but
+      # have nothing to assert.
+      #
+      # An arm whose subsystem has not landed on this tree SKIPS with the reason (never absent,
+      # never a green that asserted nothing), so this step is green through the P2 and P3a landing
+      # orders and starts asserting as each package arrives.
+      #
+      # The environment is the sibling step's, deliberately and in full: these entries run the
+      # same DirectVulkan binary through the same runner, so the three MOBILEGL_MAGMA_* fixes it
+      # needs apply here too, and a crash here has to leave a core for the same black-box flow.
+      # The step above is the only reason those lines exist in this job; a control that crashed
+      # without one would be the hardest failure in the job to diagnose.
+      #
+      # THE -R ALTERNATIVES ARE TEST-NAME PREFIXES, NOT LANE LABELS, and each one is deliberately
+      # the SHORTEST string that still selects only what it means to. `ResourceSubsystem` (not
+      # `ResourceSubsystemControl`) is what reaches the eight
+      # DirectGLES.ResourceSubsystemOn./Off.LargeArenaAdoptionScenario.* entries - the A/B lanes
+      # whose entire purpose is that the handle path and the legacy BufferBackendOps path must
+      # agree about an adopted store - as well as the two ResourceSubsystemControl. entries.
+      # `MapPersistentRoundtrip` is singular because LargeArenaAdoptionScenario's case is
+      # `AnAdoptionCostsExactlyOneMapPersistentRoundtrip`; the plural matched only the lane PREFIX
+      # of the other one. Both mistakes were silent: this is the only CI job that unpacks a push
+      # build, so an entry the filter misses is either never run under the P3a bits at all or runs
+      # only in the pull `integration` job, where a MOBILEGL_PIPE_PUSH value steers nothing
+      # (Config.h declares the field inside the push guard) and both arms are the same legacy path.
+      # A lane that cannot go red where it is installed is not a gate (ROADMAP.md:7).
+      # P4a ADDS THREE ALTERNATIVES, and each one is here because this is the only CI job that
+      # unpacks a push build:
+      #   * `ObjectSubsystem` reaches the three DirectGLES.ObjectSubsystemControl. entries - the
+      #     0x1fff-vs-0x1ff A/B and the 0x9ff dependency refusal (G12). `ResourceSubsystem` does
+      #     NOT match it: the two families are named apart on purpose, because they are different
+      #     phases' switches and a filter that merged them would hide one behind the other.
+      #   * `TextureParamsWithoutASamplerView` reaches G9's four cases, the scenario ROADMAP.md:20
+      #     names by hand. It runs in the ambient lanes, which the label already selects - but this
+      #     step is where those cases run against a PUSH library, and G9's whole claim is about the
+      #     push path. All four cases are green on the contract commit - including the one D-E3
+      #     expected to be red, for the reason the scenario's header records - so this row is green
+      #     from the day it lands and goes red only if a reachability path stops syncing texture
+      #     parameters at all, which is the coupling ARCHITECTURE.md:100 exists to remove.
+      #   * `TextureUploadShape` is RECORDED, not gated (D-D4): it asserts that the two upload-shape
+      #     counters could be read and that they agree, and prints the shape for MEASUREMENTS.md. It
+      #     is in the filter so that the number is actually collected on every run - an unmeasured
+      #     shape is not a recorded one - and because its own assertions can go red.
+      # As with the four before them, each alternative is the SHORTEST string that selects only what
+      # it means to.
+      - name: The handle-ABA, CSO, subsystem and texture-parameter controls (G8, G8b, G9, G10, G12)
+        working-directory: build-verify
+        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'
+          ctest --output-on-failure -L integration-gpu \
+            -R 'HandleRecycle|CsoContentAddressing|ResourceSubsystem|MapPersistentRoundtrip|ObjectSubsystem|TextureParamsWithoutASamplerView|TextureUploadShape' \
+            --no-tests=error -j 4
+
       # 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
@@ -694,7 +790,16 @@ jobs:
         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
+        # --expect-probes 4 (contract-v2.md 7.6): an exit code cannot tell four probes from none,
+        # so a --probe typo or a manifest edit that selected nothing would run zero probes and
+        # exit 0 - the second half of the finding that added the flag. The count is the length of
+        # scripts/check_include_closure.py's PROBES list and changing one means changing the other.
+        #
+        # --compiler stays clang++-20, which is what the step above installs (Debian's clang-20
+        # package ships /usr/bin/clang++-20). It is deliberately NOT the bare `clang++` the local
+        # campaign gate spells: that spelling exists because the WSL box has no clang++-20, and
+        # copying it here would trade a version-pinned compiler for whatever the runner has.
+        run: python3 scripts/check_include_closure.py --mode both --compiler clang++-20 --self-test --require-all --expect-probes 4
 
   benchmark:
     runs-on: ubuntu-latest
@@ -1479,10 +1584,27 @@ jobs:
     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.
+    env:
+      # THE CURRENT PHASE's base ref, for the two G5 region gates below. It is 37da3c3a - P4a's
+      # base ref, INTEGRATOR-DECISIONS ID-1 - and NOT the workflow's baseline_sha input: that input
+      # is the SYMBOL baseline (087685d1, P1's G1 reading) and it is empty on a push, whereas these
+      # gates ask "did the do-not-touch list move since the phase started".
+      #
+      # IT MOVED FROM P3a's 44c2b5cf TO P4a's 37da3c3a WITH THE PHASE, and that is a deliberate
+      # narrowing rather than a loss: P3a's eleven functions were compared against 44c2b5cf at P3a's
+      # own exit and were byte-identical there, so 37da3c3a carries the same bodies (measured: the
+      # eleven shas at 37da3c3a are the eleven shas at 44c2b5cf, and FlushPendingRangesFrom's is
+      # still the sha pinned in the script at 3e298c9a). What the two gates now both answer is "did
+      # anything on the list move during P4a", which is the question this phase can act on.
+      BASELINE: "37da3c3a"
 
     steps:
       - name: Checkout repo
         uses: actions/checkout@v6
+        with:
+          # The G5 gate reads Managers.cpp at BASELINE with `git show`, which a depth-1 checkout
+          # does not have. Nothing else in this job needs history.
+          fetch-depth: 0
 
       # 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
@@ -1521,11 +1643,93 @@ jobs:
           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
+      # A GATE as of P2, which is when MG_Pipe/DirtySurface.def exists to diff the scan against
+      # (ROADMAP.md:18 puts the first mapping round in P2). --check fails BOTH directions: a
+      # mutator the scanner finds with no row in the def, and a row naming a mutator the scan no
+      # longer finds - so a deleted mutator cannot leave a stale row behind claiming coverage.
+      #
+      # --self-test is the half that keeps --check honest, and it is not optional. A completeness
+      # check that silently stopped checking produces exactly the same green as a complete
+      # mapping; the self-test feeds it two canned negative controls (a mutator withheld from the
+      # def, a row naming a function that does not exist) and fails if either fails to trip. Same
+      # shape as gen_pipe.py --self-test and check_include_closure.py above.
+      #
+      # What this gate does NOT cover is written into DirtySurface.def's header rather than left
+      # implicit: the scanner attributes a mutation inside a lambda to the enclosing function,
+      # reads a mutation published through a helper as deferred, and scans only MG_Impl/GLImpl -
+      # so the four MGP_NOTE_MUTATION sites in MG_State are outside it entirely. This is a
+      # completeness gate over what the scanner can see; the semantic proof is the verify lane.
+      - name: MGPipe dirty-surface mapping is complete (G9)
+        run: |
+          python3 scripts/gen_pipe_dirty_surface.py --check
+          python3 scripts/gen_pipe_dirty_surface.py --self-test
+
+      # A GATE as of P3a (G5). "pool 与延迟释放原样搬" (ROADMAP.md:19) is meant literally: the
+      # buffer pool, the deferred-release drain and the three persistently mapped rings move
+      # VERBATIM, and ARCHITECTURE.md:515 says why - their retire happens only inside Present, so
+      # a batching or ordering change there starves them, and nothing else in this workflow can
+      # see it. P3a rewrites the rest of Managers.cpp by design, so a file diff says nothing; the
+      # script extracts the ELEVEN named bodies and compares their hashes on their own.
+      #
+      # Eleven and not ten (ID-15): Managers.cpp carries the three-tier flush drain TWICE, once
+      # per preprocessor arm, and a push build compiles only FlushPendingRangesFrom while the
+      # untouched FlushPendingRangesNow lives in the `#else`. Hashing the pull name alone would
+      # protect text no shipping build compiles, so both are hashed - the pull ladder against
+      # BASELINE, the push ladder against a sha pinned in the script at 3e298c9a, because that
+      # one was born in P3a and has no body at the base ref to compare with.
+      #
+      # Scoped to the disaggregation branch and to a manual dispatch, deliberately: the question
+      # is "did these eleven move since P3a started", and BASELINE is P3a's base ref. On dev,
+      # where unrelated buffer fixes land on their own schedule, the same comparison would be
+      # asking a question nobody posed - it belongs with the TEMPORARY trigger lines at the top
+      # of this file and retires with them.
+      #
+      # --self-test is the half that keeps it honest, and it is not optional: a comparison that
+      # silently stopped comparing produces exactly the same green as eleven untouched bodies. It
+      # runs six canned controls - eleven bodies extracted, an untouched copy compared equal, an
+      # edit OUTSIDE them ignored, and each of the three perturbation targets (ClearBufferPool
+      # and BOTH flush ladders) reported BY NAME - and fails if any of them does not answer.
+      # Same shape as gen_pipe.py --self-test above.
+      - name: The buffer pool, the deferred-release drain and the rings did not move (G5)
+        if: ${{ github.ref == 'refs/heads/feat/disaggregated' || github.event_name == 'workflow_dispatch' }}
+        run: bash scripts/p3a_untouched_regions.sh "${BASELINE}" HEAD
+
+      - name: The untouched-region gate can still fail (G5)
+        if: ${{ github.ref == 'refs/heads/feat/disaggregated' || github.event_name == 'workflow_dispatch' }}
+        run: bash scripts/p3a_untouched_regions.sh --self-test
+
+      # A GATE AS OF P4a (G5), and a SECOND script rather than an edit to the one above. P4a extends
+      # the same claim to the rest of ARCHITECTURE.md:318's do-not-touch list - the unpack PBO ring's
+      # staging repack and its two helpers, the attachment permutation, the D24S8 sampling-emulation
+      # core and the format-caveat handler - which is SEVENTEEN regions across THREE files
+      # (BRIEF-P4A.md D-N: P3a's eleven, which P4a must not touch either, plus P4a's six). The
+      # parent's SOURCE_PATH is a single file, so the extension needed a per-region source path and
+      # a region KIND (DepthStencilSamplingReadImpl is a namespace, not a function); everything else
+      # about the extraction is its parent's, verbatim.
+      #
+      # Both scripts run. The parent keeps answering its own question against its own eleven, so a
+      # regression in either half names itself, and neither gate can be silenced by editing the
+      # other's list.
+      #
+      # Same feat/disaggregated-or-dispatch guard as the P3a step, for the same reason: the question
+      # is "did these move since the phase started", and on dev - where unrelated buffer and texture
+      # fixes land on their own schedule - it would be a question nobody posed. It belongs with the
+      # TEMPORARY trigger lines at the top of this file and retires with them.
+      #
+      # --self-test is the half that keeps it honest and is not optional: a comparison that silently
+      # stopped comparing produces exactly the same green as seventeen untouched regions. It runs
+      # three positive controls (seventeen regions extracted, an untouched copy compared equal, an
+      # edit OUTSIDE them invisible in all three files) and FOUR negative ones - ClearBufferPool,
+      # FlushPendingRangesNow, RecomputeBackendColorSlots and StageBlocksIntoUnpackRing, each
+      # perturbed on its own and each required to be named BY NAME - and fails if any of them does
+      # not answer.
+      - name: The unpack ring, the attachment permutation, the D24S8 core and the format caveat did not move (G5)
+        if: ${{ github.ref == 'refs/heads/feat/disaggregated' || github.event_name == 'workflow_dispatch' }}
+        run: bash scripts/p4a_untouched_regions.sh "${BASELINE}" HEAD
+
+      - name: The P4a untouched-region gate can still fail (G5)
+        if: ${{ github.ref == 'refs/heads/feat/disaggregated' || github.event_name == 'workflow_dispatch' }}
+        run: bash scripts/p4a_untouched_regions.sh --self-test
 
       # 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

CMakeLists.txt

@@ -27,6 +27,13 @@ option(MOBILEGL_BUILD_SERVER_SPIKE "Build the P0 spike-A MobileGLServer delivery
 # 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)
+# Track H's old-versus-new arm (ARCHITECTURE.md 9.6). With a MOBILEGL_PIPE_PUSH bit clear
+# the backend would still run the RE-KEYED memo code, so the bitmask alone stops being a
+# valid A/B the moment a handle wave lands: this option compiles the pre-handle arm - the
+# registries, OwnerEquals, the TwinLookupMemos, g_fbSlotCache, ComputePipelineStateHash,
+# the address-keyed VaoDrawMemo - beside it, behind the same PipeInputs interface. ON for
+# the whole migration window; it retires with the pull path itself at P13.
+option(MOBILEGL_PIPE_LEGACY_MEMOS "Compile the pre-handle memo arm beside the {slot, gen} arm so Track H has a real A/B (ARCHITECTURE.md 9.6)" ON)
 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")
 
@@ -328,6 +335,7 @@ set(SOURCE_FILES
     MobileGL/MG_Util/BackendLoaders/Vulkan/Loader.cpp
 
     MobileGL/MG_Util/SelfTest/DriverBugProbes.cpp
+    MobileGL/MG_Util/SelfTest/PersistentBufferOrderingProbe.cpp
     MobileGL/MG_Util/SelfTest/DriverPost.cpp
     MobileGL/MG_Util/SelfTest/DriverPostIterationRPWitness.cpp
     MobileGL/MG_Util/SelfTest/PrimitivesGeneratedNoXfbProbe.cpp
@@ -464,11 +472,31 @@ if (MOBILEGL_PIPE_VERIFY AND NOT MOBILEGL_PIPE_PUSH)
     set(MOBILEGL_PIPE_PUSH ON)
 endif()
 
+# In a pull build the legacy arm is the ONLY arm, so the option cannot be off there.
+# A normal variable, not a forced cache write, for the same reason as the two above.
+if (NOT MOBILEGL_PIPE_PUSH AND NOT MOBILEGL_PIPE_LEGACY_MEMOS)
+    message(STATUS "MobileGL: MOBILEGL_PIPE_PUSH=OFF forces MOBILEGL_PIPE_LEGACY_MEMOS ON for this "
+                   "configure: with nothing pushed it is the only arm there is")
+    set(MOBILEGL_PIPE_LEGACY_MEMOS 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
+        # P2's contract: the chunk table and its subset hash, the in-process applier, and
+        # the client's {slot, gen} allocator. All three are push-only, which is how the
+        # pull build gains no symbol from P2 (G1) - a declaration emits nothing.
+        MobileGL/MG_Pipe/MGPipeRenderStateSpans.cpp
+        MobileGL/MG_Pipe/PipeApply.cpp
+        MobileGL/MG_Impl/Pipe/SlotAllocator.cpp
+        # P4a's contract: the reflection-archive serializer over ProgramArtifacts.h's
+        # VisitFields tables. Push-only for the same G1 reason as the three above - in
+        # monolith the archive never crosses (create_shader_state hands the two structs over
+        # by pointer beside the record), so the codec is live code only in the VERIFY lane,
+        # where the applier serialises, deserialises and field-compares before storing.
+        MobileGL/MG_State/GLState/ProgramState/ProgramArtifactsCodec.cpp
     )
 endif()
 
@@ -551,6 +579,9 @@ endif()
 if (MOBILEGL_PIPE_VERIFY)
     list(APPEND MOBILEGL_COMPILE_DEF -DMOBILEGL_PIPE_VERIFY=1)
 endif()
+if (MOBILEGL_PIPE_LEGACY_MEMOS)
+    list(APPEND MOBILEGL_COMPILE_DEF -DMOBILEGL_PIPE_LEGACY_MEMOS=1)
+endif()
 
 message(STATUS "MOBILEGL_COMPILE_DEF=${MOBILEGL_COMPILE_DEF}")
 

MobileGL/Config.h

@@ -14,7 +14,7 @@ namespace MobileGL::MG_Config {
     inline const String ProjectName = "MobileGL";
     inline const String CoreName = "MobileGL Core";
     inline const String CoreVendor = "MobileGL-Dev (BZLZHH, Swung0x48, Tungsten)";
-    inline const Version CoreVersion = {26, 8, 0, "-dev", VersionType::Development};
+    inline const Version CoreVersion = {26, 9, 0, "-dev", VersionType::Development};
     inline const VersionStringFormatAttrib DefaultVersionStringFormatAttrib = {2, 2, 0, true, true};
     inline const Uint64 CacheVersion = 0;
 
@@ -319,10 +319,34 @@ namespace MobileGL::MG_Config {
         // --- 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.
+        // 0 - the only shipped value until the migration lands - is "pull everything",
+        // i.e. exactly today's behaviour, and is the default of a PULL build, where the
+        // knob is meaningless anyway. A PUSH build defaults to every subsystem migrated so
+        // far (MG_Pipe::kMGPipeSubsystemsMigratedAtP4a), so MOBILEGL_PIPE_PUSH=0 in the
+        // environment is the all-pull control and 0x1ff (kMGPipeSubsystemsMigratedAtP3a) is
+        // the "everything before P4a" control P4a's A/B is run against - each phase's
+        // constant survives as the next phase's control, which is why none of them is ever
+        // edited. Accepts decimal or 0x-prefixed hex, and operators pass it as hex, so the
+        // bits are listed here (MG_Pipe/MGPipe.h owns them):
+        //   0x01 render state (create/bind_render_state + set_dynamic_state)
+        //   0x02 pixel pack        0x04 patch state      0x08 vertex attrib defaults
+        //   0x10 residual values   0x20 Espryt slots     0x40 Magma vertex input
+        //   0x80 resources (the resource_* family: the seven BufferBackendOps hooks)
+        //   0x100 vertex input (vertex elements / vertex buffers / index buffer)
+        //   0x200 framebuffer (set_framebuffer_state)              - requires 0x400
+        //   0x400 texture resources (texture + renderbuffer resource_*,
+        //         set_texture_params)                              - requires 0x80 AND 0x800
+        //         (the built-in sampler CSO a set_texture_params record names is minted by
+        //         the sampler family alone, ID-15; the four rows are MG_Impl/Pipe/PipeFill.cpp's
+        //         kMGPipeP4aFamilyDependencies, mirrored bit for bit by Espryt's resolvers)
+        //   0x800 samplers (sampler CSO, sampler view, set_sampler_views /
+        //         bind_sampler_states / set_shader_images)         - requires 0x400
+        //   0x1000 programs (shader CSO, set_draw/dispatch_program, global constants)
+        //   A dependency that is not met is REFUSED with one ERROR naming both bits and the
+        //   family runs its legacy arm; it is never half-run.
+        //   1<<63 NOT a subsystem, a BEHAVIOUR: turn OFF client-side content addressing of
+        //         CSOs, so every pipeline-version change mints a fresh CSO and the map is
+        //         never probed. The negative control the CSO design is measured against.
         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
@@ -349,6 +373,18 @@ namespace MobileGL::MG_Config {
         // Fatal{UnmigratedPipeInput} (negative control B). Unknown name is
         // Fatal{PipeVerifyBadKnob}.
         String PipePoisonOmit;
+        // MOBILEGL_PIPE_HANDLE_ABA_CONTROL (negative control C, P2 brief D18): replace the
+        // OBJECT IDENTITY in every DirectVulkan vertex-input memo key with a constant, on
+        // whichever arm the run is on - the pre-handle (address, lifetime id) pair AND the
+        // handle arm's {slot, gen} generation - so a replacement object inherits its dead
+        // predecessor's resolved vertex bindings and HandleRecycleScenario.AbaControl asserts
+        // the WRONG pixels. That is what proves the reproducer still reproduces. D18 wrote
+        // this as "hash the raw BufferObject* instead of its lifetime id"; measured, the heap
+        // block is never handed back, so that spelling collided with nothing and the control
+        // went vacuous - see MagmaPipeArms.h's MagmaPipeAbaControlDefeatsIdentity for the
+        // measurement and for what the control still leaves standing. Under
+        // MOBILEGL_PIPE_PUSH only, so it cannot exist in a shipping pull build.
+        Bool PipeHandleAbaControl = false;
 #endif
         // MOBILEGL_PIPE_STATS: dump the boundary counters (bytes, calls, roundtrips,
         // texture pulls, upload shapes, residual-block bytes, index mirror bytes).

MobileGL/ConfigLoader.cpp

@@ -7,6 +7,12 @@
 // End of Source File Header
 
 #include "Config.h"
+#if MOBILEGL_PIPE_PUSH
+// For kMGPipeSubsystemsMigratedAtP4a, the push build's PipePush default (the P2 and P3a
+// constants beside it are the phase-by-phase controls, not the default). Push-only, so the
+// pull build's translation unit is unchanged.
+#include <MG_Pipe/MGPipe.h>
+#endif
 
 #include <cerrno>
 #include <cstdlib>
@@ -242,7 +248,18 @@ namespace MobileGL::MG_ConfigLoader {
         // 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.
+#if MOBILEGL_PIPE_PUSH
+        // A push build with the knob unset runs every subsystem migrated so far, so the
+        // shipped path is the one the gates measure; MOBILEGL_PIPE_PUSH=0 in the
+        // environment is the all-subsystems-pull control that reproduces P1 exactly, and
+        // kMGPipeSubsystemsMigratedAtP3a (0x1ff) is the phase-by-phase control - P4a's four
+        // subsystems off, everything P3a landed still on.
+        features.PipePush = QueryEnvUint64("MOBILEGL_PIPE_PUSH", MG_Pipe::kMGPipeSubsystemsMigratedAtP4a);
+#else
+        // Meaningless in a pull build: there is nothing to push. Config.h documents 0 as
+        // "pull everything" and that stays literally true.
         features.PipePush = QueryEnvUint64("MOBILEGL_PIPE_PUSH", 0);
+#endif
         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.
@@ -250,6 +267,7 @@ namespace MobileGL::MG_ConfigLoader {
             QueryEnvQuirkOverride("MOBILEGL_PIPE_VERIFY_FATAL") != MG_Config::QuirkOverride::ForceOff;
         QueryEnvVariable("MOBILEGL_PIPE_VERIFY_CORRUPT", features.PipeVerifyCorrupt, "");
         QueryEnvVariable("MOBILEGL_PIPE_POISON_OMIT", features.PipePoisonOmit, "");
+        features.PipeHandleAbaControl = QueryEnvFlag("MOBILEGL_PIPE_HANDLE_ABA_CONTROL");
 #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

MobileGL/MG_Backend/DirectGLES/SlotTables.h

@@ -0,0 +1,577 @@
+// MobileGL - MobileGL/MG_Backend/DirectGLES/SlotTables.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>
+
+#include <MG_Pipe/MGPipeHandles.h>
+
+#if MOBILEGL_PIPE_PUSH
+#include <MG_Impl/Pipe/SlotAllocator.h>
+#endif
+
+// Espryt 0b, the first Track H slice: the DENSE, {slot, gen}-keyed twin table that replaces
+// StateBackendObjectRegistry's UnorderedMap<StateObject*, Entry>.
+//
+// What changes, and why each of them is the point:
+//
+//  * The KEY stops being a frontend heap address. It is MGPipeHandle{Slot, Gen}, minted by the
+//    client's MGPipeSlotAllocator off the frontend object's GetLifetimeId(). A recycled heap
+//    address cannot reproduce a handle, so the weak_ptr the registry carried per entry purely
+//    to catch that (its Entry::stateRef, used as an IDENTITY test) stops being an identity
+//    mechanism, and OwnerEquals / TwinLookupMemo x3 / UnitSamplerLookupMemo's owner compare all
+//    lose their reason to exist.
+//  * The lookup stops being a hash probe into an open-addressed map and becomes one bounds
+//    check plus one array index, so a returned BackendPtr* is NOT invalidated by the next Find
+//    on the table. That kills the hazard Managers.h documents at length, and with it the
+//    by-value copy plus second Find that SyncTextureObjectToBackend paid to survive it.
+//  * Slots are dense per kind, which is what lets the server side (ARCHITECTURE.md 10.1,
+//    MG_Remote/Server/PipeObjectTables) be an array rather than an object graph.
+//
+// Death is ANNOUNCED, and that is what lets this table have no garbage collector - the
+// deliverable ROADMAP.md:18 spells "GC" in and the one D13 makes a precondition of the switch-
+// over. All six re-keyed object classes raise MG_State::GLState::NotifyStateObjectDestroyed()
+// from their destructor (BufferBackendOps' shape, one entry point for six kinds), the backend
+// consumes it in Managers.cpp, and OnFrontendObjectDestroyed() below drops the twin in EVERY
+// table of the kind and returns the slot, at the moment the frontend object's last SharedPtr
+// goes. So:
+//   * there is NO draw-path tick, NO creation tick and NO sweep of any kind on this arm. The
+//     seven CollectGarbageIfNeeded call sites in DirectGLES.cpp drive the LEGACY registry only;
+//   * a twin, and the driver storage it owns, is freed when the application lets go of the
+//     object rather than up to 64 creations or 1024 draw ticks later. That is what
+//     Managers.h's "dead gigabytes" note asked for.
+//
+// EVERY HOLDER OF THE KIND, not one. Two live tables of one kind is a real configuration - the
+// ScopedDirectGLESTextureBindings fixture keeps a by-value copy of the Texture registry for the
+// length of a test, and a context reset does the same in reverse - and the slot allocator
+// erases its lifetimeId -> slot mapping on Free, so a notice delivered to one holder and
+// resolved again by the next would find nothing to resolve. Every table therefore links itself
+// into a per-table-type list at construction and out at destruction, and one notice resolves
+// the handle ONCE, drops the twin in each holder BY HANDLE, and frees the slot once, last. No
+// holder can be left naming a live entry for a dead object, and there is nothing a sweep could
+// still find. (The list is per table TYPE; the kind is the type's template parameter, and each
+// of the six kinds has exactly one table type in this backend. Magma's subsystem-4 table mints
+// out of its own per-renderer allocator, not MGPipeSlots(), so it is not a holder here.)
+//
+// The weak_ptr per entry survives for exactly one reason: ForEachLive() hands the callee a
+// STRONG reference to the frontend object, which the one direct-iteration site
+// (ScopedDetachedTextureFramebufferAttachments) needs. It is never an identity test - that is
+// what Gen is for - and it is never read to decide whether an entry is dead: a destructor that
+// runs after exit() has begun has its notice dropped by InProcessTeardown(), and that twin is
+// then a DELIBERATE leak (the process is exiting, the driver reclaims the object, and a twin
+// destructor must not call into a driver that may already be unloaded), not something to be
+// collected later.
+//
+// P3+ DEBT, recorded rather than hidden: this header is under MG_Backend/ and it MINTS
+// handles (MGPipeSlots().Acquire below) off a frontend SharedPtr's GetLifetimeId().
+// MGPipeHandles.h:13-16 says a handle is minted by the CLIENT and never by the server, and
+// under a real split neither the frontend object nor its lifetime id exists on this side of
+// the wire. This is monolith glue: the minting and the lifetimeId -> handle resolution both
+// belong on the client, and the backend should receive the handle in the verb payload. It is
+// NOT part of "Track H done" and check_include_closure.py does not probe MG_Backend headers,
+// so nothing catches it automatically.
+namespace MobileGL::MG_Backend::DirectGLES {
+
+#if MOBILEGL_PIPE_PUSH
+
+    // Declared in Managers.h as well; repeated here because this header is included from it
+    // before that declaration, and the table below is the arming site on this arm (D13: "the
+    // arming site moves to the slot table's first insertion").
+    void EnsureProcessTeardownSentinel();
+
+    // What the two knobs add up to. Split out as a PURE function of them so a test can drive
+    // every combination without needing a process per combination.
+    enum class EsprytSlotArmVerdict {
+        Handles, // kMGPipeSubsystemEsprytSlots is set: the {slot, gen} tables run.
+        Legacy,  // the bit is clear and the legacy address-keyed registry is reachable.
+        NoArm,   // the bit is clear AND MOBILEGL_PIPE_LEGACY_MEMOS=0 made the legacy arm
+                 // unreachable, so the operator asked for a configuration with no arm at all.
+    };
+
+    EsprytSlotArmVerdict ClassifyEsprytSlotArm(Bool subsystemBitSet, Bool legacyMemosEnabled);
+
+    // This process's verdict, read off MG_Config::Features. Latches nothing and stops nothing.
+    EsprytSlotArmVerdict CurrentEsprytSlotArmVerdict();
+
+    // Says, at backend bring-up, that the knobs leave no arm - and does NOT stop.
+    //
+    // The stop cannot live here, and that is the whole point of the split. Backend context
+    // creation runs inside eglMakeCurrent, and the integration harness pre-flights exactly that
+    // sequence in a FORKED CHILD (MG_IntegrationTest/Harness/HeadlessGL.cpp): a child that dies
+    // on a signal is reported as "no usable GPU/display/ICD" and every scenario in the lane is
+    // SKIPPED - i.e. the lane goes green having run nothing, on the very pair of env vars the
+    // D14/D18 A/B is driven with, which is what ROADMAP.md:7 forbids. So bring-up only
+    // DIAGNOSES; the stop is raised by ResolveEsprytSlotTablesArm() at the first twin lookup,
+    // which happens in the test body where the harness reports it as a failure.
+    //
+    // The CALL SITE (InitDisplayAndContext in DirectGLES.cpp) is pinned by
+    // DirectGLESSlotTable.EglBringUpUnderTheArmlessKnobPairReturnsInsteadOfStopping, which runs
+    // the real bring-up entry point under the pair in a forked child: edit that site back to
+    // ResolveEsprytSlotTablesArm() and the case fails naming both knobs.
+    void DiagnoseEsprytSlotArm();
+
+    // Reads the config, logs, installs the death-notice consumer, and STOPS when the operator
+    // left no arm at all. Cold: called exactly once per process, from the latch below - i.e. at
+    // the first twin lookup, which is the first moment an arm is actually needed. A process
+    // that never twins anything needs no arm and is not stopped.
+    Bool ResolveEsprytSlotTablesArm();
+
+    // True when this process runs the {slot, gen} arm. Fixed for the life of the process: the
+    // two arms hold their twins in different containers, so flipping mid-run would strand them.
+    //
+    // INLINE on purpose. Every Find / GetOrCreate / HandleOf / ForEachLive on the twin tables
+    // consults it, i.e. it is on the per-draw path several times per draw. As an out-of-line
+    // function in Managers.cpp (no LTO in any shipped configuration) that was a call through
+    // the PLT per lookup; here the caller sees a guard-variable load and a perfectly-predicted
+    // branch, and the arm dispatch folds into the caller.
+    inline Bool EsprytSlotTablesEnabled() {
+        static const Bool enabled = ResolveEsprytSlotTablesArm();
+        return enabled;
+    }
+
+    template <typename StateObject, typename BackendObject, MG_Pipe::MGPipeKind kKind>
+    class BackendSlotTable {
+    public:
+        using StatePtr = SharedPtr<StateObject>;
+        using StateWeakPtr = std::weak_ptr<StateObject>;
+        using BackendPtr = SharedPtr<BackendObject>;
+
+        // The largest slot index this table will grow to for a handle that ARRIVED in a call's
+        // payload. Slots are dense and allocated per kind, so a million of one kind is already
+        // far past any application's live object count; the cap is here because the alternative
+        // is letting a corrupt 32-bit slot decide a vector resize. See GetOrCreate(MGPipeHandle).
+        static constexpr Uint32 kMaxHandleSlot = 1u << 20;
+
+        struct Entry {
+            BackendPtr backend;
+            // LIVENESS ONLY, and only for ForEachLive(), which locks it so the callee holds a
+            // strong ref. Never compared against another object to decide identity - that is
+            // what Gen is for - never dereferenced for its address, and never read to decide
+            // whether the slot is dead: death is announced, not discovered.
+            StateWeakPtr stateRef;
+            // The generation this entry's twin was built for. An entry whose Gen no longer
+            // matches the allocator's is a twin of the slot's PREVIOUS owner.
+            Uint32 Gen = 0;
+            Bool Live = false;
+        };
+
+        // Every constructor links the table into the per-type holder list and the destructor
+        // unlinks it, so a by-value copy (the ScopedDirectGLESTextureBindings fixture's saved
+        // registry) is a holder for exactly as long as it exists. Copy and move carry the
+        // ENTRIES and the memo; the links are the table's own and are never copied.
+        BackendSlotTable() { LinkHolder(); }
+        BackendSlotTable(const BackendSlotTable& other):
+            m_slots(other.m_slots),
+            m_nullTwin(other.m_nullTwin),
+            m_memoLifetimeId(other.m_memoLifetimeId),
+            m_memoHandle(other.m_memoHandle) {
+            LinkHolder();
+        }
+        BackendSlotTable(BackendSlotTable&& other) noexcept:
+            m_slots(std::move(other.m_slots)),
+            m_nullTwin(std::move(other.m_nullTwin)),
+            m_memoLifetimeId(other.m_memoLifetimeId),
+            m_memoHandle(other.m_memoHandle) {
+            other.m_slots.clear();
+            other.ForgetHandle();
+            LinkHolder();
+        }
+        BackendSlotTable& operator=(const BackendSlotTable& other) {
+            if (this != &other) {
+                m_slots = other.m_slots;
+                m_nullTwin = other.m_nullTwin;
+                m_memoLifetimeId = other.m_memoLifetimeId;
+                m_memoHandle = other.m_memoHandle;
+            }
+            return *this;
+        }
+        BackendSlotTable& operator=(BackendSlotTable&& other) noexcept {
+            if (this != &other) {
+                m_slots = std::move(other.m_slots);
+                m_nullTwin = std::move(other.m_nullTwin);
+                m_memoLifetimeId = other.m_memoLifetimeId;
+                m_memoHandle = other.m_memoHandle;
+                other.m_slots.clear();
+                other.ForgetHandle();
+            }
+            return *this;
+        }
+        ~BackendSlotTable() { UnlinkHolder(); }
+
+        // Resolve-or-create. The handle comes from the client allocator keyed on the frontend
+        // object's lifetime id, so two calls for the same live object always land on the same
+        // slot, and a successor object at the same heap address never does.
+        BackendPtr& GetOrCreate(const StatePtr& stateObj) {
+            // No assert on null here, unlike the map arm: null is TOLERATED, so a DEBUG build
+            // must not trap where the release build quietly does the documented thing.
+            if (stateObj == nullptr) {
+                // The registry this replaces inserted a null KEY and handed back that entry's
+                // twin (DirectGLES.cpp's SyncTextureObjectToBackend documents relying on
+                // exactly that tolerance), so a release build never dereferenced null here.
+                // Keep the shape exactly, INCLUDING across calls: the map kept its null-keyed
+                // entry, so a second null call was handed the same twin the first one got.
+                // Resetting here instead would have destroyed it - an arm difference in the one
+                // path that documents relying on this. One per-table parking slot, never live,
+                // never handed a handle, because a null object has no identity and
+                // therefore cannot have a {slot, gen}.
+                return m_nullTwin;
+            }
+
+            // D13: the teardown sentinel is armed by the slot table's first insertion. Twin
+            // creation is the moment a driver-owned id starts needing a guarded destructor;
+            // this is the cold path, so the once-guard costs nothing per draw. On the legacy
+            // arm StateBackendObjectRegistry::GetOrCreate arms it itself.
+            EnsureProcessTeardownSentinel();
+
+            const MG_Pipe::MGPipeHandle handle =
+                MG_Pipe::MGPipeSlots().Acquire(kKind, stateObj->GetLifetimeId());
+            MOBILEGL_ASSERT(!MG_Pipe::MGPipeHandleIsNull(handle),
+                            "MGPipe slot space of kind %u is exhausted",
+                            static_cast<Uint32>(kKind));
+            Entry& entry = EntryAt(handle.Slot);
+            if (entry.Live && entry.Gen != handle.Gen) {
+                // The slot was reclaimed and handed to a new object: the twin at it describes
+                // driver ids the new state object never made.
+                entry.backend.reset();
+            }
+            entry.Gen = handle.Gen;
+            entry.Live = true;
+            entry.stateRef = stateObj;
+            // No creation tick and no sweep here. The registry this replaces needed both,
+            // because nothing told it a texture or a renderbuffer had been DELETED and object
+            // CHURN rather than draw count is what made that urgent. Every one of the six kinds
+            // now announces its own death from its destructor, so a dead twin's slot is already
+            // back before the next creation asks for one.
+            RememberHandle(stateObj->GetLifetimeId(), handle);
+            return entry.backend;
+        }
+
+        // P3a: resolve-or-create BY HANDLE, and it is the shape that discharges the debt this
+        // header records against itself at the top of the file.
+        //
+        // The overload above mints - it calls MGPipeSlots().Acquire off a frontend object's
+        // lifetime id, from inside MG_Backend - which is monolith glue: a handle is minted by
+        // the CLIENT, and under a real split neither the object nor its lifetime id exists on
+        // this side. This overload never touches the allocator at all. The handle ARRIVED, in
+        // the call's payload, already minted by the side that owns minting; all this does is
+        // index the slot, notice a generation that no longer matches (the slot was recycled,
+        // so the twin at it describes driver ids the new resource never made) and hand back
+        // the twin pointer. FindByHandle beside it is the same shape and already existed.
+        //
+        // No StatePtr, therefore no Entry::stateRef: the weak pointer is liveness for
+        // ForEachLive() and a handle-keyed entry has no frontend object to weakly hold. Such
+        // an entry is therefore invisible to ForEachLive, which is correct - the one direct
+        // iteration site walks texture twins, and it is not one of these tables.
+        //
+        // Death stays ANNOUNCED, as it is on the other overload: for a handle-keyed kind the
+        // announcement is the family's own destroy call, not the shared death notice, and the
+        // slot is freed by the CLIENT after that call returns.
+        //
+        // UNUSED AT THE CONTRACT COMMIT, deliberately: it is a member of a class template, so
+        // an uninstantiated one costs nothing anywhere, and the backend package is what gives
+        // it its first caller.
+        BackendPtr& GetOrCreate(MG_Pipe::MGPipeHandle handle) {
+            MOBILEGL_ASSERT(!MG_Pipe::MGPipeHandleIsNull(handle),
+                            "GetOrCreate(handle) named the reserved null handle");
+            if (MG_Pipe::MGPipeHandleIsNull(handle)) return m_nullTwin;
+
+            // A slot index that ARRIVED in a payload indexes a vector this call would RESIZE,
+            // and nothing between the payload and here bounds it: the applier's blob gates sit
+            // in front of the vertex-input family, not in front of the resource family, which
+            // dispatches ops->Create(record.Res, ...) straight through. There is no allocator
+            // constant to check against on this side - the allocator is the client's - so this
+            // is a sanity cap and is documented as one: kMaxHandleSlot entries of one kind is
+            // already orders of magnitude past any real GL object count, while a corrupt 32-bit
+            // slot asks for a four-billion-entry resize.
+            if (handle.Slot >= kMaxHandleSlot) {
+                MOBILEGL_ASSERT(false, "GetOrCreate(handle) named slot %u, past this table's %u bound",
+                                handle.Slot, kMaxHandleSlot);
+                return m_nullTwin;
+            }
+
+            // Same arming as the minting overload, and for the same reason: twin creation is
+            // the moment a driver-owned id starts needing a guarded destructor.
+            EnsureProcessTeardownSentinel();
+
+            // THE TWO DIRECTIONS ARE NOT SYMMETRIC HERE, where they are on the minting overload.
+            // There the handle comes straight out of MGPipeSlots().Acquire and can never be
+            // BEHIND the entry, so a bare `!=` only ever means "the slot was recycled forward".
+            // Here the handle arrived in a payload, so `handle.Gen < entry.Gen` is a reachable
+            // input, and adopting it would destroy the INCUMBENT LIVE twin - a driver buffer id,
+            // a persistent map, a pooled store, released by a defaulted destructor that issues
+            // no glDeleteBuffers and no pool enrolment - and then stamp the slot back to the
+            // dead resource's generation, after which the incumbent's own FindByHandle refuses
+            // it and it is silently handed a fresh, empty twin. That is a leak AND a resource
+            // that loses its storage with no diagnostic, i.e. the shape commit d7655247 fixed
+            // and the thing MGPipeHandle::Gen exists to prevent. So: forward is a recycle and
+            // resets the twin, BACKWARD is refused - which is the same answer FindByHandle
+            // below already gives the same input.
+            Entry& entry = EntryAt(handle.Slot);
+            if (entry.Live && entry.Gen > handle.Gen) {
+                MOBILEGL_ASSERT(false,
+                                "GetOrCreate(handle) named generation %u at slot %u, which is BEHIND "
+                                "the live entry's %u - refusing rather than destroying the incumbent",
+                                handle.Gen, handle.Slot, entry.Gen);
+                return m_nullTwin;
+            }
+            if (entry.Live && entry.Gen != handle.Gen) entry.backend.reset();
+            entry.Gen = handle.Gen;
+            entry.Live = true;
+            return entry.backend;
+        }
+
+        // The generation of the LIVE entry at this slot, or 0 when the slot is out of range or
+        // holds no live entry. It exists so a caller can DIAGNOSE - in a release build, where
+        // MOBILEGL_ASSERT is inert - the refusal GetOrCreate(handle) above performs silently.
+        Uint32 LiveGenAt(Uint32 slot) const {
+            if (slot >= m_slots.size()) return 0;
+            const Entry& entry = m_slots[slot];
+            return entry.Live ? entry.Gen : 0;
+        }
+
+        // P3a: the death half of the overload above, for a kind whose announcement is its own
+        // destroy CALL rather than the shared death notice (D-L). Hands the twin OUT rather
+        // than destroying it in place, because the caller may still have to decide what
+        // happens to the driver id it owns - Espryt pools it, deletes it, or parks it on the
+        // deferred-release list when no context is current on this thread - and every one of
+        // those outcomes has to be reached with the entry already retired, so a re-entrant
+        // GetOrCreate from a twin destructor cannot resurrect it.
+        //
+        // The slot itself is NOT freed here: it belongs to the kind, and for a handle-keyed
+        // kind the CLIENT frees it after the destroy call returns (SlotAllocator.h:60 - the
+        // Gen bump rides the next handout, so a double free cannot skip a generation). An
+        // entry whose Gen no longer matches is a twin of the slot's previous owner and is
+        // left alone: the successor's own GetOrCreate resets it.
+        BackendPtr ReleaseByHandle(MG_Pipe::MGPipeHandle handle) {
+            if (MG_Pipe::MGPipeHandleIsNull(handle)) return BackendPtr{};
+            if (m_memoHandle.Slot == handle.Slot) ForgetHandle();
+            if (handle.Slot >= m_slots.size()) return BackendPtr{};
+            Entry& entry = m_slots[handle.Slot];
+            if (!entry.Live || entry.Gen != handle.Gen) return BackendPtr{};
+            BackendPtr dead = std::move(entry.backend);
+            entry.backend.reset();
+            entry.stateRef.reset();
+            entry.Live = false;
+            return dead;
+        }
+
+        // Null when no live twin of this object exists. Unlike the registry's Find this NEVER
+        // mutates the table, so the returned pointer survives any later Find on it; only a
+        // GetOrCreate that grows the vector can move it, and callers that hold one across a
+        // possible insertion still copy the BackendPtr out.
+        BackendPtr* Find(StateObject* stateObj) {
+            if (stateObj == nullptr) return nullptr;
+            return FindByHandle(HandleOf(stateObj));
+        }
+
+        const BackendPtr* Find(StateObject* stateObj) const {
+            return const_cast<BackendSlotTable*>(this)->Find(stateObj);
+        }
+
+        BackendPtr* FindByHandle(MG_Pipe::MGPipeHandle handle) {
+            if (MG_Pipe::MGPipeHandleIsNull(handle)) return nullptr;
+            if (handle.Slot >= m_slots.size()) return nullptr;
+            Entry& entry = m_slots[handle.Slot];
+            if (!entry.Live || entry.Gen != handle.Gen) return nullptr;
+            return &entry.backend;
+        }
+
+        // The handle this object's twin is keyed on, or the null handle. This is what a backend
+        // memo stores instead of a raw pointer, a GL name or a bare lifetime id.
+        //
+        // A NULL answer is never memoised. The memo is per table and the allocator is per
+        // kind, so with two holders of one kind the OTHER table can be the one that acquires;
+        // a cached "no handle" here would then outlive the twin's creation over there, and
+        // nothing on this table's own acquire path would ever refresh it. A miss costs the
+        // allocator probe it always cost; a hit is refreshed the moment anyone acquires.
+        MG_Pipe::MGPipeHandle HandleOf(const StateObject* stateObj) const {
+            if (stateObj == nullptr) return MG_Pipe::kMGPipeNullHandle;
+            const Uint64 lifetimeId = stateObj->GetLifetimeId();
+            if (lifetimeId == m_memoLifetimeId) return m_memoHandle;
+            const MG_Pipe::MGPipeHandle handle =
+                MG_Pipe::MGPipeSlots().FindByLifetimeId(kKind, lifetimeId);
+            if (!MG_Pipe::MGPipeHandleIsNull(handle)) RememberHandle(lifetimeId, handle);
+            return handle;
+        }
+
+        // P2 step e2's backend half. The frontend object with this lifetime id has just been
+        // DESTROYED: resolve its handle ONCE, drop its twin in EVERY table of this type, and
+        // return the slot to the allocator - in that order, because the allocator forgets the
+        // lifetime id on Free and a holder told second could no longer resolve it.
+        //
+        // The slot is returned whether or not any holder still had a twin at it: the lifetime
+        // id is dead and MG_State never hands one out twice, so nothing can acquire it again,
+        // and a slot minted for it that no table holds (a table reset with `= {}` drops its
+        // entries without freeing) would otherwise stay allocated for the life of the process.
+        //
+        // STATIC, and deliberately so: a notice is about an object, not about a table, and
+        // "which table holds it" is exactly the question that produced the two-holder leak.
+        // Returns whether the object had a slot of this kind, i.e. whether anything was freed;
+        // a second call for the same id answers false because the allocator no longer maps it.
+        static Bool OnFrontendObjectDestroyed(Uint64 lifetimeId) {
+            const MG_Pipe::MGPipeHandle handle =
+                MG_Pipe::MGPipeSlots().FindByLifetimeId(kKind, lifetimeId);
+            if (MG_Pipe::MGPipeHandleIsNull(handle)) return false;
+            for (BackendSlotTable* holder = s_firstHolder; holder != nullptr;) {
+                // The successor is read BEFORE the release: ReleaseTwinAt runs the twin's
+                // destructor, which is a driver call, and nothing that outlives it may be a
+                // reference into this holder.
+                BackendSlotTable* const next = holder->m_nextHolder;
+                holder->ReleaseTwinAt(handle);
+                holder = next;
+            }
+            MG_Pipe::MGPipeSlots().Free(kKind, handle);
+            return true;
+        }
+
+        // How many tables of this type exist right now. For the tests that pin the holder
+        // list; nothing on a shipping path asks.
+        static Uint32 HolderCount() {
+            Uint32 count = 0;
+            for (const BackendSlotTable* holder = s_firstHolder; holder != nullptr;
+                 holder = holder->m_nextHolder) {
+                ++count;
+            }
+            return count;
+        }
+
+        // fn(const StatePtr& state, const BackendPtr& twin) over every live, still-owned entry.
+        // Replaces the registry's begin()/end(), whose iterator exposed the raw frontend
+        // address as the map key - the one place the backend read an identity it must not have.
+        // The state object is handed over as a STRONG reference, so the callee cannot be handed
+        // a dangling key the way the old iteration could.
+        template <typename Fn>
+        void ForEachLive(Fn&& fn) const {
+            // Index loop and a COPIED twin, not a range-for over references: fn is arbitrary
+            // backend code, and a nested GetOrCreate on this table would resize m_slots and
+            // invalidate both the iterator and any reference into the vector that outlives the
+            // call. The one caller today happens not to insert; that is not a property the
+            // walk should depend on.
+            for (SizeT slot = 0; slot < m_slots.size(); ++slot) {
+                const Entry& entry = m_slots[slot];
+                if (!entry.Live || !entry.backend) continue;
+                const StatePtr state = entry.stateRef.lock();
+                if (!state) continue;
+                const BackendPtr twin = entry.backend;
+                fn(state, twin);
+            }
+        }
+
+        Uint32 LiveCount() const {
+            Uint32 count = 0;
+            for (const Entry& entry : m_slots) {
+                if (entry.Live) ++count;
+            }
+            return count;
+        }
+
+    private:
+        // Drop the twin at `handle` if THIS table holds it. Frees nothing: the slot belongs to
+        // the kind, not to the table, and OnFrontendObjectDestroyed returns it once, after
+        // every holder has let go.
+        Bool ReleaseTwinAt(MG_Pipe::MGPipeHandle handle) {
+            // Forget the memo whenever it names this slot, even if this table has no entry
+            // there: a memo can be a handle learned from the allocator for an object another
+            // holder twinned, and it must not survive the slot's next handout.
+            if (m_memoHandle.Slot == handle.Slot) ForgetHandle();
+            if (handle.Slot >= m_slots.size()) return false;
+            // The twin's destructor is a driver call and could, in principle, re-enter
+            // GetOrCreate on this table and resize m_slots. So NOTHING that outlives the
+            // destructor may be a reference into m_slots: the twin is moved out into a local,
+            // the entry is finished with, and only then is the local released.
+            BackendPtr dead;
+            {
+                Entry& entry = m_slots[handle.Slot];
+                if (!entry.Live || entry.Gen != handle.Gen) return false;
+                dead = std::move(entry.backend);
+                entry.backend.reset();
+                entry.stateRef.reset();
+                entry.Live = false;
+            }
+            dead.reset();
+            return true;
+        }
+
+        // Grows the table to hold `slot`. Every caller bounds `slot` first - the minting
+        // overload because the allocator produced it, the handle overload against
+        // kMaxHandleSlot - because this is the one place a client-supplied number decides an
+        // allocation size.
+        Entry& EntryAt(Uint32 slot) {
+            if (slot >= m_slots.size()) m_slots.resize(static_cast<SizeT>(slot) + 1);
+            return m_slots[slot];
+        }
+
+        void RememberHandle(Uint64 lifetimeId, MG_Pipe::MGPipeHandle handle) const {
+            m_memoLifetimeId = lifetimeId;
+            m_memoHandle = handle;
+        }
+        void ForgetHandle() const {
+            m_memoLifetimeId = 0;
+            m_memoHandle = MG_Pipe::kMGPipeNullHandle;
+        }
+
+        // The holder list: intrusive and doubly linked, so registering and unregistering are
+        // two pointer writes with no allocation, and its head is a constant-initialised
+        // static - which is what lets the process-lifetime registry globals in Managers.cpp
+        // link themselves in from their own constructors with no initialisation-order
+        // question to answer. Single-threaded, like every table it links (the tables live and
+        // die on the context thread, as the notice they answer does).
+        void LinkHolder() {
+            m_prevHolder = nullptr;
+            m_nextHolder = s_firstHolder;
+            if (s_firstHolder != nullptr) s_firstHolder->m_prevHolder = this;
+            s_firstHolder = this;
+        }
+        void UnlinkHolder() {
+            if (m_prevHolder != nullptr) {
+                m_prevHolder->m_nextHolder = m_nextHolder;
+            } else {
+                s_firstHolder = m_nextHolder;
+            }
+            if (m_nextHolder != nullptr) m_nextHolder->m_prevHolder = m_prevHolder;
+            m_prevHolder = nullptr;
+            m_nextHolder = nullptr;
+        }
+
+        static inline BackendSlotTable* s_firstHolder = nullptr;
+        BackendSlotTable* m_prevHolder = nullptr;
+        BackendSlotTable* m_nextHolder = nullptr;
+
+        // Indexed by MGPipeHandle::Slot; [0] is the reserved slot and is never live.
+        Vector<Entry> m_slots;
+        // Handed back by GetOrCreate for a null state object. Never live, never handed a handle.
+        BackendPtr m_nullTwin;
+
+        // ONE-entry resolution memo, lifetimeId -> handle. It exists because without it every
+        // resolution goes through the allocator's ByLifetimeId hash, which the deleted
+        // TwinLookupMemos existed to avoid and which D13 promises to replace with "direct slot
+        // indexing".
+        //
+        // It is one entry and therefore only helps a caller that asks for the SAME object twice
+        // running - ResolveVaoTwin and SyncCurrentProgram do, once per draw each. Two callers
+        // it does NOT help, recorded rather than claimed away: BindCurrentFBO resolves BOTH
+        // targets in a frame, and ResolveUnitSamplerBackend asks for a different sampler per
+        // texture unit, so both thrash a single-entry memo and pay the probe P1 did not (P1 had
+        // a per-unit memo and a direct-mapped 6-slot array there). Making the memo per-unit /
+        // per-target is the fix, and G11 - the device-side gate that would price it - is owed.
+        //
+        // It cannot serve a stale answer, by three independent arguments:
+        //   * the key is a lifetime id, which MG_State never hands out twice, so a recycled
+        //     heap address cannot hit this memo the way it could hit an address-keyed one;
+        //   * a null answer is never stored, so another holder's acquire cannot be hidden by
+        //     a "no handle" this table remembered earlier; and
+        //   * even a hit for a slot that has since been freed and re-handed is caught, because
+        //     the caller resolves the handle through FindByHandle, which compares Gen.
+        // Cleared anyway when a death notice names the memoised slot. 0 is never a live
+        // lifetime id (MG_State's counters start at 1), so a zeroed memo is a guaranteed miss.
+        mutable Uint64 m_memoLifetimeId = 0;
+        mutable MG_Pipe::MGPipeHandle m_memoHandle = MG_Pipe::kMGPipeNullHandle;
+    };
+
+#endif // MOBILEGL_PIPE_PUSH
+} // namespace MobileGL::MG_Backend::DirectGLES

MobileGL/MG_Backend/DirectVulkan/Renderer/MagmaPipeArms.h

@@ -0,0 +1,603 @@
+// MobileGL - MobileGL/MG_Backend/DirectVulkan/Renderer/MagmaPipeArms.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>
+
+#include <Config.h>
+#if MOBILEGL_PIPE_PUSH
+// kMGPipeSubsystem* - the runtime bitmask's named bits - and MGPipeHandle itself. Both are
+// header-only constant/POD declarations, and both are push-only, so the pull build's include
+// graph is unchanged (G1).
+#include <MG_Pipe/MGPipe.h>
+#include <MG_Pipe/MGPipeHandles.h>
+#endif
+
+#include <cstdlib>
+
+// Magma's arm selector for the P2 Track H / render-state re-keys (P2 brief D14), and the
+// {slot, gen} mint the re-keyed sites are written against.
+//
+// Two switches decide which arm a re-keyed site runs, and they are NOT the same switch:
+//
+//   MOBILEGL_PIPE_PUSH (compile)          - is the pushed state there to be keyed on at all
+//   Features.PipePush  (runtime bitmask)  - is THIS subsystem migrated in THIS run
+//   MOBILEGL_PIPE_LEGACY_MEMOS (compile)  - is the pre-handle arm compiled beside it
+//   Features.PipeLegacyMemos (runtime)    - may the pre-handle arm be ENTERED in this run
+//
+// ARCHITECTURE.md 9.6's point: once a handle wave lands, a clear MOBILEGL_PIPE_PUSH bit is
+// only a valid A/B while the legacy arm is still compiled, because with the bit clear the
+// backend would otherwise still run the re-keyed code. So a clear bit selects the legacy
+// arm, and a run that has explicitly disabled the legacy arm may not fall into it.
+//
+// D14 spends that last sentence at STARTUP, not per draw: "a Track-H subsystem whose bit is
+// clear is a startup Fatal{PipeLegacyMemosDisabled}". Nothing in the draw path aborts, and
+// nothing outside Track H consults the legacy-memo lever at all - see
+// MagmaPipeValidateSubsystemConfiguration below for both halves of that rule.
+//
+// The whole header is inert in a pull build: MOBILEGL_PIPE_PUSH is 0 there, every helper
+// below is behind it, and the pull build's translation units are byte-identical (G1).
+namespace MobileGL::MG_Backend::DirectVulkan {
+
+#if MOBILEGL_PIPE_PUSH
+    // Is `subsystemBit` (MG_Pipe/MGPipe.h's kMGPipeSubsystem*) migrated in this run?
+    inline Bool MagmaPipeSubsystemOn(Uint64 subsystemBit) {
+        return (MG_Config::Features.PipePush & subsystemBit) != 0;
+    }
+
+    // ---------------------------------------------------------------------------------
+    // D14's startup gate
+    // ---------------------------------------------------------------------------------
+    //
+    // Called once from VulkanRenderer::Initialize(), i.e. only when Magma is the backend
+    // that is actually running. It answers exactly one question and it answers it before the
+    // first draw: is there an arm for Magma's Track-H subsystem in this configuration?
+    //
+    // Three deliberate boundaries, each of which the per-draw shape this replaces got wrong:
+    //
+    //  * ONLY Magma's own Track-H bit is checked. Espryt's bit 5 is Espryt's business (a
+    //    DirectVulkan run does not execute one line of DirectGLES' re-key), so
+    //    MOBILEGL_PIPE_PUSH=0x20 must not kill a Magma run, and MOBILEGL_PIPE_PUSH=0x40 must
+    //    not kill an Espryt one.
+    //  * bit 0 (kMGPipeSubsystemRenderState) is NOT Track H and is NOT fatal. It is not a
+    //    memo re-key at all: it decides where the pipeline memo's STATE KEY comes from, and
+    //    a clear bit there simply means the client is not pushing render-state CSOs in this
+    //    run, which GetOrCreatePipeline answers with its own state hash. D14 labels bits 5
+    //    and 6 "Track H" and labels bit 0 nothing of the sort.
+    //  * it is Fatal at STARTUP, once, not on a draw. A per-draw abort inside
+    //    GetOrCreatePipeline turns a configuration mistake into a mid-frame crash and puts a
+    //    branch nobody needs on the hottest path in the backend.
+    //
+    // [declared deviation from D14, review v2 minor 2] D14's runtime row reads "false: the
+    // legacy arm is never entered", and D14's compile-switch row names ComputePipelineStateHash
+    // as part of the pre-handle arm. Those two together would make MOBILEGL_PIPE_LEGACY_MEMOS=0
+    // with bit 0 CLEAR a contradiction: the pipeline memo has no CSO handle to key on, so it
+    // keys on a state hash, and in a build that compiles the pre-handle arm that hash IS
+    // ComputePipelineStateHash. Magma does not make that fatal - bit 0 is not Track H, and
+    // there is a correct answer (the state hash) where for bits 5/6 there is none - but it no
+    // longer does it SILENTLY: the combination is named once, at startup, right here.
+    inline void MagmaPipeValidateSubsystemConfiguration() {
+        if (!MG_Config::Features.PipeLegacyMemos &&
+            !MagmaPipeSubsystemOn(MG_Pipe::kMGPipeSubsystemRenderState)) {
+            MGLOG_W("MGPipe: MOBILEGL_PIPE_LEGACY_MEMOS=0 with kMGPipeSubsystemRenderState (bit 0 "
+                    "of MOBILEGL_PIPE_PUSH) clear - Magma's pipeline memo has no CSO handle to key "
+                    "on, so every draw whose pipeline-state version moved runs the pre-handle STATE "
+                    "HASH instead. That is not a Track-H subsystem and not fatal, but it is not the "
+                    "handle arm either: set bit 0 (MOBILEGL_PIPE_PUSH=0x%llx) if this run was meant "
+                    "to measure it.",
+                    static_cast<unsigned long long>(MG_Config::Features.PipePush |
+                                                    MG_Pipe::kMGPipeSubsystemRenderState));
+        }
+#if MOBILEGL_PIPE_LEGACY_MEMOS
+        // The pre-handle arm is compiled AND the operator has not forbidden entering it, so a
+        // clear bit is an ordinary, valid A/B: the site takes the legacy arm.
+        if (MG_Config::Features.PipeLegacyMemos) return;
+#endif
+        if (MagmaPipeSubsystemOn(MG_Pipe::kMGPipeSubsystemMagmaVertexInput)) return;
+#if MOBILEGL_PIPE_LEGACY_MEMOS
+        const char* const why = "this run has MOBILEGL_PIPE_LEGACY_MEMOS=0";
+#else
+        const char* const why =
+            "this build has cmake -DMOBILEGL_PIPE_LEGACY_MEMOS=OFF, which compiles no such arm";
+#endif
+        MGLOG_F("MGPipe: Fatal{PipeLegacyMemosDisabled} Magma's Track-H subsystem "
+                "(kMGPipeSubsystemMagmaVertexInput, bit 6 of MOBILEGL_PIPE_PUSH) is clear, so the "
+                "vertex-input cache and the VAO draw memo want the pre-handle arm - but %s. Set "
+                "bit 6 (MOBILEGL_PIPE_PUSH=0x%llx, or the default 0x%llx), or allow the legacy arm.",
+                why,
+                static_cast<unsigned long long>(MG_Config::Features.PipePush |
+                                                MG_Pipe::kMGPipeSubsystemMagmaVertexInput),
+                static_cast<unsigned long long>(MG_Pipe::kMGPipeSubsystemsMigratedAtP2));
+        std::abort();
+    }
+
+    // "Does this Track-H site run the handle arm?" - the ONE question every re-keyed Track-H
+    // site asks, so that they cannot disagree with each other or with the startup gate.
+    inline Bool MagmaPipeTrackHArmIsHandles(Uint64 trackHBit) {
+#if MOBILEGL_PIPE_LEGACY_MEMOS
+        return MagmaPipeSubsystemOn(trackHBit);
+#else
+        // No pre-handle arm exists in this build, and MagmaPipeValidateSubsystemConfiguration
+        // has already made a clear bit a startup Fatal, so the handle arm is the only arm a
+        // running process can be on.
+        (void)trackHBit;
+        return true;
+#endif
+    }
+
+    // ---------------------------------------------------------------------------------
+    // Negative control C (P2 brief D18): MOBILEGL_PIPE_HANDLE_ABA_CONTROL
+    // ---------------------------------------------------------------------------------
+    //
+    // "Is the object-identity half of every vertex-input memo key deliberately defeated in
+    // this run?" - the ONE question the control's sites ask, for the same reason
+    // MagmaPipeTrackHArmIsHandles exists: three sites deciding separately could disagree,
+    // and a control that defeats two of three guards proves nothing.
+    //
+    // WHAT IT DEFEATS, AND WHY IT IS SPELLED AS "REPLACE THE IDENTITY WITH A CONSTANT"
+    // RATHER THAN "USE THE HEAP ADDRESS".
+    //
+    // D18 wrote the control as "hash attr.Buffer.get() instead of GetLifetimeId(), and skip
+    // the vaoLifetimeId compare", on the theory that a deleted object's replacement lands at
+    // the freed heap block and so reproduces the key. Measured, it does not: in
+    // HandleRecycleScenario the GL NAMES come back (glGen* hands the deleted name straight
+    // out) but the C++ heap blocks do not - a VertexArrayObject is 3920 bytes, too large for
+    // glibc's tcache, so its chunk goes to the unsorted bin and is split by the very next
+    // allocation the replacement path makes. Four create/delete cycles in one run produced
+    // four distinct addresses, ~1 MiB apart. With no address reuse there is nothing for
+    // "hash the address" to collide with: the replacement hashes differently, indexes a
+    // different memo slot, and inherits nothing - so the arm asserted stale pixels and saw
+    // fresh ones, which is a FAILING negative control that had stopped controlling anything.
+    //
+    // So the control no longer asks the allocator for the collision; it manufactures it. On
+    // both arms the object identity is replaced by a constant, which is the strongest form of
+    // "the allocator handed the block back" and is deterministic. That covers strictly more
+    // than D18's spelling, and in particular it reaches the arm P2 SHIPS: on the handle arm
+    // the constant defeats the OBJECT IDENTITY THAT SELECTS THE SLOT - the key the handle arm
+    // ships - so the replacement VAO is handed the dead one's memo entry and its content hash.
+    // Defeating only the retired lifetime-id/address guards would leave that key untested,
+    // which is exactly the vacuity this control exists to catch.
+    //
+    // WHAT IT DOES NOT COVER, AND WHY NO REPRODUCER OF THIS SHAPE CAN [fix-aba review v1,
+    // MAJOR 1]. It does NOT exercise the GENERATION half of {slot, gen}:
+    //
+    //   * this mint has no death notification - nothing in MG_Backend/DirectVulkan consumes
+    //     NotifyStateObjectDestroyed - so a slot returns to the free list only through
+    //     OnFrameBoundary's age sweep (kSweepInterval 256, kRetireAgeBoundaries 1024, below);
+    //   * HandleRecycleScenario issues five frame boundaries, so the free list is empty when
+    //     the replacement VAO acquires and it gets a BRAND-NEW slot at Gen 1 (measured:
+    //     redVao slot=2 gen=1, greenVao slot=3 gen=1). The knob-off FRESH verdict there is
+    //     decided by the SLOT alone, and deleting the ++Gen below leaves all four arms green;
+    //   * a genuine slot REUSE needs >= 1024 idle boundaries after the dead object's last
+    //     draw, which necessarily puts the two draws in different frames - and the only memo
+    //     that carries a GPU slice rather than a layout, ResolvedVertexBindings, declines
+    //     across frames by design. The two requirements are mutually exclusive, so the
+    //     generation is out of reach of any same-frame pixel reproducer for this memo.
+    //
+    // The generation is covered where it IS expressible, over this mint and the claim rule
+    // MagmaPipeClaimSlotMemos below: MG_Test/Pipe/MagmaPipeIdentityTest.cpp drives a real
+    // retire -> reuse and asserts that a memo stamped at {slot, gen=N} is not served at
+    // {slot, gen=N+1} with the knob off and IS served with it on. Deleting the ++Gen reds that
+    // suite; it is the only place in the tree where that deletion is caught.
+    //
+    // Everything the control does NOT defeat is as load-bearing as what it does. It never
+    // touches a guard that is not an IDENTITY guard: the resolved-bindings memo's frame
+    // serial, its slice-epoch compares and its host-map check all stay in force, so a green
+    // AbaControl arm still means "a replacement object was handed its dead predecessor's
+    // resolved vertex bindings because the identity halves of the keys were defeated", not
+    // "every safety net was switched off until something broke".
+    //
+    // Off by default (Config.h), set only by the HandleRecycle AbaControl ctest lanes, and
+    // #if MOBILEGL_PIPE_PUSH throughout, so no shipping pull build can even parse it.
+    // P4a (BRIEF-P4A.md D-I2, G8): WHICH KINDS THIS ANSWER COVERS, and it is not "all of them".
+    //
+    // P4a mints six more client-side kinds - Texture, Renderbuffer, Framebuffer, SamplerCso,
+    // SamplerViewCso and ShaderCso - and requires the ABA control to defeat "the identity half
+    // of P4a's memo keys as well", because a control that only defeats the guards a phase
+    // RETIRED says nothing about the key that phase SHIPS.
+    //
+    // On Magma there is no such key to defeat, and that is a fact about the roadmap rather than
+    // an omission here. MagmaPipeIdentityTables below mints exactly TWO kinds,
+    // VertexElementsCso and Buffer; a texture, a framebuffer, a sampler, a view and a program
+    // are all still reached from their frontend objects on this backend, and moving them onto
+    // handles is P7's work (ROADMAP.md:24 - "Magma anything"; P4a leaves MG_Backend/DirectVulkan
+    // untouched apart from this file). So the honest statement is per KIND, and it is spelled as
+    // code rather than as a comment so that a caller cannot read the blanket answer above and
+    // conclude the knob covers its kind:
+    //
+    //   * for the two kinds this backend really keys on {slot, gen}, the knob defeats the
+    //     identity exactly as it always has (MagmaPipeClaimSlotMemos);
+    //   * for P4a's six there is nothing here to defeat, so the answer is FALSE - and
+    //     MG_IntegrationTest's HandleRecycleScenario reads that through its own build probe and
+    //     makes those cases' AbaControl arm assert the CORRECT pixels while SAYING that it is
+    //     not controlling anything for that kind. It does not assert a corruption that no code
+    //     on this tree can produce, which would be a permanently red always-on lane.
+    //
+    // WHAT MAKES IT TRUE LATER, in one sentence, so the next reader does not have to derive it:
+    // when a backend grows a Features.PipeHandleAbaControl consumer over its P4a object slot
+    // tables - one `if` in GetOrCreate / FindByHandle, the shape MagmaPipeClaimSlotMemos already
+    // has for vertex input - this function's per-kind answer becomes that consumer's, the
+    // integration probe finds the consumer, and the six cases flip to expecting the corruption.
+    inline Bool MagmaPipeAbaControlDefeatsIdentity() {
+        return MG_Config::Features.PipeHandleAbaControl;
+    }
+
+    // WHICH KINDS THIS BACKEND ACTUALLY KEYS ON {slot, gen}, and therefore which kinds the knob
+    // above has an identity to defeat at all. `kind` is MG_Pipe::MGPipeKind.
+    //
+    // EXHAUSTIVE, WITH NO `default:`, for MG_IntegrationTest/Harness/PipeSlotPeek.cpp's reason:
+    // a kind added to MGPipeKind without a decision here must be a -Wswitch warning in this
+    // file rather than a row that silently inherits somebody else's answer. Being wrong in the
+    // "covered" direction is the expensive one - a control asserting a corruption nobody can
+    // produce is a permanently red always-on lane - so an undecided kind must never read true,
+    // and with no `default:` there is no arm for it to read true from.
+    //
+    // constexpr AND PINNED BY static_assert BELOW, which is what stops it rotting the way a
+    // predicate with no caller does: MagmaPipeIdentityTables mints exactly two kinds, the
+    // asserts say so in both directions, and the file no longer compiles if the tables and this
+    // statement of them ever part company. (Review F-m5: the earlier form had no caller at all
+    // and could not make anything red or green.)
+    inline constexpr Bool MagmaPipeAbaControlKindIsRekeyedHere(MG_Pipe::MGPipeKind kind) {
+        switch (kind) {
+            // The two MagmaPipeIdentityTables really mints.
+            case MG_Pipe::MGPipeKind::VertexElementsCso:
+            case MG_Pipe::MGPipeKind::Buffer:
+                return true;
+            // P4a's six object classes: still reached from their frontend objects on this
+            // backend (Magma's object paths are P7, ROADMAP.md:24), so there is no key here for
+            // the knob to defeat.
+            case MG_Pipe::MGPipeKind::Texture:
+            case MG_Pipe::MGPipeKind::Renderbuffer:
+            case MG_Pipe::MGPipeKind::Framebuffer:
+            case MG_Pipe::MGPipeKind::SamplerCso:
+            case MG_Pipe::MGPipeKind::SamplerViewCso:
+            case MG_Pipe::MGPipeKind::ShaderCso:
+            // ...and everything else this backend does not mint a handle for.
+            case MG_Pipe::MGPipeKind::None:
+            case MG_Pipe::MGPipeKind::Xfb:
+            case MG_Pipe::MGPipeKind::RenderStateCso:
+            case MG_Pipe::MGPipeKind::Fence:
+            case MG_Pipe::MGPipeKind::Query:
+            case MG_Pipe::MGPipeKind::Context:
+            case MG_Pipe::MGPipeKind::KindCount:
+                return false;
+        }
+        return false;
+    }
+
+    static_assert(MagmaPipeAbaControlKindIsRekeyedHere(MG_Pipe::MGPipeKind::VertexElementsCso),
+                  "MagmaPipeIdentityTables mints VertexElementsCso: the knob has an identity to "
+                  "defeat for it");
+    static_assert(MagmaPipeAbaControlKindIsRekeyedHere(MG_Pipe::MGPipeKind::Buffer),
+                  "MagmaPipeIdentityTables mints Buffer: the knob has an identity to defeat for it");
+    static_assert(!MagmaPipeAbaControlKindIsRekeyedHere(MG_Pipe::MGPipeKind::Texture) &&
+                      !MagmaPipeAbaControlKindIsRekeyedHere(MG_Pipe::MGPipeKind::Renderbuffer) &&
+                      !MagmaPipeAbaControlKindIsRekeyedHere(MG_Pipe::MGPipeKind::Framebuffer) &&
+                      !MagmaPipeAbaControlKindIsRekeyedHere(MG_Pipe::MGPipeKind::SamplerCso) &&
+                      !MagmaPipeAbaControlKindIsRekeyedHere(MG_Pipe::MGPipeKind::SamplerViewCso) &&
+                      !MagmaPipeAbaControlKindIsRekeyedHere(MG_Pipe::MGPipeKind::ShaderCso),
+                  "P4a's six object classes are not keyed on {slot, gen} on this backend, so "
+                  "HandleRecycleScenario's six AbaControl arms must NOT expect a corruption here. "
+                  "Wiring one of them is what flips this assert, this predicate and that arm - and "
+                  "MG_IntegrationTest's two-symbol probe over MG_Backend/DirectVulkan is what "
+                  "carries the answer into the lane");
+
+    // THERE IS DELIBERATELY NO PER-KIND WRAPPER HERE, and review F-v2-m3 is why. An earlier
+    // round carried `MagmaPipeAbaControlCoversKind(kind)` - the conjunction of the two
+    // statements above - and it had no caller anywhere in the tree: the knob's only two
+    // consumers (VulkanRenderer.cpp's VAO draw memo and VertexInputStateFactory.cpp's pipeline
+    // key) each hold ONE kind, VertexElementsCso, by construction, so the kind is not a
+    // variable at either site. A conjunction no build ever evaluates cannot be pinned the way
+    // the predicate above is pinned - it is not constexpr, because it reads MG_Config::Features,
+    // so no static_assert can reach it - which makes it exactly the rot F-m5 was raised about,
+    // one level up: an `&&` whose operands could be inverted or dropped with nothing to say so.
+    //
+    // The two pieces stand alone instead, and each is pinned by something that runs:
+    // MagmaPipeAbaControlKindIsRekeyedHere is constexpr and asserted in BOTH directions by the
+    // three static_asserts above, which compile in every Magma build; MagmaPipeAbaControlDefeats
+    // Identity is the knob, and its two consumers are what make it true or false. A call site
+    // that ever does hold a variable kind writes the `&&` there, where a build will run it.
+
+    // The single consumer-table entry every VAO collapses onto while the control is on. Slot
+    // 0 is a real, ordinary entry of both tables (MagmaPipeSlotIndex maps the first allocatable
+    // handle onto it), so nothing about the tables changes shape for the control's sake.
+    inline constexpr Uint32 kMagmaPipeAbaControlSlotIndex = 0;
+
+    // ---------------------------------------------------------------------------------
+    // The {slot, gen} mint
+    // ---------------------------------------------------------------------------------
+    //
+    // Maps a frontend object's never-reused lifetime id to a dense {slot, gen}. Three
+    // properties, and the third is the one review v2 got wrong:
+    //
+    //   1. exact identity - Gen moves whenever a slot changes owner, so a stale handle can
+    //      never match a live object even if the allocator hands back the same heap address
+    //      (the ABA HandleRecycleScenario reproduces);
+    //   2. dense slots - the slot IS an index, so a consumer's per-slot table needs no hash,
+    //      no probe and no mix;
+    //   3. NO CAPACITY CLIFF. A live object's handle never changes while the object is being
+    //      drawn, whatever the working set size.
+    //
+    // Property 3 is why this is not the fixed 2-way set-associative LRU the previous round
+    // shipped. That structure evicted a LIVE object once the working set passed its capacity,
+    // and every consumer memo keyed on the handle died with it: measured on a verbatim
+    // transcription, 54% of uses lost their handle at 2500 live VAOs against 2048 entries, and
+    // 20% at 1024 live VAOs once the lifetime ids are sparse (an app that creates and destroys
+    // VAOs, which is the Minecraft chunk shape this exists for). Two of the three memos it
+    // fed - the content-hash memo and the resolved-state memo - had NO capacity before this
+    // package: they were unbounded mutable fields on VertexArrayObject. Introducing eviction
+    // there turns one ComputeHash per VAO reconfiguration into one per DRAW, and, once the
+    // buffer table thrashes too, makes the vertex-input content hash a per-draw value that
+    // inserts a fresh heap-allocated BackendVertexInputState into an unbounded map on every
+    // draw. That is a worse leak than the one it was introduced to avoid.
+    //
+    // So: grow on demand, and reclaim by AGE instead of by capacity.
+    //
+    //   * Acquire hits an UnorderedMap<lifetimeId, slotIndex>, in front of which sits a
+    //     one-entry memo. Every re-keyed site in a draw asks about the SAME VAO, so the memo
+    //     turns the five-or-six acquisitions a draw makes into one map probe plus five Uint64
+    //     compares - less than the address multiply plus two-way probe the pre-handle arm ran.
+    //   * OnFrameBoundary retires slots whose object has not been drawn for
+    //     kRetireAgeBoundaries boundaries and returns them to a free list, so the table's
+    //     footprint tracks the LIVE DRAWN working set, not objects ever created. That is the
+    //     property MG_Impl/Pipe/SlotAllocator cannot have here: nothing in P2 can call its
+    //     Free (the tracker emits no object-class state, BufferBackendOps::OnDestroy is handed
+    //     a BackendBufferResource rather than the BufferObject, and VertexArrayObject has no
+    //     death hook at all - adding one is D13's explicit-destroy work, which covers Espryt's
+    //     six kinds, not VertexElementsCso), so an allocator here would grow by one SlotState
+    //     plus one map node per object EVER created, for the life of the process, on a
+    //     platform with an LMK. Age-based reclamation is the stand-in for the death
+    //     notification, and it is exactly as ABA-proof, because reuse bumps Gen.
+    //   * A retire costs at most one memo recompute if the object is drawn again - the same
+    //     price a cache miss costs - and it is charged only to objects that went idle for
+    //     ~1024 frames, never to a hot one.
+    //
+    // Memory: one map node plus one 24-byte Entry per live object, i.e. tens of bytes against
+    // the kilobyte a VertexArrayObject or a BufferObject already costs the frontend. There is
+    // no capacity to size off a device measurement because there is no capacity; what the
+    // device run in D.4.2 can still want is the number itself, so the high-water mark is
+    // logged at MGLOG_D on the allocate-a-new-slot branch (once per new object, never on a
+    // draw - ROADMAP.md:7).
+    //
+    // Single-threaded, like the rest of the renderer. Owned per VulkanRenderer (see
+    // MagmaPipeIdentityTables): a process-global would share one table, and one reclamation
+    // clock, across two live contexts.
+    class MagmaPipeIdentityTable {
+    public:
+        explicit MagmaPipeIdentityTable(const char* kindName) : m_kindName(kindName) {}
+
+        // Slots ever minted. A consumer table indexed by MagmaPipeSlotIndex() needs this many
+        // entries; MagmaPipeSlotTable below grows itself, so nobody has to ask.
+        Uint32 Count() const { return static_cast<Uint32>(m_entries.size()); }
+        // Objects currently holding a slot - the live working set this table tracks.
+        Uint32 LiveCount() const { return static_cast<Uint32>(m_index.size()); }
+
+        MG_Pipe::MGPipeHandle Acquire(Uint64 lifetimeId) {
+            // Unreachable: MG_State hands out lifetime ids from 1 precisely so that a
+            // zero-initialised memo slot cannot carry a live object's id. Guarded anyway so
+            // that a zero can never be minted into a slot and then indexed with.
+            if (lifetimeId == 0) return MG_Pipe::kMGPipeNullHandle;
+            // The one-entry front memo. Cleared by any retire, so it can never serve a slot
+            // that has been handed back to the free list.
+            if (lifetimeId == m_lastLifetimeId) {
+                m_entries[m_lastIndex].LastUse = m_boundary;
+                return m_lastHandle;
+            }
+            Uint32 index = 0;
+            const auto it = m_index.find(lifetimeId);
+            if (it != m_index.end()) {
+                index = it->second;
+            } else {
+                index = ClaimSlot();
+                m_entries[index].LifetimeId = lifetimeId;
+                m_index.emplace(lifetimeId, index);
+            }
+            Entry& entry = m_entries[index];
+            entry.LastUse = m_boundary;
+            m_lastLifetimeId = lifetimeId;
+            m_lastIndex = index;
+            m_lastHandle = MG_Pipe::MGPipeHandle{index + MG_Pipe::kMGPipeFirstAllocatableSlot,
+                                                 entry.Gen};
+            return m_lastHandle;
+        }
+
+        // Ages the table and returns idle slots to the free list. Same shape and the same
+        // self-gating as VertexInputStateFactory::OnFrameBoundary, which is what the reclaimed
+        // slots' consumers use.
+        void OnFrameBoundary() {
+            ++m_boundary;
+            if ((m_boundary % kSweepInterval) != 0) return;
+            SizeT retired = 0;
+            for (auto it = m_index.begin(); it != m_index.end();) {
+                Entry& entry = m_entries[it->second];
+                if ((m_boundary - entry.LastUse) > kRetireAgeBoundaries) {
+                    entry.LifetimeId = 0;
+                    m_freeSlots.push_back(it->second);
+                    it = m_index.erase(it);
+                    ++retired;
+                } else {
+                    ++it;
+                }
+            }
+            if (retired != 0) {
+                // A retired slot's Gen has not moved yet - it moves when the slot is reused -
+                // so a front memo pointing at one would still hand out a handle the consumer
+                // tables would accept. Drop it.
+                m_lastLifetimeId = 0;
+                m_lastHandle = MG_Pipe::kMGPipeNullHandle;
+                MGLOG_D("MagmaPipeIdentityTable(%s): retired %zu idle slots, %u live of %u minted",
+                        m_kindName, retired, LiveCount(), Count());
+            }
+        }
+
+    private:
+        // Sweep cadence and retirement age, deliberately the same numbers
+        // VertexInputStateFactory::OnFrameBoundary uses for the entries these slots key: a slot
+        // retired earlier than its cache entry would mint a new handle for an object whose
+        // entry is still live and still correct, which is a pure waste.
+        static constexpr Uint64 kSweepInterval = 256;
+        static constexpr Uint64 kRetireAgeBoundaries = 1024;
+
+        struct Entry {
+            Uint64 LifetimeId = 0;
+            Uint64 LastUse = 0;
+            // Moves ONLY on slot reuse, never on respecify: an object that keeps its slot keeps
+            // its generation, which is what makes a memo survive a reconfiguration.
+            Uint32 Gen = 0;
+        };
+
+        Uint32 ClaimSlot() {
+            while (!m_freeSlots.empty()) {
+                const Uint32 index = m_freeSlots.back();
+                m_freeSlots.pop_back();
+                // MGPipeHandles.h:52-58 defends the Gen wrap only in a debug allocator, and
+                // MOBILEGL_ASSERT is compiled out of every build P2 runs (Defines.h: asserts are
+                // live only at MOBILEGL_LOG_ACTIVE_LEVEL == DEBUG). So the wrap is handled on the
+                // RELEASE path instead of asserted: a slot that has been reused 2^32 times is
+                // permanently retired rather than wrapped, because a wrapped Gen would let a
+                // stale handle match a live object. It costs one slot.
+                if (m_entries[index].Gen == ~Uint32{0}) {
+                    MGLOG_W("MagmaPipeIdentityTable(%s): slot %u reached generation 2^32-1 and is "
+                            "retired for good; {slot, gen} stays unique",
+                            m_kindName, index + MG_Pipe::kMGPipeFirstAllocatableSlot);
+                    continue;
+                }
+                ++m_entries[index].Gen;
+                return index;
+            }
+            const Uint32 index = static_cast<Uint32>(m_entries.size());
+            m_entries.push_back(Entry{});
+            m_entries[index].Gen = 1;
+            // The high-water mark, at powers of two from 1024 up: at most a handful of lines
+            // for a whole session, emitted from the allocate-a-NEW-slot branch, i.e. once per
+            // object this backend has ever seen and never on a draw (ROADMAP.md:7).
+            //
+            // [narrow, declared deviation from D20's "MGLOG_D for anything non-critical"] This
+            // one is I, not D, because D is compiled out of every build that ships and of every
+            // build P2 measures, and this line IS the measurement review v2's MAJOR 1 asks for:
+            // the live-object high-water mark of minecraft-1.21.4-in-world and
+            // ...-sodium-in-world, which nothing on desktop reaches and no gate here can see.
+            // The structure no longer has a capacity to size off it, so the number is evidence
+            // rather than a tuning input - but D.4.2 should still read it out of the device log,
+            // and it cannot read a line that was compiled away.
+            const SizeT minted = m_entries.size();
+            if (minted >= 1024 && (minted & (minted - 1)) == 0) {
+                MGLOG_I("MagmaPipeIdentityTable(%s): high-water %zu slots minted, %u live",
+                        m_kindName, minted, LiveCount());
+            }
+            return index;
+        }
+
+        const char* m_kindName = "";
+        Uint64 m_boundary = 0;
+        Vector<Entry> m_entries;
+        Vector<Uint32> m_freeSlots;
+        UnorderedMap<Uint64, Uint32> m_index;
+        // One-entry front memo (see Acquire). m_lastLifetimeId == 0 means "empty": a live
+        // object's lifetime id is never 0.
+        Uint64 m_lastLifetimeId = 0;
+        Uint32 m_lastIndex = 0;
+        MG_Pipe::MGPipeHandle m_lastHandle = MG_Pipe::kMGPipeNullHandle;
+    };
+
+    // The two mints one renderer owns. Per renderer, NOT process-global: two live contexts (or
+    // a context recreation, which destroys and rebuilds the renderer) would otherwise share one
+    // table and one reclamation clock, and both consumer tables are per-instance already.
+    class MagmaPipeIdentityTables {
+    public:
+        // A VAO is kind VertexElementsCso: that is the gallium-shaped CSO a vertex array
+        // resolves to, and the only kind in MGPipeKind that names vertex-input state.
+        MG_Pipe::MGPipeHandle HandleOf(MG_Pipe::MGPipeKind kind, Uint64 lifetimeId) {
+            return kind == MG_Pipe::MGPipeKind::Buffer ? m_buffers.Acquire(lifetimeId)
+                                                       : m_vaos.Acquire(lifetimeId);
+        }
+        void OnFrameBoundary() {
+            m_vaos.OnFrameBoundary();
+            m_buffers.OnFrameBoundary();
+        }
+        const MagmaPipeIdentityTable& Vaos() const { return m_vaos; }
+        const MagmaPipeIdentityTable& Buffers() const { return m_buffers; }
+
+    private:
+        MagmaPipeIdentityTable m_vaos{"VertexElementsCso"};
+        MagmaPipeIdentityTable m_buffers{"Buffer"};
+    };
+
+    // The table entry a handle names. Every per-slot table Magma keeps is indexed by this.
+    //
+    // A null handle has no slot, and it is unreachable here: both lifetime-id sources start at
+    // 1 (VertexArrayObject.cpp, BufferObject.cpp), so Acquire's zero guard never fires. The
+    // ternary, not the assertion, is what has effect in a shipped build (Defines.h compiles
+    // MOBILEGL_ASSERT out at INFO), and slot 0 of a consumer table is a real entry that a null
+    // handle can never match, because MGPipeHandleIsNull is also what the consumers compare.
+    inline Uint32 MagmaPipeSlotIndex(const MG_Pipe::MGPipeHandle& handle) {
+        MOBILEGL_ASSERT(!MG_Pipe::MGPipeHandleIsNull(handle),
+                        "a null MGPipeHandle has no slot to index a per-slot table with");
+        return MG_Pipe::MGPipeHandleIsNull(handle)
+                   ? 0u
+                   : handle.Slot - MG_Pipe::kMGPipeFirstAllocatableSlot;
+    }
+
+    // A grow-on-demand per-slot table whose ENTRY ADDRESSES NEVER MOVE.
+    //
+    // D12.4 asks for a grow-on-demand Vector, and with an unbounded mint that is what a
+    // consumer needs - but a Vector that grows relocates its elements, and the draw path holds
+    // references into these entries across nested calls. Chunks of kChunkEntries are appended
+    // instead: the Vector of owning pointers reallocates, the chunks never do, so an entry
+    // reference is valid for the life of the table. That is the same guarantee the fixed table
+    // it replaces gave, without the fixed capacity.
+    template <typename T, Uint32 kChunkEntries = 256>
+    class MagmaPipeSlotTable {
+    public:
+        T& operator[](Uint32 index) {
+            const Uint32 chunk = index / kChunkEntries;
+            while (m_chunks.size() <= chunk) {
+                m_chunks.push_back(MakeUnique<Chunk>());
+            }
+            return m_chunks[chunk]->Entries[index % kChunkEntries];
+        }
+        SizeT Capacity() const { return m_chunks.size() * kChunkEntries; }
+
+    private:
+        struct Chunk {
+            T Entries[kChunkEntries] = {};
+        };
+        Vector<UniquePtr<Chunk>> m_chunks;
+    };
+
+    // The claim rule every per-slot memo table uses, in one place so that the rule and the
+    // negative control that defeats it cannot drift apart between consumers - and so that the
+    // unit suite which drives a REAL slot reuse (MG_Test/Pipe/MagmaPipeIdentityTest.cpp) tests
+    // this code rather than a copy of it.
+    //
+    // The SLOT picks the entry; the WHOLE handle - Gen included - decides whether the entry is
+    // this object's. A slot the mint recycled for a different object comes back with a moved
+    // Gen, so the compare fails and the entry is cleared rather than inherited. That is the
+    // half HandleRecycleScenario cannot reach (see MagmaPipeAbaControlDefeatsIdentity).
+    //
+    // With negative control C on, every object collapses onto one entry and the entry is handed
+    // back UNCLEARED and UNCLAIMED - at once "the replacement reproduced its predecessor's
+    // slot" and "the slot was reused and Gen did not move".
+    //
+    // `Memos` needs a MG_Pipe::MGPipeHandle member named Owner and a default constructor that
+    // means "empty"; VertexInputStateFactory::VaoBackendMemos is the one production instance.
+    template <typename Memos, Uint32 kChunkEntries>
+    inline Memos& MagmaPipeClaimSlotMemos(MagmaPipeSlotTable<Memos, kChunkEntries>& table,
+                                          const MG_Pipe::MGPipeHandle& handle) {
+        if (MagmaPipeAbaControlDefeatsIdentity()) {
+            return table[kMagmaPipeAbaControlSlotIndex];
+        }
+        Memos& memos = table[MagmaPipeSlotIndex(handle)];
+        if (!(memos.Owner == handle)) {
+            memos = Memos{};
+            memos.Owner = handle;
+        }
+        return memos;
+    }
+#endif // MOBILEGL_PIPE_PUSH
+} // namespace MobileGL::MG_Backend::DirectVulkan

MobileGL/MG_Backend/DirectVulkan/Renderer/VertexInputStateFactory.cpp

@@ -7,6 +7,7 @@
 // End of Source File Header
 
 #include "VertexInputStateFactory.h"
+#include "MagmaPipeArms.h"
 #include "MG_Util/Converters/MGToStr/DataTypeConverter.h"
 #include <MG_Backend/BackendObjects.h>
 #include <utility>
@@ -45,25 +46,149 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             // capture came back holding a dead VAO's vertex data (0,0,0,1 - the previous
             // test's positions) instead of its own.
             // Zero for client memory (no buffer), which is a distinct identity of its own.
-            const Uint64 bufferKey = attr.Buffer ? attr.Buffer->GetLifetimeId() : 0;
+            //
+            // P2 D12.4 / ARCHITECTURE.md 9.5: under the handle arm the identity is the
+            // buffer's {slot, gen} rather than its lifetime id - "lifetimeId -> gen mixed
+            // into every server-side content hash". The two are equally ABA-proof (the
+            // allocator maps one onto the other and bumps Gen only on slot REUSE); what
+            // changes is that the key is now the identity the SERVER will be handed once
+            // buffers travel as handles, instead of a number only the client can mint.
+            Uint64 bufferKey = attr.Buffer ? attr.Buffer->GetLifetimeId() : 0;
+#if MOBILEGL_PIPE_PUSH
+            if (attr.Buffer) {
+                // The SAME arm question the other four re-keyed sites ask, through the same
+                // helper: a site that decided for itself could silently key on the pre-handle
+                // identity while its neighbours keyed on the handle.
+                if (MagmaPipeTrackHArmIsHandles(MG_Pipe::kMGPipeSubsystemMagmaVertexInput)) {
+                    const MG_Pipe::MGPipeHandle handle =
+                        m_identity->HandleOf(MG_Pipe::MGPipeKind::Buffer, attr.Buffer->GetLifetimeId());
+                    bufferKey = static_cast<Uint64>(handle.Slot) | (static_cast<Uint64>(handle.Gen) << 32);
+                }
+                if (MagmaPipeAbaControlDefeatsIdentity()) {
+                    // Negative control C (P2 brief D18), on WHICHEVER arm this run is on - the
+                    // pre-handle lifetime id and the handle's {slot, gen} are the same guard
+                    // wearing two hats, and a control that defeated only the retired one would
+                    // say nothing about the key P2 ships.
+                    //
+                    // The identity is replaced by a constant rather than by the raw
+                    // BufferObject*, because the address is not recycled in practice and so
+                    // never collides (see MagmaPipeAbaControlDefeatsIdentity). Zero is what a
+                    // key with NO buffer identity in it looks like - the exact defect this
+                    // hash was fixed for: "the hash is what TryBindResolvedVertexBindings
+                    // accepts as proof that a memoised binding still reads the buffer it was
+                    // resolved from", and with the identity gone it accepts a binding resolved
+                    // from a different buffer. HandleRecycleScenario.AbaControl then draws a
+                    // replacement VAO and gets its dead predecessor's vertex data.
+                    bufferKey = 0;
+                }
+            }
+#endif
             XXHASH_VERIFY(XXH64_update(m_hashState, &bufferKey, sizeof(bufferKey)));
         }
 
         return XXH64_digest(m_hashState);
     }
 
+#if MOBILEGL_PIPE_PUSH
+    VertexInputStateFactory::VaoBackendMemos& VertexInputStateFactory::MemosFor(
+        const MG_State::GLState::VertexArrayObject& vao) const {
+        const MG_Pipe::MGPipeHandle handle =
+            m_identity->HandleOf(MG_Pipe::MGPipeKind::VertexElementsCso, vao.GetLifetimeId());
+        // One entry per mintable slot, grown on demand: the mint has no capacity, so neither
+        // does this, and no two live VAOs can share an entry however large the working set is.
+        // There is no probe in front of it because the mint itself is one - a one-entry memo
+        // hit for every acquisition after this draw's first, and a hash probe otherwise.
+        //
+        // The claim rule - the slot picks the entry, the whole handle (Gen included) decides
+        // whose it is - and negative control C's defeat of it are MagmaPipeArms.h's
+        // MagmaPipeClaimSlotMemos, so that the unit suite which drives a REAL slot reuse
+        // (MG_Test/Pipe/MagmaPipeIdentityTest.cpp) exercises this code and not a copy of it.
+        // What the control defeats HERE is the identity that SELECTS the entry: every VAO
+        // collapses onto one, handed back uncleared, so the replacement inherits the dead
+        // VAO's content hash and its resolved-entry pointer. The GENERATION half is the unit
+        // suite's business, for the reason MagmaPipeAbaControlDefeatsIdentity spells out.
+        return MagmaPipeClaimSlotMemos(m_vaoMemos, handle);
+    }
+#endif
+
+#if MOBILEGL_PIPE_PUSH
+    Bool VertexInputStateFactory::TryGetMemoizedHash(const MG_State::GLState::VertexArrayObject& vao,
+                                                     Uint64& outHash) const {
+        if (MagmaPipeTrackHArmIsHandles(MG_Pipe::kMGPipeSubsystemMagmaVertexInput)) {
+            const VaoBackendMemos& memos = MemosFor(vao);
+            if (memos.HashConfigVersion != vao.GetConfigVersion()) return false;
+            outHash = memos.Hash;
+            return true;
+        }
+#if MOBILEGL_PIPE_LEGACY_MEMOS
+        return vao.GetBackendHashMemo(outHash);
+#else
+        return false;
+#endif
+    }
+#endif
+
     VertexInputStateFactory::HashType VertexInputStateFactory::GetOrComputeHash(
         const MG_State::GLState::VertexArrayObject& vao) const {
         HashType hash = 0;
+#if MOBILEGL_PIPE_PUSH
+        // P2 D12.5: the same memo, on the backend's side of the boundary.
+        if (MagmaPipeTrackHArmIsHandles(MG_Pipe::kMGPipeSubsystemMagmaVertexInput)) {
+            VaoBackendMemos& memos = MemosFor(vao);
+            if (memos.HashConfigVersion == vao.GetConfigVersion()) {
+                return memos.Hash;
+            }
+            hash = ComputeHash(vao);
+            memos.Hash = hash;
+            memos.HashConfigVersion = vao.GetConfigVersion();
+            return hash;
+        }
+#endif
+#if MOBILEGL_PIPE_LEGACY_MEMOS
         if (!vao.GetBackendHashMemo(hash)) {
             hash = ComputeHash(vao);
             vao.SetBackendHashMemo(hash);
         }
+#endif
         return hash;
     }
 
     const VertexInputStateFactory::BackendVertexInputState& VertexInputStateFactory::GetOrCreateVertexInputState(
         const MG_State::GLState::VertexArrayObject& vao) {
+#if MOBILEGL_PIPE_PUSH
+        // P2 D12.5: the same per-draw fast path, but the resolved-entry pointer lives in this
+        // factory's slot-indexed table instead of on the frontend VAO. The eviction epoch
+        // survives the move and is still what stops a stale pointer being dereferenced: the
+        // POINTEE is a cache entry this factory can erase at a frame boundary, and moving the
+        // memo does not change that.
+        if (MagmaPipeTrackHArmIsHandles(MG_Pipe::kMGPipeSubsystemMagmaVertexInput)) {
+            VaoBackendMemos& memos = MemosFor(vao);
+            if (memos.StateConfigVersion == vao.GetConfigVersion() && memos.State != nullptr &&
+                memos.StateEpoch == m_evictionEpoch) {
+                const auto* memoEntry = static_cast<const BackendVertexInputState*>(memos.State);
+                memoEntry->lastUsedFrameBoundary = m_frameBoundaryCounter;
+                return *memoEntry;
+            }
+            const BackendVertexInputState& resolved =
+                GetOrCreateVertexInputState(vao, GetOrComputeHash(vao));
+            // MemosFor is re-taken rather than kept live across GetOrCreateVertexInputState:
+            // the reference is not worth holding across a call that can resize the table.
+            VaoBackendMemos& stamp = MemosFor(vao);
+            stamp.State = &resolved;
+            stamp.StateEpoch = m_evictionEpoch;
+            stamp.StateConfigVersion = vao.GetConfigVersion();
+            // The AUX memo is deliberately NOT stamped here: its two words already live in
+            // VulkanRenderer::VaoDrawMemo (layoutHash / layoutAuxMasks) and its getter has no
+            // live reader anywhere, so the handle arm retires it rather than moving it.
+            return resolved;
+        }
+#endif
+#if !MOBILEGL_PIPE_LEGACY_MEMOS
+        // Unreachable: with no legacy arm compiled MagmaPipeTrackHArmIsHandles is a compile-
+        // time true, so the handle arm above always returns. Written out rather than left to
+        // fall off the end so the function still has a return on every path a compiler sees.
+        return GetOrCreateVertexInputState(vao, GetOrComputeHash(vao));
+#else
         // Per-draw fast path: the VAO carries a pointer to its resolved entry,
         // valid while its config version and the cache's eviction epoch both
         // match - no re-hash, no map lookup.
@@ -83,6 +208,7 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         vao.SetBackendAuxMemo(entry.layoutHash,
                               PackVertexInputAuxMasks(entry.unsupportedAttribMask, entry.attributeLocationMask));
         return entry;
+#endif // MOBILEGL_PIPE_LEGACY_MEMOS
     }
 
     const VertexInputStateFactory::BackendVertexInputState& VertexInputStateFactory::GetOrCreateVertexInputState(
@@ -316,8 +442,14 @@ namespace MobileGL::MG_Backend::DirectVulkan {
                 // Invalidate every VAO's state-pointer memo: the erased node's
                 // 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).
+                // instances (see the member comment). With no legacy arm the memos
+                // live in this factory and die with it, so a per-instance bump is
+                // enough - P2 D12.5.
+#if MOBILEGL_PIPE_LEGACY_MEMOS
                 m_evictionEpoch = ++s_evictionEpochSource;
+#else
+                ++m_evictionEpoch;
+#endif
             } else {
                 ++it;
             }

MobileGL/MG_Backend/DirectVulkan/Renderer/VertexInputStateFactory.h

@@ -7,8 +7,12 @@
 // End of Source File Header
 
 #pragma once
+// MG_Pipe::MGPipeHandle for the P2 D12.5 memo table below. A header of constexpr constants,
+// so the pull build gains nothing from it.
+#include <MG_Pipe/MGPipeHandles.h>
 
 #include "Config.h"
+#include "MagmaPipeArms.h"
 #include "VertexInputStateBuilder.h"
 #include "MG_State/GLState/VertexArrayState/VertexArrayObject.h"
 #include <Includes.h>
@@ -70,8 +74,18 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             };
         };
 
+#if MOBILEGL_PIPE_PUSH
+        // The mint is the RENDERER's (MagmaPipeIdentityTables), not a process-global and not
+        // this factory's: VulkanRenderer::LookupVaoDrawMemo has to derive the same {slot, gen}
+        // for the same VAO, and a table that outlived the context it was minted for would share
+        // one reclamation clock across two live contexts (review v2 minor 4).
+        VertexInputStateFactory(const VulkanRendererConfig& config, VkPhysicalDevice physicalDevice,
+                                MagmaPipeIdentityTables& identity):
+            m_config(config), m_physicalDevice(physicalDevice), m_identity(&identity) {}
+#else
         VertexInputStateFactory(const VulkanRendererConfig& config, VkPhysicalDevice physicalDevice):
             m_config(config), m_physicalDevice(physicalDevice) {}
+#endif
         ~VertexInputStateFactory() = default;
         VertexInputStateFactory(const VertexInputStateFactory&) = delete;
 
@@ -86,6 +100,13 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // Memoized ComputeHash: reuses the VAO's cached hash while its config version
         // is unchanged. Use this on per-draw paths.
         HashType GetOrComputeHash(const MG_State::GLState::VertexArrayObject& vao) const;
+#if MOBILEGL_PIPE_PUSH
+        // The VAO's content hash IF it has already been memoized, without computing one.
+        // P2 D12.5: the three draw-path readers that used to ask the VAO object this
+        // question ask the factory instead, because that is where the memo lives once the
+        // frontend object stops carrying the backend's state.
+        Bool TryGetMemoizedHash(const MG_State::GLState::VertexArrayObject& vao, Uint64& outHash) const;
+#endif
         const BackendVertexInputState& GetOrCreateVertexInputState(
             const MG_State::GLState::VertexArrayObject& vao, HashType hash);
         const BackendVertexInputState& GetOrCreateVertexInputState(const MG_State::GLState::VertexArrayObject& vao);
@@ -112,6 +133,44 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         static VkFormat ToFloat32VertexFormat(Int componentCount);
         Bool SupportsVertexBufferFormat(VkFormat format) const;
 
+#if MOBILEGL_PIPE_PUSH
+        // ---- P2 D12.5: the backend's memos, off the frontend VAO and into the backend ----
+        //
+        // The two facts that used to live as `mutable` fields on VertexArrayObject
+        // (Get/SetBackendHashMemo and Get/SetBackendStateMemo), kept here instead, keyed on
+        // the VAO's {slot, gen} and guarded by exactly the same config version. A frontend
+        // state object holding the backend's raw pointer is what P2 retires: under split the
+        // backend is in another process and its cache entry has no address a client could
+        // store, so the memo has to live on the side that owns the pointee.
+        //
+        // The AUX memo is not carried over: its two words moved into VaoDrawMemo::layoutHash
+        // and layoutAuxMasks long ago and its getter has no live reader anywhere in the tree,
+        // so the handle arm simply stops writing it (D12.5 says delete rather than move).
+        struct VaoBackendMemos {
+            // Whose memos these are. The identity table can recycle a slot for a different
+            // VAO under LRU pressure, and the handle compare - Gen included - is what says
+            // the contents are this object's and not its predecessor's.
+            MG_Pipe::MGPipeHandle Owner = MG_Pipe::kMGPipeNullHandle;
+            Uint64 Hash = 0;
+            Uint32 HashConfigVersion = ~0u;
+            const void* State = nullptr;
+            Uint64 StateEpoch = 0;
+            Uint32 StateConfigVersion = ~0u;
+        };
+        // Grow-on-demand (D12.4), one entry per slot the renderer's mint has ever handed
+        // out, and NO CAPACITY: these two memos had none before this package either - they
+        // were unbounded mutable fields on the VertexArrayObject itself - and re-introducing
+        // eviction here is what review v2 rejected. MagmaPipeSlotTable grows in chunks so an
+        // entry reference stays valid across the nested GetOrCreateVertexInputState call.
+        // 48 B per live VAO, reclaimed with the slot when the object goes idle.
+        mutable MagmaPipeSlotTable<VaoBackendMemos> m_vaoMemos;
+        // The renderer's {slot, gen} mint (see the constructor). Never null under push.
+        MagmaPipeIdentityTables* m_identity = nullptr;
+        // The entry belonging to `vao`, claimed (and cleared) if the slot currently holds
+        // someone else's.
+        VaoBackendMemos& MemosFor(const MG_State::GLState::VertexArrayObject& vao) const;
+#endif
+
         const VulkanRendererConfig& m_config;
         VkPhysicalDevice m_physicalDevice = VK_NULL_HANDLE;
         // Values are heap-allocated: UnorderedMap is open-addressing, so INSERT
@@ -137,8 +196,17 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // 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).
+        //
+        // P2 D12.5: the process-wide source is the LEGACY arm's need. It exists because the
+        // memos live on the frontend VAOs and therefore outlive the factory. The handle arm's
+        // memo table is owned by this factory and dies with it, so a per-instance counter is
+        // enough there and the epoch shrinks back to what it looks like it should be.
+#if MOBILEGL_PIPE_LEGACY_MEMOS
         static inline Uint64 s_evictionEpochSource = 0;
         Uint64 m_evictionEpoch = ++s_evictionEpochSource;
+#else
+        Uint64 m_evictionEpoch = 1;
+#endif
         static inline XXH64_state_t* m_hashState = XXH64_createState();
     };
 } // namespace MobileGL::MG_Backend::DirectVulkan

MobileGL/MG_Backend/DirectVulkan/Renderer/VulkanRenderer.cpp

@@ -397,6 +397,101 @@ namespace MobileGL::MG_Backend::DirectVulkan {
     };
     static DynamicStateShadow g_dynamicStateShadow;
 
+#if MOBILEGL_PIPE_PUSH
+    // ---- D12.3: DynamicTailKey's inputs against the P2 chunk table ----
+    //
+    // DynamicTailKey's inventory (declared above, one line per reader) is an exact,
+    // hand-maintained enumeration of what the six Apply* in the tail read. The P2 chunk table
+    // (MG_Pipe/MGPipeRenderStateSpans.h) is an independent, offsetof-derived statement of
+    // which bytes of RenderStateParameters are dynamic state. The two were written for
+    // different reasons, so making them check each other is free evidence: if a later chunk
+    // edit demotes or promotes one of these members, the mismatch is a BUILD BREAK here rather
+    // than a tail that silently stops being re-run when its input moves.
+    //
+    // The brief (P2 D12.3) expects every input to be dynamic; the tree says otherwise for
+    // exactly one, and the tree is right - see the ScissorTestEnabledMask note below.
+    //
+    // These assertions ARE D19's DynamicChunksCoverMagmasDynamicTailKey, in the only file this
+    // package owns. D19 names it as a case in MG_Test/Pipe/RenderStateSpansTest.cpp, which
+    // belongs to package A (C.5). INTEGRATOR: make sure the outcome is not "neither" - if
+    // package A did not land that case, this static_assert block is the whole gate, and if it
+    // did, the two are redundant on purpose and both should stay.
+    namespace {
+        // Is [begin, begin + size) covered entirely by DYNAMIC chunks?
+        constexpr Bool MagmaRenderStateRangeIsDynamic(SizeT begin, SizeT size) {
+            const SizeT end = begin + size;
+            for (SizeT i = 0; i < MG_Pipe::kMGPipeRenderStateChunkCount; ++i) {
+                const SizeT chunkBegin = MG_Pipe::kMGPipeRenderStateChunkBoundaries[i];
+                const SizeT chunkEnd = MG_Pipe::kMGPipeRenderStateChunkBoundaries[i + 1];
+                if (end <= chunkBegin || begin >= chunkEnd) continue; // disjoint
+                if (MG_Pipe::MGPipeRenderStateChunkIsPipeline(i)) return false;
+            }
+            return true;
+        }
+        using MagmaTailRsp = RenderStateParameters;
+
+#define MAGMA_TAIL_INPUT_IS_DYNAMIC(Member)                                                                            \
+    static_assert(MagmaRenderStateRangeIsDynamic(offsetof(MagmaTailRsp, Member),                                       \
+                                                 sizeof(MagmaTailRsp::Member)),                                        \
+                  "ApplyDynamicDrawStateTail reads " #Member                                                           \
+                  ", which the P2 chunk table no longer calls dynamic state: a change to it would "                    \
+                  "move the pipeline version, not the parameters version, and the tail would stop "                    \
+                  "being re-run for it")
+
+        // Viewports, DepthRanges and ScissorBoxes are asserted over the WHOLE array while the
+        // tail reads only element 0. That is deliberately stricter than the reader needs: the
+        // chunk table has no per-element granularity today, so an array that is dynamic at all
+        // is dynamic entirely, and asserting the whole of it says so. If a later phase ever
+        // splits a per-viewport chunk out, this is a build break by design - narrow the assert
+        // to element 0 then, and say why in the same commit.
+        MAGMA_TAIL_INPUT_IS_DYNAMIC(Viewports);        // ApplyGLViewportState: Viewports[0]
+        MAGMA_TAIL_INPUT_IS_DYNAMIC(DepthRanges);      // ApplyGLViewportState: DepthRanges[0]
+        MAGMA_TAIL_INPUT_IS_DYNAMIC(BlendColor);       // ApplyBlendConstants
+        MAGMA_TAIL_INPUT_IS_DYNAMIC(PolygonOffsetFactor); // ApplyPolygonOffsetState
+        MAGMA_TAIL_INPUT_IS_DYNAMIC(PolygonOffsetUnits);  // ApplyPolygonOffsetState
+        MAGMA_TAIL_INPUT_IS_DYNAMIC(LineWidth);        // ApplyLineWidthState
+        MAGMA_TAIL_INPUT_IS_DYNAMIC(ScissorBoxes);     // the scissor rect: ScissorBoxes[0]
+#undef MAGMA_TAIL_INPUT_IS_DYNAMIC
+
+        // ApplyStencilState reads three of the seven members of each face, and D6 splits
+        // StencilFaceState at sub-member granularity for exactly this reason: Ref, ValueMask
+        // and WriteMask are VK_DYNAMIC_STATE_STENCIL_{REFERENCE,COMPARE_MASK,WRITE_MASK}, while
+        // Func and the three ops are baked into the pipeline. Asserted per member, per face,
+        // because the split runs THROUGH the struct rather than around it.
+        constexpr SizeT kMagmaStencilFace1 = offsetof(MagmaTailRsp, StencilStates) + sizeof(StencilFaceState);
+#define MAGMA_TAIL_STENCIL_IS_DYNAMIC(Member)                                                                          \
+    static_assert(MagmaRenderStateRangeIsDynamic(offsetof(MagmaTailRsp, StencilStates) +                               \
+                                                     offsetof(StencilFaceState, Member),                               \
+                                                 sizeof(StencilFaceState::Member)),                                    \
+                  "ApplyStencilState reads the FRONT face's " #Member " as dynamic state");                            \
+    static_assert(MagmaRenderStateRangeIsDynamic(kMagmaStencilFace1 + offsetof(StencilFaceState, Member),               \
+                                                 sizeof(StencilFaceState::Member)),                                    \
+                  "ApplyStencilState reads the BACK face's " #Member " as dynamic state")
+
+        MAGMA_TAIL_STENCIL_IS_DYNAMIC(Ref);
+        MAGMA_TAIL_STENCIL_IS_DYNAMIC(ValueMask);
+        MAGMA_TAIL_STENCIL_IS_DYNAMIC(WriteMask);
+#undef MAGMA_TAIL_STENCIL_IS_DYNAMIC
+
+        // THE ONE INPUT THAT IS NOT DYNAMIC, and the brief's D12.3 says it should be.
+        // The tree wins, and it is right: the split's only rule is "a byte is pipeline state
+        // iff a public setter that calls BumpVersions() writes it", and ScissorTestEnabledMask
+        // is written by SetCapability(ScissorTest), which does. It sits in pipeline chunk P6
+        // with the other capability bools. The tail reads it only to decide between the
+        // scissor box and a full-extent rect, and it is HARMLESS there for a reason worth
+        // stating: a pipeline-half write moves the pipeline version, and the pipeline version
+        // moves only together with the parameters version (BumpVersions bumps both), so the
+        // tail's version gate is invalidated by it just the same. A DYNAMIC member promoted
+        // into the pipeline half would break that direction, which is what the asserts above
+        // are for; this one is pinned in the opposite direction so that DEMOTING it - which
+        // would be a real G7 violation - is also a build break.
+        static_assert(!MagmaRenderStateRangeIsDynamic(offsetof(MagmaTailRsp, ScissorTestEnabledMask),
+                                                      sizeof(MagmaTailRsp::ScissorTestEnabledMask)),
+                      "ScissorTestEnabledMask is written by SetCapability(ScissorTest), which calls "
+                      "BumpVersions(), so the chunk table must keep it in the pipeline half");
+    } // namespace
+#endif // MOBILEGL_PIPE_PUSH
+
     static void ResetDynamicStateShadow() {
         g_dynamicStateShadow = {};
     }
@@ -3015,6 +3110,13 @@ void main() {
     }
 
     void VulkanRenderer::Initialize() {
+#if MOBILEGL_PIPE_PUSH
+        // P2 D14, and it belongs HERE rather than on a draw: "a Track-H subsystem whose bit is
+        // clear is a STARTUP Fatal{PipeLegacyMemosDisabled}". Checks Magma's own bit only, and
+        // only once this backend is the one being brought up, so an Espryt-side bitmask cannot
+        // kill a Magma run and vice versa.
+        MagmaPipeValidateSubsystemConfiguration();
+#endif
         CreateInstance();
         CreateSurface();
         PickPhysicalDevice();
@@ -3194,7 +3296,12 @@ void main() {
             m_physicalDevice.properties.limits.minUniformBufferOffsetAlignment, m_config.MaxFramesInFlight,
             maxProgramBindings, kDescriptorSetsPerFrame, m_textureManager.get(), m_samplerManager.get());
         MOBILEGL_ASSERT(succeeded, "UniformDescriptorBinder initialization failed.");
+#if MOBILEGL_PIPE_PUSH
+        m_vertexInputStateFactory =
+            MakeUnique<VertexInputStateFactory>(m_config, m_physicalDevice.handle, m_pipeIdentity);
+#else
         m_vertexInputStateFactory = MakeUnique<VertexInputStateFactory>(m_config, m_physicalDevice.handle);
+#endif
         MOBILEGL_ASSERT(m_vertexInputStateFactory != nullptr, "VertexInputStateFactory creation failed.");
 
         // Prime the first frame so Render() always targets an acquired swapchain image.
@@ -3542,6 +3649,76 @@ void main() {
         if (m_vaoDrawMemoTable.empty()) {
             m_vaoDrawMemoTable.resize(kVaoDrawMemoSlotCount);
         }
+#if MOBILEGL_PIPE_PUSH
+        if (MagmaPipeAbaControlDefeatsIdentity()) {
+            // Negative control C (P2 brief D18), ahead of BOTH arms because it defeats the
+            // identity half of both keys at once: the legacy arm's (address, lifetime id) pair
+            // and the handle arm's {slot, gen}. Every VAO lands on one entry and the entry is
+            // handed back without an identity compare and WITHOUT being cleared - which is
+            // exactly what this table would do if a replacement object reproduced its dead
+            // predecessor's address, or reused its slot without the generation moving.
+            //
+            // Nothing else about the entry is relaxed: whether the resolved bindings it holds
+            // are then USED is still decided by TryBindResolvedVertexBindings' frame serial,
+            // content hash, active-attribute mask and slice epochs. That is what keeps the arm
+            // an assertion about identity rather than about the memo as a whole.
+            VaoDrawMemo& aliased = m_vaoDrawMemoTable[kMagmaPipeAbaControlSlotIndex];
+            aliased.vaoKey = vao;
+            aliased.vaoLifetimeId = vao->GetLifetimeId();
+            return &aliased;
+        }
+        // ---- P2 D12.4, the handle arm ----
+        //
+        // The slot PICKS the entry, and the handle DECIDES whether the entry is this VAO's -
+        // the same division of labour the legacy arm below gives the address and the lifetime
+        // id, with two differences that are both improvements:
+        //
+        //   * the slot is dense from 1, so below kVaoDrawMemoSlotCount live slots the map is a
+        //     bijection and the two-way probe never collides at all, where an address hash
+        //     collides by the birthday rule from the first few dozen VAOs;
+        //   * the handle is an exact identity - Gen moves whenever a slot changes owner - so
+        //     neither a deleted VAO's successor at the same heap address nor a VAO whose slot
+        //     was recycled can match a predecessor's entry, even byte-identically configured.
+        //     That is what makes the lifetime-id half of the legacy compare unnecessary here.
+        //
+        // The capacity and the victim rule are deliberately the base ref's, unchanged: this is
+        // the one memo of the three that HAD a capacity before P2, and an entry lost to a
+        // collision costs exactly what it cost then (one vertex-binding re-resolve). Above
+        // kVaoDrawMemoSlotCount live VAOs a set of two ways serves four slots, and degrades
+        // from there - never worse than the address-hashed table it replaces, which was already
+        // colliding.
+        if (MagmaPipeTrackHArmIsHandles(MG_Pipe::kMGPipeSubsystemMagmaVertexInput)) {
+            const MG_Pipe::MGPipeHandle handle = ResolveVaoHandle(*vao);
+            const Uint32 index = MagmaPipeSlotIndex(handle) & (kVaoDrawMemoSlotCount - 1u);
+            VaoDrawMemo& first = m_vaoDrawMemoTable[index];
+            if (first.vaoHandle == handle) {
+                return &first;
+            }
+            VaoDrawMemo& second = m_vaoDrawMemoTable[index ^ 1u];
+            if (second.vaoHandle == handle) {
+                return &second;
+            }
+            // Miss: recycle a slot. Prefer an unclaimed one; otherwise evict the entry whose
+            // bindings memo is older (its VAO is the one drawn less recently).
+            VaoDrawMemo* victim = &first;
+            if (!MG_Pipe::MGPipeHandleIsNull(first.vaoHandle) &&
+                (MG_Pipe::MGPipeHandleIsNull(second.vaoHandle) ||
+                 second.bindings.frameSerial < first.bindings.frameSerial)) {
+                victim = &second;
+            }
+            victim->vaoHandle = handle;
+            victim->vaoKey = vao;
+            victim->vaoLifetimeId = vao->GetLifetimeId();
+            victim->contentHash = 0;
+            victim->layoutFactsValid = false;
+            // Unmatchable until a resolve completes (same rule as the legacy arm: a bailed-out
+            // resolve must never leave stale contents matchable).
+            victim->bindings.frameSerial = 0;
+            victim->bindings.indexFrameSerial = 0;
+            victim->bindings.indexBuffer = nullptr;
+            return victim;
+        }
+#endif
         // Multiplicative mix of the (16-byte-aligned) address; take high bits, they
         // carry the most entropy of a multiply.
         const Uint64 mixed = static_cast<Uint64>(reinterpret_cast<SizeT>(vao) >> 4) * 0x9E3779B97F4A7C15ull;
@@ -3551,6 +3728,13 @@ void main() {
         // its own is recycled, and a slot matched on a recycled address hands the new VAO
         // the dead one's resolved bindings.
         const Uint64 lifetimeId = vao->GetLifetimeId();
+        // Negative control C has NO consumer here. It is answered once, ahead of both arms, by
+        // the early return above, so a run that reaches this line has the knob off and the
+        // lifetime-id half of the compare is unconditional. A fourth consumer here would be a
+        // second site deciding the same question - what MagmaPipeAbaControlDefeatsIdentity
+        // exists to prevent - and a trap: narrow that early return later and this one would
+        // silently return to D18's retired semantics. If it is ever narrowed, ask the accessor
+        // here rather than re-reading MG_Config::Features.
         VaoDrawMemo& first = m_vaoDrawMemoTable[index];
         if (first.vaoKey == vao && first.vaoLifetimeId == lifetimeId) {
             return &first;
@@ -3619,7 +3803,7 @@ void main() {
         VaoDrawMemo* slot = nullptr;
         ResolvedVertexBindings* memo = nullptr;
         Uint64 vaoContentHash = 0;
-        const Bool vaoHashKnown = vao.GetBackendHashMemo(vaoContentHash);
+        const Bool vaoHashKnown = VaoContentHashIfKnown(vao, vaoContentHash);
         if (vaoHashKnown) {
             slot = LookupVaoDrawMemo(&vao);
             memo = &slot->bindings;
@@ -4818,6 +5002,7 @@ void main() {
         return MGB_CTX->GetRenderStateParameters().SampleMaskValue;
     }
 
+#if MOBILEGL_PIPE_LEGACY_MEMOS
     Uint64 VulkanRenderer::ComputePipelineStateHash(Uint32 colorAttachmentCount,
                                                     VkSampleCountFlagBits rasterizationSamples) const {
         // One bulk fetch instead of ~17 per-field accessor calls into MG_State: every
@@ -4908,6 +5093,23 @@ void main() {
         }
         return hash;
     }
+#endif // MOBILEGL_PIPE_LEGACY_MEMOS
+
+#if MOBILEGL_PIPE_PUSH && !MOBILEGL_PIPE_LEGACY_MEMOS
+    Uint64 VulkanRenderer::ComputePipelineSubsetStateHashFallback() const {
+        // The client's own hash, over the client's own definition of the pipeline subset - the
+        // seven pipeline chunks of the P2 chunk table, which is a strict SUPERSET of what
+        // ComputePipelineStateHash enumerated by hand. The render-pass facts it does not carry
+        // (colorAttachmentCount, the rasterization sample count, and through them the effective
+        // sample mask) are exactly the facts entry.renderPassHash separates, which is why the
+        // CSO handle can key this memo in the first place; this fallback inherits that argument
+        // unchanged.
+        //
+        // Only reached with no render-state CSO bound, and only in a build with no pre-handle
+        // arm to fall back to instead.
+        return MG_Pipe::MGPipeComputePipelineSubsetHash(MGB_CTX->GetRenderStateParameters());
+    }
+#endif
 
     // A program that runs a geometry shader AND captures transform feedback. Both halves are
     // link-time properties, so this is safe to fold into a pipeline keyed on the program hash.
@@ -4983,6 +5185,25 @@ void main() {
         // per-draw state flips (GL_BLEND toggles) would otherwise miss entries the memo holds.
         // The version only guards recomputing the hash - unchanged version, unchanged bytes.
         const Uint renderStateVersion = MGB_CTX->GetPipelineStateVersion();
+#if MOBILEGL_PIPE_PUSH
+        // P2 D12.1. Non-null means the client's render-state CSO handle is this draw's state
+        // key and the hash below is not computed at all; null means the pre-handle arm. The
+        // two arms' entries can never match each other: the handle arm stores hash 0 and a
+        // real handle, the legacy arm a real hash and the null handle, and the probe compares
+        // both components.
+        const MG_Pipe::MGPipeHandle renderStateCso = ResolveBoundRenderStateCso();
+        // Exactly one of the two state keys is live per draw, and the ternary short-circuits,
+        // so a draw on the handle arm neither hashes nor touches the fallback cache.
+        const Uint64 pipelineStateHash =
+            !MG_Pipe::MGPipeHandleIsNull(renderStateCso)
+                ? 0
+                : ResolveFallbackPipelineStateHash(renderStateVersion,
+                                                   renderPassEntry.colorAttachmentCount,
+                                                   renderPassEntry.sampleCount);
+#else
+        // THE PULL BUILD'S TEXT, statement for statement what the base ref has: G1 admits no
+        // resize of this function, and a helper the compiler merely inlines is not the same
+        // instruction schedule.
         if (!m_pipelineStateHashValid || m_pipelineStateHashVersion != renderStateVersion ||
             m_pipelineStateHashColorCount != renderPassEntry.colorAttachmentCount ||
             m_pipelineStateHashSampleCount != renderPassEntry.sampleCount) {
@@ -4994,12 +5215,16 @@ void main() {
             m_pipelineStateHashValid = true;
         }
         const Uint64 pipelineStateHash = m_pipelineStateHash;
+#endif
         for (Uint32 i = 0; i < m_pipelineMemoCount; ++i) {
             const PipelineMemoEntry& entry = m_pipelineMemo[i];
             if (entry.pipeline != VK_NULL_HANDLE && entry.mode == mode &&
                 entry.programHash == programObj.hash && entry.vertexInputHash == vertexLayoutHash &&
                 entry.renderPassHash == renderPassHash &&
                 entry.pipelineStateHash == pipelineStateHash &&
+#if MOBILEGL_PIPE_PUSH
+                entry.renderStateCso == renderStateCso &&
+#endif
                 entry.primitiveRestartEnable == primitiveRestartEnable &&
                 entry.transformFlags == transformFlags) {
                 if (MG_Util::PipeStats::Enabled()) {
@@ -5668,6 +5893,9 @@ void main() {
             entry.vertexInputHash = vertexLayoutHash;
             entry.renderPassHash = renderPassHash;
             entry.pipelineStateHash = pipelineStateHash;
+#if MOBILEGL_PIPE_PUSH
+            entry.renderStateCso = renderStateCso;
+#endif
             entry.primitiveRestartEnable = primitiveRestartEnable;
             entry.transformFlags = transformFlags;
             entry.pipeline = pipeline;
@@ -5917,6 +6145,21 @@ void main() {
         // One compare for the whole tail: see the gate's declaration in
         // DynamicStateShadow for why (version, extent, default-FBO flag) pins every
         // input the six Apply* below read.
+        //
+        // P2 D12.3: this read is RE-SOURCED, not re-shaped. Under MOBILEGL_PIPE_PUSH the
+        // accessor no longer walks into GLContext's RenderState - it returns
+        // PipeInputs::m_renderStateParametersVersion, which the applier publishes from
+        // MGPDynamicState::Version (set_dynamic_state) and MGPBindRenderState::Version
+        // (bind_render_state). So the gate now reads what the client PUSHED.
+        //
+        // What it does NOT do, and the P2 brief expects it to, is stop moving on a
+        // pipeline-only change. The tree settles that against the brief: bind_render_state
+        // carries m_version too and the applier publishes it, and it has to - Espryt's
+        // SyncRenderState uses the very same counter as its all-state change detector and G5
+        // forbids touching it, so a bind that rewrote the pipeline half while leaving the
+        // counter still would make Espryt skip re-syncing the blend state it just changed.
+        // The second-level DynamicTailKey compare below is therefore what actually absorbs a
+        // pipeline-only change, exactly as it did before P2: one key build, no vkCmd*.
         const Uint paramsVersion = MGB_CTX->GetRenderStateParametersVersion();
         if (shadow.dynamicTailValid && shadow.dynamicTailParamsVersion == paramsVersion &&
             shadow.dynamicTailExtentX == extent.x() && shadow.dynamicTailExtentY == extent.y() &&
@@ -6104,9 +6347,23 @@ void main() {
         // draw of a VAO-cycling stream (Minecraft chunk rendering) through the full
         // path, re-resolving descriptors and texture layouts nothing invalidated.
         const auto& vao = *MGB_CTX->GetBoundVertexArray();
+#if MOBILEGL_PIPE_PUSH
+        // P2 D12.4: the handle replaces the (address, lifetime id) pair here too - one
+        // compare instead of two, and the same identity the VAO draw memo is keyed on, so
+        // the two cannot disagree about whether "the VAO moved". The config version stays:
+        // it answers a different question (did this same object's layout change).
+        const Bool vaoMoved =
+            MagmaPipeTrackHArmIsHandles(MG_Pipe::kMGPipeSubsystemMagmaVertexInput)
+                ? (!(ResolveVaoHandle(vao) == snap.vaoHandle) ||
+                   vao.GetConfigVersion() != snap.vaoConfigVersion)
+                : (static_cast<const void*>(&vao) != snap.vao ||
+                   vao.GetLifetimeId() != snap.vaoLifetimeId ||
+                   vao.GetConfigVersion() != snap.vaoConfigVersion);
+#else
         const Bool vaoMoved =
             static_cast<const void*>(&vao) != snap.vao || vao.GetLifetimeId() != snap.vaoLifetimeId ||
             vao.GetConfigVersion() != snap.vaoConfigVersion;
+#endif
         const auto& drawFbo =
             MGB_CTX->GetFramebufferBindingSlot(FramebufferTarget::Draw).GetBoundObject();
         if (static_cast<const void*>(drawFbo.get()) != snap.drawFbo ||
@@ -6194,7 +6451,7 @@ void main() {
             Uint64 auxMasks = 0;
             Bool factsKnown = false;
             Uint64 contentHash = 0;
-            if (vao.GetBackendHashMemo(contentHash)) {
+            if (VaoContentHashIfKnown(vao, contentHash)) {
                 const VaoDrawMemo* vaoMemo = LookupVaoDrawMemo(&vao);
                 if (vaoMemo->layoutFactsValid && vaoMemo->contentHash == contentHash) {
                     vaoLayoutHash = vaoMemo->layoutHash;
@@ -6211,7 +6468,7 @@ void main() {
                 auxMasks = VertexInputStateFactory::PackVertexInputAuxMasks(
                     vertexInputState.unsupportedAttribMask, vertexInputState.attributeLocationMask);
                 Uint64 stampedHash = 0;
-                if (vao.GetBackendHashMemo(stampedHash)) {
+                if (VaoContentHashIfKnown(vao, stampedHash)) {
                     VaoDrawMemo* vaoMemo = LookupVaoDrawMemo(&vao);
                     vaoMemo->contentHash = stampedHash;
                     vaoMemo->layoutHash = vaoLayoutHash;
@@ -6332,6 +6589,18 @@ void main() {
             // what lets a per-draw GL_BLEND toggle alternate between two memo entries
             // instead of missing forever on a monotonic version. A miss falls through
             // to the full lookup.
+#if MOBILEGL_PIPE_PUSH
+            // Same arm selector as GetOrCreatePipeline's probe (P2 D12.1); this site is the
+            // fast path's copy of it, and the two must key identically or the fast path would
+            // hand back a pipeline the full path would not have matched.
+            const MG_Pipe::MGPipeHandle renderStateCso = ResolveBoundRenderStateCso();
+            const Uint64 pipelineStateHash =
+                !MG_Pipe::MGPipeHandleIsNull(renderStateCso)
+                    ? 0
+                    : ResolveFallbackPipelineStateHash(renderStateVersion, snap.renderPassColorCount,
+                                                       snap.renderPassSampleCount);
+#else
+            // The pull build's text, statement for statement (see GetOrCreatePipeline).
             if (!m_pipelineStateHashValid || m_pipelineStateHashVersion != renderStateVersion ||
                 m_pipelineStateHashColorCount != snap.renderPassColorCount ||
                 m_pipelineStateHashSampleCount != snap.renderPassSampleCount) {
@@ -6342,6 +6611,8 @@ void main() {
                 m_pipelineStateHashSampleCount = snap.renderPassSampleCount;
                 m_pipelineStateHashValid = true;
             }
+            const Uint64 pipelineStateHash = m_pipelineStateHash;
+#endif
             const auto memoTransformFlags =
                 ProgramFactory::CompileOptionFlags(snap.resolvedTransformFlags);
             for (Uint32 i = 0; i < m_pipelineMemoCount; ++i) {
@@ -6349,7 +6620,10 @@ void main() {
                 if (entry.pipeline != VK_NULL_HANDLE && entry.mode == mode &&
                     entry.programHash == programObj.hash && entry.vertexInputHash == vaoLayoutHash &&
                     entry.renderPassHash == snap.renderPassHash &&
-                    entry.pipelineStateHash == m_pipelineStateHash &&
+                    entry.pipelineStateHash == pipelineStateHash &&
+#if MOBILEGL_PIPE_PUSH
+                    entry.renderStateCso == renderStateCso &&
+#endif
                     entry.primitiveRestartEnable == drawPrimitiveRestartEnable &&
                     entry.transformFlags == memoTransformFlags) {
                     pipeline = entry.pipeline;
@@ -6383,6 +6657,15 @@ void main() {
         snap.bindGeneration = bindGeneration;
         snap.vao = static_cast<const void*>(&vao);
         snap.vaoLifetimeId = vao.GetLifetimeId();
+#if MOBILEGL_PIPE_PUSH
+        // Guarded by the SUBSYSTEM, not only by the build switch: with bit 6 clear the field
+        // is dead (vaoMoved takes the address/lifetime-id branch), and minting a handle for it
+        // would put this package's cost inside MOBILEGL_PIPE_PUSH=0 - the all-pull control arm
+        // D14 defines as reproducing P1 exactly, and the arm D.4.3's T2 is measured on.
+        if (MagmaPipeTrackHArmIsHandles(MG_Pipe::kMGPipeSubsystemMagmaVertexInput)) {
+            snap.vaoHandle = ResolveVaoHandle(vao);
+        }
+#endif
         snap.vaoConfigVersion = vao.GetConfigVersion();
         snap.vaoLayoutHash = vaoLayoutHash;
         snap.pipeline = pipeline;
@@ -6964,6 +7247,13 @@ void main() {
                 snap.programVersion = program.GetBackendStateVersion();
                 snap.vao = &vao;
                 snap.vaoLifetimeId = vao.GetLifetimeId();
+#if MOBILEGL_PIPE_PUSH
+                // Subsystem-guarded for the same reason as the other stamping site: the field
+                // is dead with bit 6 clear, and MOBILEGL_PIPE_PUSH=0 has to be P1 exactly.
+                if (MagmaPipeTrackHArmIsHandles(MG_Pipe::kMGPipeSubsystemMagmaVertexInput)) {
+                    snap.vaoHandle = ResolveVaoHandle(vao);
+                }
+#endif
                 snap.vaoConfigVersion = vao.GetConfigVersion();
                 snap.drawFbo = drawFbo.get();
                 snap.drawFboLifetimeId = drawFbo->GetLifetimeId();
@@ -12524,6 +12814,14 @@ void main() {
         if (m_vertexInputStateFactory) {
             m_vertexInputStateFactory->OnFrameBoundary();
         }
+#if MOBILEGL_PIPE_PUSH
+        // Reclaim {slot, gen} for objects that have not been drawn for a long time, on the same
+        // cadence and the same retirement age as the entries those slots key. This is the
+        // stand-in for the frontend death notification P2 has no hook for, and it is what keeps
+        // the mint's footprint the LIVE working set rather than every object ever created
+        // (review v2 MAJOR 1 / MAJOR 3).
+        m_pipeIdentity.OnFrameBoundary();
+#endif
         if (m_samplerManager) {
             m_samplerManager->OnFrameBoundary();
         }
@@ -12889,6 +13187,9 @@ void main() {
             InvalidateSetupDrawSnapshots();
         }
         m_vertexInputStateFactory->OnFrameBoundary();
+#if MOBILEGL_PIPE_PUSH
+        m_pipeIdentity.OnFrameBoundary();
+#endif
         m_samplerManager->OnFrameBoundary();
         auto& frame = m_frameContext.GetCurrent();
         auto* activeRenderPass = VkRenderPassManager::GetActiveRenderPass();

MobileGL/MG_Backend/DirectVulkan/Renderer/VulkanRenderer.h

@@ -9,6 +9,7 @@
 #pragma once
 #include "Config.h"
 #include "FrameContext.h"
+#include "MagmaPipeArms.h"
 #include "PipelineFactory.h"
 #include "ProgramFactory.h"
 #include "SwapchainObject.h"
@@ -24,7 +25,13 @@
 #include "MG_Util/Math/VectorTypes.h"
 #include <Includes.h>
 #include <MG_Backend/BackendObject.h>
+#include <MG_Pipe/MGPipeHandles.h>
 #include <MG_Util/SelfTest/PrimitivesGeneratedNoXfbProbe.h>
+#if MOBILEGL_PIPE_PUSH
+// The applier's CSO store: MGPipeApplier().BoundRenderStateCso is what the pipeline memo
+// keys on after P2 (D12.1). Push-only, so the pull build's include graph is unchanged.
+#include <MG_Pipe/PipeApply.h>
+#endif
 #include <vk_mem_alloc.h>
 
 #include "../VkIncludes.h"
@@ -820,12 +827,31 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             Uint64 programHash = 0;
             Uint64 vertexInputHash = 0;
             Uint64 renderPassHash = 0;
-            // VALUE hash of the pipeline-relevant fixed-function state (see
-            // ComputePipelineStateHash), not the monotonic pipeline-state version:
-            // the version never repeats, so a per-draw GL_BLEND toggle would miss
-            // all entries forever even though the state alternates between two
-            // values the memo already holds.
+            // The PRE-HANDLE arm's key component (P2 brief D12.1), and 0 in every entry the
+            // handle arm mints. VALUE hash of the pipeline-relevant fixed-function state (see
+            // ComputePipelineStateHash), not the monotonic pipeline-state version: the version
+            // never repeats, so a per-draw GL_BLEND toggle would miss all entries forever even
+            // though the state alternates between two values the memo already holds.
             Uint64 pipelineStateHash = 0;
+#if MOBILEGL_PIPE_PUSH
+            // The HANDLE arm's key component, and the whole of D12.1: the CLIENT already
+            // hashed the pipeline subset of RenderStateParameters and minted a content-
+            // addressed CSO for it (MG_Pipe/MGPipeRenderStateSpans.h, MG_Impl/Pipe/CsoCache),
+            // so re-hashing the same 396 bytes here was work the boundary had already done.
+            // Two draws share a CSO handle exactly when their pipeline bytes are equal, and
+            // the client's subset is a strict SUPERSET of what ComputePipelineStateHash read,
+            // so the handle discriminates at least as finely as the hash it replaces.
+            //
+            // renderPassHash STAYS beside it and is what keeps this key complete: the CSO
+            // carries GL state only, while colorAttachmentCount and the rasterization sample
+            // count - which ComputePipelineStateHash folded in through its signature and
+            // through ResolveEffectiveSampleMask - are render-pass facts that the render-pass
+            // hash already separates.
+            //
+            // Null in an entry minted by the legacy arm, so entries of the two arms can never
+            // match each other: the compare below tests BOTH components.
+            MG_Pipe::MGPipeHandle renderStateCso = MG_Pipe::kMGPipeNullHandle;
+#endif
             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
@@ -839,19 +865,142 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         PipelineMemoEntry m_pipelineMemo[kPipelineMemoSize];
         Uint32 m_pipelineMemoCount = 0;
         Uint32 m_pipelineMemoNext = 0;
-        // Hash of every fixed-function GL state the pipeline payload reads that the
-        // memo key's other fields (mode / program / vertex input / render pass /
-        // transform flags) do not already pin down. Equal hash under an equal rest
-        // of key => byte-identical PipelineCreatePayload. Cached per pipeline-state
+
+#if MOBILEGL_PIPE_PUSH
+        // P2 D12.1's arm selector, and the whole of the pipeline memo's re-key. Returns the
+        // render-state CSO this draw is keyed on, or the null handle when the pre-handle arm
+        // is the one that runs.
+        //
+        // Under the handle arm the memo's state key IS this handle. The client hashed those
+        // 396 pipeline bytes when it minted the CSO (MGPipeComputePipelineSubsetHash), so
+        // recomputing an overlapping hash here was work the boundary had already done; the
+        // client's pipeline subset is a strict SUPERSET of what ComputePipelineStateHash read,
+        // so the handle discriminates at least as finely as the hash it replaces. What the
+        // handle does NOT carry is the render-pass side - colorAttachmentCount and the
+        // rasterization sample count, which ComputePipelineStateHash folded in through its
+        // signature and through ResolveEffectiveSampleMask - and that is exactly why
+        // entry.renderPassHash stays in the key beside it.
+        //
+        // The arm is live only when the render-state subsystem is migrated in this run AND the
+        // client has actually bound a CSO. The second half is not belt and braces: a tree whose
+        // tracker does not emit create/bind_render_state yet has no handle to key on, and
+        // delete_render_state clears the binding (MG_Pipe/PipeApply.cpp), so the null handle is
+        // reachable on any tree. Keying every draw on it would alias every render state onto
+        // one memo entry, so a null handle means "fall back to a state hash" - never an abort,
+        // and never a per-draw consultation of the legacy-memo lever: bit 0 is not a Track-H
+        // subsystem (D14 labels only bits 5 and 6 that), and the lever's Fatal is a STARTUP
+        // one, in MagmaPipeValidateSubsystemConfiguration.
+        //
+        // The fallback is warned ONCE rather than logged at debug, and that is deliberate: a
+        // silent fallback is what makes "the CSO arm never ran" easy to miss. W is compiled in
+        // at every shipped log level.
+        //
+        // The latch is a plain member bool, NOT MGLOG_W_ONCE. MOBILEGL_LOG_ONCE_INTERNAL
+        // (MG_Util/Debug/Log.h) is an UNCONDITIONAL std::atomic_flag::test_and_set - a locked
+        // xchg, executed on every evaluation, not "one static bool test" as an earlier round of
+        // this comment claimed - and this site is on the per-draw pipeline path in the very
+        // configuration that reaches it (no tracker: every draw). ROADMAP.md:7 forbids leaving
+        // instrumentation on a hot path, so the once-ness is one non-atomic, always-predicted
+        // load of a member that is false exactly once. Single-threaded like the rest of the
+        // renderer, and per renderer rather than per process, which is also the right scope: a
+        // second context that never binds a CSO deserves to say so.
+        //
+        // What the absence of this warning from a run's log proves, EXACTLY: that no draw took
+        // the fallback WHILE bit 0 was set. With kMGPipeSubsystemRenderState clear the function
+        // returns before the latch, so absence proves nothing at all - and no draw is keyed on a
+        // handle either. Grep the mask out of the log beside it (review v2 minor 3).
+        //
+        // Push-only by construction: the pull build does not compile this function at all, so
+        // its two callers are statement-for-statement what they were (G1).
+        //
+        // [routed to the integrator, review v2 minor 11] MG_Pipe::MGPipeApplier() is ONE
+        // process-global applier (MG_Pipe/PipeApply.cpp), not the per-context CSO store D2
+        // specifies. In a multi-context process this reads whatever CSO another context last
+        // bound. The defect is package A's and the fix belongs there; Magma is its only P2
+        // consumer, so it is named here rather than left for both reviews to assume the other
+        // caught it.
+        MG_Pipe::MGPipeHandle ResolveBoundRenderStateCso() const {
+            if (!MagmaPipeSubsystemOn(MG_Pipe::kMGPipeSubsystemRenderState)) {
+                return MG_Pipe::kMGPipeNullHandle;
+            }
+            const MG_Pipe::MGPipeHandle boundCso = MG_Pipe::MGPipeApplier().BoundRenderStateCso;
+            if (MG_Pipe::MGPipeHandleIsNull(boundCso) && !m_pipelineCsoFallbackWarned) {
+                m_pipelineCsoFallbackWarned = true;
+                MGLOG_W("MGPipe: kMGPipeSubsystemRenderState is on but no render-state CSO is "
+                        "bound; the pipeline memo is running on a state hash, not on the CSO "
+                        "handle (no tracker on this build, or a draw between "
+                        "delete_render_state and the next bind)");
+            }
+            return boundCso;
+        }
+        // Latch for the warning above. Mutable because the resolve is const and the latch is
+        // not part of the renderer's observable state.
+        mutable Bool m_pipelineCsoFallbackWarned = false;
+        // The memo key's STATE-HASH half, for a draw that has no CSO handle to key on: the
+        // pre-handle arm, and the fallback of D12.1's handle arm. Cached on the pipeline-state
+        // version plus the two render-pass facts the hash's inputs depend on, so an unchanged
+        // (version, colorAttachmentCount, sampleCount) proves the bytes are unchanged.
+        //
+        // [deviation from D12.1] The brief deletes this gate and its cached fields outright.
+        // They cannot go while a no-CSO draw is reachable - and it is, on any tree: a draw
+        // between delete_render_state and the next bind has no handle. On a tree whose tracker
+        // binds a CSO these five words are written once and never read again; they retire for
+        // real when the pull path does, at P13.
+        Uint64 ResolveFallbackPipelineStateHash(Uint renderStateVersion, Uint32 colorAttachmentCount,
+                                                VkSampleCountFlagBits rasterizationSamples) {
+            if (!m_pipelineStateHashValid || m_pipelineStateHashVersion != renderStateVersion ||
+                m_pipelineStateHashColorCount != colorAttachmentCount ||
+                m_pipelineStateHashSampleCount != rasterizationSamples) {
+#if MOBILEGL_PIPE_LEGACY_MEMOS
+                m_pipelineStateHash =
+                    ComputePipelineStateHash(colorAttachmentCount, rasterizationSamples);
+#else
+                m_pipelineStateHash = ComputePipelineSubsetStateHashFallback();
+#endif
+                m_pipelineStateHashVersion = renderStateVersion;
+                m_pipelineStateHashColorCount = colorAttachmentCount;
+                m_pipelineStateHashSampleCount = rasterizationSamples;
+                m_pipelineStateHashValid = true;
+            }
+            return m_pipelineStateHash;
+        }
+#endif // MOBILEGL_PIPE_PUSH
+#if MOBILEGL_PIPE_PUSH && !MOBILEGL_PIPE_LEGACY_MEMOS
+        // The same answer as ComputePipelineStateHash, computed from the P2 chunk table
+        // instead of from a hand-written field list, for the build that compiles no
+        // pre-handle arm (cmake -DMOBILEGL_PIPE_LEGACY_MEMOS=OFF). It is the CLIENT's own
+        // hash function - MGPipeComputePipelineSubsetHash over the 396 pipeline bytes - so a
+        // draw keyed on it and a draw keyed on a CSO handle are keyed on the same equivalence
+        // class of state, and the render-pass facts stay separated by renderPassHash either
+        // way. This is what makes the no-legacy build RUNNABLE rather than a configuration
+        // that aborts on the first draw that arrives without a CSO.
+        Uint64 ComputePipelineSubsetStateHashFallback() const;
+#endif
+#if MOBILEGL_PIPE_LEGACY_MEMOS
+        // THE PRE-HANDLE ARM (P2 brief D12.1 / D14). Hash of every fixed-function GL state the
+        // pipeline payload reads that the memo key's other fields (mode / program / vertex
+        // input / render pass / transform flags) do not already pin down. Equal hash under an
+        // equal rest of key => byte-identical PipelineCreatePayload. Cached per pipeline-state
         // 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.
+        //
+        // The handle arm computes none of this: the client hashed the same bytes when it
+        // minted the CSO, so all five cached-hash members below exist only to avoid a
+        // re-hash the handle arm never performs.
         Uint64 ComputePipelineStateHash(Uint32 colorAttachmentCount,
                                         VkSampleCountFlagBits rasterizationSamples) const;
+#endif
         // 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.
+        // and the pipeline-state memo word so the two cannot disagree. NOT part of the legacy
+        // arm: it is a PAYLOAD computation that depends on rasterizationSamples, so it survives
+        // the re-key and keeps reading Multisample / SampleMask / SampleMaskValue out of the
+        // working block.
         Uint32 ResolveEffectiveSampleMask(VkSampleCountFlagBits rasterizationSamples) const;
+        // ResolveFallbackPipelineStateHash's cache. Written once and never read again on a
+        // build whose client binds a render-state CSO; see that function for why it survives
+        // the re-key at all.
         Uint m_pipelineStateHashVersion = 0;
         Uint32 m_pipelineStateHashColorCount = 0;
         // The sample count the cached hash was computed at. A pipeline-state input now depends on
@@ -877,7 +1026,8 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // Drops every memoized pipeline handle. Required at command-buffer
         // boundaries and whenever any pipeline may have been destroyed. Also drops
         // the cached pipeline-state hash: the same boundaries can retire the GL
-        // context whose monotonic version the cache is keyed on.
+        // context whose monotonic version the cache is keyed on. The handle arm has no
+        // such cache to drop - a CSO handle is not derived from a monotonic version.
         void InvalidatePipelineMemo() {
             m_pipelineMemoCount = 0;
             m_pipelineMemoNext = 0;
@@ -968,6 +1118,13 @@ namespace MobileGL::MG_Backend::DirectVulkan {
             // common shape), and "the VAO did not move" would then skip the layout
             // re-resolve for a different VAO.
             Uint64 vaoLifetimeId = 0;
+#if MOBILEGL_PIPE_PUSH
+            // P2 D12.4: the handle arm's answer to the same question, and one compare rather
+            // than the pair above. Kept BESIDE them rather than replacing them because the
+            // pre-handle arm is still compiled (MOBILEGL_PIPE_LEGACY_MEMOS) and this snapshot
+            // is a value struct, not a wire type.
+            MG_Pipe::MGPipeHandle vaoHandle = MG_Pipe::kMGPipeNullHandle;
+#endif
             Uint32 vaoConfigVersion = 0;
             const void* drawFbo = nullptr;
             // Never-reused lifetime id beside the raw pointer + Uint16 version: a
@@ -1242,6 +1399,13 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         //  - bindings revalidates per draw exactly as before (frame serial, content
         //    hash, per-binding live buffer pointers and slice epochs).
         struct alignas(64) VaoDrawMemo {
+#if MOBILEGL_PIPE_PUSH
+            // P2 D12.4: the handle arm's key, and the ONLY key it needs. {slot, gen} is an
+            // identity, so the pointer-plus-lifetime-id pair below stops being a key here;
+            // the slot also picks the table entry, so the address hash and the two-way probe
+            // go with it. Null in an entry that has never been claimed.
+            MG_Pipe::MGPipeHandle vaoHandle = MG_Pipe::kMGPipeNullHandle;
+#endif
             const MG_State::GLState::VertexArrayObject* vaoKey = nullptr;
             // The VAO's never-reused lifetime id, checked alongside vaoKey. The pointer
             // ALONE is not an identity: a deleted VAO's heap address is handed straight
@@ -1268,8 +1432,48 @@ namespace MobileGL::MG_Backend::DirectVulkan {
         // fixed table also makes every VaoDrawMemo/ResolvedVertexBindings pointer
         // stable for the duration of a draw, which the EBO memo handoff
         // (m_currentDrawResolvedEntry) relies on.
+        //
+        // [deviation from D12.4, deliberate and narrow] The brief asks for a grow-on-demand
+        // Vector. This one stays FIXED at exactly the capacity and exactly the 2-way victim
+        // rule it has on the base ref, and only its KEY changes (a {slot, gen} handle instead
+        // of a hashed heap address plus a lifetime id). Two reasons, and the second is the
+        // whole of review v2's MAJOR 1:
+        //   * a VaoDrawMemo is ~450 B (ResolvedVertexBindings dominates), so growing this
+        //     table with the live VAO set is megabytes on a platform with an LMK, where the
+        //     other two memos are 48 B and can afford it;
+        //   * this is the ONLY one of the three memos that had a capacity before this package.
+        //     Losing an entry here costs a vertex-binding re-resolve, exactly what losing it
+        //     cost on the base ref, so at any working-set size this table is no worse than what
+        //     it replaces - and strictly better below capacity, where the handle is a bijection
+        //     with the slot and the two-way probe never collides at all. The other two memos
+        //     (VertexInputStateFactory::m_vaoMemos) had NO capacity, so they keep having none.
         static constexpr Uint32 kVaoDrawMemoSlotCount = 2048; // power of two
         Vector<VaoDrawMemo> m_vaoDrawMemoTable;
+#if MOBILEGL_PIPE_PUSH
+        // The renderer's {slot, gen} mint, shared with its VertexInputStateFactory so both
+        // derive the same handle for the same VAO. Per renderer, never a process-global: a
+        // global would share one table and one reclamation clock across two live contexts and
+        // outlive every one of them (review v2 minor 4).
+        MagmaPipeIdentityTables m_pipeIdentity;
+        // The VAO's {slot, gen}. A one-entry memo hit for every acquisition after a draw's
+        // first, so there is no second memo in front of it here.
+        MG_Pipe::MGPipeHandle ResolveVaoHandle(const MG_State::GLState::VertexArrayObject& vao) {
+            return m_pipeIdentity.HandleOf(MG_Pipe::MGPipeKind::VertexElementsCso,
+                                           vao.GetLifetimeId());
+        }
+#endif
+        // "Is this VAO's content hash already memoized?", asked of whichever side owns the
+        // memo (P2 D12.5). Force-inlined and defined in the class body so that the PULL
+        // build's three readers keep compiling to the very same two loads they always did -
+        // G1 admits no resize, and an out-of-line call here would be one.
+        [[gnu::always_inline]] inline Bool VaoContentHashIfKnown(
+            const MG_State::GLState::VertexArrayObject& vao, Uint64& outHash) const {
+#if MOBILEGL_PIPE_PUSH
+            return m_vertexInputStateFactory->TryGetMemoizedHash(vao, outHash);
+#else
+            return vao.GetBackendHashMemo(outHash);
+#endif
+        }
         // Finds the slot holding `vao`, or recycles the older of its two candidate
         // slots into an empty memo keyed on `vao`. Never returns null.
         VaoDrawMemo* LookupVaoDrawMemo(const MG_State::GLState::VertexArrayObject* vao);

MobileGL/MG_Backend/MGPipe/PipeInputs.h

@@ -609,6 +609,13 @@ namespace MobileGL::MG_Pipe {
         // 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;
+        // The other door, and the one that exists because of what this block IS after P2:
+        // the server's working RenderStateParameters. MG_Pipe/PipeApply.cpp scatters
+        // bind_render_state's and set_dynamic_state's chunks straight into m_renderState,
+        // which is why DirectGLES' SyncRenderState is not one line changed. It deliberately
+        // does NOT stamp the poison generations - a stamp says "the filler published this
+        // for THIS verb", which is the walk's statement, not the applier's.
+        friend struct MGPipeApplyAccess;
 
         // ---- identity ----
         const void* m_contextIdentity = nullptr;
@@ -681,7 +688,22 @@ namespace MobileGL::MG_Pipe {
 
     // 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{};
+    //
+    // LEAK-AT-EXIT STORAGE, and it is the same rule Init.cpp and GlobalObjects.cpp state for
+    // pGLContext and pActiveBackendObject: "a process that exits without eglTerminate simply
+    // leaks the global singletons to the OS instead of running destructors during static
+    // teardown". This block breaks that rule if it is a value, because its O-class members
+    // are SharedPtrs to FRONTEND objects: a VertexArrayObject that the application deleted
+    // while it was bound has its last reference here, and destroying this block from
+    // __run_exit_handlers therefore runs ~VertexArrayObject -> ~BufferObject at exit. Those
+    // destructors are not exit-safe and cannot be made so - they reach the client's slot
+    // allocator, the resource tracker, the vertex-input emitter, the applier AND, through
+    // MGPipeApplyResourceDestroy, the backend's own twin tables, deferred-release queue,
+    // buffer pool and driver entry points, every one of which is either already destroyed or
+    // about to be. So the reference is never dropped: nothing here can start such a chain.
+    // A live context releases these SharedPtrs the ordinary way, at the fill point.
+    // (P3a; the exit-time heap corruption this closes is p3a-results/exit-order-v1.md.)
+    inline PipeInputs& gPipeInputs = *new PipeInputs();
 
     // Every field has storage or is forwarded, and nothing else.
 #define MGP_INPUT_COUNT_ONE(Field, Member) +1

MobileGL/MG_Benchmark/Driver/CMakeLists.txt

@@ -11,5 +11,51 @@ endif()
 add_executable(DriverBench DriverBench.c)
 target_link_libraries(DriverBench PRIVATE dl)
 
+# WHY EVERY ENTRY HERE CARRIES A PASS_REGULAR_EXPRESSION.
+#
+# DriverBench prints one CSV row per case it ran and exits 0 whatever it ran. Before this, a ctest
+# entry naming a case therefore could not answer the only question it exists to ask: an argument
+# matching nothing in kBenchCases selected no case, printed only the header row, and still exited
+# 0. DriverBench.c now refuses an unknown case name (exit 2), which closes it at the source - but
+# the entry must be able to go red for the reason it exists WITHOUT depending on that check
+# staying in the binary, so each entry also requires the case's own output row to appear.
+#
+# The regex is what a healthy run of that case prints and nothing else does: the case name at the
+# start of a line, then the frames / ops-per-frame / median-ms / ns-per-op / fps columns
+# (run_case()). A rename, a drop from kBenchCases, a boot_egl() failure or
+# a crash part-way through the case all remove that row and turn the entry red.
+#
+# Note that a PASS_REGULAR_EXPRESSION makes ctest ignore the process exit code (cmCTestRunTest:
+# success is `retVal == 0 || !RequiredRegularExpressions.empty()`), which is why the row itself
+# has to be the evidence rather than a companion to the rc.
 add_test(NAME DriverBench COMMAND DriverBench draw_tiny)
-set_tests_properties(DriverBench PROPERTIES LABELS benchmark)
+# draw_tiny's a/ops scale with $DRIVERBENCH_DRAWS (main()), so only the shape of
+# the row is pinned here, not the column values.
+set_tests_properties(DriverBench PROPERTIES
+    LABELS benchmark
+    PASS_REGULAR_EXPRESSION "(^|\n)draw_tiny,[0-9]+,[0-9]+,[0-9.]+,[0-9.]+,[0-9.]+")
+
+# The Blaze3D blend toggle, as its own entry.
+#
+# mc_state_toggle is glEnable(GL_BLEND) / glBlendFuncSeparate / glDrawElements /
+# glDisable(GL_BLEND) / glDrawElements, 46 times - the measured vanilla-frame rate, and the exact
+# shape ROADMAP.md writes down as the microbenchmark P2 owes the GO/NO-GO. It is the workload the
+# whole "push at validate, not in the setter" decision was made for: a per-setter design pays for
+# every toggle, and a CSO that is minted twice and then reused pays for none of them.
+#
+# The case has existed in kBenchCases since P0 and nothing ran it, so nothing noticed if it broke.
+# Exposing it costs about 1.2 s inside an existing three-minute job, and it means the number the
+# P2 report quotes comes from a case CI has been executing all along rather than from a code path
+# whose first run is the day it is measured.
+#
+# Like the entry above, this runs against whatever $DRIVERBENCH_EGL_LIB names (the system driver
+# when unset) - the ctest entry is a "does this case still run" gate, not the measurement. The
+# measurement is run_driver_bench.sh against each of {native, espryt, magma}.
+add_test(NAME DriverBenchStateToggle COMMAND DriverBench mc_state_toggle)
+# The ops-per-frame column is pinned to 46 here, unlike the entry above: the mc_* cases are
+# excluded from the $DRIVERBENCH_DRAWS scaling on purpose ("the mc_* rates are measured and must
+# not move, or the numbers stop being comparable", main()), so 46 toggles per frame
+# is part of what "this case still runs" means. Change the workload and this entry says so.
+set_tests_properties(DriverBenchStateToggle PROPERTIES
+    LABELS benchmark
+    PASS_REGULAR_EXPRESSION "(^|\n)mc_state_toggle,[0-9]+,46,[0-9.]+,[0-9.]+,[0-9.]+")

MobileGL/MG_Benchmark/Driver/DriverBench.c

@@ -476,6 +476,28 @@ int main(int argc, char** argv) {
     if (getenv("DRIVERBENCH_FRAMES")) g_frames = atoi(getenv("DRIVERBENCH_FRAMES"));
     if (getenv("DRIVERBENCH_SPRITES")) g_mixSprites = atol(getenv("DRIVERBENCH_SPRITES"));
 
+    /* A requested case name that matches nothing used to select nothing, print the header row and
+     * exit 0 - so a caller that names a case (run_driver_bench.sh, and the two ctest entries in
+     * CMakeLists.txt) could not tell "the case ran" from "the case has been renamed or deleted".
+     * Refuse it here, before any GL work, so the refusal reaches a caller that has no display
+     * either, and name what does exist so the fix is obvious. */
+    int unknownCases = 0;
+    for (int j = 1; j < argc; ++j) {
+        int known = 0;
+        for (int i = 0; i < kBenchCaseCount; ++i)
+            if (strcmp(argv[j], kBenchCases[i].name) == 0) known = 1;
+        if (!known) {
+            fprintf(stderr, "DriverBench: no case named '%s'\n", argv[j]);
+            unknownCases = 1;
+        }
+    }
+    if (unknownCases) {
+        fprintf(stderr, "DriverBench: the %d cases in kBenchCases are:\n", kBenchCaseCount);
+        for (int i = 0; i < kBenchCaseCount; ++i)
+            fprintf(stderr, "  %s\n", kBenchCases[i].name);
+        return 2;
+    }
+
     if (boot_egl()) return 1;
     build_resources();
 

MobileGL/MG_Impl/GLImpl/Drawing/GL_Drawing.cpp

@@ -659,6 +659,7 @@ namespace MobileGL::MG_Impl::GLImpl {
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
         if (ConditionalRenderDiscardsCommand()) return;
+        MGP_SET_BASE_INSTANCE(baseinstance);
         MGP_FILL(DrawElementsInstancedBaseVertexBaseInstance);
         MG_Backend::gBackendFunctionsTable.GL.DrawElementsInstancedBaseVertexBaseInstance(
             mode, count, type, indices, instancecount, basevertex, baseinstance);
@@ -681,6 +682,7 @@ namespace MobileGL::MG_Impl::GLImpl {
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
         if (ConditionalRenderDiscardsCommand()) return;
+        MGP_SET_BASE_INSTANCE(baseinstance);
         MGP_FILL(DrawElementsInstancedBaseInstance);
         MG_Backend::gBackendFunctionsTable.GL.DrawElementsInstancedBaseInstance(mode, count, type, indices,
                                                                                 instancecount, baseinstance);
@@ -710,6 +712,7 @@ namespace MobileGL::MG_Impl::GLImpl {
         ZoneScopedC(TRACY_ZONECOLOR_BACKEND);
 #endif
         if (ConditionalRenderDiscardsCommand()) return;
+        MGP_SET_BASE_INSTANCE(baseinstance);
         MGP_FILL(DrawArraysInstancedBaseInstance);
         MG_Backend::gBackendFunctionsTable.GL.DrawArraysInstancedBaseInstance(mode, first, count, instancecount,
                                                                               baseinstance);

MobileGL/MG_Impl/GLImpl/Framebuffer/GL_Framebuffer.cpp

@@ -16,6 +16,14 @@
 #include <MG_Impl/GLImpl/Getter/GL_Getter.h>
 #include <MG_State/GLState/ErrorState/Error.h>
 #include <MG_Impl/Pipe/PipeFill.h>
+#if MOBILEGL_PIPE_PUSH
+// P4a, ID-19(c). This file is the ONLY place every DSA framebuffer entry point lives, and the
+// emitter it reaches is this package's own header rather than a declaration in one of the
+// contract's: MG_Pipe/PipeMutation.h is the door MG_State has into the client and carries no
+// framebuffer row, and MG_Impl/GLImpl and MG_Impl/Pipe are the same layer (this file already
+// includes MG_Impl/Pipe/PipeFill.h for MGP_FILL).
+#include <MG_Impl/Pipe/FramebufferEmit.h>
+#endif
 #include <MG_Util/Converters/GLToStr/GLEnumConverter.h>
 #include <MG_Util/Converters/GLToMG/TextureEnumConverter.h>
 #include <MG_Util/Converters/MGToMG/TextureEnumConverter.h>
@@ -613,6 +621,35 @@ namespace MobileGL::MG_Impl::GLImpl {
 
             framebufferObject->AttachTexture(attachmentType, textureObject, textureUploadTarget, level, 0, layered);
         }
+#if MOBILEGL_PIPE_PUSH
+        // P4a, ID-19(c): ANY FRAMEBUFFER THE SERVER IS ABOUT TO RECEIVE BY NAME HAS A RECORD.
+        //
+        // The applier keeps framebuffer records PER OBJECT, keyed by the handle - but before
+        // ID-19 it held only the two BOUND-target records, and the emitter only ever built them
+        // at the validate point out of the two bindings. So glClearNamedFramebufferfv(fbo) or
+        // glBlitNamedFramebuffer(..., fbo, ...) on an fbo bound to NEITHER binding reached a
+        // backend that minted a fresh driver framebuffer with no attachments, found no record
+        // for it, declined, and issued the clear against it anyway: GL_INVALID_FRAMEBUFFER_-
+        // OPERATION and nothing cleared, where the legacy arm cleared correctly.
+        //
+        // TWO CLASSES OF SITE call this, and both are "the point at which the object is final
+        // for this call": the five CONSUMERS (blit and the four clears) publish immediately
+        // before MGP_FILL, so the record precedes the verb that hands the object over and a
+        // later bound-target record for the same object still wins; the ten MUTATORS (the DSA
+        // attachment, draw-buffer and read-buffer setters) publish immediately after the
+        // frontend mutation, because they have no validate point at all - FillPoints.def has no
+        // verb for any of them, so there is no MGP_FILL to sit in front of.
+        //
+        // A CALL THAT MOVED NOTHING IS FREE: the record's ContentHash is the emitter's own
+        // suppressor and it is keyed per framebuffer object, so a redundant publish emits zero
+        // bytes. EmitFramebufferByName picks Draw/Read/Both over Named when the object IS
+        // bound, so a Named record can never overwrite a bound record's Target underneath the
+        // binding that resolves through it.
+        void PipePublishFramebufferByName(const SharedPtr<MG_State::GLState::FramebufferObject>& fbo) {
+            if (!fbo) return;
+            MG_Pipe::MGPipeFramebufferEmitterInstance().EmitFramebufferByName(*fbo);
+        }
+#endif
     } // namespace
 
     void BlitFramebuffer_Backend(GLint srcX0, GLint srcY0, GLint srcX1, GLint srcY1, GLint dstX0, GLint dstY0,
@@ -631,6 +668,10 @@ namespace MobileGL::MG_Impl::GLImpl {
             MGLOG_E_ONCE("glBlitNamedFramebuffer skipped: backend does not implement explicit framebuffer blit.");
             return;
         }
+#if MOBILEGL_PIPE_PUSH
+        PipePublishFramebufferByName(readFramebuffer);
+        PipePublishFramebufferByName(drawFramebuffer);
+#endif
         MGP_FILL(BlitNamedFramebuffer);
         blitNamedFramebuffer(readFramebuffer, drawFramebuffer, srcX0, srcY0, srcX1, srcY1, dstX0, dstY0, dstX1,
                              dstY1, mask, filter);
@@ -643,6 +684,9 @@ namespace MobileGL::MG_Impl::GLImpl {
             MGLOG_E_ONCE("glClearNamedFramebufferfv skipped: backend does not implement explicit framebuffer clear.");
             return;
         }
+#if MOBILEGL_PIPE_PUSH
+        PipePublishFramebufferByName(framebuffer);
+#endif
         MGP_FILL(ClearNamedFramebufferfv);
         clearNamedFramebufferfv(framebuffer, buffer, drawbuffer, value);
     }
@@ -654,6 +698,9 @@ namespace MobileGL::MG_Impl::GLImpl {
             MGLOG_E_ONCE("glClearNamedFramebufferfi skipped: backend does not implement explicit framebuffer clear.");
             return;
         }
+#if MOBILEGL_PIPE_PUSH
+        PipePublishFramebufferByName(framebuffer);
+#endif
         MGP_FILL(ClearNamedFramebufferfi);
         clearNamedFramebufferfi(framebuffer, buffer, drawbuffer, depth, stencil);
     }
@@ -665,6 +712,9 @@ namespace MobileGL::MG_Impl::GLImpl {
             MGLOG_E_ONCE("glClearNamedFramebufferiv skipped: backend does not implement explicit framebuffer clear.");
             return;
         }
+#if MOBILEGL_PIPE_PUSH
+        PipePublishFramebufferByName(framebuffer);
+#endif
         MGP_FILL(ClearNamedFramebufferiv);
         clearNamedFramebufferiv(framebuffer, buffer, drawbuffer, value);
     }
@@ -676,6 +726,9 @@ namespace MobileGL::MG_Impl::GLImpl {
             MGLOG_E_ONCE("glClearNamedFramebufferuiv skipped: backend does not implement explicit framebuffer clear.");
             return;
         }
+#if MOBILEGL_PIPE_PUSH
+        PipePublishFramebufferByName(framebuffer);
+#endif
         MGP_FILL(ClearNamedFramebufferuiv);
         clearNamedFramebufferuiv(framebuffer, buffer, drawbuffer, value);
     }
@@ -1404,6 +1457,9 @@ namespace MobileGL::MG_Impl::GLImpl {
 
         if (texture == 0) {
             framebufferObject->Detach(attachmentType);
+#if MOBILEGL_PIPE_PUSH
+        PipePublishFramebufferByName(framebufferObject);
+#endif
             return;
         }
 
@@ -1430,6 +1486,9 @@ namespace MobileGL::MG_Impl::GLImpl {
         }
 
         framebufferObject->AttachTexture(attachmentType, textureObject, textureUploadTarget, level, 0, layered);
+#if MOBILEGL_PIPE_PUSH
+        PipePublishFramebufferByName(framebufferObject);
+#endif
     }
 
     void NamedFramebufferTextureWithUploadTarget_State(const char* functionName, GLuint framebuffer, GLenum attachment,
@@ -1453,6 +1512,9 @@ namespace MobileGL::MG_Impl::GLImpl {
 
         if (texture == 0) {
             framebufferObject->Detach(attachmentType);
+#if MOBILEGL_PIPE_PUSH
+        PipePublishFramebufferByName(framebufferObject);
+#endif
             return;
         }
 
@@ -1480,6 +1542,9 @@ namespace MobileGL::MG_Impl::GLImpl {
         }
 
         framebufferObject->AttachTexture(attachmentType, textureObject, textureUploadTarget, level);
+#if MOBILEGL_PIPE_PUSH
+        PipePublishFramebufferByName(framebufferObject);
+#endif
     }
 
     void NamedFramebufferTexture1D_State(GLuint framebuffer, GLenum attachment, GLenum textarget, GLuint texture,
@@ -1534,6 +1599,9 @@ namespace MobileGL::MG_Impl::GLImpl {
 
         if (texture == 0) {
             framebufferObject->Detach(attachmentType);
+#if MOBILEGL_PIPE_PUSH
+        PipePublishFramebufferByName(framebufferObject);
+#endif
             return;
         }
 
@@ -1636,6 +1704,9 @@ namespace MobileGL::MG_Impl::GLImpl {
 
         framebufferObject->AttachTexture(attachmentType, textureObject, textureUploadTarget, level, layer,
                                          /*layered=*/false);
+#if MOBILEGL_PIPE_PUSH
+        PipePublishFramebufferByName(framebufferObject);
+#endif
     }
 
     void FramebufferRenderbuffer_State(GLenum target, GLenum attachment, GLenum renderbuffertarget,
@@ -1705,6 +1776,9 @@ namespace MobileGL::MG_Impl::GLImpl {
 
         if (renderbuffer == 0) {
             framebufferObject->Detach(attachmentType);
+#if MOBILEGL_PIPE_PUSH
+        PipePublishFramebufferByName(framebufferObject);
+#endif
             return;
         }
 
@@ -1714,6 +1788,9 @@ namespace MobileGL::MG_Impl::GLImpl {
         if (!renderbufferObject) return;
 
         framebufferObject->AttachRenderbuffer(attachmentType, renderbufferObject);
+#if MOBILEGL_PIPE_PUSH
+        PipePublishFramebufferByName(framebufferObject);
+#endif
     }
 
     void DrawBuffersForFramebuffer_State(const SharedPtr<MG_State::GLState::FramebufferObject>& fbo, Bool isDefaultFBO,
@@ -1913,6 +1990,9 @@ namespace MobileGL::MG_Impl::GLImpl {
             : GetNamedFramebufferObject_State(framebuffer, "NamedFramebufferDrawBuffers_State");
         if (!framebufferObject) return;
         DrawBuffersForFramebuffer_State(framebufferObject, framebuffer == 0, n, bufs, false);
+#if MOBILEGL_PIPE_PUSH
+        PipePublishFramebufferByName(framebufferObject);
+#endif
     }
 
     void NamedFramebufferDrawBuffer_State(GLuint framebuffer, GLenum buf) {
@@ -1927,6 +2007,9 @@ namespace MobileGL::MG_Impl::GLImpl {
             const GLenum bufs[] = {buf};
             DrawBuffersForFramebuffer_State(framebufferObject, framebuffer == 0, 1, bufs, true);
         }
+#if MOBILEGL_PIPE_PUSH
+        PipePublishFramebufferByName(framebufferObject);
+#endif
     }
 
     void NamedFramebufferReadBuffer_State(GLuint framebuffer, GLenum src) {
@@ -1936,6 +2019,9 @@ namespace MobileGL::MG_Impl::GLImpl {
         if (!framebufferObject) return;
         ReadBufferForFramebuffer_State(framebufferObject, framebuffer == 0, src,
                                        "NamedFramebufferReadBuffer_State");
+#if MOBILEGL_PIPE_PUSH
+        PipePublishFramebufferByName(framebufferObject);
+#endif
     }
 
     SharedPtr<MG_State::GLState::FramebufferObject> GetFramebufferObjectForNamedClear(GLuint framebuffer,

MobileGL/MG_Impl/GLImpl/Texture/GL_Texture.cpp

@@ -31,6 +31,9 @@
 #include <MG_Util/Math/FixedPointConversion.h>
 #include <MG_State/GLState/TextureState/TextureObjectBuffer.h>
 #include <MG_Impl/Pipe/PipeFill.h>
+// P4a, ID-18 M2. The ONE door MG_State and MG_Impl have into the client's emitters; the three
+// call sites below are declarations only, exactly as the frontend's mutators are.
+#include <MG_Pipe/PipeMutation.h>
 
 namespace MobileGL::MG_Impl::GLImpl {
     static SharedPtr<MG_State::GLState::ITextureObject> nullTextureObject;
@@ -1337,6 +1340,30 @@ namespace MobileGL::MG_Impl::GLImpl {
                     std::format("pname {} is not a valid texture parameter.", MG_Util::ConvertGLEnumToString(pname))));
             return;
         }
+#if MOBILEGL_PIPE_PUSH
+        // P4a, ID-18 M2 - THE THIRTEENTH MGP_NOTE_AGGREGATE(TextureParams) SITE, and the one
+        // no publisher reached. Nine of the thirteen are TextureObject.cpp's own mutators and
+        // publish through TextureObjectBase::PipePublishParams; the tenth is
+        // SetDepthStencilTextureMode; two more move fields MGPTextureParams does not carry. The
+        // last is SamplerObject::BumpVersion, whose own comment calls it "the one choke point
+        // every setter reaches" - and MGPTextureParams takes MinLod, MaxLod and LodBias off that
+        // object, so every glTexParameter that writes GL_TEXTURE_MIN_LOD / MAX_LOD / LOD_BIAS
+        // landed on state nothing watched and the applier's record kept saying MinLod = 0.
+        // Wrong pixels, not a lost optimisation.
+        //
+        // THE HOOK IS HERE RATHER THAN ON BumpVersion because MG_State/GLState/SamplerState is
+        // package C's after the tag; C.7 grants this file for exactly this class of path ("the
+        // grant is one call site per path"), and this switch IS the path - every arm of it
+        // either writes the built-in SamplerObject or writes a texture field that publishes for
+        // itself. Placed after the switch, so the error arms above return without emitting.
+        //
+        // IT IS ALSO ID-14's RE-EMIT HOOK. C's sampler CSO cache is content-addressed, so the
+        // handle MGPTextureParams::BuiltinSampler names MOVES WITH THE CONTENT; the emitter
+        // re-Acquires from the cache and releases the previous handle here. An over-call is
+        // free: the emitter's version-first skip reads GetTextureParamsVersion() AND
+        // SamplerObject::GetVersion() and returns without hashing anything when neither moved.
+        MobileGL::MG_Pipe::MGPipeEmitTextureParams(*textureObject);
+#endif
     }
 
     void TextureParameterObjectf_State(const SharedPtr<MG_State::GLState::ITextureObject>& textureObject, GLenum pname,
@@ -1415,6 +1442,30 @@ namespace MobileGL::MG_Impl::GLImpl {
                     std::format("pname {} is not a valid texture parameter.", MG_Util::ConvertGLEnumToString(pname))));
             return;
         }
+#if MOBILEGL_PIPE_PUSH
+        // P4a, ID-18 M2 - THE THIRTEENTH MGP_NOTE_AGGREGATE(TextureParams) SITE, and the one
+        // no publisher reached. Nine of the thirteen are TextureObject.cpp's own mutators and
+        // publish through TextureObjectBase::PipePublishParams; the tenth is
+        // SetDepthStencilTextureMode; two more move fields MGPTextureParams does not carry. The
+        // last is SamplerObject::BumpVersion, whose own comment calls it "the one choke point
+        // every setter reaches" - and MGPTextureParams takes MinLod, MaxLod and LodBias off that
+        // object, so every glTexParameter that writes GL_TEXTURE_MIN_LOD / MAX_LOD / LOD_BIAS
+        // landed on state nothing watched and the applier's record kept saying MinLod = 0.
+        // Wrong pixels, not a lost optimisation.
+        //
+        // THE HOOK IS HERE RATHER THAN ON BumpVersion because MG_State/GLState/SamplerState is
+        // package C's after the tag; C.7 grants this file for exactly this class of path ("the
+        // grant is one call site per path"), and this switch IS the path - every arm of it
+        // either writes the built-in SamplerObject or writes a texture field that publishes for
+        // itself. Placed after the switch, so the error arms above return without emitting.
+        //
+        // IT IS ALSO ID-14's RE-EMIT HOOK. C's sampler CSO cache is content-addressed, so the
+        // handle MGPTextureParams::BuiltinSampler names MOVES WITH THE CONTENT; the emitter
+        // re-Acquires from the cache and releases the previous handle here. An over-call is
+        // free: the emitter's version-first skip reads GetTextureParamsVersion() AND
+        // SamplerObject::GetVersion() and returns without hashing anything when neither moved.
+        MobileGL::MG_Pipe::MGPipeEmitTextureParams(*textureObject);
+#endif
     }
 
     void GetTextureParameterObjectiv_State(const SharedPtr<MG_State::GLState::ITextureObject>& textureObject,
@@ -2105,6 +2156,30 @@ namespace MobileGL::MG_Impl::GLImpl {
                     std::format("pname {} is not a valid texture parameter.", MG_Util::ConvertGLEnumToString(pname))));
             return;
         }
+#if MOBILEGL_PIPE_PUSH
+        // P4a, ID-18 M2 - THE THIRTEENTH MGP_NOTE_AGGREGATE(TextureParams) SITE, and the one
+        // no publisher reached. Nine of the thirteen are TextureObject.cpp's own mutators and
+        // publish through TextureObjectBase::PipePublishParams; the tenth is
+        // SetDepthStencilTextureMode; two more move fields MGPTextureParams does not carry. The
+        // last is SamplerObject::BumpVersion, whose own comment calls it "the one choke point
+        // every setter reaches" - and MGPTextureParams takes MinLod, MaxLod and LodBias off that
+        // object, so every glTexParameter that writes GL_TEXTURE_MIN_LOD / MAX_LOD / LOD_BIAS
+        // landed on state nothing watched and the applier's record kept saying MinLod = 0.
+        // Wrong pixels, not a lost optimisation.
+        //
+        // THE HOOK IS HERE RATHER THAN ON BumpVersion because MG_State/GLState/SamplerState is
+        // package C's after the tag; C.7 grants this file for exactly this class of path ("the
+        // grant is one call site per path"), and this switch IS the path - every arm of it
+        // either writes the built-in SamplerObject or writes a texture field that publishes for
+        // itself. Placed after the switch, so the error arms above return without emitting.
+        //
+        // IT IS ALSO ID-14's RE-EMIT HOOK. C's sampler CSO cache is content-addressed, so the
+        // handle MGPTextureParams::BuiltinSampler names MOVES WITH THE CONTENT; the emitter
+        // re-Acquires from the cache and releases the previous handle here. An over-call is
+        // free: the emitter's version-first skip reads GetTextureParamsVersion() AND
+        // SamplerObject::GetVersion() and returns without hashing anything when neither moved.
+        MobileGL::MG_Pipe::MGPipeEmitTextureParams(*textureObject);
+#endif
     }
 
     void TexParameteri_State(GLenum target, GLenum pname, GLint param) {

MobileGL/MG_Impl/Pipe/CompositeResolver.h

@@ -0,0 +1,270 @@
+// MobileGL - MobileGL/MG_Impl/Pipe/CompositeResolver.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>
+
+// P4a's PROGRAM-PIPELINE COMPOSITE, on the client side.
+//
+// GLContext::GetProgramForDraw() already flattens a bound pipeline into one hidden composite
+// ProgramObject entirely in the frontend - it joins every graphics stage, computes the
+// pipeline's draw-program signature, looks it up in the pipeline's own cache and, on a miss,
+// attaches each stage's LINKED SNAPSHOT into a fresh ProgramObject and links it. All of that
+// is frontend work and none of it moves. What this file adds is the one thing the wire needs:
+// the composite gets ONE handle, out of the ShaderCso reserved high band, and
+// create_shader_state goes out for it exactly as for an ordinary program. THE SERVER NEVER
+// LEARNS IT IS A COMPOSITE and needs no "resolved draw program" hook at all.
+//
+// WHY A BAND RATHER THAN A FLAG ON THE HANDLE: a flag would have to be carried, honoured and
+// masked off by every consumer of a ShaderCso handle, on both sides; a reserved slot range is
+// a property of the allocator instead, so "an ordinary program can never be handed a composite
+// slot" is true by construction. MGPipeSlotAllocator::Allocate refuses the band outright and
+// AllocateComposite is the only door in.
+//
+// WHAT THIS FILE IS ACTUALLY FOR: the composite's slot has TWO INDEPENDENT RELEASE PATHS and
+// either order has to free it exactly once.
+//   * the pipeline cache drops the composite when the draw-program signature moves. In the
+//     frontend that overwrite drops the last SharedPtr, so the composite's own destructor
+//     usually runs first; the resolver still speaks the release, because "usually" is not a
+//     contract and a client that only reacted to destructors would leak a slot the moment the
+//     frontend started holding a second reference.
+//   * the composite ProgramObject's own ~ProgramObject, which is an ordinary program's death
+//     path and takes the same helper.
+// Both go through MGPipeEmitShaderCsoDestroyAndFree, and whichever runs second is a PROVEN
+// no-op: MGPipeSlotAllocator::Free refuses a slot that is not live at that generation and
+// bumps no generation of its own, so a double release cannot skip a generation either.
+//
+// THE MEMO's KEY IS (CONTEXT ID, PIPELINE GL NAME) AND THE CONTEXT HALF IS NOT OPTIONAL.
+// This resolver is a PROCESS singleton while a pipeline's GL name is per context: GLContext
+// owns m_programPipelines AND its own name generator m_programPipelineNames (Core.h), so name
+// N names two different ProgramPipelineObjects in two contexts, each with its own composite
+// and its own handle. Keyed on the name alone, the first emission after a make-current found
+// the OTHER context's entry, matched nothing - two composites are two ProgramObjects with two
+// lifetime ids, so the handles differ even when the stage set and the signature are identical
+// - and released it: a delete_shader_state and a cleared publication latch for a composite
+// whose frontend ProgramObject is alive, its band slot handed back and re-issued at gen + 1,
+// and the server rebuilding that program (glslang + SPIR-V + spirv-opt, the very cost the
+// signature below exists to avoid) once per context switch.
+//
+// THE CONTEXT ID IS GLContext::GetTextureContextId() AND NOTHING ELSE - the tree's existing
+// never-reused per-context id (TextureState::AllocateContextId; PipeInputs carries it as
+// m_textureContextId at seven fill points and the backends' own per-context memos key on it).
+// Deliberately NOT the GLContext ADDRESS that MGB_CTX_IDENTITY and MGPipeTracker::m_context
+// compare, because Core.h states the reason that id exists at all: a context freed and remade
+// lands on the old heap address, which would put this same defect back one context recreation
+// later.
+//
+// WHAT RELEASES A DESTROYED CONTEXT's ENTRIES: nothing in this file, and that is the correct
+// answer rather than an omission. Destroying a context drops m_programPipelines, which drops
+// each ProgramPipelineObject, which drops the composite it cached; ~ProgramObject then runs
+// MGPipeEmitShaderCsoDestroyAndFree - the composite's OWN release path, the second of the two
+// above - and the slot goes back exactly once. The entries those composites leave behind can
+// never be found again (no future Observe can carry a dead context id) and could not release
+// anything if they were (the allocator erases the lifetime-id mapping on Free), so Reset()
+// DROPS them instead of releasing them. That is also what bounds the vector; see Reset().
+//
+// THE SIGNATURE IS ComputeDrawProgramSignature(), the per-graphics-stage {lifetimeId,
+// GetLinkVersion()} array - and DELIBERATELY NOT GetBackendStateVersion(), which is what made
+// the SSO conformance loop rebuild the composite (glslang + SPIR-V + spirv-opt) on every draw,
+// because a glUniform1i to a sampler moves it.
+//
+// HEADER-ONLY, for the ownership reason Tracker.h states: a new .cpp would need the root
+// CMakeLists.txt, which is the contract package's.
+//
+// IT IS INCLUDED BY ProgramEmit.h AND NOT THE OTHER WAY ROUND, deliberately: the composite is
+// a special case of the program family's own emission, so the family header depends on this
+// one and this one depends on nothing of the family's. The reverse arrangement would make the
+// resolver reachable only from a translation unit that had already decided to use it, i.e.
+// dead in the build that matters and live only in the tests.
+#if MOBILEGL_PIPE_PUSH
+#include <MG_Impl/Pipe/SlotAllocator.h>
+#include <MG_Pipe/MGPipe.h>
+#include <MG_Pipe/PipeMutation.h>
+#include <MG_State/GLState/ProgramState/ProgramObject.h>
+#include <MG_State/GLState/ProgramState/ProgramPipelineObject.h>
+
+namespace MobileGL::MG_Pipe {
+
+    // IS THIS PROGRAM A PIPELINE COMPOSITE? A composite is the one ProgramObject in the system
+    // constructed with external index 0 (Core.cpp's MakeShared<ProgramObject>(0u)), and that is
+    // not an accident of implementation: it is deliberately not a named program, so it must not
+    // answer glIsProgram and must not consume a GL name, and glCreateProgram never returns 0.
+    //
+    // ASKED THIS WAY RATHER THAN CARRIED ON THE OBJECT because a Bool member on ProgramObject
+    // would resize the pull build's object and break G1 outright - the phase's admitted-resize
+    // set is empty - and a hook in Core.cpp would have to be maintained on a path that already
+    // states the invariant in its own comment.
+    inline Bool MGPipeProgramIsPipelineComposite(const MG_State::GLState::ProgramObject& program) {
+        return program.GetExternalIndex() == 0;
+    }
+
+    class MGPipeCompositeResolver {
+    public:
+        using ProgramObject = MG_State::GLState::ProgramObject;
+        using ProgramPipelineObject = MG_State::GLState::ProgramPipelineObject;
+        using DrawProgramSignature = ProgramPipelineObject::DrawProgramSignature;
+
+        struct Counters {
+            Uint64 Mints = 0;    // signatures this resolver has seen minted
+            Uint64 Reuses = 0;   // a signature that had not moved
+            Uint64 Releases = 0; // signature-move releases, i.e. the pipeline-cache path
+            // Entries dropped by Reset() because the composite's slot was already gone - the
+            // shape every entry of a DESTROYED CONTEXT ends in. A dropped entry is not a
+            // release: nothing is emitted and nothing is freed, the obligation having been
+            // discharged by the composite's own ~ProgramObject.
+            Uint64 Sweeps = 0;
+        };
+
+        // Told, at every emission, which composite the frontend handed out for which pipeline.
+        // Returns the handle the emitter should use, which is always the one already minted off
+        // the composite's own lifetime id - the resolver never mints a second identity for an
+        // object that has one.
+        //
+        // WHEN THE SIGNATURE MOVES the previous composite's slot is released here, through the
+        // one death helper and in its fixed order. That is the pipeline-cache release path; the
+        // composite's own destructor is the other one and the second of the two is the proven
+        // no-op.
+        MGPipeHandle Observe(Uint64 contextId, const ProgramPipelineObject& pipeline,
+                             const ProgramObject& composite, MGPipeHandle handle) {
+            const DrawProgramSignature signature = pipeline.ComputeDrawProgramSignature();
+            const Uint pipelineName = pipeline.GetExternalIndex();
+            Entry* entry = Find(contextId, pipelineName);
+            if (entry != nullptr) {
+                if (entry->Signature == signature && entry->Handle == handle) {
+                    // THE SAME COMPOSITE. Not merely "the same signature": the handle is minted
+                    // off the composite ProgramObject's own lifetime id, so an identical handle
+                    // IS an identical object and there is nothing to release. Live is
+                    // deliberately NOT touched - it is the release obligation and it is still
+                    // owed for exactly this handle.
+                    ++m_counters.Reuses;
+                    return handle;
+                }
+                // A MOVED SIGNATURE ON THIS CONTEXT's OWN ENTRY, which is the only thing that
+                // can reach here now: another context's pipeline of the same name is not found
+                // above and therefore not released, its obligation staying owed to the context
+                // that took it.
+                ReleaseEntry(*entry);
+            } else {
+                m_entries.push_back(Entry{});
+                entry = &m_entries.back();
+                entry->ContextId = contextId;
+                entry->PipelineName = pipelineName;
+            }
+            entry->Signature = signature;
+            entry->Handle = handle;
+            entry->CompositeLifetimeId = composite.GetLifetimeId();
+            entry->Live = true;
+            ++m_counters.Mints;
+            return handle;
+        }
+
+        // A make-current, and it RELEASES NOTHING. The entries name composites that belong to
+        // the frontend objects of the context being left, those objects outlive the switch, and
+        // releasing them would emit a delete for a live program.
+        //
+        // NOR IS ANY MEMO INVALIDATED, and that is what the context key bought. This used to
+        // clear a per-entry `Fresh` flag beside `Live`, because with a name-only key an entry
+        // could not say whether it described "my own pipeline before the switch" or "another
+        // context's pipeline of the same name" - and exactly one of those two properties could
+        // hold at a time. The key answers the question directly now, so the freshness flag and
+        // its one reader (a HandleFor() accessor that had no caller anywhere in the tree) are
+        // both gone rather than left as scaffolding: `Live`, the release obligation, is the
+        // entry's only state and nothing but ReleaseEntry may clear it.
+        //
+        // WHAT IS LEFT TO DO HERE IS RECLAMATION, and this is the one moment the client is told
+        // that a context boundary was crossed. An entry whose composite slot is no longer live
+        // has had its obligation discharged elsewhere - by that composite's own ~ProgramObject,
+        // which is precisely what happened to EVERY entry of a context that has just been
+        // destroyed - so it is DROPPED rather than released: a release would resolve nothing
+        // anyway (the allocator erases the lifetime-id mapping on Free) and no reader is left.
+        // Without this the vector would grow by one per (context, pipeline name) pair the
+        // process ever used, where the name-only key bounded it by the highest pipeline name;
+        // with it, it is bounded by the pairs whose composite slot is actually live.
+        void Reset() {
+            SizeT kept = 0;
+            for (SizeT i = 0; i < m_entries.size(); ++i) {
+                if (!m_entries[i].Live || !MGPipeSlots().IsLive(MGPipeKind::ShaderCso, m_entries[i].Handle)) {
+                    ++m_counters.Sweeps;
+                    continue;
+                }
+                if (kept != i) m_entries[kept] = m_entries[i];
+                ++kept;
+            }
+            m_entries.resize(kept);
+        }
+
+        void ResetCounters() { m_counters = Counters{}; }
+
+        // Diagnostics and unit cases only; nothing on the emission path asks. There is no
+        // HandleFor(name) accessor and there must not be one: the emitter takes the handle from
+        // the composite ProgramObject it already holds, so a lookup by name would be a second
+        // authority on an identity the allocator already owns.
+        SizeT Size() const { return m_entries.size(); }
+        const Counters& GetCounters() const { return m_counters; }
+
+    private:
+        struct Entry {
+            // NO FRONTEND SharedPtr, and that is the exit-order rule rather than a style
+            // choice: a static that held one would put a frontend destructor on an exit
+            // handler's path into a torn-down pipe. A GL name, a signature of plain integers,
+            // a handle and a lifetime id are all this needs.
+            // KEYED ON (CONTEXT ID, GL NAME), and the name half is the GL name because a
+            // ProgramPipelineObject has no lifetime id - ComputeDrawProgramSignature reads the
+            // STAGE programs' ids and the pipeline itself carries none. The context half is
+            // GLContext::GetTextureContextId(); see the file header for why the name alone was
+            // wrong and why the context ADDRESS would be too.
+            //
+            // WITHIN ONE CONTEXT glGenProgramPipelines recycles names, so a deleted-and-
+            // recreated pipeline can still inherit its predecessor's entry; that is bounded and
+            // self-correcting rather than a hazard. The first Observe on the new object finds a
+            // signature and a handle that do not match and releases the old entry, and that
+            // release resolves NOTHING - the allocator erases the lifetime-id mapping on Free,
+            // so a stale CompositeLifetimeId emits no delete and frees no slot; all it costs is
+            // one redundant, idempotent death notice, which is the same shape the composite's
+            // own second release path already has.
+            Uint64 ContextId = 0;
+            Uint PipelineName = 0;
+            DrawProgramSignature Signature{};
+            MGPipeHandle Handle = kMGPipeNullHandle;
+            Uint64 CompositeLifetimeId = 0;
+            // THE RELEASE OBLIGATION. Set when this entry takes responsibility for a composite's
+            // slot, cleared ONLY by ReleaseEntry when that responsibility is discharged.
+            Bool Live = false;
+        };
+
+        // BOTH HALVES OF THE KEY, always. An entry of another context is not this pipeline's
+        // entry: not found, not matched, not released.
+        Entry* Find(Uint64 contextId, Uint pipelineName) {
+            for (Entry& entry : m_entries) {
+                if (entry.ContextId == contextId && entry.PipelineName == pipelineName) return &entry;
+            }
+            return nullptr;
+        }
+
+        void ReleaseEntry(Entry& entry) {
+            if (!entry.Live || entry.CompositeLifetimeId == 0) return;
+            entry.Live = false;
+            MGPipeEmitShaderCsoDestroyAndFree(entry.CompositeLifetimeId);
+            entry.Handle = kMGPipeNullHandle;
+            entry.CompositeLifetimeId = 0;
+            ++m_counters.Releases;
+        }
+
+        Vector<Entry> m_entries;
+        Counters m_counters;
+    };
+
+    inline MGPipeCompositeResolver& MGPipeCompositeResolverInstance() {
+        // NEVER DESTROYED, for MGPipeTrackerInstance()' reason, and named in the phase's risk
+        // list beside the other three new client singletons: heap-constructed and intentionally
+        // leaked at exit, holding no frontend SharedPtr.
+        static MGPipeCompositeResolver* resolver = new MGPipeCompositeResolver();
+        return *resolver;
+    }
+} // namespace MobileGL::MG_Pipe
+#endif // MOBILEGL_PIPE_PUSH

MobileGL/MG_Impl/Pipe/CsoCache.h

@@ -0,0 +1,205 @@
+// MobileGL - MobileGL/MG_Impl/Pipe/CsoCache.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>
+
+// The render-state CSO cache (ARCHITECTURE.md 4.5.2 / 5.3, P2 brief D7).
+//
+// THE LOOKUP, and the first step is the whole point:
+//   1. m_pipelineStateVersion (widened) did not move -> reuse the last handle. ZERO hashing,
+//      zero probing, and nothing is emitted unless m_version also moved. That is the steady
+//      state of every frame, and it is why the tracker asks the cache at all only when the
+//      dirty walk says the pipeline version moved.
+//   2. moved -> hash the 396 pipeline bytes, probe, and on a hit CONFIRM WITH A MEMCMP
+//      before reusing the handle. ARCHITECTURE.md 4.1 says content addressing on an
+//      xxHash; a bare 64-bit equality would let a collision alias two different render
+//      states onto one CSO, which is silent wrong pixels with no gate that can see it.
+//      Mesa's cso_cache memcmps for the same reason. The memcmp only ever runs on a
+//      pipeline-version change, i.e. never in the steady state.
+//   3. miss -> mint a slot, emit create_render_state with every pipeline chunk, then bind.
+//
+// CAPACITY 64 (ROADMAP.md P2). 64 x (8 + 8 + 396 + 8) = about 26 KB per context. ROADMAP.md
+// open question 4 says 64 is provisional and the counters retune it at P13; this ships 64
+// and publishes the mint / bind / evict counters that retune reads.
+//
+// THE NEGATIVE CONTROL. kMGPipeBehaviourNoCsoContentAddressing (bit 63 of the runtime
+// MOBILEGL_PIPE_PUSH bitmask) turns off the PROBE and the handle reuse, not the records:
+// every pipeline-version change then mints a fresh CSO, binds it and evicts, which is
+// precisely "whole-block content addressing" and reproduces the regression
+// RenderState.h records. It is what separates "push is slower" from "the CSO design is
+// slower", and CsoContentAddressingScenario (package E) is the always-on ctest that stops
+// the switch from rotting.
+//
+// Header-only for the same ownership reason as Tracker.h: the root CMakeLists.txt that
+// would name a new .cpp is package A's and is frozen behind the p2/contract tag.
+#if MOBILEGL_PIPE_PUSH
+#include <Config.h>
+#include <MG_Impl/Pipe/SlotAllocator.h>
+#include <MG_Pipe/MGPipe.h>
+#include <MG_Pipe/MGPipeRenderStateSpans.h>
+#include <MG_Pipe/PipeApply.h>
+#include <MG_Util/Metrics/PipeStats.h>
+
+#include <cstring>
+
+namespace MobileGL::MG_Pipe {
+
+    inline constexpr SizeT kMGPipeCsoCacheCapacity = 64;
+
+    class MGPipeCsoCache {
+    public:
+        struct Counters {
+            Uint64 Mints = 0;     // create_render_state emissions
+            // bind_render_state emissions, mint or reuse. Counted in Acquire because Acquire
+            // has exactly ONE caller (PipeFill.cpp's EmitRenderState) and that caller binds
+            // immediately after every call - so "acquisitions" and "binds" are the same
+            // number, and counting it here keeps the count from depending on an emitter
+            // remembering to tick it. mints/binds is the cache's hit rate and it is the
+            // number the CSO content-addressing negative control moves.
+            Uint64 Binds = 0;
+            Uint64 Hits = 0;      // a probe that found a live entry and passed the memcmp
+            Uint64 Collisions = 0; // a hash hit the memcmp REJECTED - the reason it exists
+            Uint64 Evictions = 0; // LRU evictions, each one a delete_render_state
+        };
+
+        // The handle for `params`' pipeline subset. Mints and emits create_render_state on a
+        // miss; emits delete_render_state for whatever it evicts to make room. `payloadBytes`
+        // accumulates what went on the wire, for PipeStats::RecordDrawPayloadBytes.
+        MGPipeHandle Acquire(const RenderStateParameters& params, Uint64& payloadBytes) {
+            Array<Uint8, kMGPipePipelineChunkBytes> bytes;
+            MGPipeGatherPipelineBytes(params, bytes.data());
+            ++m_counters.Binds;
+
+            const Bool contentAddressed =
+                (MG_Config::Features.PipePush & kMGPipeBehaviourNoCsoContentAddressing) == 0;
+            if (contentAddressed) {
+                const Uint64 hash = s_hashForTest != nullptr ? s_hashForTest(bytes.data())
+                                                             : MGPipeHashPipelineBytes(bytes.data());
+                for (SizeT i = 0; i < m_entries.size(); ++i) {
+                    if (m_entries[i].Hash != hash) continue;
+                    if (std::memcmp(m_entries[i].Bytes.data(), bytes.data(), bytes.size()) != 0) {
+                        // A 64-bit collision between two DIFFERENT render states. Reusing the
+                        // handle here would render one state with the other's pipeline, so the
+                        // entry is dropped and the caller mints - correctness first, and the
+                        // counter says how often it happened.
+                        ++m_counters.Collisions;
+                        Evict(i);
+                        break;
+                    }
+                    m_entries[i].LastUsed = ++m_clock;
+                    ++m_counters.Hits;
+                    return m_entries[i].Cso;
+                }
+                return Mint(hash, bytes, payloadBytes);
+            }
+            // Content addressing OFF: never probe, always mint. The records still exist, so
+            // the arm differs from the default one in exactly one thing - whether a handle is
+            // reused - which is what makes it a control rather than a different design.
+            return Mint(0, bytes, payloadBytes);
+        }
+
+        // Context teardown, a server reset, a unit test's fixture. Emits nothing: the applier
+        // is reset alongside, and a delete for a record that is about to be dropped anyway
+        // would be a wire message with no reader.
+        void Reset() {
+            for (auto& entry : m_entries) MGPipeSlots().Free(MGPipeKind::RenderStateCso, entry.Cso);
+            m_entries.clear();
+            m_clock = 0;
+        }
+
+        void ResetCounters() { m_counters = Counters{}; }
+
+        SizeT Size() const { return m_entries.size(); }
+        const Counters& GetCounters() const { return m_counters; }
+
+        // TEST SEAM, and it is here because the thing it tests cannot be reached any other
+        // way. A 64-bit collision between two DIFFERENT render states is silent wrong pixels
+        // and it is exactly what the memcmp confirm above exists to stop, so
+        // CsoCacheTest.HashCollisionDoesNotAliasTwoStates has to be able to make one happen.
+        // Null in every real build - one never-taken, perfectly-predicted branch on a path
+        // that runs only when the pipeline version moved, i.e. never in the steady state.
+        using HashForTestFn = Uint64 (*)(const void* pipelineBytes);
+        inline static HashForTestFn s_hashForTest = nullptr;
+
+    private:
+        struct Entry {
+            Uint64 Hash = 0;
+            Uint64 LastUsed = 0;
+            MGPipeHandle Cso = kMGPipeNullHandle;
+            Array<Uint8, kMGPipePipelineChunkBytes> Bytes{};
+        };
+
+        MGPipeHandle Mint(Uint64 hash, const Array<Uint8, kMGPipePipelineChunkBytes>& bytes,
+                          Uint64& payloadBytes) {
+            if (m_entries.size() >= kMGPipeCsoCacheCapacity) {
+                SizeT victim = 0;
+                for (SizeT i = 1; i < m_entries.size(); ++i) {
+                    if (m_entries[i].LastUsed < m_entries[victim].LastUsed) victim = i;
+                }
+                Evict(victim);
+            }
+
+            const MGPipeHandle cso = MGPipeSlots().Allocate(MGPipeKind::RenderStateCso);
+            MGPRenderStateDesc desc{};
+            desc.Cso = cso;
+            desc.BaseCso = kMGPipeNullHandle;
+            // A brand-new CSO names every pipeline chunk; the incremental form against a
+            // BaseCso is what the applier's assertion allows and P3 will use once a CSO is
+            // minted from a neighbour rather than from nothing.
+            desc.ChunkMask = kAllPipelineChunks;
+            desc.Blob.Size = kMGPipePipelineChunkBytes;
+            MGPipeApplyCreateRenderState(desc, bytes.data());
+            payloadBytes += sizeof(MGPRenderStateDesc) + kMGPipePipelineChunkBytes;
+
+            Entry entry;
+            entry.Hash = hash;
+            entry.LastUsed = ++m_clock;
+            entry.Cso = cso;
+            entry.Bytes = bytes;
+            m_entries.push_back(entry);
+
+            ++m_counters.Mints;
+            if (MG_Util::PipeStats::Enabled()) {
+                MG_Util::PipeStats::AddCalls(MG_Util::PipeStats::CallClass::RenderStateCsoMints, 1);
+            }
+            return cso;
+        }
+
+        void Evict(SizeT index) {
+            MGPHandleOnly handle{};
+            handle.Handle = m_entries[index].Cso;
+            handle.Kind = static_cast<Uint32>(MGPipeKind::RenderStateCso);
+            MGPipeApplyDeleteRenderState(handle);
+            MGPipeSlots().Free(MGPipeKind::RenderStateCso, m_entries[index].Cso);
+            m_entries[index] = m_entries.back();
+            m_entries.pop_back();
+            ++m_counters.Evictions;
+        }
+
+        static constexpr Uint32 kAllPipelineChunks =
+            static_cast<Uint32>((Uint64{1} << kMGPipePipelineChunkCount) - 1);
+
+        Vector<Entry> m_entries;
+        Uint64 m_clock = 0;
+        Counters m_counters;
+    };
+
+    // The monolith's one cache, held beside the tracker. A Vector scan rather than a hash
+    // map on purpose: 64 entries of Uint64 is a handful of cache lines, it is probed only
+    // when the pipeline version moved, and it keeps the eviction order in the same array as
+    // the content - a map would need a second structure to answer "which is oldest".
+    inline MGPipeCsoCache& MGPipeCsoCacheInstance() {
+        // NEVER DESTROYED, for MGPipeTrackerInstance()' reason (MG_Impl/Pipe/Tracker.h): the
+        // rule covers every MGPipe process singleton, not only the ones on today's death
+        // paths.
+        static MGPipeCsoCache* cache = new MGPipeCsoCache();
+        return *cache;
+    }
+} // namespace MobileGL::MG_Pipe
+#endif // MOBILEGL_PIPE_PUSH

MobileGL/MG_Impl/Pipe/FramebufferEmit.h

@@ -0,0 +1,659 @@
+// MobileGL - MobileGL/MG_Impl/Pipe/FramebufferEmit.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>
+
+// The CLIENT side of P4a's framebuffer family: set_framebuffer_state, emitted at the validate
+// point once per bound TARGET that moved, or once with Target = Both when the two bindings
+// name the same object.
+//
+// THIS FILE IS CREATED BY THE CONTRACT COMMIT AND FILLED BY THE PACKAGE THAT OWNS IT, and the
+// split is the whole reason it exists this early. MG_Impl/Pipe/PipeFill.cpp is the contract
+// package's for the entire phase - it carries Coverage.def's enum-coupled block, the validate
+// point and the death helpers - so the emitter package must not edit it. What it edits instead
+// is this header: the emitter's BODY, and the value of kMGPipeWiredFramebufferSubsystem below.
+// That is what makes "no file is touched twice by two packages" structural rather than a
+// convention, and it is what the bb2a236d semantic-merge trap taught (two branches green
+// separately, the integrated tree not compiling).
+//
+// HEADER-ONLY, for the ownership reason Tracker.h and ResourceTracker.h both state: the root
+// CMakeLists.txt that would name a new .cpp is the contract package's and is frozen behind the
+// tag. MG_Impl/Pipe/PipeFill.cpp is the one translation unit that includes it in the library.
+#if MOBILEGL_PIPE_PUSH
+#include <MG_Impl/Pipe/SetHashSuppressor.h>
+#include <MG_Impl/Pipe/SlotAllocator.h>
+#include <MG_Impl/Pipe/TextureEmit.h>
+#include <MG_Impl/Pipe/Tracker.h>
+#include <MG_Pipe/MGPipe.h>
+#include <MG_Pipe/PipeApply.h>
+#include <MG_State/GLState/Core.h>
+#include <MG_Util/Metrics/PipeStats.h>
+
+#include <xxhash.h>
+
+#include <algorithm>
+
+namespace MobileGL::MG_Pipe {
+
+    // WHICH SUBSYSTEM BIT THIS BUILD ACTUALLY EMITS FOR. PipeFill.cpp ORs the four per-family
+    // constants into kMGPipeWiredSubsystems, so the bit is added by the commit that gives the
+    // emitters their bodies, with no file touched twice - and a Coverage.def row can never
+    // silently drop a field on the floor before the call that carries it exists.
+    //
+    // TURNING IT ON RETIRES NO PULL. GetFramebufferBindingSlot is the family's one
+    // Coverage.def emitted row and PipeFill.cpp's EmittedCallSuppliesTheWholeField answers
+    // FALSE for it, with the reason: the field's storage is a BindingSlot<FramebufferObject> -
+    // a frontend heap reference - and the call that supplies it carries eight-byte {slot, gen}
+    // handles and a fully resolved descriptor. So this bit switches the EMISSION on and the
+    // residual fill keeps writing the mirror, which is what keeps the verify lane at zero
+    // divergence.
+    inline constexpr Uint64 kMGPipeWiredFramebufferSubsystem = kMGPipeSubsystemFramebuffer;
+
+    inline Bool MGPipeFramebufferSubsystemEnabled() {
+        return (kMGPipeWiredFramebufferSubsystem & kMGPipeSubsystemFramebuffer) != 0 &&
+               (MG_Config::Features.PipePush & kMGPipeSubsystemFramebuffer) != 0;
+    }
+
+    // ---------------------------------------------------------------------------------
+    // D-C1: the MGPSurface builder, one pure function, one statement per field
+    // ---------------------------------------------------------------------------------
+
+    // MGPSurface::Kind's three constants ARE THE CONTRACT'S (ID-12 DV-4, c0c):
+    // kMGPipeSurfaceKindNone / ...Texture / ...Renderbuffer live in MG_Pipe/MGPipeTypes.h under
+    // exactly these names with the same MGPipeKind derivation and the same static_assert. This
+    // package's copies were a redefinition in the same namespace and are deleted.
+
+    // The upload target an attachment names, RESOLVED: an attachment made through an entry
+    // point that carries no face token stores TextureUploadTarget::Unknown, and the record goes
+    // out fully resolved - nothing in it may require a lookup on the far side.
+    //
+    // THE FALLBACK IS ONLY LEGAL FOR A SINGLE-TARGET TEXTURE (m1), and v1's was not. The
+    // precedent it copied - FramebufferAttachmentObject::GetSize - needs an EXTENT, which is
+    // identical across a cube map's six faces; face IDENTITY is not, so
+    // `glFramebufferTexture(GL_COLOR_ATTACHMENT0, cube, 0)` resolved to targets[0] and the
+    // record ASSERTED CubeMapPositiveX for a layered attachment that names all six. A texture
+    // with exactly one upload target has a [0] that IS the truth; anything else keeps Unknown,
+    // which is the value the field already carries for "this attachment names no single face"
+    // and which Layered = 1 tells the reader to ignore.
+    inline MobileGL::TextureUploadTarget MGPipeResolveAttachmentUploadTarget(
+        const MG_State::GLState::FramebufferAttachmentObject& attachment) {
+        MobileGL::TextureUploadTarget resolved = attachment.GetTextureUploadTarget();
+        if (resolved != MobileGL::TextureUploadTarget::Unknown) return resolved;
+        const auto& texture = attachment.GetTexture();
+        if (!texture) return MobileGL::TextureUploadTarget::Unknown;
+        const auto& targets = texture->GetUploadTargets();
+        return targets.size() == 1 ? targets[0] : MobileGL::TextureUploadTarget::Unknown;
+    }
+
+    // ONE PURE FUNCTION, ONE STATEMENT PER FIELD, and that shape is a gate requirement rather
+    // than taste: G7's scripted control stops this conversion copying exactly one member
+    // (MGPSurface::Layered) and expects the framebuffer suite to go red NAMING that field. A
+    // loop or a memcpy would make the control unanswerable.
+    //
+    // `res` is handed in because resolving it needs the slot allocator and this function stays
+    // pure; `internalFormat` is INLINE in the record on purpose, so the four cross-object masks
+    // fall out at push time with no lookup on the far side.
+    // THE EMPTY POINT IS THE ZERO-INITIALISED RECORD EXCEPT FOR ITS TWO TARGET FIELDS. Both
+    // are Uint16 enumerations whose zero is a REAL value - TextureTarget::Texture1D and
+    // TextureUploadTarget::Texture1D - so a reader that forgot to gate on Kind would read a
+    // plausible wrong answer rather than a nonsense one. Unknown (0xFFFF) is what the contract
+    // spells for TextureTarget (kMGPipeSurfaceNoTextureTarget) and m6 applies the same rule to
+    // UploadTarget, which shares the collision ID-12 DV-3 ruled on for MGPSubData::Target.
+    inline MGPSurface MGPipeEmptySurface() {
+        MGPSurface surface{};
+        surface.UploadTarget = static_cast<Uint16>(MobileGL::TextureUploadTarget::Unknown);
+        surface.TextureTarget = kMGPipeSurfaceNoTextureTarget;
+        return surface;
+    }
+
+    inline MGPSurface MGPipeBuildSurface(const MG_State::GLState::FramebufferAttachmentObject& attachment,
+                                         MGPipeHandle res) {
+        MGPSurface surface = MGPipeEmptySurface();
+        if (attachment.IsEmpty()) return surface;
+        surface.Res = res;
+        if (attachment.IsTexture()) {
+            const auto& texture = attachment.GetTexture();
+            surface.Kind = kMGPipeSurfaceKindTexture;
+            surface.InternalFormat = static_cast<Uint32>(texture->GetFormat());
+            surface.Layered = attachment.IsLayered() ? 1 : 0;
+            surface.Level = static_cast<Uint16>(std::max<Int>(attachment.GetTextureLevel(), 0));
+            surface.Layer = static_cast<Uint32>(std::max<Int>(attachment.GetTextureLayer(), 0));
+            surface.UploadTarget = static_cast<Uint16>(MGPipeResolveAttachmentUploadTarget(attachment));
+            // ID-12 DV-5: the field that WAS Pad0, and the size did not move. The four
+            // cross-object masks all reduce to (format, TEXTURE TARGET) -
+            // ShouldUseCaveatTextureFormat / BackendTextureFormatAddsAlpha - and no
+            // TextureUploadTarget -> TextureTarget inverse exists anywhere in the tree, so
+            // without this the inline InternalFormat cannot make them fall out at push time and
+            // the backend keeps reading the frontend attachment objects.
+            surface.TextureTarget = static_cast<Uint16>(texture->GetTarget());
+            return surface;
+        }
+        const auto& renderbuffer = attachment.GetRenderbuffer();
+        surface.Kind = kMGPipeSurfaceKindRenderbuffer;
+        surface.InternalFormat = static_cast<Uint32>(renderbuffer->GetInternalFormat());
+        surface.Layered = 0;
+        surface.Level = 0;
+        surface.Layer = 0;
+        return surface;
+    }
+
+    // MGPFramebufferState::DrawBuffers[i]: an index INTO THIS RECORD'S OWN Color[] array, and
+    // -1 for NONE, which is the field's documented convention read literally.
+    //
+    // THE FOUR DEFAULT-FRAMEBUFFER TOKENS map to 0, and that is a deliberate narrowing rather
+    // than an oversight: a default framebuffer has one colour surface, this record carries it
+    // in Color[0] (see MGPipeBuildFramebufferState), and IsDefault is what tells the server
+    // which framebuffer it is looking at. The distinction the narrowing loses is FRONT versus
+    // BACK and LEFT versus RIGHT, which MobileGL's frontend never gives a default framebuffer
+    // in the first place - FramebufferObject's constructor seeds BackLeft and nothing writes
+    // another. A phase that needs stereo has to widen the field, not re-encode this one.
+    inline Int8 MGPipeDrawBufferIndex(MobileGL::FramebufferAttachmentType buffer) {
+        using MobileGL::FramebufferAttachmentType;
+        if (buffer == FramebufferAttachmentType::None) return -1;
+        if (buffer >= FramebufferAttachmentType::Color0 && buffer <= FramebufferAttachmentType::ColorMax) {
+            return static_cast<Int8>(static_cast<Int>(buffer) - static_cast<Int>(FramebufferAttachmentType::Color0));
+        }
+        return 0;
+    }
+
+    // m2: A DRAW-BUFFER TOKEN CAN NAME A COLOUR POINT THE RECORD CANNOT CARRY, and D-C3's
+    // refusal loop only ever scanned ATTACHMENTS. `glDrawBuffers(1, {GL_COLOR_ATTACHMENT10})`
+    // with nothing attached at 10 is legal state - draw-incomplete, but legal - and the index
+    // above would have written 10 into a record whose Color[] is 8 wide, so the server would
+    // index out of its own storage or invent a bound the record does not carry. Truncating
+    // silently is the bug class this phase is closing, so the record is refused exactly as an
+    // over-wide attachment is.
+    inline Bool MGPipeDrawBufferIsInsideTheWireWidth(MobileGL::FramebufferAttachmentType buffer) {
+        using MobileGL::FramebufferAttachmentType;
+        if (buffer < FramebufferAttachmentType::Color0 || buffer > FramebufferAttachmentType::ColorMax) {
+            return true; // None and the four default-framebuffer tokens; neither indexes Color[]
+        }
+        return static_cast<Int>(buffer) - static_cast<Int>(FramebufferAttachmentType::Color0) <
+               static_cast<Int>(kMGPipeMaxColorAttachments);
+    }
+
+    // ---------------------------------------------------------------------------------
+    // D-C4: ContentHash, and the one input it must not swallow
+    // ---------------------------------------------------------------------------------
+    //
+    // XXH64 over the WHOLE record with ContentHash itself zeroed, computed field-wise into a
+    // zero-initialised staging copy so that no padding byte can enter the hash. Two jobs: the
+    // server's render-pass memo key, and this client's emission suppressor.
+    //
+    // IT MUST COVER Fbo. A recycled framebuffer handle whose successor happens to carry an
+    // identical attachment set would otherwise be suppressed against its predecessor; Fbo
+    // carries Gen, so it cannot be.
+    //
+    // IT MUST COVER DrawBuffers[8], and this is the trap worth naming. The backend derives the
+    // fragColor BROADCAST COUNT from the draw-buffer array, and it does that at the verb, from
+    // the framebuffer state it then holds, precisely so a program can relink inside the same
+    // draw. A hash that did not cover the array would let a suppressed set_framebuffer_state
+    // mean "the draw buffers did not move" when they had, and the shader would be specialised
+    // for the previous output shape. With the array in the hash, a suppression provably means
+    // the array did not move, which provably means the broadcast count did not move.
+    inline void MGPipeCopySurfaceForHash(MGPSurface& dst, const MGPSurface& src) {
+        dst.Res = src.Res;
+        dst.InternalFormat = src.InternalFormat;
+        dst.Kind = src.Kind;
+        dst.Layered = src.Layered;
+        dst.Level = src.Level;
+        dst.Layer = src.Layer;
+        dst.UploadTarget = src.UploadTarget;
+        // MANDATORY, not optional: TextureTarget is a PipeFields.def row now, so a
+        // field-wise copy that skipped it would suppress a record whose only moved field is
+        // the attachment's texture target - and that field decides three of the four
+        // cross-object masks.
+        dst.TextureTarget = src.TextureTarget;
+    }
+
+    inline Uint64 MGPipeFramebufferStateContentHash(const MGPFramebufferState& state) {
+        MGPFramebufferState staging{};
+        staging.Fbo = state.Fbo;
+        for (SizeT i = 0; i < kMGPipeMaxColorAttachments; ++i) {
+            MGPipeCopySurfaceForHash(staging.Color[i], state.Color[i]);
+        }
+        MGPipeCopySurfaceForHash(staging.Depth, state.Depth);
+        MGPipeCopySurfaceForHash(staging.Stencil, state.Stencil);
+        MGPipeCopySurfaceForHash(staging.ReadSurface, state.ReadSurface);
+        for (SizeT i = 0; i < kMGPipeMaxColorAttachments; ++i) {
+            staging.DrawBuffers[i] = state.DrawBuffers[i];
+        }
+        staging.Width = state.Width;
+        staging.Height = state.Height;
+        staging.Layers = state.Layers;
+        staging.Samples = state.Samples;
+        staging.FixedSampleLocations = state.FixedSampleLocations;
+        staging.IsDefault = state.IsDefault;
+        staging.Complete = state.Complete;
+        staging.Target = state.Target;
+        // staging.ContentHash stays 0 - that is the whole point.
+        return XXH64(&staging, sizeof(staging), 0);
+    }
+
+    // ---------------------------------------------------------------------------------
+    // The emitter
+    // ---------------------------------------------------------------------------------
+
+    class MGPipeFramebufferEmitter {
+    public:
+        using GLContext = MG_State::GLState::GLContext;
+        using FramebufferObject = MG_State::GLState::FramebufferObject;
+        using FramebufferAttachmentType = MobileGL::FramebufferAttachmentType;
+
+        // The handle for `fbo`. kMGPipeDefaultFramebuffer ({0,1}) for the default framebuffer,
+        // which is what retires the four pDefaultFramebufferInfo->defaultFBO identity
+        // comparisons into an ordinary handle compare; a client-minted {slot, gen} otherwise.
+        //
+        // Minted, never gated: a framebuffer handle is CLIENT state and costs one free-list pop.
+        static MGPipeHandle HandleFor(const FramebufferObject& fbo) {
+            if (fbo.IsDefaultFramebuffer()) return kMGPipeDefaultFramebuffer;
+            return MGPipeSlots().Acquire(MGPipeKind::Framebuffer, fbo.GetLifetimeId());
+        }
+
+        // Returns the bytes that went on the wire, for the per-draw payload histogram.
+        Uint64 EmitFramebufferState(GLContext& ctx) {
+            if (!MGPipeFramebufferSubsystemEnabled()) return 0;
+            const auto& drawFbo = ctx.GetFramebufferBindingSlot(MobileGL::FramebufferTarget::Draw).GetBoundObject();
+            const auto& readFbo = ctx.GetFramebufferBindingSlot(MobileGL::FramebufferTarget::Read).GetBoundObject();
+            if (!drawFbo && !readFbo) return 0;
+
+            // ONE OBJECT BOUND TO BOTH TARGETS IS ONE RECORD WITH Target = Both, and that is
+            // not an optimisation: Espryt's "same FBO as draw" skip is the habitat of the
+            // read-buffer defect class, and a record that says which target it describes turns
+            // "apply the draw buffers only for the draw target" from call-site discipline into
+            // a one-line test on the far side.
+            const Bool shared = drawFbo && readFbo && drawFbo.get() == readFbo.get();
+
+            MGPFramebufferState drawState{};
+            MGPFramebufferState readState{};
+            Bool drawOk = false;
+            Bool readOk = false;
+            if (shared) {
+                drawOk = BuildFramebufferState(*drawFbo, MGPipeFramebufferTarget::Both, drawState);
+            } else {
+                if (drawFbo) {
+                    drawOk = BuildFramebufferState(*drawFbo, MGPipeFramebufferTarget::Draw, drawState);
+                }
+                if (readFbo) {
+                    readOk = BuildFramebufferState(*readFbo, MGPipeFramebufferTarget::Read, readState);
+                }
+            }
+            if (!drawOk && !readOk) return 0;
+
+            // THE SUPPRESSOR SLOT IS FED THE COMBINED ANSWER and the per-target latches decide
+            // which of the two records actually goes out. The slot exists so that
+            // InvalidateAll() on a fresh context reaches this family like every other, and so
+            // that "nothing moved" costs one compare rather than two.
+            const Uint64 drawHash = drawOk ? drawState.ContentHash : 0;
+            const Uint64 readHash = readOk ? readState.ContentHash : 0;
+            const Uint64 combined =
+                MGPipeMixShutter(MGPipeMixShutter(drawHash, readHash), shared ? 1u : 0u);
+            if (!MGPipeSetHashSuppressorInstance().ShouldEmit(MGPipeSuppressorSlot::SetFramebufferState,
+                                                             combined)) {
+                return 0;
+            }
+
+            Uint64 bytes = 0;
+            if (shared) {
+                if (drawOk && (drawHash != m_lastEmitted[kDraw] || drawHash != m_lastEmitted[kRead])) {
+                    bytes += Emit(drawState);
+                    m_lastEmitted[kDraw] = drawHash;
+                    m_lastEmitted[kRead] = drawHash;
+                }
+                return bytes;
+            }
+            if (drawOk && drawHash != m_lastEmitted[kDraw]) {
+                bytes += Emit(drawState);
+                m_lastEmitted[kDraw] = drawHash;
+            }
+            if (readOk && readHash != m_lastEmitted[kRead]) {
+                bytes += Emit(readState);
+                m_lastEmitted[kRead] = readHash;
+            }
+            return bytes;
+        }
+
+        // ID-19(c): EVERY DSA ENTRY POINT THAT HANDS A FRAMEBUFFER TO THE SERVER BY NAME IS
+        // PRECEDED BY A RECORD FOR IT, and that is the phase's main correction rather than a
+        // nicety. With only the two BOUND-target records, glClearNamedFramebufferfv(fbo) on an
+        // unbound fbo made the backend mint a fresh driver framebuffer with NO ATTACHMENTS,
+        // find no record for it, decline, and issue the clear against it anyway -
+        // GL_INVALID_FRAMEBUFFER_OPERATION and nothing cleared, where the legacy arm cleared
+        // correctly (esprytobj C-1).
+        //
+        // THE TARGET IS Named ONLY WHEN THE OBJECT IS BOUND TO NEITHER BINDING. A record always
+        // writes FramebufferRecords[Fbo.Slot]; Draw/Read/Both ADDITIONALLY set the bound
+        // handle(s). So handing a currently-bound framebuffer a Named record would overwrite
+        // the bound record's Target with one that says "no binding" while BoundFramebuffer
+        // still names it, and the server would read a record whose Target contradicts the
+        // binding it is resolved through. Re-asserting the binding the object already has is
+        // free (the content hash suppresses it) and keeps the two consistent.
+        //
+        // Returns the bytes that went on the wire.
+        Uint64 EmitFramebufferByName(const FramebufferObject& fbo) {
+            if (!MGPipeFramebufferSubsystemEnabled()) return 0;
+            MGPipeFramebufferTarget target = MGPipeFramebufferTarget::Named;
+            const Bool boundToDraw = IsBoundTo(fbo, MobileGL::FramebufferTarget::Draw);
+            const Bool boundToRead = IsBoundTo(fbo, MobileGL::FramebufferTarget::Read);
+            if (boundToDraw && boundToRead) {
+                target = MGPipeFramebufferTarget::Both;
+            } else if (boundToDraw) {
+                target = MGPipeFramebufferTarget::Draw;
+            } else if (boundToRead) {
+                target = MGPipeFramebufferTarget::Read;
+            }
+
+            MGPFramebufferState state{};
+            if (!BuildFramebufferState(fbo, target, state)) return 0;
+
+            // THE SUPPRESSOR IS KEYED BY THE FRAMEBUFFER THE RECORD NAMES, never by one global
+            // slot (MGPipeTypes.h states the rule): two different objects' Named records in a
+            // row must both go out, and a Named record must never be suppressed against the
+            // same object's bound record or the reverse. Target is a ContentHash input, so the
+            // second half holds by construction; the per-object table is what buys the first.
+            // The two BOUND latches stay what they are - "does the server's draw/read binding
+            // already hold this record" - and a bound-target emission from here consults them,
+            // because a rebind of an unchanged object must still move the binding.
+            if (target == MGPipeFramebufferTarget::Named) {
+                NamedEntry& entry = NamedEntryFor(state.Fbo);
+                if (entry.Has && entry.Gen == state.Fbo.Gen && entry.LastHash == state.ContentHash) {
+                    return 0;
+                }
+                const Uint64 bytes = Emit(state);
+                entry.Has = true;
+                entry.Gen = state.Fbo.Gen;
+                entry.LastHash = state.ContentHash;
+                return bytes;
+            }
+            if (target == MGPipeFramebufferTarget::Both) {
+                if (state.ContentHash == m_lastEmitted[kDraw] && state.ContentHash == m_lastEmitted[kRead]) {
+                    return 0;
+                }
+                const Uint64 bytes = Emit(state);
+                m_lastEmitted[kDraw] = state.ContentHash;
+                m_lastEmitted[kRead] = state.ContentHash;
+                return bytes;
+            }
+            const SizeT slot = target == MGPipeFramebufferTarget::Read ? kRead : kDraw;
+            if (state.ContentHash == m_lastEmitted[slot]) return 0;
+            const Uint64 bytes = Emit(state);
+            m_lastEmitted[slot] = state.ContentHash;
+            return bytes;
+        }
+
+        // ---- what a unit case reads. The emitter builds INTO these and hands the applier the
+        // same objects, so "what was emitted" costs no copy. ----
+        // ---- the death half (P4a final review C-2) ----
+        //
+        // Called by the contract's death helper before the slot is freed (there is no wire
+        // delete for this kind, D-I2, so this is the only client-side thing a framebuffer's
+        // death has to do). The per-object Named latch is the entry: a recycled handle's Gen
+        // already refuses the stale latch, so this is hygiene rather than a fix - the rule
+        // (ID-8) is that whatever mints a handle retires everything it keeps under it at the
+        // death, and every P4a kind takes the same shape. Gen-keyed for a late notice.
+        void NoteFramebufferDied(MGPipeHandle handle) {
+            const SizeT slot = handle.Slot;
+            if (MGPipeHandleIsNull(handle) || slot >= m_named.size()) return;
+            if (m_named[slot].Gen == handle.Gen) m_named[slot] = NamedEntry{};
+        }
+        // "Does this emitter hold a Named-record latch for this handle at its generation."
+        Bool NamedRecordIsLatched(MGPipeHandle handle) const {
+            const SizeT slot = handle.Slot;
+            if (MGPipeHandleIsNull(handle) || slot >= m_named.size()) return false;
+            return m_named[slot].Has && m_named[slot].Gen == handle.Gen;
+        }
+
+        const MGPFramebufferState& LastDraw() const { return m_lastDraw; }
+        const MGPFramebufferState& LastRead() const { return m_lastRead; }
+        const MGPFramebufferState& LastNamed() const { return m_lastNamed; }
+        Uint64 EmissionCount() const { return m_emissions; }
+        Uint64 RefusedCount() const { return m_refusals; }
+
+        // A fresh context: what the server has is no longer what this emitter last sent. Only
+        // LATCHES reset here - MGPipeApplierReset clears the applier's DrawFramebuffer and
+        // ReadFramebuffer working state, so these mirrors have to go with them or the first
+        // emission after a make-current would be suppressed as unchanged and the server would
+        // draw into the previous context's framebuffer. The suppressor slot is invalidated by
+        // the validate point's own InvalidateAll(), beside this call.
+        void Reset() {
+            m_lastEmitted[kDraw] = 0;
+            m_lastEmitted[kRead] = 0;
+            // The per-object latch goes too, and the safe direction is why: MGPipeApplierReset
+            // keeps FramebufferRecords standing (they are object state, ID-19(b)) but
+            // ReleaseObjectRecords clears the whole table, and this emitter cannot tell the two
+            // scopes apart from here. Keeping a latch across a table that may have been dropped
+            // would suppress the one record that had to go out; dropping it costs one extra
+            // 304-byte record per named framebuffer after a context switch.
+            m_named.clear();
+        }
+
+        void ResetCounters() { m_emissions = m_refusals = 0; }
+
+        void ResetForTest() {
+            Reset();
+            ResetCounters();
+            m_lastDraw = MGPFramebufferState{};
+            m_lastRead = MGPFramebufferState{};
+            m_lastNamed = MGPFramebufferState{};
+        }
+
+    private:
+        static constexpr SizeT kDraw = 0;
+        static constexpr SizeT kRead = 1;
+
+        Uint64 Emit(const MGPFramebufferState& state) {
+            if (state.Target == static_cast<Uint8>(MGPipeFramebufferTarget::Named)) {
+                m_lastNamed = state;
+            } else if (state.Target == static_cast<Uint8>(MGPipeFramebufferTarget::Read)) {
+                m_lastRead = state;
+            } else {
+                m_lastDraw = state;
+                if (state.Target == static_cast<Uint8>(MGPipeFramebufferTarget::Both)) m_lastRead = state;
+            }
+            MGPipeApplySetFramebufferState(state);
+            ++m_emissions;
+            if (MG_Util::PipeStats::Enabled()) {
+                MG_Util::PipeStats::AddCalls(MG_Util::PipeStats::CallClass::FramebufferEmissions, 1);
+            }
+            return sizeof(MGPFramebufferState);
+        }
+
+        // ONE RECORD DESCRIBES ONE FRAMEBUFFER OBJECT - the one named by `fbo` - and every
+        // field in it is a property of THAT object. Target is the only binding-specific one.
+        //
+        // ReadSurface IS RESOLVED FROM THIS FRAMEBUFFER'S OWN READ BUFFER UNDER EVERY TARGET,
+        // Named included (c0e / MGPipeTypes.h). v1 resolved a Draw record's ReadSurface from
+        // the READ-bound object, which was D-C2's letter and muddled in substance: the record
+        // then described a surface that is not part of the framebuffer its own Fbo names, and a
+        // glReadBuffer on the read FBO moved the DRAW record's ContentHash and forced a
+        // redundant draw emission. Resolving it per object is what makes the
+        // read-buffer-shared-FBO defect class unrepresentable rather than merely fixed - the
+        // record carries a surface, not an index, and no field of it refers to "whatever is
+        // bound".
+        Bool BuildFramebufferState(const FramebufferObject& fbo, MGPipeFramebufferTarget target,
+                                   MGPFramebufferState& out) {
+            // D-C3, THE CLIENT HALF OF THE BRING-UP REFUSAL. The wire array is 8 wide and
+            // GetDynamicParameters().MaxColorAttachments is the driver's raw ES cap, not
+            // clamped to 8 on the GLES path. An attachment point at or above the wire width
+            // cannot be carried at all, so the record is REFUSED and the legacy arm runs -
+            // truncating it silently is exactly the bug class this phase is closing. The
+            // backend half of the same refusal (bit 9 declined at its first lookup, with one
+            // ERROR naming the cap) rides ResolveFramebufferSubsystemArm.
+            for (Int point = static_cast<Int>(FramebufferAttachmentType::Color0) +
+                             static_cast<Int>(kMGPipeMaxColorAttachments);
+                 point <= static_cast<Int>(FramebufferAttachmentType::ColorMax); ++point) {
+                if (fbo.GetAttachment(static_cast<FramebufferAttachmentType>(point)).IsEmpty()) continue;
+                MGLOG_E_ONCE("MGPipe: framebuffer %u has an attachment at colour point %d, which is at or "
+                             "above the wire width of %u - set_framebuffer_state is refused rather than "
+                             "truncated and the legacy arm runs",
+                             fbo.GetExternalIndex(),
+                             point - static_cast<Int>(FramebufferAttachmentType::Color0),
+                             static_cast<Uint>(kMGPipeMaxColorAttachments));
+                ++m_refusals;
+                return false;
+            }
+
+            // m2, THE SAME REFUSAL ONE FIELD OVER. A draw-buffer token may name a colour point
+            // at or above the wire width with nothing attached there, which the loop above
+            // cannot see; MGPipeDrawBufferIndex would then write 8..31 into an 8-wide array.
+            {
+                const auto& tokens = fbo.GetDrawBuffers();
+                for (SizeT i = 0; i < kMGPipeMaxColorAttachments; ++i) {
+                    if (MGPipeDrawBufferIsInsideTheWireWidth(tokens[i])) continue;
+                    MGLOG_E_ONCE("MGPipe: framebuffer %u names colour point %d in draw buffer %u, which is "
+                                 "at or above the wire width of %u - set_framebuffer_state is refused "
+                                 "rather than truncated and the legacy arm runs",
+                                 fbo.GetExternalIndex(),
+                                 static_cast<Int>(tokens[i]) -
+                                     static_cast<Int>(FramebufferAttachmentType::Color0),
+                                 static_cast<Uint>(i), static_cast<Uint>(kMGPipeMaxColorAttachments));
+                    ++m_refusals;
+                    return false;
+                }
+            }
+
+            out = MGPFramebufferState{};
+            out.Fbo = HandleFor(fbo);
+            out.Target = static_cast<Uint8>(target);
+            out.IsDefault = fbo.IsDefaultFramebuffer() ? 1 : 0;
+
+            // THE COLOUR POINTS. A default framebuffer keeps its one colour surface under
+            // BackLeft rather than under Color0, and the record has exactly one place to put
+            // it: Color[0], which is also the index MGPipeDrawBufferIndex maps that token to,
+            // so the array and the draw-buffer indices agree by construction.
+            if (out.IsDefault != 0) {
+                out.Color[0] = SurfaceOf(fbo, FramebufferAttachmentType::BackLeft);
+            } else {
+                for (SizeT i = 0; i < kMGPipeMaxColorAttachments; ++i) {
+                    out.Color[i] = SurfaceOf(fbo, static_cast<FramebufferAttachmentType>(
+                                                      static_cast<Int>(FramebufferAttachmentType::Color0) +
+                                                      static_cast<Int>(i)));
+                }
+            }
+            out.Depth = SurfaceOf(fbo, FramebufferAttachmentType::Depth);
+            out.Stencil = SurfaceOf(fbo, FramebufferAttachmentType::Stencil);
+            out.ReadSurface = SurfaceOf(fbo, fbo.GetReadBuffer());
+
+            const auto& drawBuffers = fbo.GetDrawBuffers();
+            for (SizeT i = 0; i < kMGPipeMaxColorAttachments; ++i) {
+                out.DrawBuffers[i] = MGPipeDrawBufferIndex(drawBuffers[i]);
+            }
+
+            FillGeometry(fbo, out);
+            // Complete is FramebufferObject::CheckCompleteness(), the FRONTEND-ONLY answer, and
+            // never glCheckFramebufferStatus's: that entry point additionally consults the
+            // backend's probed format-capability cache, and a client emitting it would be
+            // reading the backend from the client side - the exact coupling this boundary
+            // exists to remove. glCheckFramebufferStatus keeps answering from the frontend
+            // exactly as it does today.
+            out.Complete = fbo.CheckCompleteness() ? 1 : 0;
+            out.ContentHash = MGPipeFramebufferStateContentHash(out);
+            return true;
+        }
+
+        static Bool IsBoundTo(const FramebufferObject& fbo, MobileGL::FramebufferTarget target) {
+            if (MG_State::pGLContext == nullptr) return false;
+            const auto& bound = MG_State::pGLContext->GetFramebufferBindingSlot(target).GetBoundObject();
+            return bound && bound.get() == &fbo;
+        }
+
+        struct NamedEntry {
+            Uint32 Gen = 0;
+            Uint64 LastHash = 0;
+            Bool Has = false;
+        };
+
+        NamedEntry& NamedEntryFor(MGPipeHandle fbo) {
+            const SizeT slot = fbo.Slot;
+            if (slot >= m_named.size()) m_named.resize(slot + 1);
+            return m_named[slot];
+        }
+
+        MGPSurface SurfaceOf(const FramebufferObject& fbo, FramebufferAttachmentType type) {
+            if (type == FramebufferAttachmentType::None || type == FramebufferAttachmentType::Unknown) {
+                return MGPipeEmptySurface();
+            }
+            const auto& attachment = fbo.GetAttachment(type);
+            if (attachment.IsEmpty()) return MGPipeEmptySurface();
+            MGPipeTextureEmitter& textures = MGPipeTextureEmitterInstance();
+            // D-A4's two producers: an attachment point is what sets RENDER_TARGET and
+            // DEPTH_STENCIL, the two sticky bind bits nothing set before P4a. Sticky and ORed,
+            // so a texture that was ever a colour attachment keeps saying so, and the mask is
+            // republished on the resource's next respecify.
+            const Uint16 bit = (type == FramebufferAttachmentType::Depth ||
+                                type == FramebufferAttachmentType::Stencil)
+                                   ? static_cast<Uint16>(kMGPipeBindDepthStencil)
+                                   : static_cast<Uint16>(kMGPipeBindRenderTarget);
+            MGPipeHandle res = kMGPipeNullHandle;
+            if (attachment.IsTexture()) {
+                const auto& texture = attachment.GetTexture();
+                res = textures.AcquireTexture(texture->GetLifetimeId(), texture.get());
+                textures.NoteTextureBoundAs(res, bit);
+            } else if (attachment.IsRenderbuffer()) {
+                const auto& renderbuffer = attachment.GetRenderbuffer();
+                res = textures.AcquireRenderbuffer(renderbuffer->GetLifetimeId());
+                textures.NoteRenderbufferBoundAs(res, bit);
+            }
+            return MGPipeBuildSurface(attachment, res);
+        }
+
+        // The attachments' common extent, and the ARB_framebuffer_no_attachments defaults when
+        // there is no attachment at all (GL 4.6 core table 23.24 - the shape a framebuffer with
+        // no attachments rasterizes at).
+        static void FillGeometry(const FramebufferObject& fbo, MGPFramebufferState& out) {
+            Bool found = false;
+            for (const auto& attachment : fbo.GetAllAttachmentObjects()) {
+                if (attachment.IsEmpty()) continue;
+                const IntVec3 size = attachment.GetSize();
+                if (!found) {
+                    out.Width = static_cast<Uint16>(std::clamp<Int>(size.x(), 0, 0xFFFF));
+                    out.Height = static_cast<Uint16>(std::clamp<Int>(size.y(), 0, 0xFFFF));
+                    out.Layers = static_cast<Uint16>(
+                        attachment.IsLayered() ? std::clamp<Int>(size.z(), 1, 0xFFFF) : 1);
+                    if (attachment.IsTexture()) {
+                        const auto& texture = attachment.GetTexture();
+                        out.Samples = static_cast<Uint16>(std::max<Int>(texture->GetSamples(), 0));
+                        out.FixedSampleLocations = texture->HasFixedSampleLocations() ? 1 : 0;
+                    } else {
+                        out.Samples = static_cast<Uint16>(
+                            std::max<Int>(attachment.GetRenderbuffer()->GetSamples(), 0));
+                        out.FixedSampleLocations = 1;
+                    }
+                    found = true;
+                }
+            }
+            if (found) return;
+            out.Width = static_cast<Uint16>(std::clamp<Int>(fbo.GetDefaultWidth(), 0, 0xFFFF));
+            out.Height = static_cast<Uint16>(std::clamp<Int>(fbo.GetDefaultHeight(), 0, 0xFFFF));
+            out.Layers = static_cast<Uint16>(std::clamp<Int>(fbo.GetDefaultLayers(), 0, 0xFFFF));
+            out.Samples = static_cast<Uint16>(std::clamp<Int>(fbo.GetDefaultSamples(), 0, 0xFFFF));
+            out.FixedSampleLocations = fbo.GetDefaultFixedSampleLocations() ? 1 : 0;
+        }
+
+        Array<Uint64, 2> m_lastEmitted{};
+        // The per-FRAMEBUFFER suppressor for Named records, slot-indexed with the generation
+        // checked, exactly as the applier's own table is. A framebuffer has no wire lifetime
+        // (D-I2), so a successor simply overwrites its predecessor's entry.
+        Vector<NamedEntry> m_named;
+        MGPFramebufferState m_lastDraw{};
+        MGPFramebufferState m_lastRead{};
+        MGPFramebufferState m_lastNamed{};
+        Uint64 m_emissions = 0;
+        Uint64 m_refusals = 0;
+    };
+
+    inline MGPipeFramebufferEmitter& MGPipeFramebufferEmitterInstance() {
+        // NEVER DESTROYED, for MGPipeTrackerInstance()' reason (MG_Impl/Pipe/Tracker.h): the
+        // rule covers every MGPipe process singleton, not only the ones a frontend destructor
+        // reaches today, and it is what keeps exit() out of a torn-down pipe.
+        static MGPipeFramebufferEmitter* emitter = new MGPipeFramebufferEmitter();
+        return *emitter;
+    }
+} // namespace MobileGL::MG_Pipe
+#endif // MOBILEGL_PIPE_PUSH

MobileGL/MG_Impl/Pipe/ImageEmit.h

@@ -0,0 +1,182 @@
+// MobileGL - MobileGL/MG_Impl/Pipe/ImageEmit.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>
+
+// The CLIENT side of set_shader_images, the third of P4a's kVarTail unit sets. It rides
+// SamplerEmit.h's subsystem bit (kMGPipeWiredSamplerSubsystem): one family, one A/B.
+//
+// TWO INVARIANTS THAT MUST SURVIVE INTO THE BODY, and they are the kind an optimisation
+// deletes:
+//   1. THE HIGH-WATER-ZERO EARLY-OUT. An image high-water mark of 0 emits nothing, BEFORE any
+//      hash - that is what makes every Minecraft draw pay one integer test for a feature it
+//      does not use.
+//   2. THE SWEEP'S GATE IS KEYED ON FRONTEND GENERATIONS AND DELIBERATELY NOT ON A BACKEND
+//      RE-MINT COUNTER. A texture bound ONLY to an image unit is re-minted INSIDE the sweep,
+//      so a server-side epoch would be bumped after the gate had already declined. The
+//      client's bit-14 shutter is Mix(Mix(textureContent, textureParams), programImageUnitVersion)
+//      - all three FRONTEND counters - so the property is preserved by construction, and it is
+//      written here because it is invisible from the shutter itself.
+//
+// The record carries the APPLICATION's format and access; the bind-format recast (a GL_RG32F
+// bind is INVALID_VALUE on 19 of 26 non-core formats on Adreno) and the buffer-texture split
+// view stay SERVER-side and unchanged. ContentHash therefore has to cover InternalFormat and
+// Access as well as the binding, because the format the shader was built against is live
+// glBindImageTexture state and the format-less image bake keys on it.
+//
+// THIS FILE IS CREATED BY THE CONTRACT COMMIT AND FILLED BY THE PACKAGE THAT OWNS IT - see
+// FramebufferEmit.h for why, in full.
+#if MOBILEGL_PIPE_PUSH
+#include <MG_Impl/Pipe/SamplerEmit.h>
+#include <MG_Impl/Pipe/SetHashSuppressor.h>
+#include <MG_Impl/Pipe/SlotAllocator.h>
+#include <MG_Pipe/MGPipe.h>
+#include <MG_Pipe/PipeApply.h>
+#include <MG_Pipe/PipeMutation.h>
+#include <MG_State/GLState/Core.h>
+#include <MG_State/GLState/TextureState/TextureState.h>
+#include <MG_Util/Metrics/PipeStats.h>
+
+#include <xxhash.h>
+
+namespace MobileGL::MG_Pipe {
+
+    // D-G3. Over the tail with Start and Count mixed in, the same shape the two sampler sets
+    // use - and it covers InternalFormat and Access because those are live glBindImageTexture
+    // state that the format-less image bake keys on, not decoration.
+    inline Uint64 MGPipeShaderImageSetContentHash(const MGPImageView* entries, Uint32 start, Uint32 count) {
+        Uint64 hash = XXH64(entries, static_cast<SizeT>(count) * sizeof(MGPImageView), 0);
+        hash = MGPipeMixShutter(hash, start);
+        hash = MGPipeMixShutter(hash, count);
+        return hash;
+    }
+
+    class MGPipeImageEmitter {
+    public:
+        using GLContext = MG_State::GLState::GLContext;
+
+        // set_shader_images. Start is 0 and Count is the image-unit window described below.
+        //
+        // WHERE THE HIGH-WATER MARK COMES FROM, because the frontend has none and this is the
+        // one place a reader will look for it. DirectGLES keeps g_imageUnitHighWaterMark, but
+        // that is written from inside its own per-unit sync and lives on the far side of the
+        // boundary; TextureState::NoteUnitTouched is the TEXTURE-unit path and
+        // glBindImageTexture does not reach it. Adding a counter to TextureState would edit
+        // another package's file and resize the pull build's object, which G1 forbids outright.
+        //
+        // So the window is derived instead, from the one thing that decides whether an image
+        // unit can matter at all: the highest image unit the CURRENT PROGRAM names, memoised
+        // per program state in SamplerEmit.h's shared inversion, UNIONED with a sticky mark of
+        // every unit this emitter has already described. A program with no image uniforms
+        // gives MaxImageUnit == -1 and, with nothing sticky yet, a window of 0 - which is the
+        // zero early-out, taken BEFORE any hash and before any 192-entry walk, exactly as
+        // property 1 requires. The mark is sticky so that a program which stops naming a unit
+        // does not silently stop describing it: the window only grows, and shrinking it is how
+        // a stale binding would become invisible to the server.
+        Uint64 EmitShaderImages(GLContext& ctx) {
+            const auto& program = ctx.GetProgramForDraw();
+            const auto& resolution = MGPipeProgramOpaqueUnitsShared().For(program.get());
+            const Uint32 programWindow =
+                resolution.MaxImageUnit < 0 ? 0u : static_cast<Uint32>(resolution.MaxImageUnit) + 1u;
+            if (programWindow > m_window) m_window = programWindow;
+            const Uint32 count = m_window < kMGPipeMaxImageUnits ? m_window : kMGPipeMaxImageUnits;
+            // PROPERTY 1, and it is one integer test on every draw of every application that
+            // never binds an image.
+            if (count == 0) return 0;
+
+            for (Uint32 unit = 0; unit < count; ++unit) {
+                const auto& binding = ctx.GetImageTextureBinding(static_cast<Int>(unit));
+                MGPImageView& entry = m_entries[unit];
+                entry = MGPImageView{};
+                entry.Unit = unit;
+                entry.Res = binding.Texture ? MGPipeSlots().Acquire(MGPipeKind::Texture,
+                                                                    binding.Texture->GetLifetimeId())
+                                            : kMGPipeNullHandle;
+                // D-A4: a texture named in an emitted MGPImageView is SHADER-IMAGE-bound from
+                // then on - the bit ImageBindableHint is derived from. The bind itself noted it
+                // first (TextureState.h, so the hint precedes the first sync); this is the
+                // letter of the rule and a one-compare early-out once the bit is set.
+                if (!MGPipeHandleIsNull(entry.Res)) {
+                    MGPipeNoteTextureBoundAs(entry.Res, static_cast<Uint32>(kMGPipeBindShaderImage));
+                }
+                // THE APPLICATION's format and access, verbatim. The bind-format recast and the
+                // buffer-texture split view are server-side and stay there; so does
+                // SupportsLayeredImageBinding's rule, which asks the BACKEND target after
+                // MapToBackendTextureTarget and forces layer to 0 for a non-layerable one -
+                // Adreno took a stray layer index literally. A client that pre-applied any of
+                // that would be answering a driver question from the wrong side.
+                entry.InternalFormat = static_cast<Uint32>(binding.Format);
+                entry.Layer = static_cast<Uint32>(binding.Layer);
+                entry.Level = static_cast<Uint16>(binding.Level);
+                entry.Layered = binding.Layered != GL_FALSE ? 1 : 0;
+                entry.Access = static_cast<Uint8>(MGPipeEncodeImageAccess(binding.Access));
+            }
+
+            const Uint64 hash = MGPipeShaderImageSetContentHash(m_entries.data(), 0, count);
+            if (!MGPipeSetHashSuppressorInstance().ShouldEmit(MGPipeSuppressorSlot::SetShaderImages, hash)) {
+                return 0;
+            }
+            m_lastImages = MGPShaderImages{};
+            m_lastImages.Start = 0;
+            m_lastImages.Count = count;
+            m_lastImages.ContentHash = hash;
+            MGPipeApplySetShaderImages(m_lastImages, m_entries.data());
+            ++m_imageSets;
+            if (MG_Util::PipeStats::Enabled()) {
+                MG_Util::PipeStats::AddCalls(MG_Util::PipeStats::CallClass::ShaderImageEmissions, 1);
+            }
+            return sizeof(MGPShaderImages) + static_cast<Uint64>(count) * sizeof(MGPImageView);
+        }
+
+        // The validate point's FreshlyPrimed arm. A fresh context is a fresh set of image
+        // bindings, so the sticky window starts over; the suppressor slot this set latches is
+        // invalidated beside this call. There is no record half here at all - set_shader_images
+        // is pure working state and mints no object of its own.
+        void Reset() { m_window = 0; }
+
+        void ResetCounters() { m_imageSets = 0; }
+
+        const MGPShaderImages& LastShaderImages() const { return m_lastImages; }
+        const Array<MGPImageView, kMGPipeMaxImageUnits>& LastImageViews() const { return m_entries; }
+        Uint64 ImageSetCount() const { return m_imageSets; }
+        Uint32 Window() const { return m_window; }
+
+    private:
+        // GL_READ_ONLY / GL_WRITE_ONLY / GL_READ_WRITE folded into the one byte the wire
+        // carries. A value the enum does not name would otherwise truncate silently into a
+        // Uint8, which is the class of bug the descriptors exist to close.
+        static Uint32 MGPipeEncodeImageAccess(GLenum access) {
+            switch (access) {
+            case GL_READ_ONLY:
+                return 0;
+            case GL_WRITE_ONLY:
+                return 1;
+            case GL_READ_WRITE:
+                return 2;
+            default:
+                MOBILEGL_ASSERT(false, "glBindImageTexture access 0x%x is not one of the three GL names",
+                                static_cast<Uint>(access));
+                return 0;
+            }
+        }
+
+        Array<MGPImageView, kMGPipeMaxImageUnits> m_entries{};
+        MGPShaderImages m_lastImages{};
+        Uint32 m_window = 0;
+        Uint64 m_imageSets = 0;
+    };
+
+    inline MGPipeImageEmitter& MGPipeImageEmitterInstance() {
+        // NEVER DESTROYED, for MGPipeTrackerInstance()' reason; heap-constructed and
+        // intentionally leaked at exit, like every other MGPipe process singleton.
+        static MGPipeImageEmitter* emitter = new MGPipeImageEmitter();
+        return *emitter;
+    }
+} // namespace MobileGL::MG_Pipe
+#endif // MOBILEGL_PIPE_PUSH

MobileGL/MG_Impl/Pipe/PipeFill.h

@@ -18,11 +18,21 @@
 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);
+    // PipeFill.cpp. THE VALIDATE POINT (ARCHITECTURE.md 5.1, P2 brief D1). In order:
+    //   1. bump the per-verb serial, record the verb and the context identity;
+    //   2. run the tracker's DIRTY WALK for this verb's class (MG_Impl/Pipe/Tracker.h);
+    //   3. EMIT, for each set dirty bit whose subsystem bit is on in the runtime
+    //      MOBILEGL_PIPE_PUSH bitmask, the P2 call that carries it;
+    //   4. run the P1 residual fill for every field an emitted call did NOT supply,
+    //      stamping each with the new serial exactly as before;
+    //   5. in a verify build, the entry compare against a second snapshot (P1 brief D8) -
+    //      which stops being a tautology the moment step 3 supplies a field step 4 skips.
+    //
+    // It was MGPipeFillForVerb through P1, when steps 2 and 3 did not exist. The macro
+    // spelling, the 83 call sites and the verb enum are unchanged: the dispatch is
+    // kMGPipeVerbClass's nine classes, which is the same code as nine named ValidateFor*
+    // entry points with one call site per verb instead of nine.
+    void MGPipeValidateForVerb(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
@@ -33,6 +43,63 @@ namespace MobileGL::MG_Pipe {
     // stop where it says it stops (MG_Test/ScopedPipeVerb.h).
     void MGPipeLeaveVerb();
 
+    // PipeFill.cpp. DOES THIS BUILD, ON THIS BACKEND, AT THIS MASK, EMIT FOR THIS P4a FAMILY?
+    // (ID-39, widened by S-3 / ID-41.) The four conjuncts are the operator's per-subsystem bit
+    // in MOBILEGL_PIPE_PUSH, the family's own kMGPipeWired*Subsystem constant (`wired`, which
+    // the caller passes because it lives in the family's emit header and this header may not
+    // include one), and - for the four families P4a migrates - a backend having registered
+    // MGPipeResourceOps (the same per-backend signal `MGPipeResourceSubsystemEnabled()` has
+    // applied to P3a's buffers since the phase began) and every D-K2 dependency bit of the
+    // family being set in the same mask.
+    //
+    // THE LAST TWO CONJUNCTS ARE THE ONES THIS DECLARATION EXISTS FOR, and they are the same
+    // defect twice. Magma (DirectVulkan) registers no table and has no P4a twins; at a mask like
+    // 0x7ff Espryt REFUSES the texture family server-side because D-K2's fourth row says bit 10
+    // requires bit 11. In both cases the client emitted anyway, the applier accepted, the
+    // emitters cleared their per-level dirty flags on that acceptance, and the legacy upload
+    // path that still owed those texels found nothing to upload (66 DirectVulkan cases at ID-39,
+    // 47 DirectGLES cases at ID-41). With them the four families emit NOTHING in that state and
+    // the legacy pull path runs exactly as it does on a pull build.
+    //
+    // D-K2's TABLE IS IN PipeFill.cpp, ONCE: bit 9 requires bit 10, bit 10 requires bits 7 and
+    // 11, bit 11 requires bit 10, bit 12 depends on nothing - the client mirror, bit for bit, of
+    // the four `Resolve<Family>SubsystemArm()` refusals in DirectGLES/Managers.cpp.
+    //
+    // It is exported for the unit gate and for no other caller: the gate itself is
+    // FamilyIsLive() inside PipeFill.cpp, every birth hook and every `wants()` row resolves
+    // through it, and this returns that same expression rather than a second copy of it.
+    Bool MGPipeP4aFamilyEmits(Uint64 subsystem, Uint64 wired);
+
+    // PipeFill.cpp. P3a D-H2.1: the DRAW's raw vertex-fetch base instance, which
+    // set_vertex_buffers now carries as an explicit field.
+    //
+    // It replaces an ambient process global the backend read at VAO sync time, which is a
+    // shape that cannot cross a pushed boundary. The client sends the raw value and never a
+    // pre-shifted offset: whether to emulate the fetch shift or let GL_EXT_base_instance do
+    // it is the SERVER's decision. It is also an input to set_vertex_buffers' content hash
+    // and to the tracker's bit-9 shutter, so a draw whose only change is its base instance
+    // still reaches the emitter and still goes out.
+    //
+    // DO NOT CALL IT DIRECTLY FROM A GL ENTRY POINT - use MGP_SET_BASE_INSTANCE below. This
+    // whole declaration block is inside #if MOBILEGL_PIPE_PUSH, so a bare call would not even
+    // compile in a pull build, and the three call sites are in a file that is compiled in
+    // both. The macro is the same shape MGP_FILL already has, for the same reason.
+    //
+    // The validate point consumes and clears it - on both of its exits - and MGPipeLeaveVerb
+    // clears it too, so a plain draw that follows a base-instanced one sees 0 again. The
+    // tracker's Reset() deliberately does NOT clear it (Tracker.h): a make-current happens
+    // BETWEEN the setter and the fill that reads it.
+    //
+    // The three GL entry points that make this call (ID-10's grant) are
+    // MG_Impl/GLImpl/Drawing/GL_Drawing.cpp's DrawElementsInstancedBaseVertexBaseInstance,
+    // DrawElementsInstancedBaseInstance and DrawArraysInstancedBaseInstance - one line each,
+    // immediately above the MGP_FILL, carrying the RAW baseinstance argument.
+    void MGPipeSetPendingBaseInstance(Uint32 baseInstance);
+    // What the next set_vertex_buffers will carry. The unit gate reads it to pin that a
+    // make-current between the setter and the fill does not eat it
+    // (TrackerWalk.ABaseInstanceSurvivesTheFirstWalkOnAFreshContext).
+    Uint32 MGPipePendingBaseInstance();
+
     // 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
@@ -41,6 +108,28 @@ namespace MobileGL::MG_Pipe {
     // Fatal{PipeVerifyBadKnob}.
     void MGPipeSetPoisonOmission(const char* verb, const char* field);
 
+    // PipeFill.cpp. How many times set_vertex_attrib_defaults' applier failed to reproduce
+    // the value the call carried, so the client wrote the mirror itself
+    // (EmitVertexAttribDefaults). It is the ONE observable of that repair: the window it
+    // covers is a verb whose class does not read m_currentVertexAttribute, where reading the
+    // storage to check it would be the poison violation the fill table exists to forbid. So
+    // TrackerShippedEmitter asserts on this counter instead, and the day package A's applier
+    // switches on MGPAttribValue::ValueClass the counter stops moving.
+    //
+    // Not hot-path instrumentation: it is incremented only inside the repair branch, which
+    // runs only when the call actually went out, which is only when an attribute default
+    // moved.
+    Uint64 MGPipeVertexAttribDefaultRepairCount();
+
+    // PipeFill.cpp. The header of the last set_vertex_attrib_defaults that actually went out
+    // - Mask, and Count == 0 for "none ever did", since a call naming no attribute is not
+    // emitted. Two properties of this call have no other observable, because reading
+    // m_currentVertexAttribute back at a verb whose class does not carry it is the poison
+    // violation the fill table exists to forbid: that a FRESH CONTEXT republishes all 32
+    // (the server's mirror still holds the previous context's defaults), and that one moved
+    // attribute publishes exactly one. Eight bytes, written only when a call goes out.
+    MGPVertexAttribDefaults MGPipeVertexAttribDefaultsLastHeader();
+
 #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
@@ -49,7 +138,14 @@ namespace MobileGL::MG_Pipe {
     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)
+#define MGP_FILL(Verb) ::MobileGL::MG_Pipe::MGPipeValidateForVerb(::MobileGL::MG_Pipe::MGPipeVerb::Verb)
+// P3a D-H2.1. One line immediately ABOVE the MGP_FILL of a draw entry point that takes a
+// baseinstance, carrying the argument RAW. It has to be a macro for MGP_FILL's reason: the
+// three call sites are compiled in the pull build too, where MGPipeSetPendingBaseInstance is
+// neither declared nor defined.
+#define MGP_SET_BASE_INSTANCE(BaseInstance)                                                        \
+    ::MobileGL::MG_Pipe::MGPipeSetPendingBaseInstance(static_cast<::MobileGL::Uint32>(BaseInstance))
 #else
 #define MGP_FILL(Verb) ((void)0)
+#define MGP_SET_BASE_INSTANCE(BaseInstance) ((void)0)
 #endif

MobileGL/MG_Impl/Pipe/ProgramEmit.h

@@ -0,0 +1,484 @@
+// MobileGL - MobileGL/MG_Impl/Pipe/ProgramEmit.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>
+
+// The CLIENT side of P4a's program family: create/bind/delete_shader_state,
+// set_draw_program, set_dispatch_program and set_global_constants.
+//
+// WHERE create_shader_state IS EMITTED FROM, and why it is not the tracker's business: the
+// tracker's bit-6 shutter reads GetCurrentProgram() and DELIBERATELY NOT GetProgramForDraw(),
+// because the tracker must not force a compile just to answer "did the shader move". So the
+// tracker keeps its shutter and the EMITTER joins - from the same GetProgramForDraw() /
+// GetProgramForDispatch() call the verb is about to make anyway, so no join happens that would
+// not have happened. Emitting from the compile pool's terminal continuation is a real
+// asynchronous win and is a LATER phase's: in monolith the applier is one function call away,
+// so it is unmeasurable here.
+//
+// WHAT THE SERVER STILL SPECIALISES, so nobody reads create_shader_state as self-contained
+// and produces a per-draw rebuild: the draw-FBO clamp masks, the fragColor broadcast count,
+// the storage-block binding signature, the atomic-counter set, the live image formats and the
+// patch parameters are all inputs a backend program depends on BEYOND the artefacts. This call
+// publishes the ARTEFACTS; the server specialises at the verb from the state it holds. The
+// clause count does not shrink - its inputs move.
+//
+// THE ARTEFACTS DO NOT TRAVEL IN MONOLITH. All seven of MGPProgramDesc's blob refs are
+// declared with Size 0 and the LinkArtifacts / SpirvArtifacts ride beside the record through
+// MGPipeApplyCreateShaderState's companion pointers, so the codec is never called on the hot
+// path; the verify build is where it is exercised.
+//
+// THIS FILE IS CREATED BY THE CONTRACT COMMIT AND FILLED BY THE PACKAGE THAT OWNS IT - see
+// FramebufferEmit.h for why, in full.
+#if MOBILEGL_PIPE_PUSH
+#include <MG_Impl/Pipe/CompositeResolver.h>
+#include <MG_Impl/Pipe/SlotAllocator.h>
+#include <MG_Pipe/MGPipe.h>
+#include <MG_Pipe/MGPipeHostSpan.h>
+#include <MG_Pipe/PipeApply.h>
+#include <MG_Pipe/PipeMutation.h>
+#include <MG_State/GLState/Core.h>
+#include <MG_State/GLState/ProgramState/ProgramObject.h>
+#include <MG_Util/Metrics/PipeStats.h>
+
+namespace MobileGL::MG_Pipe {
+
+    // WIRED. create/bind/delete_shader_state, set_draw_program, set_dispatch_program and
+    // set_global_constants all have bodies, so this family contributes its bit to
+    // kMGPipeWiredSubsystems.
+    //
+    // AND SINCE c0b THAT CONSTANT REALLY IS PART OF THE EMISSION GATE, so the note that used to
+    // say otherwise here was true only against the contract commit: the validate point's
+    // `wants()` asks the subsystem mapping, the operator's MOBILEGL_PIPE_PUSH mask, THIS
+    // CONSTANT and the dirty bit, and the birth hooks' `FamilyIsLive` asks the same pair one
+    // level in. It is also a compile-time contract - while it is non-zero PipeFill.cpp's
+    // `if constexpr` seam instantiates the forward to EmitShaderCso below, so a missing entry
+    // point is a build error here rather than at the merge. The RUNTIME A/B that switches the
+    // family off is still the mask. See SamplerEmit.h's twin note.
+    inline constexpr Uint64 kMGPipeWiredProgramSubsystem = kMGPipeSubsystemPrograms;
+
+    // D-H6. ~0u is the BACKENDS' "never uploaded" sentinel for a global-constants version, and
+    // ProgramObject::MarkUBOContentDirty skips it on the wrap for exactly that reason. The
+    // client must never put it on the wire either: a server that received it would read its own
+    // record as "nothing has ever been uploaded here" and re-upload for ever.
+    inline constexpr Uint32 kMGPipeGlobalConstantsNeverUploaded = ~Uint32{0};
+
+    inline constexpr Bool MGPipeGlobalConstantsVersionIsEmittable(Uint32 version) {
+        return version != kMGPipeGlobalConstantsNeverUploaded;
+    }
+
+    // A program's identity for the wire, out of the SNAPSHOT the last link consumed and never
+    // out of the live attach list: glAttachShader and glCompileShader take effect only at the
+    // NEXT link and neither moves m_linkVersion, so a stage mask built from GetAttachedShaders
+    // would describe a program that does not exist yet. GetLinkedShaderStages() is also what
+    // indexes GetGeneratedSpirv(), so the two halves of this descriptor are guaranteed to agree
+    // by construction rather than by care.
+    inline Uint32 MGPipeStageMaskOf(const MG_State::GLState::ProgramObject& program) {
+        Uint32 mask = 0;
+        for (const ShaderStage stage : program.GetLinkedShaderStages()) {
+            if (stage == ShaderStage::Unknown) continue;
+            mask |= Uint32{1} << static_cast<Uint32>(stage);
+        }
+        return mask;
+    }
+
+    class MGPipeProgramEmitter {
+    public:
+        using GLContext = MG_State::GLState::GLContext;
+        using ProgramObject = MG_State::GLState::ProgramObject;
+
+        // create_shader_state (re-issued on the SAME handle whenever the link version moves -
+        // Gen moves only on slot reuse), then bind_shader_state and set_draw_program /
+        // set_dispatch_program. Two program calls because the frontend has two joins and two
+        // PipeInputs slots.
+        //
+        // BOTH JOINS HAPPEN HERE and both are the verb's own: GetProgramForDraw flattens a
+        // bound pipeline into its composite and GetProgramForDispatch answers the compute
+        // question, and with a plain glUseProgram they are the same object, so the ordinary
+        // frame pays one join it was going to pay anyway.
+        Uint64 EmitShaderState(GLContext& ctx) {
+            Uint64 bytes = 0;
+            const auto& drawProgram = ctx.GetProgramForDraw();
+            const auto& dispatchProgram = ctx.GetProgramForDispatch();
+
+            const MGPipeHandle drawCso =
+                drawProgram ? AcquireShaderCso(*drawProgram, bytes) : kMGPipeNullHandle;
+            // THE COMPOSITE'S SECOND RELEASE PATH is spoken here, not in a destructor: when the
+            // bound pipeline's draw-program signature moves, the resolver releases the slot the
+            // previous composite held. Whichever of the two paths runs second - this one or the
+            // composite ProgramObject's own ~ProgramObject - is a proven no-op, because the slot
+            // allocator refuses a slot that is not live at that generation.
+            if (drawProgram && MGPipeProgramIsPipelineComposite(*drawProgram)) {
+                if (const auto& pipeline = ctx.GetBoundProgramPipeline()) {
+                    // THE CONTEXT IS PART OF THE RESOLVER's KEY and this is the only place that
+                    // supplies it: the resolver is a process singleton and a pipeline's GL name
+                    // is per context, so without it a make-current between two contexts holding
+                    // one pipeline name released the other context's LIVE composite.
+                    // GetTextureContextId() is the tree's never-reused per-context id, the same
+                    // one PipeInputs carries and the backends' per-context memos key on.
+                    MGPipeCompositeResolverInstance().Observe(ctx.GetTextureContextId(), *pipeline,
+                                                             *drawProgram, drawCso);
+                }
+            }
+            const MGPipeHandle dispatchCso =
+                dispatchProgram ? (dispatchProgram == drawProgram ? drawCso
+                                                                  : AcquireShaderCso(*dispatchProgram, bytes))
+                                : kMGPipeNullHandle;
+
+            // THE BOUND CSO IS THE DRAW ONE WHEN THERE IS ONE. bind_shader_state names what
+            // glUseProgram selected, and when a program pipeline is bound instead that is the
+            // composite; a compute-only pipeline has no draw program at all, and then the
+            // dispatch program is the only thing bound. A null handle is legal here and means
+            // exactly "nothing bound".
+            const MGPipeHandle boundCso = !MGPipeHandleIsNull(drawCso) ? drawCso : dispatchCso;
+            if (boundCso != m_boundCso) {
+                MGPipeApplyBindShaderState(HandleOnly(boundCso));
+                m_boundCso = boundCso;
+                ++m_binds;
+                bytes += sizeof(MGPHandleOnly);
+            }
+            if (drawCso != m_drawCso) {
+                MGPipeApplySetDrawProgram(HandleOnly(drawCso));
+                m_drawCso = drawCso;
+                ++m_drawSets;
+                bytes += sizeof(MGPHandleOnly);
+            }
+            if (dispatchCso != m_dispatchCso) {
+                MGPipeApplySetDispatchProgram(HandleOnly(dispatchCso));
+                m_dispatchCso = dispatchCso;
+                ++m_dispatchSets;
+                bytes += sizeof(MGPHandleOnly);
+            }
+            return bytes;
+        }
+
+        // set_global_constants: the DEFAULT UNIFORM BLOCK only, keyed (ShaderCso, Version) and
+        // at most once per program per frame. Version is GetUBOContentVersion() and must never
+        // be ~0u, which is the backends' "never uploaded" sentinel - the wrap skips it.
+        //
+        // NAMED uniform blocks are NOT this call's: set_shader_buffers(Uniform) is a later
+        // phase's and BindCurrentProgramWithResources' named-UBO block is untouched. What
+        // travels here is globalUboScratch, the link phase's CPU array, which has no GL name
+        // and no BufferObject behind it.
+        Uint64 EmitGlobalConstants(GLContext& ctx) {
+            const auto& program = ctx.GetProgramForDraw();
+            if (!program) return 0;
+            const Uint32 version = program->GetUBOContentVersion();
+            // THE SENTINEL IS NEVER EMITTED. A server that received ~0u would read its own
+            // record as "never uploaded" and re-upload every frame for ever.
+            if (!MGPipeGlobalConstantsVersionIsEmittable(version)) return 0;
+            const Uint size = program->GetUBOSize();
+            if (size == 0) return 0;
+
+            Uint64 bytes = 0;
+            const MGPipeHandle cso = AcquireShaderCso(*program, bytes);
+            // (ShaderCso, Version) IS the key, so the latch is the key: an unchanged pair means
+            // the server already holds these bytes and re-sending them would move the record's
+            // serial for nothing.
+            if (cso == m_constantsCso && version == m_constantsVersion) return bytes;
+
+            m_lastConstants = MGPGlobalConstants{};
+            m_lastConstants.ShaderCso = cso;
+            m_lastConstants.Version = version;
+            // THE ONE BLOB RULE: Size 0 means "this record does not declare its blob" - which
+            // is what a monolith emission is - and the bytes ride beside it as a companion
+            // pointer. Offset carries the staging address for diagnostics only; nothing reads
+            // it as a length.
+            m_lastConstants.Blob.Seg = kMGHostSpanSegNone;
+            m_lastConstants.Blob.Offset = reinterpret_cast<Uint64>(program->GetUBOData());
+            m_lastConstants.Blob.Size = 0;
+            MGPipeApplySetGlobalConstants(m_lastConstants, program->GetUBOData());
+            m_constantsCso = cso;
+            m_constantsVersion = version;
+            ++m_constantSets;
+            if (MG_Util::PipeStats::Enabled()) {
+                MG_Util::PipeStats::AddBytes(MG_Util::PipeStats::ByteClass::CsoBlobBytes, size);
+            }
+            return bytes + sizeof(MGPGlobalConstants) + size;
+        }
+
+        // D-H4's re-issue rule, and it is the CreateVertexElements shape one for one: the
+        // record goes out again on the SAME handle whenever the link version moves, which is
+        // legal because MGPipeHandle::Gen increments only on slot reuse and never on a
+        // respecify. A program that relinks is the same GL object and the server's twin table
+        // must not be asked to mint a second one.
+        MGPipeHandle AcquireShaderCso(const ProgramObject& program, Uint64& payloadBytes) {
+            const MGPipeHandle handle = AcquireShaderCsoHandle(program);
+            if (MGPipeHandleIsNull(handle)) return handle;
+            Latch& latch = LatchFor(handle);
+
+            const Uint32 linkVersion = program.GetLinkVersion();
+            if (latch.RecordLive && latch.RecordGen == handle.Gen && latch.LinkVersion == linkVersion) {
+                return handle;
+            }
+
+            const auto& link = program.GetLinkReflection();
+            const auto& spirv = program.GetSpirvReflection();
+
+            m_lastDesc = MGPProgramDesc{};
+            m_lastDesc.Cso = handle;
+            m_lastDesc.StageMask = MGPipeStageMaskOf(program);
+            m_lastDesc.GlobalUboSize = static_cast<Uint32>(program.GetUBOSize());
+            m_lastDesc.ReservedNumSamplesOffset = static_cast<Uint32>(spirv.reservedNumSamplesOffset);
+            m_lastDesc.SpirvStatus = spirv.spirvStatus ? 1 : 0;
+            m_lastDesc.NativeFloat64 = spirv.nativeFloat64 ? 1 : 0;
+            m_lastDesc.PointSizeDemoted = spirv.pointSizeDemoted ? 1 : 0;
+            m_lastDesc.EnableSpirvValidation = spirv.enableSpirvValidation ? 1 : 0;
+
+            // ONE BLOB REF PER MODULE, IN THE LINKED-SHADER-SNAPSHOT'S ORDER, which is the
+            // order GetGeneratedSpirv() is indexed in - so Spirv[i] and StageMask agree because
+            // they came out of the same snapshot. Every one of them declares Size 0 (the one
+            // Blob rule); Offset carries the module's staging address so a reader can see which
+            // slots are occupied without the record pretending to declare a length it does not
+            // own.
+            //
+            // A COUNTED REFUSAL AND NOT AN ASSERTION (D-J3). MOBILEGL_ASSERT compiles out at
+            // INFO, which is all three gate builds and every shipped build, so an assert here
+            // would leave the truncation below completely silent in exactly the builds that
+            // run - which is the idiom D-J3 exists to forbid. generatedSpirv cannot exceed six
+            // stages today, so this is a guard against a seventh; truncation is the safe
+            // direction and the counter is what makes it visible.
+            const SizeT moduleCount = spirv.generatedSpirv.size();
+            if (moduleCount > 6) ++m_moduleTruncations;
+            for (SizeT i = 0; i < moduleCount && i < 6; ++i) {
+                m_lastDesc.Spirv[i].Seg = kMGHostSpanSegNone;
+                m_lastDesc.Spirv[i].Offset = reinterpret_cast<Uint64>(spirv.generatedSpirv[i].data());
+                m_lastDesc.Spirv[i].Size = 0;
+            }
+            m_lastDesc.Reflection.Seg = kMGHostSpanSegNone;
+            m_lastDesc.Reflection.Offset = reinterpret_cast<Uint64>(&link);
+            m_lastDesc.Reflection.Size = 0;
+
+            MGPipeApplyCreateShaderState(m_lastDesc, &link, &spirv);
+            // THE CREATE WENT OUT, so the publication latch is taken here and nowhere else
+            // (contract-v2 §3.1). MGPipeEmitShaderCsoDestroyAndFree reads it, and without it
+            // delete_shader_state can never go out - for an ordinary program or for a
+            // composite, both of which take that one helper.
+            MGPipeNoteHandlePublished(MGPipeKind::ShaderCso, handle);
+            ++m_creates;
+            payloadBytes += sizeof(MGPProgramDesc);
+
+            // A RE-ISSUED create_shader_state CLEARS THE APPLIER's DEFAULT UNIFORM BLOCK (wire
+            // W6), so the (Cso, Version) latch that suppresses set_global_constants has to go
+            // with it or the block is never re-sent. The case the design worries about is a
+            // FAILED relink of a bound program - GL keeps the previous executable and its
+            // uniforms running - and the general one is any future re-issue trigger that does
+            // not happen to move the content version, of which a recycled slot is one.
+            // Invalidated rather than re-emitted here, because this function has no business
+            // deciding when the constants go out: the next EmitGlobalConstants sees an
+            // unlatched key and sends them.
+            if (m_constantsCso == handle) {
+                m_constantsCso = kMGPipeNullHandle;
+                m_constantsVersion = kMGPipeGlobalConstantsNeverUploaded;
+            }
+
+            latch.RecordLive = true;
+            latch.RecordGen = handle.Gen;
+            latch.LinkVersion = linkVersion;
+            return handle;
+        }
+
+        // ---- THE CONTRACT ENTRY POINT THIS FAMILY OWES (contract-v2 §3.4) ----
+        //
+        // PipeFill.cpp's MGPipeEmitShaderCsoCreate forwards here through the `if constexpr`
+        // seam keyed on kMGPipeWiredProgramSubsystem, so while that constant is non-zero this
+        // must exist and be spelled exactly like this. A thin wrapper on purpose:
+        // AcquireShaderCso above IS this family's handle rule - identity-addressed per
+        // ProgramObject, the composite band entered through the one door, the re-issue on the
+        // same handle and the publication - and a second copy of any of it here would be a
+        // second authority.
+        //
+        // THE HOOK HAS ALREADY APPLIED BOTH GATES (the operator's mask and the wired constant),
+        // so this body applies none of its own. The byte count is discarded: a birth is not a
+        // validate-point emission and has no payload budget to report into.
+        void EmitShaderCso(ProgramObject& program) {
+            Uint64 bytes = 0;
+            AcquireShaderCso(program, bytes);
+        }
+
+        // The emitter's OWN record memo - "have I already published a create_shader_state at
+        // this slot, for this generation, at this link version".
+        //
+        // IT IS NOT WHAT THE DEATH PATH ASKS, and that changed at c0b (contract-v2 §3.1/D17):
+        // MGPipeEmitShaderCsoDestroyAndFree reads A's publication latch, which is one answer
+        // per {kind, slot, gen} that all six death helpers share. This stays because the
+        // VERSION-FIRST SKIP needs it - it is the same latch AcquireShaderCso consults before
+        // it builds a descriptor - and because a unit case reads it.
+        //
+        // THE COMPOSITE BAND IS INDEXED SEPARATELY, for the allocator's own reason: the band
+        // base is 983040, so a slot-indexed vector would allocate ~983k latches for one program
+        // pipeline. Both spaces stay dense against their own high-water mark.
+        Bool RecordIsPublished(MGPipeHandle handle) const {
+            if (MGPipeHandleIsNull(handle)) return false;
+            const Vector<Latch>& table = TableOf(handle);
+            const SizeT slot = SlotIndexOf(handle);
+            if (slot >= table.size()) return false;
+            const Latch& latch = table[slot];
+            return latch.RecordLive && latch.RecordGen == handle.Gen;
+        }
+
+        // The memo's other half, and the bound-mirror clearing beside it.
+        //
+        // THE CALLER IS THE CONTRACT's DEATH HELPER (P4a final review C-2): the death path
+        // reads the contract's latch for the wire delete and then forwards here, before the
+        // slot is freed, so a dead handle no longer reads as published in this memo between
+        // the death and the recycle and the three bound mirrors never name a dead program.
+        // Gen-keyed, so a late notice for a slot already handed out again clears nothing of
+        // the successor's.
+        void NoteRecordDestroyed(MGPipeHandle handle) {
+            if (MGPipeHandleIsNull(handle)) return;
+            Vector<Latch>& table = TableOf(handle);
+            const SizeT slot = SlotIndexOf(handle);
+            if (slot < table.size() && table[slot].RecordGen == handle.Gen) {
+                table[slot] = Latch{};
+            }
+            if (m_boundCso == handle) m_boundCso = kMGPipeNullHandle;
+            if (m_drawCso == handle) m_drawCso = kMGPipeNullHandle;
+            if (m_dispatchCso == handle) m_dispatchCso = kMGPipeNullHandle;
+            if (m_constantsCso == handle) {
+                m_constantsCso = kMGPipeNullHandle;
+                m_constantsVersion = kMGPipeGlobalConstantsNeverUploaded;
+            }
+        }
+
+        // The validate point's FreshlyPrimed arm. MGPipeApplierReset clears DrawProgram,
+        // DispatchProgram and BoundShaderCso - all three are per-context WORKING STATE - so
+        // the three mirrors here go with them, or the first emission after a make-current
+        // would be suppressed as unchanged and the server would draw with the previous
+        // context's program bound.
+        //
+        // The RECORD half stays, and that is the rule rather than an oversight: the applier
+        // keeps its shader-CSO records across a make-current because a program lives in a share
+        // group, and re-publishing one would move its Serial for nothing. The global-constants
+        // key goes with the working state because its record's bytes are per (Cso, Version) and
+        // a fresh server has not been told them.
+        void Reset() {
+            m_boundCso = kMGPipeNullHandle;
+            m_drawCso = kMGPipeNullHandle;
+            m_dispatchCso = kMGPipeNullHandle;
+            m_constantsCso = kMGPipeNullHandle;
+            m_constantsVersion = kMGPipeGlobalConstantsNeverUploaded;
+            // The composite memo's freshness goes with them - and only its freshness. Its
+            // ENTRIES name composites whose frontend objects outlive the context switch, so
+            // releasing them here would emit a delete for a live program.
+            MGPipeCompositeResolverInstance().Reset();
+        }
+
+        void ResetCounters() {
+            m_creates = m_binds = m_drawSets = m_dispatchSets = m_constantSets = 0;
+            m_moduleTruncations = 0;
+        }
+
+        // ---- what a unit case reads ----
+        const MGPProgramDesc& LastProgramDesc() const { return m_lastDesc; }
+        const MGPGlobalConstants& LastGlobalConstants() const { return m_lastConstants; }
+        // THE (Cso, Version) KEY set_global_constants is suppressed against. Exposed so a case
+        // can pin that a re-issued create_shader_state invalidates it - the applier clears the
+        // block on the re-issue (wire W6), so a latch that survived it would never re-send.
+        MGPipeHandle GlobalConstantsCso() const { return m_constantsCso; }
+        Uint32 GlobalConstantsVersion() const { return m_constantsVersion; }
+        // D-J3's counted refusal: programs whose linked snapshot carried more modules than
+        // MGPProgramDesc::Spirv[] can name, and whose tail was therefore dropped.
+        Uint64 TruncatedModuleCount() const { return m_moduleTruncations; }
+        MGPipeHandle BoundCso() const { return m_boundCso; }
+        MGPipeHandle DrawCso() const { return m_drawCso; }
+        MGPipeHandle DispatchCso() const { return m_dispatchCso; }
+        Uint64 CreateCount() const { return m_creates; }
+        Uint64 BindCount() const { return m_binds; }
+        Uint64 DrawProgramSetCount() const { return m_drawSets; }
+        Uint64 DispatchProgramSetCount() const { return m_dispatchSets; }
+        Uint64 GlobalConstantsSetCount() const { return m_constantSets; }
+
+    private:
+        // THE ONE PLACE THE BAND CAN ENTER. An ordinary program's slot comes from the ordinary
+        // allocator door keyed on its lifetime id. CompositeResolver.h widens this to send a
+        // pipeline composite through MGPipeSlotAllocator::AllocateComposite instead, and
+        // nothing else about the emission changes - the server never learns a composite is a
+        // composite.
+        MGPipeHandle AcquireShaderCsoHandle(const ProgramObject& program) {
+            const Uint64 lifetimeId = program.GetLifetimeId();
+            const MGPipeHandle existing = MGPipeSlots().FindByLifetimeId(MGPipeKind::ShaderCso, lifetimeId);
+            if (!MGPipeHandleIsNull(existing)) return existing;
+            // A composite is minted off ITS OWN lifetime id, out of the reserved band, and is
+            // an ordinary ShaderCso handle in every other respect - the same kind, the same
+            // {slot, gen} rules, the same Free, the same death helper. Keying it on its own
+            // lifetime id rather than on the pipeline's signature is what makes ~ProgramObject
+            // able to release it at all, and it is why two pipelines that happen to have the
+            // same signature keep their own composite: sharing one handle between two frontend
+            // objects would let the first one's death free a slot the second still names.
+            return MGPipeProgramIsPipelineComposite(program)
+                       ? MGPipeSlots().AllocateComposite(lifetimeId)
+                       : MGPipeSlots().AllocateFor(MGPipeKind::ShaderCso, lifetimeId);
+        }
+
+        struct Latch {
+            Bool RecordLive = false;
+            Uint32 RecordGen = 0;
+            Uint32 LinkVersion = 0;
+        };
+
+        // TWO TABLES, NOT A WIDER ONE, and it is the allocator's own reason repeated where it
+        // bites a second time: the composite band starts at slot 983040, so folding a composite
+        // into the ordinary slot-indexed vector would allocate ~983k latches - and grow them
+        // again on every future push_back - for a single program pipeline. Both spaces stay
+        // dense against their own high-water mark, which is exactly what the allocator does one
+        // level down.
+        Vector<Latch>& TableOf(MGPipeHandle handle) {
+            return MGPipeIsCompositeShaderSlot(handle.Slot) ? m_compositeLatch : m_latch;
+        }
+        const Vector<Latch>& TableOf(MGPipeHandle handle) const {
+            return MGPipeIsCompositeShaderSlot(handle.Slot) ? m_compositeLatch : m_latch;
+        }
+        static SizeT SlotIndexOf(MGPipeHandle handle) {
+            return MGPipeIsCompositeShaderSlot(handle.Slot)
+                       ? static_cast<SizeT>(handle.Slot - kMGPipeShaderCsoCompositeSlotBase)
+                       : static_cast<SizeT>(handle.Slot);
+        }
+        Latch& LatchFor(MGPipeHandle handle) {
+            Vector<Latch>& table = TableOf(handle);
+            const SizeT slot = SlotIndexOf(handle);
+            if (slot >= table.size()) table.resize(slot + 1);
+            return table[slot];
+        }
+
+        static MGPHandleOnly HandleOnly(MGPipeHandle handle) {
+            MGPHandleOnly only{};
+            only.Handle = handle;
+            only.Kind = static_cast<Uint32>(MGPipeKind::ShaderCso);
+            return only;
+        }
+
+        MGPProgramDesc m_lastDesc{};
+        MGPGlobalConstants m_lastConstants{};
+
+        Vector<Latch> m_latch;
+        Vector<Latch> m_compositeLatch;
+        MGPipeHandle m_boundCso = kMGPipeNullHandle;
+        MGPipeHandle m_drawCso = kMGPipeNullHandle;
+        MGPipeHandle m_dispatchCso = kMGPipeNullHandle;
+        MGPipeHandle m_constantsCso = kMGPipeNullHandle;
+        Uint32 m_constantsVersion = kMGPipeGlobalConstantsNeverUploaded;
+
+        Uint64 m_creates = 0;
+        Uint64 m_binds = 0;
+        Uint64 m_drawSets = 0;
+        Uint64 m_dispatchSets = 0;
+        Uint64 m_constantSets = 0;
+        Uint64 m_moduleTruncations = 0;
+    };
+
+    inline MGPipeProgramEmitter& MGPipeProgramEmitterInstance() {
+        // NEVER DESTROYED, for MGPipeTrackerInstance()' reason; heap-constructed and
+        // intentionally leaked at exit, and it MUST NOT hold a frontend SharedPtr - that is
+        // the exit-order rule, stated over every MGPipe process singleton rather than over the
+        // ones a destructor reaches today.
+        static MGPipeProgramEmitter* emitter = new MGPipeProgramEmitter();
+        return *emitter;
+    }
+} // namespace MobileGL::MG_Pipe
+#endif // MOBILEGL_PIPE_PUSH

MobileGL/MG_Impl/Pipe/ResourceTracker.h

@@ -0,0 +1,615 @@
+// MobileGL - MobileGL/MG_Impl/Pipe/ResourceTracker.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>
+
+// The CLIENT side of P3a's resource family (brief D-A, D-B, D-C, D-D).
+//
+// WHERE IT RUNS, and it is the ONE exception to push-at-validate (ARCHITECTURE.md 5.1):
+// the seven BufferBackendOps hooks already dispatch at the GL call that causes them, so
+// their pipe calls are emitted from the same BufferObject dispatchers - not from
+// MGPipeValidateForVerb. Nothing about buffers moves to validate time in P3a.
+//
+// WHAT LIVES HERE
+//   * the sticky BindMask, one constexpr BufferTarget -> bit table with a static_assert
+//     that it covers every enumerator, so a new target cannot be silently unmapped;
+//   * the lifetimeId -> {slot, gen} mint (through MGPipeSlots(), the one allocator) and
+//     the slot -> BufferObject* INVERSE the reverse channel resolves a writeback through;
+//   * the nine MGPipeEmitResource* bodies, declared in MG_Pipe/PipeMutation.h so that
+//     MG_State sees a declaration and never this file (the same layering PipeMutation.h
+//     already has for MGP_NOTE_MUTATION: declare in MG_Pipe, define in MG_Impl);
+//   * the MGPSubData range splitter, because one record's box caps the destination at a
+//     2^31-1 offset and a 2^32-1 size;
+//   * the map-persistent-roundtrips counting site.
+//
+// HEADER-ONLY, for the ownership reason Tracker.h states in full: the root CMakeLists.txt
+// that would name a new .cpp belongs to the contract package and is frozen behind the tag.
+// MG_Impl/Pipe/PipeFill.cpp is the one translation unit that includes it in the library.
+//
+// NO TIMER, and no per-call record copy on a HOT path. The two observables a unit case
+// needs - the last emitted descriptor and the per-call counts - are written only by
+// resource_create and resource_respecify, which run once per glBufferData rather than per
+// upload; resource_subdata, the hot one, is observed through the pure builders below
+// instead (MGPipeBuildSubDataRecord / MGPipeForEachSubDataRecordRange), which is also what
+// lets a test drive the splitter at both of its bounds without a 4 GiB buffer.
+#if MOBILEGL_PIPE_PUSH
+#include <MG_Impl/Pipe/SlotAllocator.h>
+#include <MG_Pipe/MGPipe.h>
+#include <MG_Pipe/PipeApply.h>
+#include <MG_Pipe/PipeMutation.h>
+#include <MG_State/GLState/BufferState/BufferState.h>
+#include <MG_State/GLState/Core.h>
+#include <MG_Util/Metrics/PipeStats.h>
+
+#include <Config.h>
+
+#include <cstdint>
+
+namespace MobileGL::MG_Pipe {
+
+    // ---------------------------------------------------------------------------------
+    // D-A3: BindMask
+    // ---------------------------------------------------------------------------------
+
+    // MGPResourceDesc::BindMask's twelve bits MOVED TO MG_Pipe/MGPipeTypes.h AT P4a, beside
+    // the field, exactly as the note that stood here said they would when a second producer
+    // appeared: P4a's texture family sets kMGPipeBindSampler / kMGPipeBindShaderImage /
+    // kMGPipeBindRenderTarget / kMGPipeBindDepthStencil, the four bits nothing set before.
+    // No alias is written for them because none is possible or needed - both files are
+    // namespace MobileGL::MG_Pipe and this one includes that header, so every spelling below
+    // and in package B's code is unchanged.
+    //
+    // What stays here is the BUFFER half of the mapping, which is this file's own: the
+    // BufferTarget table, its sentinel and its completeness assert.
+
+    // A sentinel the table below returns for an enumerator it does not name. It is NOT a
+    // legal mask value: every enumerator must be listed, including the ones that map to no
+    // bit at all, so that ADDING a BufferTarget is a build break here rather than a bit
+    // that silently stops being published.
+    inline constexpr Uint32 kMGPipeBindUnmapped = 0x10000u;
+
+    // The one table. No `default:` arm on purpose - that is what makes the static_assert
+    // below able to see an unnamed enumerator.
+    constexpr Uint32 MGPipeBindMaskForBufferTarget(BufferTarget target) {
+        switch (target) {
+        case BufferTarget::Vertex:
+            return kMGPipeBindVertex;
+        // GL_ELEMENT_ARRAY_BUFFER is the VAO's element slot: the same bind is both "this
+        // resource is an index buffer" and "the server may need its bytes on its own side".
+        case BufferTarget::Index:
+            return kMGPipeBindIndex | kMGPipeBindElementArray;
+        case BufferTarget::Uniform:
+            return kMGPipeBindConstant;
+        case BufferTarget::ShaderStorage:
+            return kMGPipeBindShaderBuffer;
+        case BufferTarget::DispatchIndirect:
+        case BufferTarget::DrawIndirect:
+        case BufferTarget::Parameter:
+            return kMGPipeBindIndirect;
+        // A texture buffer's backing store is SAMPLED through the texture that names it.
+        case BufferTarget::Texture:
+            return kMGPipeBindSampler;
+        case BufferTarget::TransformFeedback:
+            return kMGPipeBindStreamOutput;
+        case BufferTarget::AtomicCounter:
+            return kMGPipeBindAtomic;
+        // TRANSFER AND QUERY TARGETS, which the bind mask deliberately does not name: none
+        // of them is a pipeline binding, none of them makes the server keep anything, and
+        // a bit set for them would only widen what a split server mirrors. Listed rather
+        // than defaulted, so the completeness assert still sees them.
+        case BufferTarget::CopyRead:
+        case BufferTarget::CopyWrite:
+        case BufferTarget::PixelPack:
+        case BufferTarget::PixelUnpack:
+        case BufferTarget::Query:
+            return kMGPipeBindNone;
+        case BufferTarget::BufferTargetCount:
+        case BufferTarget::Unknown:
+            return kMGPipeBindNone;
+        }
+        return kMGPipeBindUnmapped;
+    }
+
+    constexpr Bool MGPipeEveryBufferTargetIsMapped() {
+        for (SizeT i = 0; i < static_cast<SizeT>(BufferTarget::BufferTargetCount); ++i) {
+            if (MGPipeBindMaskForBufferTarget(static_cast<BufferTarget>(i)) == kMGPipeBindUnmapped) {
+                return false;
+            }
+        }
+        return true;
+    }
+    static_assert(MGPipeEveryBufferTargetIsMapped(),
+                  "a BufferTarget enumerator has no MGPResourceDesc::BindMask row: add it to "
+                  "MGPipeBindMaskForBufferTarget, including a deliberate kMGPipeBindNone, or the "
+                  "resource it is bound to stops publishing that binding (D-A3, P8 expectation 1)");
+    static_assert(MGPipeBindMaskForBufferTarget(BufferTarget::Index) & kMGPipeBindElementArray,
+                  "the ELEMENT_ARRAY bit is the index host mirror's switch (ARCHITECTURE.md 10.3)");
+
+    // ---------------------------------------------------------------------------------
+    // The discriminators MGPResourceDesc / MGPSubData carry for a BUFFER
+    // ---------------------------------------------------------------------------------
+    //
+    // P4a MINTED THE FIRST LIST: MGPipeTypes.h now carries enum MGPipeResourceTarget beside
+    // the field, and kMGPipeResourceTargetBuffer moved there with it - the narrowed
+    // resource_respecify ack predicate lives in that header and has to name the buffer target
+    // explicitly, and it may not reach into MG_Impl to do so. The second discriminator is the
+    // frontend enum, named rather than open-coded, and stays here because only this file
+    // produces it.
+    inline constexpr Uint8 kMGPipeResourceStorageKindBuffer =
+        static_cast<Uint8>(MobileGL::TextureStorageType::Buffer);
+
+    // ---------------------------------------------------------------------------------
+    // D-A2: the payload builders. Pure, so a unit case can assert field by field.
+    // ---------------------------------------------------------------------------------
+
+    // The descriptor for `buffer`. `storageDefined` is false for the create that the
+    // constructor emits - storage is defined lazily by the first respecify and a backend
+    // tolerates a resource that has none - and true for every respecify.
+    inline MGPResourceDesc MGPipeBuildResourceDesc(const MG_State::GLState::BufferObject& buffer,
+                                                   MGPipeHandle handle, Uint16 bindMask,
+                                                   Bool storageDefined) {
+        MGPResourceDesc desc{};
+        desc.Resource = handle;
+        desc.Target = static_cast<Uint8>(kMGPipeResourceTargetBuffer);
+        desc.StorageKind = kMGPipeResourceStorageKindBuffer;
+        desc.BindMask = bindMask;
+        if (storageDefined) {
+            // MGPResourceDesc::Width is a Uint32 and that is the CONTRACT's shape, not this
+            // package's, so a store of 4 GiB or more cannot be declared at all. Truncating it
+            // silently is the one answer that must not happen: the applier's range gate would
+            // then refuse the first legal write past the truncated extent as
+            // Fatal{ProtocolCorruption} and name a corruption that is really a narrowing here.
+            // So it is said out loud, once, in every build - the assertion compiles out at
+            // INFO, which is what all three gate builds are.
+            if (buffer.GetSize() > static_cast<SizeT>(0xFFFFFFFFull)) {
+                MGLOG_E_ONCE("MGPipe: buffer %u declares a store of %llu bytes, which does not fit "
+                             "MGPResourceDesc::Width - the descriptor's extent is narrowed and every "
+                             "write past 4 GiB will be refused by the applier's range gate",
+                             buffer.GetExternalIndex(),
+                             static_cast<unsigned long long>(buffer.GetSize()));
+                MOBILEGL_ASSERT(false, "MGPResourceDesc::Width cannot carry this buffer's size");
+            }
+            desc.Width = static_cast<Uint32>(buffer.GetSize());
+            desc.Usage = static_cast<Uint32>(buffer.GetUsage());
+            desc.StorageFlags = static_cast<Uint32>(buffer.GetStorageFlags());
+            desc.Immutable = buffer.IsImmutableStorage() ? 1 : 0;
+            desc.HasDefinedContent = buffer.HasDefinedContent() ? 1 : 0;
+        }
+        // Diagnostics only: a GL name is never an identity, never a memo key and never part
+        // of a content hash (ARCHITECTURE.md 4.2.1).
+        desc.GlNameForDiag = static_cast<Uint32>(buffer.GetExternalIndex());
+        return desc;
+    }
+
+    // The buffer half of MGPSubData: the destination range rides in the box's first
+    // coordinate and first extent, and MGPipeSetSubDataBufferRange is the ONLY spelling of
+    // that convention. Returns false, with the record untouched, when the range does not fit
+    // one record - which is where MGPipeForEachSubDataRecordRange comes in.
+    //
+    // `sourceIsVerbatimLevelShadow` is the record's own question - "are these bytes an
+    // untransformed level shadow?" - and it is a PARAMETER because the answer differs by
+    // caller: resource_subdata hands over the client's own shadow at an offset into it and
+    // says yes; buffer_subdata_resident hands over the application's staging store, or the
+    // locally expanded pattern FillSubData built, and both say no. Nothing reads it on the
+    // buffer path today, which is exactly why it must not be a hard-coded 1 that becomes
+    // wrong the moment something does.
+    //
+    // Blob is FILLED, exactly: Seg is kMGHostSpanSegNone (monolith - the bytes travel beside
+    // the record through the entry point's companion pointer) and Size is the piece's own
+    // byte length, which is what the applier's ONE Blob rule holds a non-zero declaration to
+    // (PipeApply.cpp's SubDataBoxFault: != 0 && != MGPipeSubDataBufferSize is refused).
+    // Leaving it 0 would be legal too; declaring it correctly is the stronger of the two.
+    inline Bool MGPipeBuildSubDataRecord(MGPipeHandle res, Uint64 offset, Uint64 size, MGPSubData& out,
+                                         Bool sourceIsVerbatimLevelShadow) {
+        out = MGPSubData{};
+        out.Res = res;
+        out.Target = kMGPipeResourceTargetBuffer;
+        out.SourceIsVerbatimLevelShadow = sourceIsVerbatimLevelShadow ? 1 : 0;
+        if (!MGPipeSetSubDataBufferRange(out, offset, size)) return false;
+        out.Blob.Seg = kMGHostSpanSegNone;
+        out.Blob.Size = size;
+        return true;
+    }
+
+    // ONE record's destination box caps the offset at 2^31-1 and the size at 2^32-1
+    // (MGPipeTypes.h), so a range beyond either has to be split. The pieces are CONTIGUOUS
+    // and in ASCENDING order, and both properties are load-bearing rather than tidy:
+    // splitting a content write into overlapping or reordered pieces would change what the
+    // backend's queue-and-drain sees, and the Mali WAR-stall fix depends on that queue being
+    // exactly the writes the application made.
+    inline constexpr Uint64 kMGPipeSubDataMaxRecordOffset = 0x7FFFFFFFull;
+    inline constexpr Uint64 kMGPipeSubDataMaxRecordSize = 0xFFFFFFFFull;
+
+    // WITH THE RECORD'S OWN BOUND THE SPLIT IS NOT REACHABLE, and saying so is better than a
+    // loop that reads as if it were: a second piece starts at least 2^32-1 bytes past the
+    // first, which is already past the OFFSET cap, so a range too big for one record is
+    // REFUSED rather than split. The offset cap cannot be split away at all - every piece of
+    // a range that starts past 2^31-1 starts past it too - and a silent truncation is the one
+    // answer that must not happen, so the walk emits nothing and its caller says so once.
+    //
+    // `maxChunk` exists because the record's bound is not the tight one for long: a transport
+    // segment is far smaller (tens of MiB), and that is where this walk starts producing real
+    // splits. It is a parameter now, and exercised at a reachable value by the unit gate, so
+    // that lowering it is one argument rather than a new code path written under pressure.
+    template <class Fn>
+    inline Bool MGPipeForEachSubDataRecordRange(Uint64 offset, Uint64 size, Fn&& piece,
+                                                Uint64 maxChunk = kMGPipeSubDataMaxRecordSize) {
+        if (offset > kMGPipeSubDataMaxRecordOffset) return false;
+        if (size == 0) return true;
+        if (maxChunk == 0) return false;
+        // Every piece has to be encodable BEFORE any of them is emitted: a half-emitted range
+        // is a partial content write the backend would land as if it were the whole one.
+        const Uint64 chunkCap = maxChunk < kMGPipeSubDataMaxRecordSize ? maxChunk : kMGPipeSubDataMaxRecordSize;
+        for (Uint64 at = offset; at < offset + size; at += chunkCap) {
+            if (at > kMGPipeSubDataMaxRecordOffset) return false;
+        }
+        for (Uint64 at = offset, left = size; left > 0;) {
+            const Uint64 chunk = left > chunkCap ? chunkCap : left;
+            piece(at, chunk);
+            at += chunk;
+            left -= chunk;
+        }
+        return true;
+    }
+
+    // ---------------------------------------------------------------------------------
+    // The tracker: handles, the inverse, the sticky mask, the reverse channel
+    // ---------------------------------------------------------------------------------
+
+    class MGPipeResourceTracker {
+    public:
+        using BufferObject = MG_State::GLState::BufferObject;
+        using GLContext = MG_State::GLState::GLContext;
+
+        // The handle for `buffer`, minted on first use. Minting is NOT gated on a backend
+        // having registered MGPipeResourceOps: the handle is CLIENT state and
+        // set_vertex_buffers names it whether or not the resource family is switched on, so
+        // gating it would make the vertex-input subsystem emit null handles whenever the
+        // resource subsystem is off. Only the CALLS are gated (D-A1).
+        MGPipeHandle Acquire(BufferObject& buffer) {
+            const MGPipeHandle handle = MGPipeSlots().Acquire(MGPipeKind::Buffer, buffer.GetLifetimeId());
+            const SizeT slot = handle.Slot;
+            if (slot >= m_bySlot.size()) m_bySlot.resize(slot + 1);
+            m_bySlot[slot].Object = &buffer;
+            m_bySlot[slot].Gen = handle.Gen;
+            return handle;
+        }
+
+        // The handle a buffer already has, or the null handle. Never mints - the emission
+        // path calls Acquire, the query paths call this.
+        MGPipeHandle Find(const BufferObject& buffer) const {
+            return MGPipeSlots().FindByLifetimeId(MGPipeKind::Buffer, buffer.GetLifetimeId());
+        }
+
+        // D-D's inverse, and a RAW pointer is exact here: the entry exists only between the
+        // create the constructor emits and the destroy the destructor emits, and a readback
+        // is only ever issued for a live, bound buffer. A WeakPtr would be wrong - the
+        // object does not own itself through a SharedPtr at those two moments. The Gen
+        // compare is what refuses a stale handle rather than resolving it to whatever now
+        // occupies the slot.
+        BufferObject* Resolve(MGPipeHandle handle) const {
+            const SizeT slot = handle.Slot;
+            if (MGPipeHandleIsNull(handle) || slot >= m_bySlot.size()) return nullptr;
+            const Entry& entry = m_bySlot[slot];
+            if (entry.Object == nullptr || entry.Gen != handle.Gen) return nullptr;
+            if (MGPipeSlots().GenOfSlot(MGPipeKind::Buffer, handle.Slot) != handle.Gen) return nullptr;
+            return entry.Object;
+        }
+
+        // Drops the inverse entry and the sticky mask. The CALLER frees the slot afterwards,
+        // in that order (D-L): MGPipeSlotAllocator::Free erases the lifetimeId -> slot
+        // mapping, so anything that has to resolve the handle must do it first.
+        void Retire(MGPipeHandle handle) {
+            const SizeT slot = handle.Slot;
+            if (slot >= m_bySlot.size()) return;
+            m_bySlot[slot] = Entry{};
+        }
+
+        // ---- D-L: was resource_create actually EMITTED for this slot? ----
+        //
+        // The create is gated at its call site (BufferObject's constructor) and the destroy
+        // is gated inside MGPipeEmitResourceDestroyAndFree, so the two ask the SAME question
+        // at two different moments. A buffer constructed while a backend's table was
+        // registered and destroyed after UnregisterBufferBackendOps() would take the second
+        // answer, free its slot, and leave the applier's record Live - on a slot the
+        // allocator is about to hand out again, with the backend's twin (a driver buffer id)
+        // still attached to it. So the answer is LATCHED at the create and the destroy uses
+        // the latched one; the two are then a pair by construction rather than by the
+        // registration outliving every buffer.
+        void NotePublished(MGPipeHandle handle) {
+            const SizeT slot = handle.Slot;
+            if (slot >= m_bySlot.size()) return;
+            m_bySlot[slot].Published = true;
+        }
+        Bool WasPublished(MGPipeHandle handle) const {
+            const SizeT slot = handle.Slot;
+            return slot < m_bySlot.size() && m_bySlot[slot].Published;
+        }
+
+        // The sticky everBoundAs mask. Sticky exactly as MGPResourceDesc::ImageBindableHint's
+        // everImageBound is: ORed, never cleared, so a buffer that was an element array once
+        // keeps saying so.
+        Uint16 BindMask(MGPipeHandle handle) const {
+            const SizeT slot = handle.Slot;
+            return slot < m_bySlot.size() ? m_bySlot[slot].BindMask : Uint16{0};
+        }
+
+        // OR one target's bit into a handle's sticky mask, without looking at the context at
+        // all. This is what closes the sampling window for the two bits anything keys on:
+        // the vertex-input emitters resolve, at EVERY draw, exactly the attribute buffers and
+        // the element-slot buffer, so any buffer ever DRAWN FROM carries its ARRAY_BUFFER /
+        // ELEMENT_ARRAY bit for the rest of its life whether or not it happened to be bound
+        // at a storage op. It grows the table rather than dropping the note: it is called
+        // from the validate point, which is GL-thread by construction, and a slot outside the
+        // table is a buffer whose mint this process has not seen (a unit fixture's
+        // ResetForTest, in practice).
+        void NoteBoundAs(MGPipeHandle handle, BufferTarget target) {
+            if (MGPipeHandleIsNull(handle)) return;
+            const SizeT slot = handle.Slot;
+            if (slot >= m_bySlot.size()) m_bySlot.resize(slot + 1);
+            m_bySlot[slot].BindMask |= static_cast<Uint16>(MGPipeBindMaskForBufferTarget(target));
+        }
+
+        // Accumulates into the sticky mask every target `buffer` is bound to RIGHT NOW, and
+        // returns the accumulated value.
+        //
+        // [DEVIATION, recorded in client-v2.md] D-A3 asks for the OR at every glBindBuffer /
+        // glBindBufferBase / glBindBufferRange / VAO element-slot bind, and C.1 points at
+        // MG_State/GLState/BufferState/BufferState.{h,cpp} for it - a file this package DOES
+        // own. The brief is wrong about where the entry points are: BufferState only VENDS
+        // BindingSlot<BufferObject>& / BindingSlotRange1D&, and the .Bind() calls are
+        // MG_Impl/GLImpl/Buffer/GL_Buffer.cpp's (BindBuffer_State, BindBufferBase_State,
+        // BindBufferRange_State), which C.5 assigns to no package. So the mask is accumulated
+        // by SAMPLING the frontend's live binding state instead - here, at every create and
+        // respecify, which is where the value is PUBLISHED - and ORed into a per-slot sticky
+        // field that is never cleared.
+        //
+        // WHAT SAMPLING ALONE CANNOT SEE is not "a bind after the last respecify" (which the
+        // specified design misses too) but a TRANSIENT bind: bind an EBO, draw, unbind, then
+        // define it through DSA - the respecify's sample sees no binding at all, and the DSA
+        // idiom makes that the common case rather than a corner (TryAdoptLargeStorage's own
+        // comment names glNamedBufferSubData as what MC 26.3 streams with). That hole is
+        // closed for the two bits anything keys on by NoteBoundAs above, called from
+        // EmitVertexBuffers / EmitIndexBuffer at every draw. What is left unpublished is a
+        // buffer that is bound, never drawn from, and never re-specified afterwards; the
+        // remaining fix is one line in each of GL_Buffer.cpp's three *_State binders, for the
+        // seven bits nothing keys on yet, and it stays handed to whoever owns that file.
+        //
+        // The scan is skipped unless a binding-slot version moved since the last one, which
+        // is one Uint16 load per global target and none per binding point. It is NOT called
+        // from the content emitters, deliberately: it walks the whole context's binding state
+        // and writes the tracker, and one of those emitters (resource_subdata) is on the path
+        // D-A2 preserves as reachable off the render thread. Extra sampling could only widen
+        // a sticky union, but not at the price of a context-wide read from the wrong thread.
+        Uint16 RefreshBindMask(GLContext& ctx, const BufferObject& buffer, MGPipeHandle handle) {
+            const SizeT slot = handle.Slot;
+            if (slot >= m_bySlot.size()) return 0;
+            Entry& entry = m_bySlot[slot];
+            const Uint64 epoch = BindEpoch(ctx);
+            if (epoch == m_bindEpoch && entry.BindMaskEpoch == epoch) return entry.BindMask;
+            m_bindEpoch = epoch;
+            entry.BindMaskEpoch = epoch;
+            Uint16 mask = entry.BindMask;
+            for (const auto target : MG_State::GLState::GlobalBufferTargets) {
+                if (ctx.GetBufferBindingSlot(target).GetBoundObject().get() == &buffer) {
+                    mask |= static_cast<Uint16>(MGPipeBindMaskForBufferTarget(target));
+                }
+            }
+            for (const auto target : MG_State::GLState::BufferBindPointTargets) {
+                const SizeT touched = ctx.GetTouchedBufferBindingPointCount(target);
+                for (SizeT i = 0; i < touched; ++i) {
+                    if (ctx.GetBufferBindingPoint(target, static_cast<Uint>(i)).GetBoundObject().get() == &buffer) {
+                        mask |= static_cast<Uint16>(MGPipeBindMaskForBufferTarget(target));
+                        break;
+                    }
+                }
+            }
+            // The index slot is the BOUND VAO's, not BufferState's, so it is not in
+            // GlobalBufferTargets and GetBufferBindingSlot(Index) asserts without a VAO.
+            if (const auto& vao = ctx.GetBoundVertexArray()) {
+                if (vao->GetIndexBufferBindingSlot().GetBoundObject().get() == &buffer) {
+                    mask |= static_cast<Uint16>(MGPipeBindMaskForBufferTarget(BufferTarget::Index));
+                }
+                for (int i = 0; i < MG_State::GLState::VertexArrayObject::MAX_VERTEX_ATTRIBS; ++i) {
+                    if (vao->GetAttribute(static_cast<Uint>(i)).Buffer.get() == &buffer) {
+                        mask |= static_cast<Uint16>(MGPipeBindMaskForBufferTarget(BufferTarget::Vertex));
+                        break;
+                    }
+                }
+            }
+            entry.BindMask = mask;
+            return mask;
+        }
+
+        // ---- the two observables a unit case reads (see the header comment) ----
+        const MGPResourceDesc& LastDesc() const { return m_lastDesc; }
+        Uint64 CreateCount() const { return m_creates; }
+        Uint64 RespecifyCount() const { return m_respecifies; }
+        Uint64 DestroyCount() const { return m_destroys; }
+        Uint64 MapPersistentCount() const { return m_mapPersistents; }
+
+        void NoteDesc(const MGPResourceDesc& desc, Bool isCreate) {
+            m_lastDesc = desc;
+            if (isCreate) {
+                ++m_creates;
+            } else {
+                ++m_respecifies;
+            }
+        }
+        void NoteDestroy() { ++m_destroys; }
+        void NoteMapPersistent() { ++m_mapPersistents; }
+
+        // A unit fixture's per-case reset, and the library never calls it. THE RULE, stated
+        // rather than left as an absence, because "nothing resets this" is not a reason:
+        //
+        //   A buffer handle and the applier record it names are SHARE-GROUP OBJECT STATE.
+        //   A GL object lives in a share group, not in a context, so a make-current changes
+        //   neither. The applier's MGPipeApplierReset() is a make-current and deliberately
+        //   keeps its Resources / VertexElementsCsos (PipeApply.h says so beside them); the
+        //   ONLY things that drop a record are the object's own death signal -
+        //   resource_destroy, which ~BufferObject raises through
+        //   MGPipeEmitResourceDestroyAndFree, and delete_vertex_elements - and
+        //   MGPipeApplierReleaseObjectRecords(), which is the SERVED CONTEXT's teardown and
+        //   is deliberately wired to nothing in the monolith (there is one applier behind
+        //   every context, so calling it on one context's destruction would drop every other
+        //   context's records).
+        //
+        // So this tracker needs no re-publication path on a fresh context and must not have
+        // one: re-emitting resource_create for a record the applier still holds would move
+        // its Serial for nothing. What the client owes instead is the destroy - which
+        // ~BufferObject already emits, in the fixed emit-then-free order (D-L) - and that is
+        // the whole of the client's side of the record lifecycle.
+        //
+        // The vertex-input emitter's latches are the OTHER half and are genuinely per
+        // context: MGPipeVertexInputEmitter::Reset() is called from the FreshlyPrimed arm
+        // because the applier's vertex-input WORKING state (the bound handle, the window, the
+        // fetch shift) IS cleared there. Its vertex-elements RECORDS are not, which is why
+        // the emitter's Reset drops the "already published" latches but no create is lost:
+        // the latch is what says "re-publish", and re-publishing an unchanged configuration
+        // is a bounded over-fire, not a dropped write.
+        void ResetForTest() {
+            m_bySlot.clear();
+            m_bindEpoch = 0;
+            m_lastDesc = MGPResourceDesc{};
+            m_creates = m_respecifies = m_destroys = m_mapPersistents = 0;
+        }
+
+    private:
+        struct Entry {
+            BufferObject* Object = nullptr;
+            Uint32 Gen = 0;
+            Uint16 BindMask = 0;
+            Bool Published = false;
+            Uint64 BindMaskEpoch = 0;
+        };
+
+        // "Has any buffer binding moved since the last scan": the sum of the binding-slot
+        // versions, which BindingSlot bumps only on a real change. A collision costs one
+        // skipped rescan of ONE buffer's mask, and the mask is re-scanned at the next
+        // emission whose epoch differs, so it can delay a bit by one storage op and never
+        // drop one - the same over-fire-is-free / under-fire-is-fatal direction every
+        // shutter in Tracker.h takes.
+        //
+        // IT DOES NOT SEE THE 84x4 INDEXED BINDING POINTS, and that is sound only because
+        // BindBufferBase_State / BindBufferRange_State also bind the GENERIC slot for the
+        // same target (GL_Buffer.cpp:1531 says why), so an indexed bind always moves one of
+        // the versions summed here. If that ever stops being true, the CONSTANT /
+        // SHADER_BUFFER / ATOMIC / STREAM_OUTPUT bits start being missed silently and the
+        // repair is to fold GetTouchedBufferBindingPointCount into the epoch.
+        static Uint64 BindEpoch(GLContext& ctx) {
+            Uint64 epoch = 1;
+            for (const auto target : MG_State::GLState::GlobalBufferTargets) {
+                epoch += ctx.GetBufferBindingSlot(target).GetVersion();
+                epoch *= 3;
+            }
+            if (const auto& vao = ctx.GetBoundVertexArray()) {
+                epoch += vao->GetIndexBufferBindingSlot().GetVersion();
+                epoch = MGPipeMixShutterValue(epoch, vao->GetLifetimeId());
+                epoch = MGPipeMixShutterValue(epoch, vao->GetConfigVersion());
+            }
+            return epoch;
+        }
+
+        // The same mix Tracker.h's composite shutters use. Spelled here rather than
+        // included so this header does not depend on the tracker.
+        static constexpr Uint64 MGPipeMixShutterValue(Uint64 accumulator, Uint64 value) {
+            accumulator ^= value + 0x9e3779b97f4a7c15ull + (accumulator << 6) + (accumulator >> 2);
+            return accumulator;
+        }
+
+        Vector<Entry> m_bySlot;
+        Uint64 m_bindEpoch = 0;
+        MGPResourceDesc m_lastDesc{};
+        Uint64 m_creates = 0;
+        Uint64 m_respecifies = 0;
+        Uint64 m_destroys = 0;
+        Uint64 m_mapPersistents = 0;
+    };
+
+    // The monolith's one resource tracker, beside the state tracker, the CSO cache and the
+    // set-hash suppressor.
+    inline MGPipeResourceTracker& MGPipeResourceTrackerInstance() {
+        // NEVER DESTROYED, for MGPipeSlots()' reason (SlotAllocator.cpp): ~BufferObject reads
+        // and writes this tracker, and the objects that own the last reference to a
+        // BufferObject outlive every function-local static.
+        static MGPipeResourceTracker* tracker = new MGPipeResourceTracker();
+        return *tracker;
+    }
+
+    // ---------------------------------------------------------------------------------
+    // D-D: the client's half of the reverse channel
+    // ---------------------------------------------------------------------------------
+
+    // The backend produced the bytes of a readback and hands them back through the channel.
+    // The client resolves the handle to its own object and writes the shadow; the epoch bump
+    // stays SERVER-side and happens AFTER this returns, never before (ARCHITECTURE.md 7.4:
+    // the reverse channel needs the same ordering guarantee as the forward one).
+    inline void MGPipeClientOnBufferWriteback(MGPipeHandle res, Uint64 offset, MGPBlobRef bytes) {
+        auto* buffer = MGPipeResourceTrackerInstance().Resolve(res);
+        if (buffer == nullptr) {
+            MGLOG_E_ONCE("MGPipe: OnBufferWriteback for a handle {%u,%u} that resolves to no buffer",
+                         res.Slot, res.Gen);
+            return;
+        }
+        if (bytes.Seg != kMGHostSpanSegNone) {
+            MGLOG_E_ONCE("MGPipe: OnBufferWriteback carried a transport segment (%u); P3a is monolith only",
+                         bytes.Seg);
+            return;
+        }
+        // Monolith: Seg is kMGHostSpanSegNone and Offset IS the address of the backend's
+        // mapped bytes (MGPipeTypes.h says so in as many words). Under a transport the
+        // segment resolves first, and that is the phase's edit, not this one's.
+        buffer->WritebackFromBackend(
+            DataPtr{reinterpret_cast<void*>(static_cast<std::uintptr_t>(bytes.Offset)),
+                    static_cast<SizeT>(bytes.Size)},
+            static_cast<SizeT>(offset));
+    }
+
+    // A draw or dispatch wrote these ranges. ARCHITECTURE.md 7.1 calls this a NARROWING
+    // channel - the client builds a conservative pending set at its own emission points and
+    // the callback only ever removes from it - so P3a's implementation marks exactly what
+    // the three Espryt MarkGpuWritten sites mark today and the observable behaviour is
+    // unchanged. The narrowing itself is P8/P9's.
+    inline void MGPipeClientOnGpuWritten(MGPipeHandle res, Uint rangeCount, const MGPRange* ranges) {
+        // THE SHAPE IS A CONTRACT POINT, not a formality: the announcement is ONE range
+        // covering kMGPipeWholeBuffer, deliberately not ZERO ranges, because zero will mean
+        // "a fully narrowed set - nothing is dirty" at P8/P9. Marking the whole buffer
+        // written for a zero-range announcement would be the narrowing channel run backwards,
+        // so the shape is asserted here rather than assumed.
+        MOBILEGL_ASSERT(rangeCount == 1 && ranges != nullptr,
+                        "OnGpuWritten {slot=%u, gen=%u}: P3a announces exactly one whole-buffer range, "
+                        "not %u",
+                        res.Slot, res.Gen, static_cast<Uint>(rangeCount));
+        (void)ranges;
+        if (rangeCount == 0) return;
+        auto* buffer = MGPipeResourceTrackerInstance().Resolve(res);
+        if (buffer == nullptr) {
+            // Loud, like its sibling above: a backend announcing a write against a handle
+            // this client cannot resolve is a dropped MarkGpuWritten, and a dropped
+            // MarkGpuWritten is a stale shadow read back as if it were current.
+            MGLOG_E_ONCE("MGPipe: OnGpuWritten for a handle {%u,%u} that resolves to no buffer", res.Slot,
+                         res.Gen);
+            return;
+        }
+        buffer->MarkGpuWritten();
+    }
+
+    // Installed once, and never over an entry a backend already claimed: these two are the
+    // CLIENT's implementations of a backend -> frontend callback, so the backend installs
+    // the rest of the table and these two answer for it.
+    inline void MGPipeInstallClientResourceCallbacks() {
+        if (gMGPipeCallbacks.OnBufferWriteback == nullptr) {
+            gMGPipeCallbacks.OnBufferWriteback = &MGPipeClientOnBufferWriteback;
+        }
+        if (gMGPipeCallbacks.OnGpuWritten == nullptr) {
+            gMGPipeCallbacks.OnGpuWritten = &MGPipeClientOnGpuWritten;
+        }
+    }
+} // namespace MobileGL::MG_Pipe
+#endif // MOBILEGL_PIPE_PUSH

MobileGL/MG_Impl/Pipe/SetHashSuppressor.h

@@ -0,0 +1,111 @@
+// MobileGL - MobileGL/MG_Impl/Pipe/SetHashSuppressor.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>
+
+// Coalescing rule 4 (ARCHITECTURE.md 5.4, P2 brief D11): every kVarTail set_* hashes the
+// RESOLVED set on the client and does not emit when the hash has not moved.
+//
+// This is the carrier for the ~175 lines of debounce that move off the backends in P3b and
+// P4b - Espryt's UnitBindingsSnapshot / CaptureUnitBindings / UnitBindingsUnchanged and
+// Magma's equivalents all answer "is this set the same set as last time", and every one of
+// them answers it against a shape the backend rediscovered. P2 lands the MECHANISM and ONE
+// real consumer (SetVertexAttribDefaults) so the shape is pinned by a test rather than by a
+// plan; the other six slots exist, are unit-tested, and are wired by the phase that moves
+// the set they name. P3a wires the second, SetVertexBuffers. P4a wires SetSamplerViews,
+// BindSamplerStates and SetShaderImages, and APPENDS an eighth slot, SetFramebufferState -
+// which leaves only SetShaderBuffers and SetStreamOutputTargets unwired, both P4b's.
+//
+// A WIRED SLOT PUTS A REQUIREMENT ON ITS HASH, and SetVertexBuffers is where that first
+// bites: the hash has to cover EVERY input the record carries, not only the set. Its
+// baseInstance is DRAW state and moves without the buffer set moving, so a hash over the
+// entries alone would suppress a record whose one changed field is the fetch shift and the
+// server would keep the previous one. MG_Impl/Pipe/VertexInputEmit.h's
+// MGPipeVertexBufferSetContentHash mixes Start, Count and BaseInstance in for exactly that
+// reason, and VertexInputEmit's base-instance pair is the test that says so.
+//
+// A hash of 0 is reserved for "never emitted", so the first emission always goes out; a
+// computed 0 is remapped to 1, which costs one collision in 2^64 an extra emission and
+// never a missed one.
+//
+// Header-only for the same ownership reason as Tracker.h and CsoCache.h: the root
+// CMakeLists.txt that would name a new .cpp belongs to package A and is frozen behind the
+// p2/contract tag.
+#if MOBILEGL_PIPE_PUSH
+#include <MG_Pipe/MGPipe.h>
+
+namespace MobileGL::MG_Pipe {
+
+    // One slot per kVarTail set_* (ARCHITECTURE.md 5.1's call list), PLUS
+    // SetFramebufferState, which is not kVarTail at all: MGPFramebufferState carries a
+    // ContentHash for TWO jobs - the server's render-pass memo key and the client's emission
+    // suppressor - and the second one needs a slot here like any other. The enum is
+    // CLIENT-ONLY and is not a wire opcode, so appending before Count is safe.
+    enum class MGPipeSuppressorSlot : Uint32 {
+        SetVertexBuffers = 0,     // P3a - wired, and its hash includes BaseInstance
+        // P4a - WIRED. The three unit sets' suppressors are not optional and were never a
+        // later phase's: MGPipeTypes.h makes the pattern mandatory for every kVarTail set_*,
+        // because GetTextureBindGeneration() bumps on a REDUNDANT rebind - MC 26.2 rebinds the
+        // same sampler at every texture-unit switch - so an unsuppressed set is a
+        // several-hundred-byte variable-length record per batch, which is the exact regression
+        // the design names. What P3b/P4b owns is the ~175-line BACKEND debounce these replace
+        // (UnitBindingsSnapshot / CaptureUnitBindings / UnitBindingsUnchanged and the two
+        // g_*SyncList tables); P4a wires the carrier, P3b/P4b deletes the backend copy.
+        SetSamplerViews,          // P4a - wired (backend debounce deletion: P3b/P4b)
+        BindSamplerStates,        // P4a - wired (backend debounce deletion: P3b/P4b)
+        SetShaderImages,          // P4a - wired (backend debounce deletion: P3b/P4b)
+        SetShaderBuffers,         // P4b
+        SetStreamOutputTargets,   // P4b
+        SetVertexAttribDefaults,  // P2 - the one consumer that is wired
+        SetFramebufferState,      // P4a - wired
+        Count,
+    };
+
+    inline constexpr SizeT kMGPipeSuppressorSlotCount = static_cast<SizeT>(MGPipeSuppressorSlot::Count);
+
+    class MGPipeSetHashSuppressor {
+    public:
+        // True when `contentHash` differs from what this slot last emitted, and LATCHES it.
+        // False means the resolved set has not moved and the call must not go out.
+        Bool ShouldEmit(MGPipeSuppressorSlot slot, Uint64 contentHash) {
+            const Uint64 latched = contentHash == 0 ? 1 : contentHash;
+            const SizeT index = static_cast<SizeT>(slot);
+            if (m_lastEmitted[index] == latched) return false;
+            m_lastEmitted[index] = latched;
+            return true;
+        }
+
+        // A context change or a server reset: what the server has is no longer what this
+        // slot last emitted, so the next resolved set must go out whatever it hashes to.
+        void Invalidate(MGPipeSuppressorSlot slot) { m_lastEmitted[static_cast<SizeT>(slot)] = 0; }
+
+        void InvalidateAll() {
+            for (SizeT i = 0; i < kMGPipeSuppressorSlotCount; ++i) m_lastEmitted[i] = 0;
+        }
+
+        // 0 == "never emitted". Exposed for the unit test, which is what pins that the
+        // reserved value really is reserved.
+        Uint64 LastEmitted(MGPipeSuppressorSlot slot) const {
+            return m_lastEmitted[static_cast<SizeT>(slot)];
+        }
+
+    private:
+        Array<Uint64, kMGPipeSuppressorSlotCount> m_lastEmitted{};
+    };
+
+    // The monolith's one suppressor, beside the tracker and the CSO cache.
+    inline MGPipeSetHashSuppressor& MGPipeSetHashSuppressorInstance() {
+        // NEVER DESTROYED, for MGPipeTrackerInstance()' reason (MG_Impl/Pipe/Tracker.h): the
+        // rule covers every MGPipe process singleton, not only the ones on today's death
+        // paths.
+        static MGPipeSetHashSuppressor* suppressor = new MGPipeSetHashSuppressor();
+        return *suppressor;
+    }
+} // namespace MobileGL::MG_Pipe
+#endif // MOBILEGL_PIPE_PUSH

MobileGL/MG_Impl/Pipe/SlotAllocator.cpp

@@ -0,0 +1,280 @@
+// MobileGL - MobileGL/MG_Impl/Pipe/SlotAllocator.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
+
+// SlotAllocator.h. Compiled only under MOBILEGL_PIPE_PUSH.
+#include <MG_Impl/Pipe/SlotAllocator.h>
+
+namespace MobileGL::MG_Pipe {
+    namespace {
+        // The ShaderCso band the ordinary allocator must never enter: the top 1/16 of the
+        // ShaderCso slot space is reserved for PROGRAM PIPELINE COMPOSITES, which are minted
+        // client-side out of the stage programs bound to a pipeline object. Reserving a band
+        // rather than a flag keeps the composite resolver's lifetime bookkeeping out of here
+        // (MGPipeHandles.h, ARCHITECTURE.md 5.6.3).
+        Bool SlotIsAllocatable(MGPipeKind kind, Uint32 slot) {
+            if (slot < kMGPipeFirstAllocatableSlot) return false;
+            if (kind != MGPipeKind::ShaderCso) return true;
+            return slot < kMGPipeShaderCsoCompositeSlotBase;
+        }
+    } // namespace
+
+    MGPipeSlotAllocator::KindState& MGPipeSlotAllocator::StateOf(MGPipeKind kind) {
+        const SizeT index = static_cast<SizeT>(kind);
+        MOBILEGL_ASSERT(index < kKindCount, "MGPipeKind %zu out of range", index);
+        return m_kinds[index < kKindCount ? index : 0];
+    }
+
+    const MGPipeSlotAllocator::KindState& MGPipeSlotAllocator::StateOf(MGPipeKind kind) const {
+        const SizeT index = static_cast<SizeT>(kind);
+        MOBILEGL_ASSERT(index < kKindCount, "MGPipeKind %zu out of range", index);
+        return m_kinds[index < kKindCount ? index : 0];
+    }
+
+    MGPipeSlotAllocator::SlotState* MGPipeSlotAllocator::EntryOf(KindState& state, MGPipeKind kind,
+                                                                 Uint32 slot) {
+        if (kind == MGPipeKind::ShaderCso && MGPipeIsCompositeShaderSlot(slot)) {
+            const SizeT index = slot - kMGPipeShaderCsoCompositeSlotBase;
+            if (index >= state.BandSlots.size()) return nullptr;
+            return &state.BandSlots[index];
+        }
+        if (slot >= state.Slots.size()) return nullptr;
+        return &state.Slots[slot];
+    }
+
+    const MGPipeSlotAllocator::SlotState*
+    MGPipeSlotAllocator::EntryOf(const KindState& state, MGPipeKind kind, Uint32 slot) {
+        return EntryOf(const_cast<KindState&>(state), kind, slot);
+    }
+
+    MGPipeHandle MGPipeSlotAllocator::Allocate(MGPipeKind kind) {
+        KindState& state = StateOf(kind);
+        if (state.Slots.empty()) {
+            // Slot 0 exists so the vector is slot-indexed, and is never handed out.
+            state.Slots.resize(kMGPipeFirstAllocatableSlot);
+        }
+
+        Uint32 slot = 0;
+        Bool reused = false;
+        while (!state.FreeList.empty()) {
+            const Uint32 candidate = state.FreeList.back();
+            state.FreeList.pop_back();
+            if (!SlotIsAllocatable(kind, candidate)) continue;
+            slot = candidate;
+            reused = true;
+            break;
+        }
+
+        if (!reused) {
+            slot = static_cast<Uint32>(state.Slots.size());
+            MOBILEGL_ASSERT(SlotIsAllocatable(kind, slot),
+                            "MGPipe slot space of kind %u is exhausted at slot %u",
+                            static_cast<Uint32>(kind), slot);
+            if (!SlotIsAllocatable(kind, slot)) return kMGPipeNullHandle;
+            state.Slots.emplace_back();
+        }
+
+        SlotState& entry = state.Slots[slot];
+        if (entry.EverHandedOut) {
+            // The one place Gen may move. 2^32 recycles of ONE slot is ~50 days of continuous
+            // churn at one recycle per frame at 1000 fps, which is why the bound is asserted
+            // in a debug allocator rather than defended in release.
+            MOBILEGL_ASSERT(entry.Gen != ~Uint32{0},
+                            "MGPipe handle generation wrapped on kind %u slot %u; {slot, gen} is "
+                            "no longer unique",
+                            static_cast<Uint32>(kind), slot);
+            ++entry.Gen;
+        }
+        entry.EverHandedOut = true;
+        entry.Live = true;
+        entry.LifetimeId = 0;
+        ++state.LiveCount;
+        return MGPipeHandle{slot, entry.Gen};
+    }
+
+    MGPipeHandle MGPipeSlotAllocator::AllocateFor(MGPipeKind kind, Uint64 lifetimeId) {
+        const MGPipeHandle handle = Allocate(kind);
+        if (MGPipeHandleIsNull(handle)) return handle;
+        KindState& state = StateOf(kind);
+        state.Slots[handle.Slot].LifetimeId = lifetimeId;
+        if (lifetimeId != 0) {
+            MOBILEGL_ASSERT(state.ByLifetimeId.find(lifetimeId) == state.ByLifetimeId.end(),
+                            "lifetime id %llu already owns a slot of kind %u",
+                            static_cast<unsigned long long>(lifetimeId), static_cast<Uint32>(kind));
+            state.ByLifetimeId[lifetimeId] = handle.Slot;
+        }
+        return handle;
+    }
+
+    MGPipeHandle MGPipeSlotAllocator::AllocateComposite(Uint64 lifetimeId) {
+        // P4a, D-H7. The mirror image of Allocate() above, restricted to the band that one
+        // refuses, and kept in a table of its own so both spaces stay DENSE: the band's base
+        // is 983040, and minting one composite into the slot-indexed vector would allocate
+        // ~23 MB of SlotState for a single program pipeline.
+        KindState& state = StateOf(MGPipeKind::ShaderCso);
+
+        Uint32 slot = 0;
+        Bool reused = false;
+        if (!state.BandFreeList.empty()) {
+            slot = state.BandFreeList.back();
+            state.BandFreeList.pop_back();
+            reused = true;
+        }
+
+        if (!reused) {
+            const SizeT next = kMGPipeShaderCsoCompositeSlotBase + state.BandSlots.size();
+            slot = static_cast<Uint32>(next);
+            // The band's own exhaustion assert, mirroring Allocate()'s: a composite that
+            // cannot be minted is a NAMED failure, not a silent fall-through into the ordinary
+            // program slots, which is exactly what reserving a band rather than setting a flag
+            // buys.
+            MOBILEGL_ASSERT(next < kMGPipeShaderCsoSlotLimit,
+                            "the MGPipe ShaderCso COMPOSITE band is exhausted at slot %zu; a "
+                            "program-pipeline composite cannot be minted and must not take an "
+                            "ordinary program's slot",
+                            next);
+            if (next >= kMGPipeShaderCsoSlotLimit) return kMGPipeNullHandle;
+            state.BandSlots.emplace_back();
+        }
+
+        SlotState* entry = EntryOf(state, MGPipeKind::ShaderCso, slot);
+        if (entry == nullptr) return kMGPipeNullHandle;
+        if (entry->EverHandedOut) {
+            MOBILEGL_ASSERT(entry->Gen != ~Uint32{0},
+                            "MGPipe handle generation wrapped on the ShaderCso composite band, "
+                            "slot %u; {slot, gen} is no longer unique",
+                            slot);
+            ++entry->Gen;
+        }
+        entry->EverHandedOut = true;
+        entry->Live = true;
+        entry->LifetimeId = lifetimeId;
+        ++state.LiveCount;
+        // The band's share of LiveCount, so CompositeLiveCount() can answer without a walk.
+        ++state.BandLiveCount;
+        if (lifetimeId != 0) {
+            MOBILEGL_ASSERT(state.ByLifetimeId.find(lifetimeId) == state.ByLifetimeId.end(),
+                            "lifetime id %llu already owns a ShaderCso slot",
+                            static_cast<unsigned long long>(lifetimeId));
+            state.ByLifetimeId[lifetimeId] = slot;
+        }
+        return MGPipeHandle{slot, entry->Gen};
+    }
+
+    MGPipeHandle MGPipeSlotAllocator::FindByLifetimeId(MGPipeKind kind, Uint64 lifetimeId) const {
+        if (lifetimeId == 0) return kMGPipeNullHandle;
+        const KindState& state = StateOf(kind);
+        const auto it = state.ByLifetimeId.find(lifetimeId);
+        if (it == state.ByLifetimeId.end()) return kMGPipeNullHandle;
+        const SlotState* entry = EntryOf(state, kind, it->second);
+        if (entry == nullptr || !entry->Live) return kMGPipeNullHandle;
+        return MGPipeHandle{it->second, entry->Gen};
+    }
+
+    MGPipeHandle MGPipeSlotAllocator::Acquire(MGPipeKind kind, Uint64 lifetimeId) {
+        const MGPipeHandle existing = FindByLifetimeId(kind, lifetimeId);
+        if (!MGPipeHandleIsNull(existing)) return existing;
+        return AllocateFor(kind, lifetimeId);
+    }
+
+    void MGPipeSlotAllocator::Free(MGPipeKind kind, MGPipeHandle handle) {
+        KindState& state = StateOf(kind);
+        SlotState* entry = EntryOf(state, kind, handle.Slot);
+        if (entry == nullptr) return;
+        // A stale handle must not free the slot its successor now owns - that is the whole
+        // reason the generation is in the key. It is also what makes the SECOND of a
+        // composite's two independent release paths a proven no-op.
+        if (!entry->Live || entry->Gen != handle.Gen) return;
+        if (entry->LifetimeId != 0) {
+            const auto it = state.ByLifetimeId.find(entry->LifetimeId);
+            if (it != state.ByLifetimeId.end() && it->second == handle.Slot) {
+                state.ByLifetimeId.erase(it);
+            }
+        }
+        entry->Live = false;
+        entry->LifetimeId = 0;
+        --state.LiveCount;
+        if (kind == MGPipeKind::ShaderCso && MGPipeIsCompositeShaderSlot(handle.Slot)) {
+            --state.BandLiveCount;
+            state.BandFreeList.push_back(handle.Slot);
+        } else {
+            state.FreeList.push_back(handle.Slot);
+        }
+    }
+
+    Bool MGPipeSlotAllocator::IsLive(MGPipeKind kind, MGPipeHandle handle) const {
+        const SlotState* entry = EntryOf(StateOf(kind), kind, handle.Slot);
+        return entry != nullptr && entry->Live && entry->Gen == handle.Gen;
+    }
+
+    Uint32 MGPipeSlotAllocator::GenOfSlot(MGPipeKind kind, Uint32 slot) const {
+        const SlotState* entry = EntryOf(StateOf(kind), kind, slot);
+        return entry != nullptr ? entry->Gen : 0;
+    }
+
+    Uint64 MGPipeSlotAllocator::LifetimeIdOfSlot(MGPipeKind kind, Uint32 slot) const {
+        const SlotState* entry = EntryOf(StateOf(kind), kind, slot);
+        return entry != nullptr ? entry->LifetimeId : 0;
+    }
+
+    Uint32 MGPipeSlotAllocator::HighWater(MGPipeKind kind) const {
+        // THE ORDINARY SPACE ONLY, and the band is reported by CompositeHighWater() below.
+        // Folding the two would pin this at ~983k from the first composite mint onward and
+        // take the ordinary space's "the high-water mark did not move" assertion away for the
+        // rest of the process - the assertion that catches a dense table that never shrinks,
+        // which is the leak shape this allocator exists to make visible. Two spaces, two
+        // numbers, two real assertions. See SlotAllocator.h.
+        return static_cast<Uint32>(StateOf(kind).Slots.size());
+    }
+
+    Uint32 MGPipeSlotAllocator::CompositeHighWater() const {
+        const KindState& state = StateOf(MGPipeKind::ShaderCso);
+        // One past the highest composite slot ever handed out; exactly the base when none ever
+        // was, so the number is monotone from the first mint and a LEAKED COMPOSITE MOVES IT.
+        return static_cast<Uint32>(kMGPipeShaderCsoCompositeSlotBase + state.BandSlots.size());
+    }
+
+    Uint32 MGPipeSlotAllocator::LiveCount(MGPipeKind kind) const { return StateOf(kind).LiveCount; }
+
+    Uint32 MGPipeSlotAllocator::CompositeLiveCount() const {
+        return StateOf(MGPipeKind::ShaderCso).BandLiveCount;
+    }
+
+    Uint32 MGPipeSlotAllocator::FreeCount(MGPipeKind kind) const {
+        const KindState& state = StateOf(kind);
+        return static_cast<Uint32>(state.FreeList.size() + state.BandFreeList.size());
+    }
+
+    Uint32 MGPipeSlotAllocator::CompositeFreeCount() const {
+        return static_cast<Uint32>(StateOf(MGPipeKind::ShaderCso).BandFreeList.size());
+    }
+
+    void MGPipeSlotAllocator::Reset() {
+        for (KindState& state : m_kinds) {
+            state.Slots.clear();
+            state.FreeList.clear();
+            state.BandSlots.clear();
+            state.BandFreeList.clear();
+            state.ByLifetimeId.clear();
+            state.LiveCount = 0;
+            state.BandLiveCount = 0;
+        }
+    }
+
+    MGPipeSlotAllocator& MGPipeSlots() {
+        // NEVER DESTROYED, deliberately (one allocation for the life of the process). A
+        // frontend object's destructor reaches this allocator - ~BufferObject through
+        // MGPipeEmitResourceDestroyAndFree, ~VertexArrayObject through the death notice - and
+        // MG_Backend/MGPipe/PipeInputs.h's gPipeInputs holds SharedPtrs to those objects at
+        // namespace scope, so they are destroyed by __run_exit_handlers AFTER this
+        // function-local static would have been. A destroyed allocator then answers
+        // FindByLifetimeId out of a freed hash table and Free() writes into freed vectors -
+        // an exit-time heap corruption whose fatality depends only on the allocator's layout.
+        static MGPipeSlotAllocator* allocator = new MGPipeSlotAllocator();
+        return *allocator;
+    }
+} // namespace MobileGL::MG_Pipe

MobileGL/MG_Impl/Pipe/SlotAllocator.h

@@ -0,0 +1,166 @@
+// MobileGL - MobileGL/MG_Impl/Pipe/SlotAllocator.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>
+
+#include <MG_Pipe/MGPipeHandles.h>
+
+// The CLIENT's slot allocator: the thing that mints every MGPipeHandle in the system
+// (ARCHITECTURE.md 4.2 - no create_* call in the catalogue returns a server-cast handle,
+// which is what lets the whole catalogue be remoted with zero creation round trips).
+//
+// Per kind: a free list plus a high-water mark, so slots stay DENSE and the server's object
+// table is an array rather than a hash map. It has nothing to do with MG_State's
+// IndexGenerator - that container's LIFO GL-name reuse is the very problem {slot, gen}
+// exists to close, and the whole point of the identity is that an ABA on the GL name, on
+// the heap address or on the lifetime id cannot reproduce a handle.
+//
+// Gen increments ONLY when a slot is reused, never on a respecify: a glBufferData on a live
+// buffer keeps the same {slot, gen}, because the object is the same object. Two generations
+// exist in the design and they are strictly separate - this is the client's answer to "is
+// this still the same GL object"; MGGen is the server's epoch for "did I recast my driver
+// object", and no MGPipe call may require the client to know it.
+//
+// The lifetimeId -> slot map is what keeps a GL NAME out of every key (ARCHITECTURE.md 4.2):
+// the frontend object's lifetime id is the client's own identity for it, so the backend key
+// is the handle and the frontend key is the lifetime id, and neither is a recyclable name.
+//
+// Lives in MG_Impl (the client side, unrestricted) and is compiled only under
+// MOBILEGL_PIPE_PUSH. It is in the P2 CONTRACT commit rather than in a Track H package
+// because both Track H slices - Espryt 0b and Magma subsystem 4 - key off it.
+namespace MobileGL::MG_Pipe {
+
+    class MGPipeSlotAllocator {
+    public:
+        static constexpr SizeT kKindCount = static_cast<SizeT>(MGPipeKind::KindCount);
+
+        // A fresh {slot, gen} of this kind, from the free list if one is waiting and from the
+        // high-water mark otherwise. Never returns slot 0 (reserved: null, and the default
+        // framebuffer for kind Framebuffer), and never returns a ShaderCso slot inside the
+        // composite band, which the program-pipeline resolver mints out of separately.
+        MGPipeHandle Allocate(MGPipeKind kind);
+        // Allocate and remember `lifetimeId` as this handle's frontend identity.
+        MGPipeHandle AllocateFor(MGPipeKind kind, Uint64 lifetimeId);
+
+        // P4a, D-H7: THE ONE ENTRY POINT INTO THE ShaderCso COMPOSITE BAND, and the only one
+        // there will ever be. Allocate() above refuses that band on purpose, so a program
+        // pipeline's flattened composite - minted client-side from the stage programs bound to
+        // the pipeline object, and indistinguishable from an ordinary program to the server -
+        // needs a door of its own rather than a flag on the handle. The kind is implied: only
+        // ShaderCso has a band.
+        //
+        // It behaves exactly like AllocateFor in every other respect (free list first, then
+        // the band's own high-water mark; Gen moves only on reuse; the lifetimeId -> slot map
+        // is written) and it carries the band's own exhaustion assert, so exhausting the
+        // composite space is a NAMED Fatal rather than silent slot theft from ordinary
+        // programs. Returns kMGPipeNullHandle when the band is full.
+        //
+        // Freed through the ordinary Free(MGPipeKind::ShaderCso, handle): a composite's slot
+        // has two independent release paths - the pipeline cache's LRU eviction and the
+        // composite ProgramObject's own destructor - and Free refusing a slot that is not live
+        // at that generation is what makes the second one a proven no-op.
+        MGPipeHandle AllocateComposite(Uint64 lifetimeId);
+        // The handle a lifetime id was allocated for, or kMGPipeNullHandle. A recycled heap
+        // address does NOT reproduce a mapping: MG_State hands out a fresh lifetime id per
+        // object, so the map key is unique for the life of the process.
+        MGPipeHandle FindByLifetimeId(MGPipeKind kind, Uint64 lifetimeId) const;
+        // FindByLifetimeId, then AllocateFor when it misses. The ordinary client path.
+        MGPipeHandle Acquire(MGPipeKind kind, Uint64 lifetimeId);
+
+        // Returns the slot to the free list. The Gen bump happens on the NEXT handout of that
+        // slot, not here, so a handle that is freed twice cannot skip a generation and the
+        // "gen moves only on reuse" contract holds for an object that is never reused.
+        void Free(MGPipeKind kind, MGPipeHandle handle);
+
+        Bool IsLive(MGPipeKind kind, MGPipeHandle handle) const;
+        // 0 for a slot that was never handed out; the generation of the LAST handout
+        // otherwise, live or not.
+        Uint32 GenOfSlot(MGPipeKind kind, Uint32 slot) const;
+        Uint64 LifetimeIdOfSlot(MGPipeKind kind, Uint32 slot) const;
+        // One past the highest ORDINARY slot ever handed out of this kind. For every kind but
+        // ShaderCso that is the whole story; for ShaderCso the composite band is a second,
+        // separately dense space and CompositeHighWater() below answers it.
+        //
+        // THE TWO SPACES ARE REPORTED SEPARATELY, and that is the point rather than a detail.
+        // Folding the band into this number pins it at ~983k from the first composite mint
+        // onward, and every later assertion of the "the high-water mark did not move over N
+        // churn rounds" shape - the one that catches a dense table that never shrinks, which
+        // is the ~1.3 KB-per-record leak C-1 produced - becomes vacuously true for ordinary
+        // ShaderCso slots for the rest of the process. A leak case per space is two real
+        // assertions; one merged number is one real assertion and one that cannot go red.
+        //
+        // It is also NOT a table size for kind ShaderCso even now: the band is sparse against
+        // the ordinary space by design, so a consumer indexing by slot must test
+        // MGPipeIsCompositeShaderSlot(slot) first and keep the band in a table of its own,
+        // exactly as this allocator does.
+        Uint32 HighWater(MGPipeKind kind) const;
+        // One past the highest COMPOSITE slot ever handed out, i.e.
+        // kMGPipeShaderCsoCompositeSlotBase + (band slots ever handed out), and exactly the
+        // base when none ever was. Kind ShaderCso is the only kind with a band, so it is
+        // implied - as it is for AllocateComposite. A LEAKED COMPOSITE MOVES THIS and moves
+        // nothing else, which is what the composite's own leak case asserts on.
+        Uint32 CompositeHighWater() const;
+        // Live slots of this kind, ORDINARY AND COMPOSITE TOGETHER for ShaderCso: a live
+        // composite is a live ShaderCso, the applier's two record tables are one object class,
+        // and a caller asking "how many shader CSOs does this client hold" wants both. The
+        // band's own count is CompositeLiveCount(); the ordinary space's is the difference.
+        Uint32 LiveCount(MGPipeKind kind) const;
+        Uint32 CompositeLiveCount() const;
+        // Slots waiting on a free list. Also BOTH SPACES for ShaderCso, for LiveCount's
+        // reason and with the same caveat: a caller that needs to know WHICH space a slot went
+        // back to reads CompositeFreeCount() and subtracts.
+        Uint32 FreeCount(MGPipeKind kind) const;
+        Uint32 CompositeFreeCount() const;
+
+        // Context teardown / server reset / a unit test's fixture.
+        void Reset();
+
+    private:
+        struct SlotState {
+            Uint32 Gen = 0;
+            Bool Live = false;
+            Bool EverHandedOut = false;
+            Uint64 LifetimeId = 0;
+        };
+
+        struct KindState {
+            // Indexed by slot; [0] is the reserved slot and is never live.
+            Vector<SlotState> Slots;
+            Vector<Uint32> FreeList;
+            // P4a: the ShaderCso COMPOSITE band, indexed by (slot - the band's base) and
+            // EMPTY for every other kind. A SECOND VECTOR RATHER THAN MORE OF THE FIRST, and
+            // it is not a micro-optimisation: the band starts at 983040, so minting one
+            // composite into the slot-indexed vector above would allocate ~983k SlotStates -
+            // ~23 MB - for a single program pipeline, and a consumer that sized a table off
+            // HighWater would pay the same shape again with a far bigger record. Both spaces
+            // stay dense against their own high-water mark, which is the property this
+            // allocator exists to give the server.
+            Vector<SlotState> BandSlots;
+            Vector<Uint32> BandFreeList;
+            UnorderedMap<Uint64, Uint32> ByLifetimeId;
+            Uint32 LiveCount = 0;
+            // The band's share of LiveCount above, so the two spaces can be reported apart
+            // without walking either table. Always 0 for every kind but ShaderCso.
+            Uint32 BandLiveCount = 0;
+        };
+
+        KindState& StateOf(MGPipeKind kind);
+        const KindState& StateOf(MGPipeKind kind) const;
+        // The SlotState a (kind, slot) names, in whichever of the two vectors holds it, or
+        // null when the slot has never been handed out. One resolver, so a caller that forgets
+        // the band cannot exist.
+        static SlotState* EntryOf(KindState& state, MGPipeKind kind, Uint32 slot);
+        static const SlotState* EntryOf(const KindState& state, MGPipeKind kind, Uint32 slot);
+
+        Array<KindState, kKindCount> m_kinds{};
+    };
+
+    // The monolith's one client allocator. Under split there is one per client context.
+    MGPipeSlotAllocator& MGPipeSlots();
+} // namespace MobileGL::MG_Pipe

MobileGL/MG_Impl/Pipe/Tracker.h

@@ -0,0 +1,864 @@
+// MobileGL - MobileGL/MG_Impl/Pipe/Tracker.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>
+
+// The frontend state tracker (ARCHITECTURE.md 5.2, P2 brief D4).
+//
+// WHERE IT RUNS. Not above MGP_FILL and not in the GL setter: MGPipeValidateForVerb, the
+// one statement MGP_FILL already expands to before every gBackendFunctionsTable.GL call
+// (PipeFill.h). Blaze3D brackets every batch with glEnable/glDisable(GL_BLEND), so a
+// setter that pushed would push twice per batch for a state the batch may not even read;
+// the validate point coalesces the whole bracket into the two draws that observe it
+// (ARCHITECTURE.md 5.1).
+//
+// WHAT IT DOES. One Uint32 dirty mask per verb, one bit per row of ARCHITECTURE.md 5.2,
+// computed by comparing a shutter against what the tracker last pushed. P2 emitted for bits
+// 0..4 (the value-class ones); P3a adds bits 5, 9 and 10 - the vertex-input family - and P4a
+// adds SEVEN: 6, 7 and 8 (the program family), 11 (the framebuffer) and 12, 13 and 14 (the
+// three unit sets). Only bits 15, 16 and 17 - the const-buffer, shader-buffer and
+// stream-output sets - are still computed, latched and counted without an emitter, so the
+// per-bit fire rate is a measurement rather than a plan and their fields go through the
+// residual fill until P4b.
+//
+// P4a NARROWS NOTHING AND WIDENS THREE THINGS, and every one of them was an UNDER-FIRE that
+// only became reachable once the bit gained an emitter:
+//   (1) bit 11's shutter gains the READ framebuffer binding slot's version, because
+//       set_framebuffer_state is emitted per bound TARGET and a glBindFramebuffer(
+//       GL_READ_FRAMEBUFFER, ...) moved no shutter at all before;
+//   (2) bit 13's gains the TEXTURE BIND generation, because glBindSampler moves that one and
+//       not the sampling-resolution one, so bind_sampler_states could not see a sampler bind;
+//   (3) bits 6/7/8 - and with them bit 14's program half - read the EFFECTIVE program source
+//       instead of GetCurrentProgram() alone, which is null for the whole life of a bound
+//       separable program pipeline, so a re-composited pipeline reached no program emitter.
+// Over-firing is free; all three of those were the other direction.
+//
+// WHY EVERY SHUTTER OVER-FIRES. A bit that fires too often costs one extra push. A bit
+// that fires too rarely renders stale, and ARCHITECTURE.md 13.2 names that as the
+// dangerous direction precisely because the P1 verify comparator cannot see it for
+// object-class state (it compares those by identity only). So each shutter below is
+// deliberately coarser than the state it guards - five bits share one buffer aggregate,
+// the framebuffer bit fires on any attachment write anywhere - and the narrowing is P3's
+// work, paid for with the fire rates this file publishes.
+//
+// NO TIMER LIVES HERE. ROADMAP.md forbids committing hot-path instrumentation; the
+// absolute ns/draw comes from DriverBench, which times whole frames from outside the
+// library (P2 brief D17). The only counting is the per-bit fire tally, behind
+// PipeStats::Enabled() like every other counting site in the tree.
+//
+// HEADER-ONLY, and that is an ownership decision rather than a design one: the P2 brief
+// asks for Tracker.{h,cpp}, but the root CMakeLists.txt that would have to name a new .cpp
+// belongs to package A and is frozen behind the p2/contract tag. Everything here is
+// included by exactly one translation unit in the library (MG_Impl/Pipe/PipeFill.cpp) plus
+// the unit tests, so inline costs nothing. Splitting it back out is one list(APPEND) line.
+#if MOBILEGL_PIPE_PUSH
+#include <MG_Pipe/MGPipe.h>
+#include <MG_Pipe/MGPipeValueTypes.h>
+#include <MG_State/GLState/Core.h>
+#include <MG_Util/Metrics/PipeStats.h>
+
+#include <cstring>
+
+namespace MobileGL::MG_Pipe {
+
+    // One bit per row of the ARCHITECTURE.md 5.2 table, hand-written rather than generated:
+    // the list is design, not derived data, and the generator has nothing to derive it from.
+    enum class MGPipeDirty : Uint32 {
+        // ---- value class: P2 emits for these five ----
+        NewRenderState = 0,      // RenderState::m_version              -> set_dynamic_state
+        NewPipelineState,        // RenderState::m_pipelineStateVersion -> create/bind_render_state
+        NewPixelPack,            // PixelStoreParameters (pack)         -> set_pixel_pack_state
+        NewPatchState,           // the patch trio, NaN legal           -> set_patch_state
+        NewVertexAttribDefaults, // glVertexAttrib* defaults            -> set_vertex_attrib_defaults
+        // ---- value class: NEW_VERTEX_ELEMENTS is emitted from P3a and the other three from
+        // P4a - the program family, one subsystem, three bits because the frontend moves them
+        // as three separate events ----
+        NewVertexElements,       // the bound VAO's attribute configuration -> create/bind_vertex_elements
+        NewShader,               // the current program's link version -> create/bind_shader_state,
+                                 //    set_draw_program, set_dispatch_program (P4a)
+        NewShaderBindings,       // image units, block bindings, uniform write set (P4a)
+        NewGlobalConstants,      // the default-uniform-block image -> set_global_constants (P4a)
+        // ---- object class. THE FIRST TWO ARE P3a's, not P3b/P4b's: the roadmap puts
+        // set_vertex_buffers and set_index_buffer in the same phase as the vertex-elements
+        // trio, and this comment said otherwise until the commit that wired them. THE NEXT
+        // FOUR ARE P4a's. The last three are still computed and counted only, until P4b. ----
+        NewVertexBuffers,        // -> set_vertex_buffers (P3a)
+        NewIndexBuffer,          // -> set_index_buffer   (P3a)
+        NewFramebuffer,          // -> set_framebuffer_state, per bound target (P4a)
+        NewSamplerViews,         // -> set_sampler_views   (P4a)
+        NewSamplers,             // -> bind_sampler_states (P4a)
+        NewShaderImages,         // -> set_shader_images   (P4a)
+        NewConstBuffers,
+        NewShaderBuffers,
+        NewSoTargets,
+        Count,
+    };
+
+    inline constexpr SizeT kMGPipeDirtyCount = static_cast<SizeT>(MGPipeDirty::Count);
+    static_assert(kMGPipeDirtyCount <= 32, "the dirty mask is a Uint32");
+
+    inline constexpr Uint32 MGPipeDirtyBit(MGPipeDirty bit) {
+        return Uint32{1} << static_cast<Uint32>(bit);
+    }
+
+    // The five P2 emits for. Each phase's constant survives as the next phase's A/B control
+    // and as what a test compares the subsystem map against, so none of them is edited in
+    // place when a later phase takes more bits over.
+    inline constexpr Uint32 kMGPipeDirtyEmittedAtP2 =
+        MGPipeDirtyBit(MGPipeDirty::NewRenderState) | MGPipeDirtyBit(MGPipeDirty::NewPipelineState) |
+        MGPipeDirtyBit(MGPipeDirty::NewPixelPack) | MGPipeDirtyBit(MGPipeDirty::NewPatchState) |
+        MGPipeDirtyBit(MGPipeDirty::NewVertexAttribDefaults);
+
+    // The three P3a adds: the vertex-input family, all on one subsystem.
+    inline constexpr Uint32 kMGPipeDirtyEmittedAtP3a =
+        kMGPipeDirtyEmittedAtP2 | MGPipeDirtyBit(MGPipeDirty::NewVertexElements) |
+        MGPipeDirtyBit(MGPipeDirty::NewVertexBuffers) | MGPipeDirtyBit(MGPipeDirty::NewIndexBuffer);
+
+    // The SEVEN P4a adds, across FOUR subsystems: bits 6/7/8 are the program family, 11 the
+    // framebuffer, and 12/13/14 the sampler-view / sampler-state / image-unit sets. Added
+    // rather than edited into the two above, for the reason those two exist: each phase's
+    // constant survives as the next phase's A/B control and as what a test compares the
+    // subsystem map against.
+    //
+    // EVERY ONE OF THESE SHUTTERS WAS ALREADY COMPUTED, LATCHED AND COUNTED before P4a; what
+    // P4a adds is an emitter for them. That is why this is a one-line constant and not seven
+    // new shutters - and it is also why the two narrowings below are stated as requirements.
+    inline constexpr Uint32 kMGPipeDirtyEmittedAtP4a =
+        kMGPipeDirtyEmittedAtP3a | MGPipeDirtyBit(MGPipeDirty::NewShader) |
+        MGPipeDirtyBit(MGPipeDirty::NewShaderBindings) |
+        MGPipeDirtyBit(MGPipeDirty::NewGlobalConstants) |
+        MGPipeDirtyBit(MGPipeDirty::NewFramebuffer) | MGPipeDirtyBit(MGPipeDirty::NewSamplerViews) |
+        MGPipeDirtyBit(MGPipeDirty::NewSamplers) | MGPipeDirtyBit(MGPipeDirty::NewShaderImages);
+
+    inline constexpr const char* kMGPipeDirtyNames[kMGPipeDirtyCount] = {
+        "NEW_RENDER_STATE",
+        "NEW_PIPELINE_STATE",
+        "NEW_PIXEL_PACK",
+        "NEW_PATCH_STATE",
+        "NEW_VERTEX_ATTRIB_DEFAULTS",
+        "NEW_VERTEX_ELEMENTS",
+        "NEW_SHADER",
+        "NEW_SHADER_BINDINGS",
+        "NEW_GLOBAL_CONSTANTS",
+        "NEW_VERTEX_BUFFERS",
+        "NEW_INDEX_BUFFER",
+        "NEW_FRAMEBUFFER",
+        "NEW_SAMPLER_VIEWS",
+        "NEW_SAMPLERS",
+        "NEW_SHADER_IMAGES",
+        "NEW_CONST_BUFFERS",
+        "NEW_SHADER_BUFFERS",
+        "NEW_SO_TARGETS",
+    };
+
+    // Which runtime MOBILEGL_PIPE_PUSH subsystem bit gates a dirty bit's emission. Zero for
+    // a bit P2 does not emit, which is what makes "the bitmask is a true per-subsystem A/B"
+    // literally true rather than approximately.
+    inline constexpr Uint64 MGPipeSubsystemForDirty(MGPipeDirty bit) {
+        switch (bit) {
+        case MGPipeDirty::NewRenderState:
+        case MGPipeDirty::NewPipelineState:
+            return kMGPipeSubsystemRenderState;
+        case MGPipeDirty::NewPixelPack:
+            return kMGPipeSubsystemPixelPack;
+        case MGPipeDirty::NewPatchState:
+            return kMGPipeSubsystemPatchState;
+        case MGPipeDirty::NewVertexAttribDefaults:
+            return kMGPipeSubsystemVertexAttribDefaults;
+        // P3a's three, all one subsystem: create/bind_vertex_elements, set_vertex_buffers
+        // and set_index_buffer are the vertex-input family and an operator switching it off
+        // has to get the whole family's legacy arm, not two thirds of it.
+        // PipeFill.cpp's SubsystemForEmitter carries the pairing static_asserts.
+        case MGPipeDirty::NewVertexElements:
+        case MGPipeDirty::NewVertexBuffers:
+        case MGPipeDirty::NewIndexBuffer:
+            return kMGPipeSubsystemVertexInput;
+        // P4a's seven, across four subsystems. FOUR AND NOT ONE for P3a's reason one level
+        // out: a framebuffer path that regressed, a texture path that regressed, a sampler
+        // path that regressed and a program path that regressed are four different findings.
+        //
+        // The program family is three bits because the frontend moves them separately - a
+        // relink, a binding change and a uniform write are three events - but one subsystem,
+        // because an operator switching programs off has to get the whole family's legacy arm.
+        // Same for the three unit sets: create_sampler_state, create_sampler_view and the
+        // three kVarTail sets are one family, and half of it is not a control.
+        case MGPipeDirty::NewShader:
+        case MGPipeDirty::NewShaderBindings:
+        case MGPipeDirty::NewGlobalConstants:
+            return kMGPipeSubsystemPrograms;
+        case MGPipeDirty::NewFramebuffer:
+            return kMGPipeSubsystemFramebuffer;
+        case MGPipeDirty::NewSamplerViews:
+        case MGPipeDirty::NewSamplers:
+        case MGPipeDirty::NewShaderImages:
+            return kMGPipeSubsystemSamplers;
+        // NO BIT NAMES kMGPipeSubsystemTextureResources, and that is deliberate rather than an
+        // omission: the texture and renderbuffer resource_* calls and set_texture_params are
+        // dispatched from the GL entry points that cause them - a constructor, a storage
+        // definition, a glTexParameter - not from a dirty walk, exactly as P3a's buffer family
+        // is. Bit 10 gates those dispatch sites; there is no dirty bit to map onto it and
+        // there must not be one, or the emission would be gated twice and disagree with itself.
+        default:
+            // The remaining bits have no call of their own until P4b, so there is no
+            // subsystem to switch and the residual fill keeps supplying their fields.
+            return 0;
+        }
+    }
+
+    // A COMPOSITE shutter, for the bits whose "did anything move" is more than one counter.
+    // It is a hash, so two different states can in principle collide and cost a MISSED fire.
+    // The five bits P2 emits for are never composed - they are widened counters and byte
+    // compares, neither of which can collide.
+    //
+    // P3a's three ARE composed, so the risk is now real rather than academic, and it is
+    // accepted with its size stated: each mix takes a 64-bit input into a 64-bit
+    // accumulator, so two DIFFERENT vertex configurations collide with probability ~2^-64
+    // per pair, and the inputs are a monotone lifetime id, a monotone configuration version
+    // and a widened slot version - none of which an application can steer. The alternative,
+    // comparing the whole 32-attribute configuration byte for byte on every verb, is the
+    // per-draw cost the shutter exists to avoid. The narrowing that removes the composition
+    // for bit 10 - its own slot version plus the bound object's identity - is what this
+    // phase already did to the one shutter that was composed over an unrelated aggregate.
+    inline constexpr Uint64 MGPipeMixShutter(Uint64 accumulator, Uint64 value) {
+        accumulator ^= value + 0x9e3779b97f4a7c15ull + (accumulator << 6) + (accumulator >> 2);
+        return accumulator;
+    }
+
+    // A Uint16 counter widened at the TRACKER boundary, never in MG_State
+    // (ARCHITECTURE.md 5.2: MG_State is not changed for this). A decrease is a wrap and adds
+    // 65536. A wrap is harmless locally - one extra re-push, never a missed one - which is
+    // exactly what TrackerTest.WrapAroundRePushesButNeverMisses pins.
+    //
+    // THE ONE CASE IT CANNOT SEE, stated because "never a missed push" is otherwise stronger
+    // than what is true: the wrap test is `now < m_last`, so a counter that advances by
+    // EXACTLY 65536 (or a multiple) between two walks reads as unchanged. That needs 65536
+    // render-state mutations inside one verb boundary, and it is pre-existing in class -
+    // both backends already compare raw Uint16 versions the same way - so P2 records it
+    // rather than widening MG_State's counters, which ARCHITECTURE.md 5.2 rules out.
+    class MGPipeWidenedCounter {
+    public:
+        Uint64 Observe(Uint16 now) {
+            if (m_started && now < m_last) m_high += 0x10000ull;
+            m_started = true;
+            m_last = now;
+            return m_high + now;
+        }
+        void Reset() {
+            m_high = 0;
+            m_last = 0;
+            m_started = false;
+        }
+
+    private:
+        Uint64 m_high = 0;
+        Uint16 m_last = 0;
+        Bool m_started = false;
+    };
+
+    class MGPipeTracker {
+    public:
+        using GLContext = MG_State::GLState::GLContext;
+
+        // The dirty walk. Compares every shutter against what was last pushed, LATCHES the
+        // new values, counts the fires per verb class, and returns the mask. Latching here
+        // rather than after emission is deliberate: a bit whose subsystem is switched off is
+        // not emitted, but its fields are then still pulled by the residual fill, so the
+        // pushed block is correct either way and a bit can never fire twice for one change.
+        Uint32 Update(GLContext& ctx, MGPipeVerbClass verbClass) {
+            // A different context is a different server: nothing the tracker latched about
+            // the old one says anything about this one, and the first walk on a fresh
+            // context must publish a COMPLETE state rather than an increment.
+            if (m_context != &ctx) {
+                Reset();
+                m_context = &ctx;
+            }
+            const Bool wasPrimed = m_primed;
+
+            Uint64 now[kMGPipeDirtyCount];
+            const RenderStateParameters& render = ctx.GetRenderStateParameters();
+
+            // ---- bits 0..1: the two Uint16 render-state counters, widened HERE ----
+            now[Index(MGPipeDirty::NewRenderState)] =
+                m_renderStateVersion.Observe(static_cast<Uint16>(ctx.GetRenderStateParametersVersion()));
+            now[Index(MGPipeDirty::NewPipelineState)] =
+                m_pipelineStateVersion.Observe(static_cast<Uint16>(ctx.GetPipelineStateVersion()));
+
+            // ---- bit 4 and the value-class bits 5..8 ----
+            now[Index(MGPipeDirty::NewVertexAttribDefaults)] = ctx.GetAnyVertexAttribDefaultGeneration();
+
+            const auto& vao = ctx.GetBoundVertexArray();
+            const Uint64 vaoIdentity =
+                vao ? MGPipeMixShutter(vao->GetLifetimeId(), vao->GetConfigVersion()) : 0;
+            now[Index(MGPipeDirty::NewVertexElements)] = vaoIdentity;
+
+            // Deliberately NOT GetProgramForDraw: that joins a pending link, and the tracker
+            // must not force a compile just to answer "did the shader move". These version
+            // counters are plain members and are exactly what the backends already read
+            // without joining (Core.cpp, the glUseProgram half of join site J1).
+            //
+            // BUT GetCurrentProgram() ALONE IS NOT THE PROGRAM SOURCE, AND AT P4a THAT IS AN
+            // UNDER-FIRE. Under GL_ARB_separate_shader_objects an application drives
+            // `glUseProgram(0); glBindProgramPipeline(P)`, and m_currentProgram is then null
+            // for the whole life of that pipeline (Core.cpp, GetProgramForDraw's second half):
+            // all three of these shutters read 0 == 0 forever, so after the first walk on a
+            // fresh context - the one !m_primed fires unconditionally - bits 6, 7 and 8 never
+            // fire again however the pipeline is restaged.
+            //
+            // WHILE NOTHING WAS EMITTED FOR THEM THAT WAS INVISIBLE, which is how it survived
+            // to P4a: GetProgramForDraw is emitted-and-still-pulled, the residual fill copies
+            // it at every verb, and DirtySurface.def rules BindProgramPipelineObject
+            // kPulledEveryVerb for exactly that reason - the backend still receives the right
+            // SharedPtr and nothing renders wrong. The moment P4a emits off these bits it
+            // stops being invisible: glUseProgramStages rebuilds the composite, EmitShaderState
+            // is never called again, so the new composite gets no ShaderCso handle and no
+            // create_shader_state while set_draw_program keeps naming the previous one - a
+            // program the handle protocol never announced, which is exactly the seam-defect
+            // class P3a spent a phase closing. And bit 8 never firing means
+            // set_global_constants is never sent for a pipeline draw at all, where the pull
+            // rescues nothing.
+            //
+            // SO THE SHUTTER READS THE EFFECTIVE SOURCE: the program in use when there is one,
+            // and the bound pipeline when there is not. What it reads OF that pipeline is the
+            // pair ComputeDrawProgramSignature() is built from - each stage program's lifetime
+            // id and LINK version - so bit 6 fires exactly when GetProgramForDraw would hand
+            // back a different composite, which is exactly when a new ShaderCso handle has to
+            // be minted. Those are the same non-artefact fields the plain-program arm above
+            // reads, and the ones Core.cpp calls out as not passing through ProgramObject's
+            // join gate, so the "must not force a compile" rule survives intact: no join, no
+            // flatten, no Link().
+            //
+            // THE PIPELINE NAME IS MIXED IN because two pipelines can carry the same stage set
+            // and each caches its OWN composite object, so the signature alone would let a
+            // glBindProgramPipeline between two such pipelines pass without a fire. What that
+            // does NOT close is a name RECYCLED (glDeleteProgramPipelines +
+            // glGenProgramPipelines) back onto the same stage programs at the same link
+            // versions with no other program-family change in between: a ProgramPipelineObject
+            // has no lifetime id and no wire object at all - DirtySurface.def says so where it
+            // rules MarkProgramPipelineForDeletion kUnpublishedDestroy - so there is nothing
+            // else here to mix it with. Recorded rather than quietly left: closing it needs a
+            // generation counter on the frontend object, which is an MG_State change and not
+            // this file's to make.
+            const auto& program = ctx.GetCurrentProgram();
+            Uint64 shader = 0;
+            Uint64 bindings = 0;
+            Uint64 constants = 0;
+            Uint64 programImages = 0;
+            // THE PROGRAM INPUT OF THE PROGRAM-RESOLVED VIEW SET (P4a fable seam F-1).
+            // set_sampler_views is resolved for the program in use (SamplerEmit.h: the sampler
+            // uniform's TYPE picks which of a unit's targets is the view) and the emitter
+            // memoises that resolution on (lifetime id, link version, backend state version). A
+            // shutter that read only the texture generations therefore missed a glUseProgram:
+            // `glBindTexture x N; glUseProgram(P1); draw; glUseProgram(P2); draw` moved nothing
+            // bit 12 read, so the view set stayed P1's - and E's record epoch, keyed on the two
+            // set serials, then never rebuilt the texture sync list for P2 either. This value is
+            // that memo key, and bit 12 mixes it in below: over-firing costs one re-resolution
+            // the set-hash suppressor absorbs, under-firing left the record describing the
+            // previous program's units.
+            Uint64 opaqueUnits = 0;
+            if (program) {
+                shader = MGPipeMixShutter(program->GetLifetimeId(), program->GetLinkVersion());
+                bindings = MGPipeMixShutter(
+                    MGPipeMixShutter(MGPipeMixShutter(program->GetImageUnitVersion(),
+                                                      program->GetBackendStateVersion()),
+                                     program->GetBlockBindingVersion()),
+                    program->GetUniformWriteSetVersion());
+                constants = MGPipeMixShutter(program->GetLifetimeId(), program->GetUBOContentVersion());
+                // THE IDENTITY IS MIXED IN (P4a fable seam F-2), exactly as the pipeline arm
+                // below mixes stageLinks into its half: the counter alone is a per-program
+                // number two programs routinely share - 0 == 0 for any pair that never moved an
+                // image unit through glUniform1i, and 0 == 0 against no program at all - so a
+                // glUseProgram between them fired nothing, set_shader_images' window stayed the
+                // previous program's, and a program whose only image is a BUFFER image (E's
+                // SD-4: nothing else moves between the bind and the dispatch) never reached the
+                // record at all.
+                programImages = MGPipeMixShutter(shader, program->GetImageUnitVersion());
+                opaqueUnits = MGPipeMixShutter(shader, program->GetBackendStateVersion());
+            } else if (const auto& pipeline = ctx.GetBoundProgramPipeline(); pipeline) {
+                using Pipeline = MG_State::GLState::ProgramPipelineObject;
+                // THE FIELDS ARE READ DIRECTLY RATHER THAN THROUGH THE TWO FUNCTIONS THAT
+                // ALREADY PACK THEM, and that is a gate constraint, not a preference. Calling
+                // ComputeDrawProgramSignature() / ComputeUniformMirrorVersions() would say
+                // "the same pairs the composite cache and the uniform-mirror gate compare"
+                // far better than this loop does - but gen_pipe_dirty_surface.py derives a
+                // shutter by following each accessor to the member it returns, and both of
+                // those build a LOCAL array and return that, which it cannot place. A shutter
+                // naming them is UNRESOLVED, and then every DirtySurface.def row that names
+                // bits 6, 7, 8 or 14 loses its verdict - including the derivation that is the
+                // only mechanism able to catch the next under-fire here. So the pairs are
+                // spelled out, and the two static_asserts below are what say they must stay in
+                // step with the functions they mirror.
+                static_assert(sizeof(Pipeline::DrawProgramSignature) ==
+                                  2 * Pipeline::kGraphicsStageCount * sizeof(Uint64),
+                              "bit 6 reads the {lifetimeId, linkVersion} pair per graphics "
+                              "stage that ComputeDrawProgramSignature packs");
+                static_assert(sizeof(Pipeline::UniformMirrorVersions) ==
+                                  2 * Pipeline::kGraphicsStageCount * sizeof(Uint64),
+                              "bits 7 and 8 read the four counters per graphics stage that "
+                              "ComputeUniformMirrorVersions packs");
+
+                // Bit 6 is the pipeline's identity plus the composite cache key. Bits 7 and 8
+                // add the per-program state, which under a pipeline is written to the STAGE
+                // programs - glUniform* addresses the pipeline's active program,
+                // glProgramUniform* and the two block-binding calls address a named one - and
+                // only reaches the composite through RefreshCompositeUniforms. Bit 14's half
+                // takes the image-unit generation, which is its own counter for the reason
+                // ProgramObject gives (ES forbids glUniform1i on an image uniform, so Espryt
+                // BAKES the unit into the ESSL it generates and only a regeneration honours a
+                // change) and which D-G4 asks this shutter to keep reading as a FRONTEND
+                // counter rather than any server-side epoch.
+                //
+                // STAGELINKS IS MIXED INTO ALL THREE OF THE OTHERS, ON PURPOSE. A composite
+                // REBUILD hands back a brand-new ProgramObject with an empty default uniform
+                // block and no backend state at all - SetCachedDrawProgram clears the mirror
+                // versions with it - so a shutter watching only the per-stage state counters
+                // would let a rebuilt composite inherit the bindings, the constants and the
+                // image units of the one it replaced.
+                Uint64 stageLinks = static_cast<Uint64>(ctx.GetBoundProgramPipelineName());
+                Uint64 stageState = 0;
+                Uint64 stageImages = 0;
+                // The per-stage sampler/image unit assignments alone (glUniform1i on a stage
+                // program's sampler moves its backend state version and reaches the composite
+                // through the uniform mirror), for bit 12's program input below.
+                Uint64 stageOpaque = 0;
+                for (SizeT stage = 0; stage < Pipeline::kGraphicsStageCount; ++stage) {
+                    const auto& staged = pipeline->GetStageProgram(static_cast<ShaderStage>(stage));
+                    if (!staged) continue;
+                    stageLinks = MGPipeMixShutter(
+                        MGPipeMixShutter(stageLinks, staged->GetLifetimeId()), staged->GetLinkVersion());
+                    stageState = MGPipeMixShutter(
+                        MGPipeMixShutter(MGPipeMixShutter(stageState, staged->GetBackendStateVersion()),
+                                         MGPipeMixShutter(staged->GetUBOContentVersion(),
+                                                          staged->GetBlockBindingVersion())),
+                        staged->GetUniformWriteSetVersion());
+                    stageImages = MGPipeMixShutter(stageImages, staged->GetImageUnitVersion());
+                    stageOpaque = MGPipeMixShutter(stageOpaque, staged->GetBackendStateVersion());
+                }
+                shader = stageLinks;
+                stageState = MGPipeMixShutter(stageLinks, stageState);
+                bindings = MGPipeMixShutter(stageState, stageImages);
+                constants = stageState;
+                programImages = MGPipeMixShutter(stageLinks, stageImages);
+                opaqueUnits = MGPipeMixShutter(stageLinks, stageOpaque);
+            }
+            now[Index(MGPipeDirty::NewShader)] = shader;
+            now[Index(MGPipeDirty::NewShaderBindings)] = bindings;
+            now[Index(MGPipeDirty::NewGlobalConstants)] = constants;
+
+            // ===========================================================================
+            // THE RECORD-FIELD -> SETTER -> SHUTTER TABLE FOR THE SEVEN P4a BITS.
+            //
+            // THE RULE (P4a fable seam audit, section C.1): every field of every emitted
+            // record names the frontend setter that changes it, and that setter moves a
+            // counter the emitting bit's shutter reads - or the emission is unconditional at
+            // the setter (the resource_* family, set_texture_params). A record field whose
+            // setter moves no shutter input is a stale record with nothing to refuse: c0d
+            // (bit 13 without the bind generation), SD-0 (an image re-bind), F-1 (the
+            // program behind the view set), F-2 (the program behind the image window) and
+            // F-3 (an attached object's storage) were all this one class. DirtySurface.def
+            // cannot catch it - it maps MUTATORS to bits and cannot see that a DERIVED field
+            // depends on a mutator whose row is another family's - so the table lives here,
+            // beside the shutters, and a row is added whenever a record gains a field.
+            //
+            // bit 6  create/bind_shader_state, set_draw/dispatch_program (ProgramEmit.h)
+            //        fields: Cso, StageMask, GlobalUboSize, the artefact blob refs, the two
+            //                bound handles
+            //        setters: glUseProgram (m_currentProgram), glLinkProgram (link version),
+            //                glBindProgramPipeline / glUseProgramStages (pipeline name +
+            //                per-stage {lifetime id, link version})
+            //        shutter: lifetime id x link version, or stageLinks under a pipeline
+            // bit 7  the program's bindings (image units, block bindings, uniform write set)
+            //        setters: glUniform1i on an opaque uniform (backend state version, image
+            //                unit version), glUniformBlockBinding / glShaderStorageBlockBinding
+            //                (block binding version), any glUniform* (uniform write set)
+            //        shutter: the four per-program counters, x stageLinks under a pipeline
+            // bit 8  set_global_constants: ShaderCso, Version, the default-block image
+            //        setters: any glUniform* on the default block (UBO content version),
+            //                glUseProgram (lifetime id)
+            //        shutter: lifetime id x UBO content version, or stageState
+            // bit 11 set_framebuffer_state: Fbo, Color[8]/Depth/Stencil/ReadSurface
+            //        (Res, Kind, InternalFormat, TextureTarget, Layered, Level, Layer,
+            //        UploadTarget), DrawBuffers[8], Width/Height/Layers/Samples/
+            //        FixedSampleLocations, IsDefault, Complete, Target
+            //        setters: glFramebufferTexture*/glFramebufferRenderbuffer, glDrawBuffer(s),
+            //                glReadBuffer, glFramebufferParameteri (the attachment
+            //                aggregate); glBindFramebuffer (the two binding slot versions);
+            //                AND a storage redefinition of an ATTACHED texture or
+            //                renderbuffer - glTexImage*/glTexStorage*/glTexBuffer/
+            //                glTextureView/glRenderbufferStorage* - because InternalFormat,
+            //                TextureTarget, the extent, Samples and Complete are INLINED at
+            //                emission (D-C1): those bump the attachment aggregate from the
+            //                object's PipePublishDescriptor (F-3)
+            //        shutter: attachment aggregate x draw bind version x read bind version
+            // bit 12 set_sampler_views: per unit {View, Texture}
+            //        setters: glBindTexture / glActiveTexture (bind generation), a texture's
+            //                or a sampler object's parameters (SamplesAsIncompleteTexture -
+            //                the params aggregate), an upload that defines a level (content
+            //                aggregate), the default texture's image appearing (bind
+            //                generation, TextureObject.cpp); AND the program in use -
+            //                glUseProgram, a relink, glUniform1i on a sampler uniform (which
+            //                unit a uniform's TYPE resolves) - F-1
+            //        shutter: content x params x bind generation x opaqueUnits
+            // bit 13 bind_sampler_states: per unit the sampler CSO handle
+            //        setters: glBindSampler (bind generation, c0d), glSamplerParameter* /
+            //                glTexParameter* (params aggregate + sampling resolution),
+            //                glDeleteSamplers (bind generation)
+            //        shutter: params x sampling resolution x bind generation
+            // bit 14 set_shader_images: per unit {Res, InternalFormat, Layer, Level,
+            //        Layered, Access} over the program's image-unit window
+            //        setters: glBindImageTexture (bind generation, SD-0), a texture's
+            //                content/params, glUniform1i on an image uniform (image unit
+            //                version); AND the program in use - glUseProgram, a relink -
+            //                F-2
+            //        shutter: content x params x bind generation x programImages
+            //                (lifetime id x link version x image unit version)
+            // ===========================================================================
+
+            // ---- the object-class bits 9..17 ----
+            const Uint64 textureContent = ctx.GetAnyTextureContentGeneration();
+            const Uint64 textureParams = ctx.GetAnyTextureParamsGeneration();
+            const Uint64 buffers = ctx.GetAnyBufferChangeGeneration();
+
+            // Bit 9. The VAO attribute aggregate mixed with the bound VAO's identity is
+            // already exact for the SET - it is bumped by all three Bump*Version functions,
+            // which are the only writers of an attribute's format, buffer or enable state -
+            // and a driver-id re-mint that moves no client counter is caught server-side by
+            // the backend's own id generation.
+            //
+            // THE PENDING BASE INSTANCE IS MIXED IN, and this is a deviation from the design
+            // note that said "keep the shutter" (recorded in client-v1.md): the draw's
+            // baseInstance is now an EXPLICIT field of set_vertex_buffers and a
+            // ContentHash input, and it moves neither the attribute aggregate nor the VAO
+            // identity. Without it here, a draw whose only change is its base instance would
+            // never reach the emitter at all and the server would keep the previous fetch
+            // shift - which is the same silently-wrong-geometry the backend's
+            // baseInstanceDirty flag exists to prevent, one level further out. It fires
+            // extra only on the draws that actually carry one.
+            now[Index(MGPipeDirty::NewVertexBuffers)] = MGPipeMixShutter(
+                MGPipeMixShutter(ctx.GetAnyVaoAttributeGeneration(), vaoIdentity), m_pendingBaseInstance);
+            // Bit 10, NARROWED (P3a, D-I). It used to mix the whole buffer-CONTENT aggregate
+            // with the VAO identity and therefore fired on any buffer write anywhere; what
+            // it guards is one binding slot, so it now reads that slot's own version and the
+            // identity of what is bound to it. The version is a WRAPPING Uint16 bumped only
+            // on a real change, so it goes through the widened counter at this boundary; the
+            // bound object's lifetime id joins it because identity is what closes the wrap
+            // hole. The VAO identity stays in the mix because the element slot BELONGS to
+            // the bound VAO - switching VAOs switches slots.
+            Uint64 indexShutter = 0;
+            if (vao) {
+                const auto& indexSlot = vao->GetIndexBufferBindingSlot();
+                const auto& indexObject = indexSlot.GetBoundObject();
+                indexShutter = MGPipeMixShutter(m_indexSlotVersion.Observe(indexSlot.GetVersion()),
+                                                indexObject ? indexObject->GetLifetimeId() : 0);
+            }
+            now[Index(MGPipeDirty::NewIndexBuffer)] = MGPipeMixShutter(vaoIdentity, indexShutter);
+            // Bit 11, WIDENED AT P4a AND THIS IS A REQUIREMENT RATHER THAN AN OPTION. The
+            // shutter observed the DRAW binding slot only, so glBindFramebuffer(
+            // GL_READ_FRAMEBUFFER, ...) moved nothing at all - which was harmless while
+            // nothing was emitted for the bit and is an UNDER-FIRE the moment P4a emits
+            // set_framebuffer_state per bound target (D-C2): the read record would never be
+            // sent and the server's ReadSurface would stay the previous framebuffer's. Over-
+            // firing costs one extra push; under-firing renders stale, and this file's own
+            // rule is that under-firing is the dangerous direction.
+            //
+            // A STORAGE REDEFINITION OF AN ATTACHED OBJECT MOVES THIS SHUTTER (P4a fable seam
+            // F-3), and the sentence that stood here - "a renderbuffer respecify is still
+            // invisible here, and deliberately so ... closed by emitting resource_respecify
+            // straight from the storage entry point" - was true of the RESOURCE record only.
+            // set_framebuffer_state inlines each attachment's InternalFormat, TextureTarget,
+            // extent, Samples and Complete (D-C1: "so the four cross-object masks fall out at
+            // push time with no lookup"), so `glTexImage2D(tex, RGB8); attach; draw;
+            // glTexImage2D(tex, RGBA8); draw` left the FRAMEBUFFER record saying RGB8 while the
+            // resource record said RGBA8, and the handle arm answered its alpha-widening,
+            // snorm-clamp and integer masks from the stale copy where the legacy arm re-read
+            // the frontend at the same re-sync - a proven arm divergence on a public-GL
+            // sequence. The fix is at the SETTER, not here: TextureObjectBase::PipePublish
+            // Descriptor and RenderbufferObject::PipePublishDescriptor - the one funnel every
+            // storage-defining entry point of either object takes, push-only - bump the
+            // attachment aggregate this shutter already reads. No counter is added to either
+            // object (G1), nothing widens this shutter onto the texture-content aggregate (which
+            // would fire the 304-byte record build on every glTexSubImage2D), and a storage
+            // definition of an UNATTACHED object over-fires it exactly once at load time.
+            //
+            // AND A TRAP THE NEXT NARROWING WOULD WALK INTO, recorded here because it is
+            // invisible from the shutter: FramebufferObject::SetDrawBuffer versions the VALUE
+            // being written rather than the index being written TO - it calls
+            // BumpAttachmentVersion(buffer). The object version and the aggregate still move,
+            // so THIS shutter is safe; a narrower one built on m_attachmentVersions would not
+            // be, and P4a must not build one.
+            now[Index(MGPipeDirty::NewFramebuffer)] = MGPipeMixShutter(
+                MGPipeMixShutter(
+                    ctx.GetAnyFramebufferAttachmentGeneration(),
+                    m_framebufferBind.Observe(
+                        ctx.GetFramebufferBindingSlot(FramebufferTarget::Draw).GetVersion())),
+                m_readFramebufferBind.Observe(
+                    ctx.GetFramebufferBindingSlot(FramebufferTarget::Read).GetVersion()));
+            // Bit 12 reads FOUR things (F-1): the two texture aggregates, the bind generation
+            // and the program input computed above. The params aggregate is here because
+            // SamplerEmit.h drops a unit's view to null when SamplesAsIncompleteTexture says so,
+            // and that predicate reads the effective sampler's filters - a glTexParameteri(
+            // MIN_FILTER) that completes a texture fired bit 13 and not this one, so the entry
+            // stayed null. The program input is here because the set is resolved FOR THE
+            // PROGRAM IN USE, and a glUseProgram alone moved nothing this shutter read.
+            now[Index(MGPipeDirty::NewSamplerViews)] = MGPipeMixShutter(
+                MGPipeMixShutter(MGPipeMixShutter(textureContent, textureParams), ctx.GetTextureBindGeneration()),
+                opaqueUnits);
+            // Bit 13, WIDENED AT P4a FOR BIT 11's REASON and found the same way. glBindSampler
+            // moves NEITHER half of what this used to read: GL_Sampler.cpp's BindSampler_State
+            // goes through NoteTextureUnitTouched and TextureUnit::SetSamplerObject, and both
+            // of those bump the TEXTURE BIND generation - bit 12's. The only two writers of
+            // BumpSamplingResolutionGeneration are PARAMETER changes (SamplerObject.cpp,
+            // TextureObject.cpp). So `glBindSampler(3, a); draw; glBindSampler(3, b); draw`
+            // fired bit 12 twice and bit 13 not once, and the server's BoundSamplerStates[3]
+            // went on naming a's CSO: wrong filtering, with nothing able to see it, because
+            // bind_sampler_states has no pulled twin to fall back on the way the view set does.
+            //
+            // MIXING THE GENERATION IN IS THE FIX RATHER THAN A SECOND GATE ON THE EMITTER,
+            // because that generation is what the unit SET is derived from: a sampler bind
+            // changes which sampler state applies at a unit, and a texture bind changes it too
+            // whenever the unit carries no sampler object and the texture's BUILT-IN sampler is
+            // what applies. Keeping it one shutter per bit is also what keeps the per-subsystem
+            // A/B and the per-bit fire tallies meaning what they say - a bit gated on another
+            // bit's shutter measures neither. The extra fires a plain texture bind now costs
+            // are swallowed by the emitter's own set-hash suppressor, which MGPipeTypes.h makes
+            // mandatory for every kVarTail set for this exact traffic.
+            now[Index(MGPipeDirty::NewSamplers)] = MGPipeMixShutter(
+                MGPipeMixShutter(textureParams, ctx.GetSamplingResolutionGeneration()),
+                ctx.GetTextureBindGeneration());
+            now[Index(MGPipeDirty::NewShaderImages)] = MGPipeMixShutter(
+            MGPipeMixShutter(MGPipeMixShutter(textureContent, textureParams), programImages),
+            ctx.GetTextureBindGeneration());
+            now[Index(MGPipeDirty::NewConstBuffers)] = buffers;
+            now[Index(MGPipeDirty::NewShaderBuffers)] = buffers;
+            now[Index(MGPipeDirty::NewSoTargets)] =
+                MGPipeMixShutter(buffers, ctx.GetTransformFeedbackGeneration());
+
+            Uint32 dirty = 0;
+            for (SizeT i = 0; i < kMGPipeDirtyCount; ++i) {
+                // Bits 2 and 3 are handled below: they are BitwiseEqual shutters, not
+                // counters, so they have no entry in `now`.
+                if (i == Index(MGPipeDirty::NewPixelPack) || i == Index(MGPipeDirty::NewPatchState)) {
+                    continue;
+                }
+                if (!m_primed || now[i] != m_lastPushed[i]) dirty |= Uint32{1} << static_cast<Uint32>(i);
+                m_lastPushed[i] = now[i];
+            }
+
+            // ---- bit 2: the PACK half of the pixel store, BitwiseEqual ----
+            const PixelStoreParameters pack = ctx.GetPixelStoreParameters(false);
+            if (!m_primed || std::memcmp(&pack, &m_pack, sizeof(pack)) != 0) {
+                dirty |= MGPipeDirtyBit(MGPipeDirty::NewPixelPack);
+                m_pack = pack;
+            }
+
+            // ---- bit 3: the patch trio, BitwiseEqual, and NaN IS LEGAL ----
+            // A NaN outer level is a legal glPatchParameterfv value and must compare equal to
+            // itself (ARCHITECTURE.md 5.2). Float equality says it is not; memcmp says it is,
+            // which is the whole reason this is a byte compare.
+            PatchTrio patch{};
+            patch.PatchVertices = render.PatchVertices;
+            for (SizeT i = 0; i < 4; ++i) patch.Outer[i] = render.PatchDefaultOuterLevel[i];
+            for (SizeT i = 0; i < 2; ++i) patch.Inner[i] = render.PatchDefaultInnerLevel[i];
+            if (!m_primed || std::memcmp(&patch, &m_patch, sizeof(patch)) != 0) {
+                dirty |= MGPipeDirtyBit(MGPipeDirty::NewPatchState);
+                m_patch = patch;
+            }
+
+            m_primed = true;
+            m_freshlyPrimed = !wasPrimed;
+            m_lastDirty = dirty;
+
+            if (MG_Util::PipeStats::Enabled()) {
+                const SizeT cls = static_cast<SizeT>(verbClass);
+                ++m_walks[cls];
+                for (SizeT i = 0; i < kMGPipeDirtyCount; ++i) {
+                    if (dirty & (Uint32{1} << static_cast<Uint32>(i))) ++m_fires[i][cls];
+                }
+            }
+            return dirty;
+        }
+
+        // Context teardown, server reset, a unit test's fixture. The next Update returns
+        // every bit set, which is what makes the first verb on a fresh context publish a
+        // complete state rather than an increment. Deliberately does NOT clear the fire
+        // tallies: they are a per-run measurement, not per-context state.
+        //
+        // AND IT DELIBERATELY DOES NOT CLEAR m_pendingBaseInstance. Everything else this
+        // function clears is a LATCH describing what the server was last told; the pending
+        // base instance is THIS CALL'S ARGUMENT, written by the draw entry point one
+        // statement before MGP_FILL and not yet read by anybody. Update() calls Reset() from
+        // inside itself whenever the current GLContext pointer moves, so clearing it here
+        // meant that `eglMakeCurrent(ctxB); glDrawArraysInstancedBaseInstance(..., 7)` put a
+        // BaseInstance of 0 on the wire - one silently mis-shifted instanced draw per context
+        // switch, on the emulation path, with nothing to catch it. The value is cleared by the
+        // verb that consumes it (PipeFill.cpp's step 3, and its no-context early return) and
+        // by MGPipeLeaveVerb, which is where a per-call argument belongs.
+        void Reset() {
+            std::memset(m_lastPushed, 0, sizeof(m_lastPushed));
+            m_renderStateVersion.Reset();
+            m_pipelineStateVersion.Reset();
+            m_framebufferBind.Reset();
+            m_readFramebufferBind.Reset();
+            m_indexSlotVersion.Reset();
+            m_pack = PixelStoreParameters{};
+            m_patch = PatchTrio{};
+            m_staged = RenderStateParameters{};
+            m_stagedAttribs = AttribDefaults{};
+            m_context = nullptr;
+            m_lastDirty = 0;
+            m_primed = false;
+            m_freshlyPrimed = false;
+        }
+
+        void ResetCounters() {
+            std::memset(m_fires, 0, sizeof(m_fires));
+            std::memset(m_walks, 0, sizeof(m_walks));
+        }
+
+        Uint64 FireCount(MGPipeDirty bit, MGPipeVerbClass verbClass) const {
+            return m_fires[Index(bit)][static_cast<SizeT>(verbClass)];
+        }
+        Uint64 FireCount(MGPipeDirty bit) const {
+            Uint64 total = 0;
+            for (SizeT i = 0; i < kMGPipeVerbClassCount; ++i) total += m_fires[Index(bit)][i];
+            return total;
+        }
+        Uint64 WalkCount(MGPipeVerbClass verbClass) const {
+            return m_walks[static_cast<SizeT>(verbClass)];
+        }
+        Uint64 WalkCount() const {
+            Uint64 total = 0;
+            for (SizeT i = 0; i < kMGPipeVerbClassCount; ++i) total += m_walks[i];
+            return total;
+        }
+
+        Uint32 LastDirty() const { return m_lastDirty; }
+        Bool Primed() const { return m_primed; }
+        // True when the LAST Update was the first one after a Reset - a fresh context, or a
+        // server reset. The emission step reads it to send a COMPLETE state rather than an
+        // increment against a staging mirror that describes a context that is gone.
+        Bool FreshlyPrimed() const { return m_freshlyPrimed; }
+
+        // "What the server has" (P2 brief D8). set_dynamic_state sends the dynamic chunks
+        // that differ from this, which is the chunk-level suppressor; a chunk that
+        // memcmp-matches is not sent at all.
+        RenderStateParameters& Staged() { return m_staged; }
+        const RenderStateParameters& Staged() const { return m_staged; }
+
+        // The same mirror for the 32 glVertexAttrib* defaults: set_vertex_attrib_defaults
+        // names only the attributes that differ from it, which is the var-tail's own
+        // suppressor underneath D11's set-hash one.
+        using AttribDefaults = Array<MG_State::GLState::CurrentVertexAttributeValue,
+                                     MG_State::GLState::VertexArrayObject::MAX_VERTEX_ATTRIBS>;
+        AttribDefaults& StagedAttribDefaults() { return m_stagedAttribs; }
+        const AttribDefaults& StagedAttribDefaults() const { return m_stagedAttribs; }
+
+        // ---- P3a D-H2: the draw's vertex-FETCH base instance ----
+        //
+        // It lives HERE rather than in a file static because bit 9's shutter has to see it:
+        // an ambient process global cannot cross a pushed boundary, and the value is now an
+        // explicit field of set_vertex_buffers and an input to its content hash, so a draw
+        // whose only change is its base instance has to reach the emitter. Set immediately
+        // before the fill at the three *BaseInstance draw entry points; CONSUMED and cleared
+        // by the validate point once it has been emitted, so a plain draw that follows one
+        // sees 0 again.
+        //
+        // THE CLEAR THAT ACTUALLY RUNS IN PRODUCTION IS THE VALIDATE POINT'S. MGPipeLeaveVerb
+        // clears it too, but no GL entry point calls MGPipeLeaveVerb - only MG_Test's
+        // ScopedPipeVerb and TrackerTest do - so the production guarantee is entirely
+        // PipeFill.cpp's, on BOTH of its exits: the end of step 3, and the no-live-context
+        // early return that skips step 3 altogether. Reset() deliberately does not clear it
+        // (see there): it is this call's argument, not a latch.
+        void SetPendingBaseInstance(Uint32 baseInstance) { m_pendingBaseInstance = baseInstance; }
+        Uint32 PendingBaseInstance() const { return m_pendingBaseInstance; }
+        void ClearPendingBaseInstance() { m_pendingBaseInstance = 0; }
+
+    private:
+        static constexpr SizeT Index(MGPipeDirty bit) { return static_cast<SizeT>(bit); }
+
+        struct PatchTrio {
+            Uint PatchVertices;
+            Float Outer[4];
+            Float Inner[2];
+        };
+
+        Uint64 m_lastPushed[kMGPipeDirtyCount]{};
+        MGPipeWidenedCounter m_renderStateVersion;
+        MGPipeWidenedCounter m_pipelineStateVersion;
+        // The draw framebuffer BINDING slot version, widened for the same reason: a Uint16
+        // that wrapped would let a composite shutter repeat and cost a missed fire.
+        MGPipeWidenedCounter m_framebufferBind;
+        // P4a: the READ framebuffer binding slot's version, its own counter for the same
+        // reason the draw one exists. Two counters rather than one over both slots: a single
+        // widened counter fed two independent Uint16s reads a decrease as a wrap on every
+        // alternation and would add 65536 per switch, which costs nothing in correctness
+        // (over-firing) but makes the high word meaningless.
+        MGPipeWidenedCounter m_readFramebufferBind;
+        // The BOUND VAO's element-array slot version, widened for the same reason. One
+        // counter over a slot that changes with the bound VAO: a stale high word can only
+        // ADD a fire, never drop one, and the VAO identity in the same mix is what makes a
+        // switch between two VAOs differ whatever their slot versions read.
+        MGPipeWidenedCounter m_indexSlotVersion;
+        Uint32 m_pendingBaseInstance = 0;
+        // Bits 2 and 3 are BitwiseEqual shutters, not counters.
+        PixelStoreParameters m_pack{};
+        PatchTrio m_patch{};
+
+        RenderStateParameters m_staged{};
+        AttribDefaults m_stagedAttribs{};
+
+        const void* m_context = nullptr;
+        Uint32 m_lastDirty = 0;
+        Bool m_primed = false;
+        Bool m_freshlyPrimed = false;
+
+        Uint64 m_fires[kMGPipeDirtyCount][kMGPipeVerbClassCount]{};
+        Uint64 m_walks[kMGPipeVerbClassCount]{};
+    };
+
+    // ONE attribute default, flattened onto the wire (P2 brief D10). A named function rather
+    // than four lines inside the emitter because this flattening is the whole correctness
+    // question of set_vertex_attrib_defaults: a CurrentVertexAttributeValue is one value in
+    // three views and GLContext converts NUMERICALLY between them, so four words alone are
+    // not the value - glVertexAttrib4f(loc, 1.5f, ...) leaves 1 in intValue and 0x3FC00000 in
+    // floatValue. MGPAttribValue::ValueClass is what makes the four words readable again, and
+    // TrackerAttribPayload pins that here instead of leaving it to the emitter's shape.
+    inline void MGPipeFillAttribValue(Uint32 location,
+                                      const MG_State::GLState::CurrentVertexAttributeValue& value,
+                                      Uint32 writtenClass, MGPAttribValue& out) {
+        out = MGPAttribValue{};
+        out.Location = location;
+        out.ValueClass = static_cast<Uint8>(writtenClass);
+        static_assert(sizeof(out.Data) == sizeof(value.floatValue), "MGPAttribValue::Data is four words");
+        switch (writtenClass) {
+        case MG_State::GLState::kVertexAttribValueClassInt:
+            std::memcpy(out.Data, value.intValue.data(), sizeof(out.Data));
+            break;
+        case MG_State::GLState::kVertexAttribValueClassUint:
+            std::memcpy(out.Data, value.uintValue.data(), sizeof(out.Data));
+            break;
+        default:
+            std::memcpy(out.Data, value.floatValue.data(), sizeof(out.Data));
+            break;
+        }
+    }
+
+    // The monolith's one tracker. Under split there is one per client context; the context
+    // identity check inside Update is what makes the single instance safe today.
+    inline MGPipeTracker& MGPipeTrackerInstance() {
+        // NEVER DESTROYED, for MGPipeSlots()' reason (MG_Impl/Pipe/SlotAllocator.cpp). The
+        // rule is stated over the SET of MGPipe process singletons rather than over the two
+        // that a frontend destructor reaches today: which of them a destructor reaches is a
+        // property of the emitters, and the emitters change (C-1 added a second reaching
+        // path in one commit). One allocation per process, no destructor to lose - this type
+        // has none - and nothing can then answer a late call out of freed storage.
+        static MGPipeTracker* tracker = new MGPipeTracker();
+        return *tracker;
+    }
+} // namespace MobileGL::MG_Pipe
+#endif // MOBILEGL_PIPE_PUSH

MobileGL/MG_Impl/Pipe/VertexInputEmit.h

@@ -0,0 +1,444 @@
+// MobileGL - MobileGL/MG_Impl/Pipe/VertexInputEmit.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>
+
+// The CLIENT side of P3a's vertex-input family (brief D-G, D-H, D-I): the bound VAO's
+// format as create/bind_vertex_elements, its buffers as set_vertex_buffers with an explicit
+// baseInstance, and its element binding as set_index_buffer.
+//
+// UNLIKE THE RESOURCE FAMILY, these three emit at the VALIDATE POINT, from
+// MGPipeValidateForVerb's step 3 in the fixed order elements -> buffers -> index. That is
+// the ordinary rule (ARCHITECTURE.md 5.1); the resource family is the one exception to it.
+//
+// THE CSO IS IDENTITY-ADDRESSED, NOT CONTENT-ADDRESSED (D-G1, a recorded deviation from
+// ARCHITECTURE.md's 1024-entry content-addressed scheme). One handle per frontend
+// VertexArrayObject, minted off its lifetime id, and create_vertex_elements is RE-ISSUED on
+// the same handle whenever the configuration moves - legal, because MGPipeHandle::Gen
+// increments only on slot reuse and never on a respecify. Espryt has no vertex-elements CSO
+// to share: its twin owns one driver VAO name plus 64 scratch buffer ids, which two frontend
+// VAOs cannot share, so content addressing would be strictly slower on the only backend this
+// phase touches. P7 adds the hash-probe-memcmp layer above these same three calls when
+// Magma's VertexInputStateFactory takes the CSO over.
+//
+// WHAT THE UNIT GATE READS. G6 is "the emitted blob + set + index record reproduce exactly
+// what the backend's VAO twin reads from the frontend today, field by field, for all 32
+// slots", and G7 is a scripted control that stops the conversion copying ONE field and
+// expects the suite to go red NAMING it. So the conversion is a pure function per field
+// (MGPipeBuildVertexAttribWire / MGPipeBuildVertexBindingPointWire) and the staging buffers
+// the emitter builds into are readable afterwards - the emitter passes m_blob and m_entries
+// straight to the applier, so "what was emitted" costs no copy at all.
+//
+// HEADER-ONLY, for the ownership reason Tracker.h states in full.
+#if MOBILEGL_PIPE_PUSH
+#include <MG_Impl/Pipe/ResourceTracker.h>
+#include <MG_Impl/Pipe/SetHashSuppressor.h>
+#include <MG_Impl/Pipe/SlotAllocator.h>
+#include <MG_Impl/Pipe/Tracker.h>
+#include <MG_Pipe/MGPipe.h>
+#include <MG_Pipe/PipeApply.h>
+#include <MG_State/GLState/Core.h>
+#include <MG_Util/Metrics/PipeStats.h>
+
+#include <xxhash.h>
+
+#include <cstring>
+
+namespace MobileGL::MG_Pipe {
+
+    // ---------------------------------------------------------------------------------
+    // D-G2: the wire conversion, one pure function per view
+    // ---------------------------------------------------------------------------------
+
+    // EVERY FIELD OF VertexAttribute THE WIRE FORM CARRIES, and nothing else:
+    //
+    //   Divisor is deliberately absent - it is resolved per binding point and travels in
+    //     MGPVertexBuffer::Divisor, which is where the backend's glVertexAttribDivisor reads
+    //     it. Carrying it twice would let a malformed record disagree with itself.
+    //   LegacyStride / LegacyPointer are deliberately absent - they are the
+    //     glGetVertexAttrib* query answers and nothing but the query path reads them, so
+    //     they stay client-side.
+    //   Buffer is deliberately absent - identity travels in set_vertex_buffers, which is
+    //     what keeps this record stable while the buffers under it change.
+    //   Stride is the RESOLVED distance and a surviving 0 is MEANINGFUL: a pointer call's 0
+    //     was already resolved to the element size by the frontend, so a 0 here can only
+    //     have come from the binding model, where it means every vertex reads the SAME
+    //     element. Collapsing it back into the element size is what made
+    //     KHR-GL43.vertex_attrib_binding.basic-input-case7/8 read past the buffer.
+    //   IsLong travels SEPARATELY from Type == Float64: VertexAttribFormat(GL_DOUBLE) reads
+    //     doubles and asks for them converted to float, VertexAttribLFormat keeps all 64
+    //     bits, and the backend's fp64 narrowing and its Adreno disabled-attribute
+    //     workaround both key on telling the two apart.
+    inline MGPVertexAttribWire MGPipeBuildVertexAttribWire(const MG_State::GLState::VertexAttribute& attrib,
+                                                           Uint32 bindingIndex) {
+        // ASSERT RATHER THAN ASSUME, in both directions, because the three narrowing casts
+        // below cross a package boundary: VertexArrayObject is another package's file and its
+        // 32-slot bound is its invariant, not this one's, so a BindingIndex of 256 would wrap
+        // to 0 and silently point every attribute at binding 0, and a negative Stride (the
+        // frontend field is a signed int) would arrive as a ~4 GiB unsigned distance.
+        MOBILEGL_ASSERT(bindingIndex < 256u,
+                        "MGPVertexAttribWire::BindingIndex is a Uint8 and cannot carry %u",
+                        static_cast<Uint>(bindingIndex));
+        MOBILEGL_ASSERT(attrib.Size >= 0 && attrib.Size <= 255,
+                        "MGPVertexAttribWire::Size is a Uint8 and cannot carry %d", attrib.Size);
+        MGPVertexAttribWire wire{};
+        wire.Offset = static_cast<Uint64>(attrib.Offset);
+        wire.Stride = static_cast<Int32>(attrib.Stride);
+        wire.Type = static_cast<Uint32>(attrib.Type);
+        wire.Size = static_cast<Uint8>(attrib.Size);
+        wire.Enabled = attrib.Enabled ? 1 : 0;
+        wire.Normalized = attrib.Normalized ? 1 : 0;
+        wire.IsInteger = attrib.IsInteger ? 1 : 0;
+        wire.IsLong = attrib.IsLong ? 1 : 0;
+        wire.IsBgra = attrib.IsBgra ? 1 : 0;
+        wire.BindingIndex = static_cast<Uint8>(bindingIndex);
+        return wire;
+    }
+
+    // The ARB_vertex_attrib_binding view. Its initial Stride is 16, not 0 (GL 4.6 core table
+    // 23.4), which is why the wire form keeps it signed and copies it verbatim.
+    inline MGPVertexBindingPointWire
+    MGPipeBuildVertexBindingPointWire(const MG_State::GLState::VertexBufferBindingPoint& point) {
+        MGPVertexBindingPointWire wire{};
+        wire.Offset = static_cast<Uint64>(point.Offset);
+        wire.Stride = static_cast<Int32>(point.Stride);
+        wire.Divisor = static_cast<Uint32>(point.Divisor);
+        return wire;
+    }
+
+    // ---------------------------------------------------------------------------------
+    // D-H2.3: the content hash, WITH BaseInstance in it
+    // ---------------------------------------------------------------------------------
+    //
+    // A HARD REQUIREMENT, not a nicety. set_vertex_buffers is suppressed on an unchanged
+    // hash (SetHashSuppressor.h's SetVertexBuffers slot), so a baseInstance that moved while
+    // the buffer set did not would be suppressed and the server would keep the previous
+    // fetch shift - exactly the bug the backend's baseInstanceDirty flag exists to prevent.
+    inline Uint64 MGPipeVertexBufferSetContentHash(const MGPVertexBuffer* entries, Uint32 start, Uint32 count,
+                                                   Uint32 baseInstance) {
+        Uint64 hash = XXH64(entries, static_cast<SizeT>(count) * sizeof(MGPVertexBuffer), 0);
+        hash = MGPipeMixShutter(hash, start);
+        hash = MGPipeMixShutter(hash, count);
+        hash = MGPipeMixShutter(hash, baseInstance);
+        return hash;
+    }
+
+    // ---------------------------------------------------------------------------------
+    // The emitter
+    // ---------------------------------------------------------------------------------
+
+    class MGPipeVertexInputEmitter {
+    public:
+        using GLContext = MG_State::GLState::GLContext;
+        using VertexArrayObject = MG_State::GLState::VertexArrayObject;
+        static constexpr SizeT kAttribs = static_cast<SizeT>(VertexArrayObject::MAX_VERTEX_ATTRIBS);
+        static constexpr SizeT kBindings = static_cast<SizeT>(VertexArrayObject::MAX_VERTEX_ATTRIB_BINDINGS);
+        static_assert(kAttribs <= kMGPipeMaxVertexAttribs && kBindings <= kMGPipeMaxVertexAttribs,
+                      "both declared counts are bounded by kMGPipeMaxVertexAttribs");
+
+        // create/bind_vertex_elements. D-G3's three arms, verbatim:
+        //
+        //   no VAO bound                     -> bind the null handle (legal, and it means
+        //                                       exactly "no VAO bound")
+        //   the bound VAO CHANGED            -> (re)create if its configuration moved since
+        //                                       this handle last published one, then bind
+        //   the same VAO, configuration MOVED-> create on the SAME handle, and do NOT rebind
+        //
+        // The latch is PER HANDLE, in a slot-indexed table, so ping-ponging between two VAOs
+        // re-binds but never re-creates either. A Uint32 configuration version does not wrap
+        // in any realistic run and is compared directly; the tracker's widened counter is
+        // for the Uint16s and is not needed here.
+        Uint64 EmitVertexElements(GLContext& ctx) {
+            const auto& vao = ctx.GetBoundVertexArray();
+            if (!vao) {
+                if (!MGPipeHandleIsNull(m_boundHandle)) {
+                    MGPipeApplyBindVertexElements(HandleOnly(kMGPipeNullHandle));
+                    ++m_binds;
+                    m_boundHandle = kMGPipeNullHandle;
+                    m_boundLifetimeId = 0;
+                }
+                return 0;
+            }
+
+            const Uint64 lifetimeId = vao->GetLifetimeId();
+            const Uint32 configVersion = vao->GetConfigVersion();
+            const MGPipeHandle handle = MGPipeSlots().Acquire(MGPipeKind::VertexElementsCso, lifetimeId);
+            const SizeT slot = handle.Slot;
+            if (slot >= m_latch.size()) m_latch.resize(slot + 1);
+            Latch& latch = m_latch[slot];
+
+            Uint64 bytes = 0;
+            const Bool configMoved = !latch.Published || latch.ConfigVersion != configVersion ||
+                                     latch.Gen != handle.Gen;
+            if (configMoved) bytes += EmitCreate(*vao, handle, latch, configVersion);
+            if (lifetimeId != m_boundLifetimeId || m_boundHandle != handle) {
+                MGPipeApplyBindVertexElements(HandleOnly(handle));
+                ++m_binds;
+                bytes += sizeof(MGPHandleOnly);
+                m_boundHandle = handle;
+                m_boundLifetimeId = lifetimeId;
+            }
+            return bytes;
+        }
+
+        // set_vertex_buffers. Espryt consumes RESOLVED attributes, so the set is one entry
+        // per attribute slot with BindingIndex == the attribute index; Start is 0 and Count
+        // is the highest ENABLED attribute plus one, which is the 32-slot prefix walk the
+        // dirty bit is specified over.
+        //
+        // A client-memory array is Res == kMGPipeNullHandle, and that is not a hole: it is
+        // exactly how the server learns "this attribute is client-sourced, upload it
+        // yourself". Its store genuinely does not exist at this moment - the client-array
+        // uploader runs after PrepareForDraw, at the draw entry point - and moving that
+        // resolution to the client is P8's.
+        Uint64 EmitVertexBuffers(GLContext& ctx, Uint32 baseInstance) {
+            const auto& vao = ctx.GetBoundVertexArray();
+            Uint32 count = 0;
+            if (vao) {
+                for (SizeT i = 0; i < kAttribs; ++i) {
+                    if (vao->GetAttribute(static_cast<Uint>(i)).Enabled) count = static_cast<Uint32>(i) + 1;
+                }
+                for (SizeT i = 0; i < count; ++i) {
+                    const auto& attrib = vao->GetAttribute(static_cast<Uint>(i));
+                    MGPVertexBuffer& entry = m_entries[i];
+                    entry = MGPVertexBuffer{};
+                    entry.Res = attrib.Buffer ? MGPipeSlots().Acquire(MGPipeKind::Buffer,
+                                                                     attrib.Buffer->GetLifetimeId())
+                                              : kMGPipeNullHandle;
+                    // D-A3's sticky mask, ORed HERE rather than only sampled at a storage op.
+                    // This is the bit that survives the DSA idiom: a buffer defined through
+                    // glNamedBuffer* may never be bound at any resource emission, but a draw
+                    // that fetches from it resolves it right here, on the GL thread, at every
+                    // draw. Sticky, so one draw is enough for the rest of its life.
+                    MGPipeResourceTrackerInstance().NoteBoundAs(entry.Res, BufferTarget::Vertex);
+                    // The attribute's own byte offset lives in MGPVertexAttribWire::Offset,
+                    // so the entry's is the BINDING's, which the frontend already folded in.
+                    entry.Offset = 0;
+                    // Signed on the frontend, unsigned on the wire, and a negative one would
+                    // arrive as a ~4 GiB fetch distance rather than as an error.
+                    MOBILEGL_ASSERT(attrib.Stride >= 0, "a resolved vertex stride is never negative (%d)",
+                                    attrib.Stride);
+                    entry.Stride = static_cast<Uint32>(attrib.Stride);
+                    entry.Divisor = static_cast<Uint32>(attrib.Divisor);
+                    entry.BindingIndex = static_cast<Uint32>(i);
+                }
+            }
+
+            const Uint64 hash = MGPipeVertexBufferSetContentHash(m_entries.data(), 0, count, baseInstance);
+            if (!MGPipeSetHashSuppressorInstance().ShouldEmit(MGPipeSuppressorSlot::SetVertexBuffers, hash)) {
+                return 0;
+            }
+            m_lastBuffers = MGPVertexBuffers{};
+            m_lastBuffers.Start = 0;
+            m_lastBuffers.Count = count;
+            // THE DRAW'S RAW value. The client never pre-shifts an offset and never learns
+            // whether the server emulated the shift or let GL_EXT_base_instance do it -
+            // emulation is server-owned.
+            m_lastBuffers.BaseInstance = baseInstance;
+            m_lastBuffers.ContentHash = hash;
+            MGPipeApplySetVertexBuffers(m_lastBuffers, m_entries.data());
+            ++m_bufferSets;
+            return sizeof(MGPVertexBuffers) + static_cast<Uint64>(count) * sizeof(MGPVertexBuffer);
+        }
+
+        // set_index_buffer. An INDEPENDENT call, not a subset of the vertex-elements
+        // configuration version (D5) - the index slot is explicitly outside the VAO's
+        // m_configVersion, and the shutter for it is bit 10's, narrowed in Tracker.h.
+        //
+        // Offset and IndexSize are 0 here and the draw verb overrides them: at the validate
+        // point there is no draw to read them from, and the applier stores what it is given.
+        Uint64 EmitIndexBuffer(GLContext& ctx) {
+            const auto& vao = ctx.GetBoundVertexArray();
+            m_lastIndex = MGPIndexBuffer{};
+            if (vao) {
+                if (const auto& bound = vao->GetIndexBufferBindingSlot().GetBoundObject()) {
+                    m_lastIndex.Res = MGPipeSlots().Acquire(MGPipeKind::Buffer, bound->GetLifetimeId());
+                    // The ELEMENT_ARRAY bit, and it is the one the split path keys on
+                    // (kCapNeedsHostIndexBytes -> restart rewriting, multi-draw flattening).
+                    // Noted at every draw for RefreshBindMask's reason: an EBO defined through
+                    // DSA and unbound before its last respecify would otherwise never publish
+                    // it, and getting that bit wrong is invisible in monolith.
+                    MGPipeResourceTrackerInstance().NoteBoundAs(m_lastIndex.Res, BufferTarget::Index);
+                }
+            }
+            MGPipeApplySetIndexBuffer(m_lastIndex);
+            ++m_indexSets;
+            return sizeof(MGPIndexBuffer);
+        }
+
+        // ---- what a unit case reads. None of it costs a copy: the emitter builds INTO
+        // these and hands the applier the same pointers. ----
+        const Array<MGPVertexAttribWire, kMGPipeMaxVertexAttribs>& LastAttributes() const { return m_attributes; }
+        const Array<MGPVertexBindingPointWire, kMGPipeMaxVertexAttribs>& LastBindingPoints() const {
+            return m_bindingPoints;
+        }
+        const MGPVertexElements& LastElements() const { return m_lastElements; }
+        const MGPVertexBuffers& LastVertexBuffers() const { return m_lastBuffers; }
+        const Array<MGPVertexBuffer, kMGPipeMaxVertexAttribs>& LastEntries() const { return m_entries; }
+        const MGPIndexBuffer& LastIndexBuffer() const { return m_lastIndex; }
+        MGPipeHandle BoundHandle() const { return m_boundHandle; }
+        Uint64 CreateCount() const { return m_creates; }
+        Uint64 BindCount() const { return m_binds; }
+        Uint64 VertexBufferSetCount() const { return m_bufferSets; }
+        Uint64 IndexBufferSetCount() const { return m_indexSets; }
+
+        // ---- C-1: "does the applier hold a record for exactly this handle?" ----
+        //
+        // The CSO's death path (MGPipeEmitVertexElementsDestroyAndFree) needs that answer and
+        // MUST NOT GUESS IT FROM THE SLOT. A VertexElementsCso slot can exist with no record
+        // behind it, because a backend that keys its twins on the handle mints the slot itself
+        // (DirectGLES' BackendSlotTable::GetOrCreate -> MGPipeSlots().Acquire) whether or not
+        // bit 8 ever asked this client to emit anything - which is exactly what a
+        // MOBILEGL_PIPE_PUSH=0x7f lane runs. delete_vertex_elements on such a handle is a
+        // REFUSED call, and the applier's resolver asserts on a refusal
+        // (PipeApply.cpp's ResolveVertexElements), i.e. a stop in a verify build.
+        //
+        // Kept OUT of Reset(), unlike the create/bind latch beside it, and for the mirror
+        // image of Reset()'s own reason: "a fresh context is a fresh server" is true of the
+        // per-context half of this table, and object RECORDS are precisely what
+        // MGPipeApplierReset does not clear (PipeApply.h's two halves). This half tracks those
+        // records, so it lives exactly as long as they do.
+        Bool RecordIsPublished(MGPipeHandle handle) const {
+            if (MGPipeHandleIsNull(handle)) return false;
+            const SizeT slot = handle.Slot;
+            if (slot >= m_latch.size()) return false;
+            const Latch& latch = m_latch[slot];
+            return latch.RecordLive && latch.RecordGen == handle.Gen;
+        }
+
+        // The record named by `handle` is gone from the applier. Also drops the bound-handle
+        // memo when it named it, so the client's idea of BoundVertexElements and the applier's
+        // (which MGPipeApplyDeleteVertexElements just cleared for the same handle) stay in
+        // step rather than diverging until the next bind happens to correct it.
+        void NoteRecordDestroyed(MGPipeHandle handle) {
+            if (MGPipeHandleIsNull(handle)) return;
+            const SizeT slot = handle.Slot;
+            if (slot < m_latch.size() && m_latch[slot].RecordGen == handle.Gen) {
+                m_latch[slot] = Latch{};
+            }
+            if (m_boundHandle == handle) {
+                m_boundHandle = kMGPipeNullHandle;
+                m_boundLifetimeId = 0;
+            }
+        }
+
+        // A fresh context is a fresh server: the applier's records are gone, so every latch
+        // this emitter holds describes objects the server no longer has. Called from the
+        // validate point's FreshlyPrimed arm beside MGPipeApplierReset and the suppressor's
+        // InvalidateAll, for the same reason they are.
+        //
+        // The PER-CONTEXT half only - see RecordIsPublished above for why RecordLive/RecordGen
+        // survive. Re-creating a configuration the applier already holds is a bounded
+        // over-fire (MGPipeApplyCreateVertexElements starts the record over); forgetting that
+        // it holds one at all would leak the record and its slot at the object's death.
+        void Reset() {
+            for (Latch& latch : m_latch) {
+                latch.Published = false;
+                latch.Gen = 0;
+                latch.ConfigVersion = 0;
+            }
+            m_boundHandle = kMGPipeNullHandle;
+            m_boundLifetimeId = 0;
+        }
+
+        void ResetCounters() { m_creates = m_binds = m_bufferSets = m_indexSets = 0; }
+
+    private:
+        struct Latch {
+            // The PER-CONTEXT half: "has this emitter told THIS server about this handle's
+            // configuration". Cleared by Reset() at every make-current.
+            Bool Published = false;
+            Uint32 Gen = 0;
+            Uint32 ConfigVersion = 0;
+            // The RECORD half: "does the applier hold a create_vertex_elements record at this
+            // slot, for this generation". Lives as long as the record does - see
+            // RecordIsPublished.
+            Bool RecordLive = false;
+            Uint32 RecordGen = 0;
+        };
+
+        static MGPHandleOnly HandleOnly(MGPipeHandle handle) {
+            MGPHandleOnly only{};
+            only.Handle = handle;
+            only.Kind = static_cast<Uint32>(MGPipeKind::VertexElementsCso);
+            return only;
+        }
+
+        Uint64 EmitCreate(const VertexArrayObject& vao, MGPipeHandle handle, Latch& latch, Uint32 configVersion) {
+            // ALL 32 OF EACH, deliberately. The record DECLARES both counts and the applier
+            // refuses one whose counts do not describe its own blob, so a self-describing
+            // record is the cheap shape - and G6 is stated over all 32 slots, which a
+            // truncated set could not answer. It rides create_vertex_elements only, i.e.
+            // once per configuration change, never per draw.
+            for (SizeT i = 0; i < kAttribs; ++i) {
+                m_attributes[i] = MGPipeBuildVertexAttribWire(vao.GetAttribute(static_cast<Uint>(i)),
+                                                              vao.GetAttributeBindingIndex(static_cast<Uint>(i)));
+            }
+            for (SizeT i = 0; i < kBindings; ++i) {
+                m_bindingPoints[i] = MGPipeBuildVertexBindingPointWire(vao.GetBindingPoint(static_cast<Uint>(i)));
+            }
+            // Attributes first, then binding points, both ascending and contiguous.
+            constexpr SizeT kAttribBytes = kAttribs * sizeof(MGPVertexAttribWire);
+            constexpr SizeT kBindingBytes = kBindings * sizeof(MGPVertexBindingPointWire);
+            std::memcpy(m_blob.data(), m_attributes.data(), kAttribBytes);
+            std::memcpy(m_blob.data() + kAttribBytes, m_bindingPoints.data(), kBindingBytes);
+
+            m_lastElements = MGPVertexElements{};
+            m_lastElements.Cso = handle;
+            m_lastElements.AttributeCount = static_cast<Uint32>(kAttribs);
+            m_lastElements.BindingPointCount = static_cast<Uint32>(kBindings);
+            m_lastElements.Blob.Seg = kMGHostSpanSegNone;
+            m_lastElements.Blob.Offset = 0;
+            m_lastElements.Blob.Size = kAttribBytes + kBindingBytes;
+            MGPipeApplyCreateVertexElements(m_lastElements, m_blob.data());
+            ++m_creates;
+            latch.Published = true;
+            latch.Gen = handle.Gen;
+            latch.ConfigVersion = configVersion;
+            // THE ONE PRODUCER of the record half: a create that reached the applier is the
+            // only thing that makes delete_vertex_elements a legal call for this handle.
+            latch.RecordLive = true;
+            latch.RecordGen = handle.Gen;
+            return sizeof(MGPVertexElements) + kAttribBytes + kBindingBytes;
+        }
+
+        Array<MGPVertexAttribWire, kMGPipeMaxVertexAttribs> m_attributes{};
+        Array<MGPVertexBindingPointWire, kMGPipeMaxVertexAttribs> m_bindingPoints{};
+        Array<Uint8, kMGPipeMaxVertexAttribs *(sizeof(MGPVertexAttribWire) + sizeof(MGPVertexBindingPointWire))>
+            m_blob{};
+        Array<MGPVertexBuffer, kMGPipeMaxVertexAttribs> m_entries{};
+
+        MGPVertexElements m_lastElements{};
+        MGPVertexBuffers m_lastBuffers{};
+        MGPIndexBuffer m_lastIndex{};
+
+        Vector<Latch> m_latch;
+        MGPipeHandle m_boundHandle = kMGPipeNullHandle;
+        Uint64 m_boundLifetimeId = 0;
+
+        Uint64 m_creates = 0;
+        Uint64 m_binds = 0;
+        Uint64 m_bufferSets = 0;
+        Uint64 m_indexSets = 0;
+    };
+
+    // The monolith's one vertex-input emitter, beside the tracker, the CSO cache, the
+    // set-hash suppressor and the resource tracker.
+    inline MGPipeVertexInputEmitter& MGPipeVertexInputEmitterInstance() {
+        // NEVER DESTROYED, for MGPipeSlots()' reason (MG_Impl/Pipe/SlotAllocator.cpp), and
+        // this one is not hypothetical: C-1 put this emitter DIRECTLY on ~VertexArrayObject's
+        // path - MGPipeEmitVertexElementsDestroyAndFree asks RecordIsPublished(handle) and
+        // then NoteRecordDestroyed(handle), which read and WRITE m_latch. A destroyed
+        // emitter answers out of a freed Vector and the write grows it, i.e. an operator
+        // new + memcpy + operator delete on an already-freed block.
+        static MGPipeVertexInputEmitter* emitter = new MGPipeVertexInputEmitter();
+        return *emitter;
+    }
+} // namespace MobileGL::MG_Pipe
+#endif // MOBILEGL_PIPE_PUSH

MobileGL/MG_IntegrationTest/Harness/P4aFinalFixPeek.cpp

@@ -0,0 +1,71 @@
+// MobileGL - MobileGL/MG_IntegrationTest/Harness/P4aFinalFixPeek.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
+
+#include "P4aFinalFixPeek.h"
+
+#if !defined(__ANDROID__)
+#include <MG_Pipe/MGPipe.h>
+#if MOBILEGL_PIPE_PUSH
+#include <MG_Pipe/MGPipeTypes.h>
+#include <MG_Pipe/PipeApply.h>
+#include <MG_Util/Metrics/PipeStats.h>
+#define MGITEST_P4A_FINALFIX_PEEK_LIVE 1
+#endif
+#endif
+
+namespace MGITest {
+
+#if defined(MGITEST_P4A_FINALFIX_PEEK_LIVE)
+    namespace {
+        namespace MGP = MobileGL::MG_Pipe;
+    } // namespace
+
+    bool PeekPipeTextureResourceRecord(unsigned glTextureName, PipeTextureResourceRecordPeek* out) {
+        if (out == nullptr) return false;
+        const MGP::MGPipeApplierState& applier = MGP::MGPipeApplier();
+        // Slot 0 is the reserved null slot; the walk is the same shape PipeApplyPeek.cpp's
+        // params reading takes. A GL name is never an identity on the wire, which is exactly
+        // why it is the right key for a harness that starts from the application's view.
+        for (MobileGL::SizeT slot = 1; slot < applier.TextureResources.size(); ++slot) {
+            const MGP::MGPipeResourceRecord& record = applier.TextureResources[slot];
+            if (!record.Live) continue;
+            if (record.Desc.GlNameForDiag != static_cast<MobileGL::Uint32>(glTextureName)) continue;
+            out->Slot = static_cast<unsigned>(slot);
+            out->Gen = static_cast<unsigned>(record.Gen);
+            out->Serial = static_cast<unsigned long long>(record.Serial);
+            out->BindMask = static_cast<unsigned>(record.Desc.BindMask);
+            out->ImageBindableHint = static_cast<unsigned>(record.Desc.ImageBindableHint);
+            out->Levels = static_cast<unsigned>(record.Desc.Levels);
+            out->PendingUploads = static_cast<unsigned>(record.PendingUploads.size());
+            return true;
+        }
+        return false;
+    }
+
+    bool PeekPipeStatsTextureRemintPulls(unsigned long long* out) {
+        if (out == nullptr) return false;
+        namespace Stats = MobileGL::MG_Util::PipeStats;
+        if (!Stats::Enabled()) Stats::SetEnabledForTesting(true);
+        *out = static_cast<unsigned long long>(Stats::TotalCalls(Stats::CallClass::TextureRemintPulls));
+        return true;
+    }
+
+    bool PeekPipeStatsTextureUploadEmissions(unsigned long long* out) {
+        if (out == nullptr) return false;
+        namespace Stats = MobileGL::MG_Util::PipeStats;
+        if (!Stats::Enabled()) Stats::SetEnabledForTesting(true);
+        *out = static_cast<unsigned long long>(Stats::TotalCalls(Stats::CallClass::TextureUploadEmissions));
+        return true;
+    }
+#else
+    bool PeekPipeTextureResourceRecord(unsigned, PipeTextureResourceRecordPeek*) { return false; }
+    bool PeekPipeStatsTextureRemintPulls(unsigned long long*) { return false; }
+    bool PeekPipeStatsTextureUploadEmissions(unsigned long long*) { return false; }
+#endif
+
+} // namespace MGITest

MobileGL/MG_IntegrationTest/Harness/P4aFinalFixPeek.h

@@ -0,0 +1,41 @@
+// MobileGL - MobileGL/MG_IntegrationTest/Harness/P4aFinalFixPeek.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
+//
+// The white-box readings P4aFinalFixScenario.cpp takes, in a translation unit of their own for
+// P4aSeamPeek.h's reason: a scenario TU includes the GL prototype headers and cannot include
+// MG_Pipe/PipeApply.h or the Espryt managers beside them, and PipeApplyPeek.cpp is the gates
+// package's file. Every entry point answers false where the reading cannot be taken (a pull
+// build, Android, or an applier that holds no record for the name), and a false teaches the
+// caller nothing - the case declines that half by name and keeps its public-GL verdict.
+#pragma once
+
+namespace MGITest {
+
+    // The applier's resource record for a texture, found by its GL name (GlNameForDiag - a
+    // diagnostics-only field, which is exactly what a test harness is).
+    struct PipeTextureResourceRecordPeek {
+        unsigned Slot;
+        unsigned Gen;
+        unsigned long long Serial;
+        unsigned BindMask;
+        unsigned ImageBindableHint;
+        unsigned Levels;
+        unsigned PendingUploads;
+    };
+    bool PeekPipeTextureResourceRecord(unsigned glTextureName, PipeTextureResourceRecordPeek* out);
+
+    // The process-wide texture-remint pull count (PipeStats "tex-remint-pulls", `trp=` on the
+    // summary line; ROADMAP open question 2). Arms the PipeStats counters for this process on
+    // the first call, which is what lets a case read the number without a stats-enabled lane.
+    bool PeekPipeStatsTextureRemintPulls(unsigned long long* out);
+    // Espryt's count of texture uploads it actually issued (PipeStats "tex-upload-emissions"):
+    // what tells a CONSUMED pending upload apart from a DROPPED one, since the record's set is
+    // empty either way. Arms the counters the same way.
+    bool PeekPipeStatsTextureUploadEmissions(unsigned long long* out);
+
+} // namespace MGITest

MobileGL/MG_IntegrationTest/Harness/P4aSeamPeek.cpp

@@ -0,0 +1,107 @@
+// MobileGL - MobileGL/MG_IntegrationTest/Harness/P4aSeamPeek.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
+
+#include "P4aSeamPeek.h"
+
+#if !defined(__ANDROID__)
+#include <MG_Pipe/MGPipe.h>
+#if MOBILEGL_PIPE_PUSH
+#include <MG_Pipe/PipeApply.h>
+#include <MG_State/GLState/Core.h>
+#include <MG_Backend/DirectGLES/Managers.h>
+#include <MG_Backend/DirectGLES/DirectGLES.h>
+#define MGITEST_P4A_SEAM_PEEK_LIVE 1
+#endif
+#endif
+
+namespace MGITest {
+
+#if defined(MGITEST_P4A_SEAM_PEEK_LIVE)
+    namespace {
+        namespace MGP = MobileGL::MG_Pipe;
+        namespace MGB = MobileGL::MG_Backend::DirectGLES;
+
+        // "Is Espryt the backend running" - the same test PipeApplyPeek.cpp makes through a twin:
+        // on Magma no ES entry point was ever resolved and every member of g_GLESFuncs is null.
+        // It is asked BEFORE SamplerSubsystemEnabled(), which is Espryt's own latch and must not
+        // be resolved on a process whose backend is not Espryt.
+        bool EsprytIsRunning() { return MGB::g_GLESFuncs.glBindSampler != nullptr; }
+    } // namespace
+
+    bool PeekEsprytSamplerHandleArmIsLive(bool* outLive) {
+        if (outLive == nullptr) return false;
+        if (!EsprytIsRunning()) return false;
+        *outLive = MGB::SamplerSubsystemEnabled();
+        return true;
+    }
+
+    bool PeekEsprytFramebufferHandleArmIsLive(bool* outLive) {
+        if (outLive == nullptr) return false;
+        if (!EsprytIsRunning()) return false;
+        *outLive = MGB::FramebufferSubsystemEnabled();
+        return true;
+    }
+
+    bool PeekPipeShaderImageWindow(PipeShaderImageWindowPeek* out) {
+        if (out == nullptr) return false;
+        const MGP::MGPipeApplierState& applier = MGP::MGPipeApplier();
+        out->Start = static_cast<unsigned>(applier.ShaderImageStart);
+        out->Count = static_cast<unsigned>(applier.ShaderImageCount);
+        out->Serial = static_cast<unsigned long long>(applier.ShaderImagesSerial);
+        return true;
+    }
+
+    bool PeekEsprytUnitSampler(unsigned unit, unsigned glSamplerName, EsprytUnitSamplerPeek* out) {
+        if (out == nullptr) return false;
+        if (!EsprytIsRunning()) return false;
+        if (!MobileGL::MG_State::pGLContext) return false;
+        const MGP::MGPipeApplierState& applier = MGP::MGPipeApplier();
+        if (unit >= applier.BoundSamplerStates.size() || unit >= MGB::SamplerImpl::g_boundSamplersCache.size()) {
+            return false;
+        }
+        *out = EsprytUnitSamplerPeek{};
+
+        // Espryt's own binding shadow: every glBindSampler this backend issues routes through it
+        // (BackendSamplerObject::Bind / UnbindSampler), so it IS what the driver holds.
+        if (MGB::SamplerImpl::BackendSamplerObject* const bound = MGB::SamplerImpl::g_boundSamplersCache[unit]) {
+            out->BoundSamplerId = static_cast<unsigned>(bound->GetBackendSamplerId());
+        }
+
+        const MGP::MGPipeHandle cso = applier.BoundSamplerStates[unit];
+        out->CsoHandleSlot = static_cast<unsigned>(cso.Slot);
+        out->CsoHandleGen = static_cast<unsigned>(cso.Gen);
+        out->UnitInsideWindow = unit >= applier.SamplerStateStart &&
+                                unit - applier.SamplerStateStart < applier.SamplerStateCount;
+        // The twin AT THE CSO HANDLE, asked of the same table Espryt asks (FindByHandle): a null
+        // here with a live handle is the F-4 shape - a content-addressed handle looked up in a
+        // table that only ever held identity-minted slots.
+        if (!MGP::MGPipeHandleIsNull(cso)) {
+            if (auto* const slot = MGB::SamplerImpl::g_backendSamplerObjects.FindByHandle(cso); slot && *slot) {
+                out->CsoTwinSamplerId = static_cast<unsigned>((*slot)->GetBackendSamplerId());
+            }
+        }
+
+        // And the twin keyed on the frontend OBJECT, which is what the pre-handle program pass
+        // used to mint and bind, so a scenario can say which of the two the driver holds.
+        const auto& object = MobileGL::MG_State::pGLContext->GetSamplerObject(
+            static_cast<MobileGL::Uint>(glSamplerName));
+        if (object) {
+            if (auto* const slot = MGB::SamplerImpl::g_backendSamplerObjects.Find(object.get()); slot && *slot) {
+                out->IdentityTwinSamplerId = static_cast<unsigned>((*slot)->GetBackendSamplerId());
+            }
+        }
+        return true;
+    }
+#else
+    bool PeekEsprytSamplerHandleArmIsLive(bool*) { return false; }
+    bool PeekEsprytFramebufferHandleArmIsLive(bool*) { return false; }
+    bool PeekPipeShaderImageWindow(PipeShaderImageWindowPeek*) { return false; }
+    bool PeekEsprytUnitSampler(unsigned, unsigned, EsprytUnitSamplerPeek*) { return false; }
+#endif
+
+} // namespace MGITest

MobileGL/MG_IntegrationTest/Harness/P4aSeamPeek.h

@@ -0,0 +1,78 @@
+// MobileGL - MobileGL/MG_IntegrationTest/Harness/P4aSeamPeek.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
+//
+// The three readings P4aSeamAuditScenario.cpp takes from the inside, for the two seams the fable
+// seam audit proved that PUBLIC GL CANNOT SEE: F-4 (the record arm's sampler bind is a permanent
+// no-op, hidden by the pre-handle program pass binding the same values) and F-2 / SD-4 (the
+// shader-image window does not follow a program switch, hidden by the server's window/high-water
+// union taking the pre-handle bind for the units outside it). Both are correct pictures over a
+// permanent silent fallback, which is precisely the class ROADMAP.md:20 says a gate has to be
+// able to make red - and the only place the difference exists is inside.
+//
+// A SEPARATE TRANSLATION UNIT for PipeApplyPeek.h's reason, verbatim: this file includes
+// Espryt's own Managers.h, which may not meet a scenario's GL headers in one file. It is NOT
+// PipeApplyPeek.cpp because that file is package F's (gates v3) and this round may not edit it.
+//
+// EVERY ENTRY POINT RETURNS false, TOUCHING NOTHING, WHERE IT CANNOT LOOK - a pull build, Android,
+// a backend that is not Espryt - and a caller that gets false has learned NOTHING: "could not
+// look" is not "was bound". The scenario declines the reading BY NAME and keeps its public-GL
+// half, which is the shape TextureParamsWithoutASamplerViewScenario.cpp argues for.
+
+#pragma once
+
+namespace MGITest {
+
+    // ---- is Espryt's sampler family on its HANDLE arm in this process? -------------------
+    //
+    // The gate for every other reading here. True only on DirectGLES, in a push build, with
+    // Espryt's own resolver answering "handle" for kMGPipeSubsystemSamplers (bit 11 set and its
+    // dependency satisfied) - i.e. exactly when bind_sampler_states / set_shader_images are
+    // consumed, so a white-box assertion about them can be red for its own reason and for no
+    // other. Written only on true.
+    bool PeekEsprytSamplerHandleArmIsLive(bool* outLive);
+
+    // The same question for the FRAMEBUFFER family (bit 9): true when Espryt consumes
+    // set_framebuffer_state in this process. The renderbuffer half of the F-3 case asserts only
+    // there - on the pre-handle arm a renderbuffer re-storaged while attached moves nothing the
+    // FBO memo reads (D-D2's documented hole, pre-P4a code), and the record is what closes it.
+    bool PeekEsprytFramebufferHandleArmIsLive(bool* outLive);
+
+    // ---- the applier's shader-image window, as last received ------------------------------
+    //
+    // MGPipeApplierState::ShaderImageStart / ShaderImageCount / ShaderImagesSerial. Count is
+    // "how many units set_shader_images last described" - 0 means the set has NEVER arrived
+    // (MGPipeApplierReset advances the serial whether or not anything was emitted, so the serial
+    // is not that test). Push build only.
+    struct PipeShaderImageWindowPeek {
+        unsigned Start;
+        unsigned Count;
+        unsigned long long Serial;
+    };
+
+    bool PeekPipeShaderImageWindow(PipeShaderImageWindowPeek* out);
+
+    // ---- which driver sampler a texture unit is bound to, and whose twin it is -------------
+    //
+    // For F-4. `BoundSamplerId` is the ES sampler name Espryt's own binding shadow says unit
+    // `unit` carries (0 = none). `CsoHandleSlot/Gen` is bind_sampler_states' handle for the unit,
+    // `CsoTwinSamplerId` the ES name of the twin Espryt holds AT THAT HANDLE (0 = no twin at the
+    // content-addressed slot - the F-4 shape), and `IdentityTwinSamplerId` the ES name of a twin
+    // keyed on the frontend SamplerObject named `glSamplerName` (0 = none). On a correct handle
+    // arm the unit's driver sampler IS the CSO twin. Push build, DirectGLES only.
+    struct EsprytUnitSamplerPeek {
+        unsigned BoundSamplerId;
+        unsigned CsoHandleSlot;
+        unsigned CsoHandleGen;
+        bool UnitInsideWindow;
+        unsigned CsoTwinSamplerId;
+        unsigned IdentityTwinSamplerId;
+    };
+
+    bool PeekEsprytUnitSampler(unsigned unit, unsigned glSamplerName, EsprytUnitSamplerPeek* out);
+
+} // namespace MGITest

MobileGL/MG_IntegrationTest/Harness/PipeApplyPeek.cpp

@@ -0,0 +1,188 @@
+// MobileGL - MobileGL/MG_IntegrationTest/Harness/PipeApplyPeek.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
+
+#include "PipeApplyPeek.h"
+
+#if !defined(__ANDROID__)
+#include <MG_Pipe/MGPipe.h>
+#if MOBILEGL_PIPE_PUSH
+#include <MG_Pipe/PipeApply.h>
+#include <MG_Pipe/MGPipeTypes.h>
+#include <MG_State/GLState/Core.h>
+#include <MG_Util/Converters/MGToGL/TextureEnumConverter.h>
+#include <MG_Backend/DirectGLES/Managers.h>
+#include <MG_Backend/DirectGLES/DirectGLES.h>
+#define MGITEST_PIPE_APPLY_PEEK_LIVE 1
+#endif
+#endif
+
+namespace MGITest {
+
+#if defined(MGITEST_PIPE_APPLY_PEEK_LIVE)
+    namespace {
+        namespace MGP = MobileGL::MG_Pipe;
+        namespace MGB = MobileGL::MG_Backend::DirectGLES;
+
+        // The frontend texture object a GL name denotes in the CURRENT context, or null. This is
+        // a LOOKUP KEY and nothing else: every value this file reports comes from the applier or
+        // from Espryt, never from the object found here. (Reading the frontend's own parameter
+        // state would answer the question the scenario is asking with the input to it.)
+        MobileGL::MG_State::GLState::ITextureObject* FrontendTexture(unsigned glTextureName) {
+            if (!MobileGL::MG_State::pGLContext) return nullptr;
+            const auto& object = MobileGL::MG_State::pGLContext->GetTextureObject(
+                static_cast<MobileGL::Uint>(glTextureName));
+            return object ? object.get() : nullptr;
+        }
+
+        // Espryt's twin for that texture, or null - which is also this file's "is Espryt even the
+        // backend running" answer. On Magma no Espryt twin was ever built, so every entry point
+        // below stops here rather than reaching for g_GLESFuncs, whose members are null there.
+        MGB::TextureImpl::BackendTextureObject* EsprytTwin(unsigned glTextureName) {
+            MobileGL::MG_State::GLState::ITextureObject* const object = FrontendTexture(glTextureName);
+            if (object == nullptr) return nullptr;
+            auto* const found = MGB::TextureImpl::g_backendTextureObjects.Find(object);
+            if (found == nullptr || !*found) return nullptr;
+            return found->get();
+        }
+
+        int SwizzleToGLEnum(MobileGL::Uint8 encoded) {
+            return static_cast<int>(MobileGL::MG_Util::ConvertTextureSwizzleParamToGLEnum(
+                static_cast<MobileGL::TextureSwizzleParam>(encoded)));
+        }
+
+        // MGPipeTypes.h owns the two numbers and says why depth is 0 (a zeroed record must decode
+        // to what an untouched texture already has). This is that decode, and nothing else in
+        // this module may open-code it.
+        int DepthStencilModeToGLEnum(MobileGL::Uint8 encoded) {
+            return encoded == MGP::kMGPipeDepthStencilModeStencil ? GL_STENCIL_INDEX
+                                                                  : GL_DEPTH_COMPONENT;
+        }
+
+        // The GL_TEXTURE_BINDING_* query for a target, or 0 where this file has no answer. A
+        // guess would be worse than a refusal: the binding is what gets RESTORED, so a wrong
+        // pname would leave the driver bound to this test's texture.
+        int BindingQueryFor(unsigned glTarget) {
+            switch (glTarget) {
+                case GL_TEXTURE_2D: return GL_TEXTURE_BINDING_2D;
+                default: return 0;
+            }
+        }
+    } // namespace
+
+    bool PeekPipeTextureParamsRecord(unsigned glTextureName, PipeTextureParamsRecordPeek* out) {
+        if (out == nullptr) return false;
+        const MGP::MGPipeApplierState& applier = MGP::MGPipeApplier();
+        // Slot 0 is the reserved null handle and is never live (MGPipeHandles.h), so the scan
+        // starts at 1 and a match at 0 is impossible rather than merely unlikely.
+        for (MobileGL::SizeT slot = 1; slot < applier.TextureResources.size(); ++slot) {
+            const MGP::MGPipeResourceRecord& record = applier.TextureResources[slot];
+            if (!record.Live) continue;
+            if (record.Desc.GlNameForDiag != static_cast<MobileGL::Uint32>(glTextureName)) continue;
+            out->Slot = static_cast<unsigned>(slot);
+            out->Gen = static_cast<unsigned>(record.Gen);
+            out->ParamsSerial = static_cast<unsigned long long>(record.ParamsSerial);
+            for (int channel = 0; channel < 4; ++channel) {
+                out->Swizzle[channel] = SwizzleToGLEnum(record.Params.Swizzle[channel]);
+            }
+            out->DepthStencilMode = DepthStencilModeToGLEnum(record.Params.DepthStencilMode);
+            return true;
+        }
+        return false;
+    }
+
+    bool PeekEsprytAppliedTextureParams(unsigned glTextureName, unsigned glTarget,
+                                        EsprytAppliedTextureParamsPeek* out) {
+        if (out == nullptr) return false;
+        const int bindingQuery = BindingQueryFor(glTarget);
+        if (bindingQuery == 0) return false;
+        MGB::TextureImpl::BackendTextureObject* const twin = EsprytTwin(glTextureName);
+        if (twin == nullptr) return false;
+        const MobileGL::Uint backendId = twin->GetBackendTextureId();
+        if (backendId == 0) return false;
+        if (MGB::g_GLESFuncs.glGetTexParameteriv == nullptr ||
+            MGB::g_GLESFuncs.glBindTexture == nullptr || MGB::g_GLESFuncs.glGetIntegerv == nullptr ||
+            MGB::g_GLESFuncs.glGetError == nullptr) {
+            return false;
+        }
+
+        // SAVE / QUERY / RESTORE ON THE UNIT THAT IS ALREADY ACTIVE. No glActiveTexture, so the
+        // only driver state this touches is one unit's binding, and it is put back byte for byte
+        // - which is what keeps Espryt's own g_boundTexturesCache true rather than merely
+        // consistent. (Binding through the twin's own Bind() would update that shadow and would
+        // therefore CHANGE what the scenario measures next; this does not.)
+        GLint previousBinding = 0;
+        MGB::g_GLESFuncs.glGetIntegerv(static_cast<GLenum>(bindingQuery), &previousBinding);
+        MGB::g_GLESFuncs.glBindTexture(static_cast<GLenum>(glTarget), backendId);
+
+        out->BackendTextureId = static_cast<unsigned>(backendId);
+        static const GLenum kSwizzlePnames[4] = {GL_TEXTURE_SWIZZLE_R, GL_TEXTURE_SWIZZLE_G,
+                                                 GL_TEXTURE_SWIZZLE_B, GL_TEXTURE_SWIZZLE_A};
+        for (int channel = 0; channel < 4; ++channel) {
+            GLint value = 0;
+            MGB::g_GLESFuncs.glGetTexParameteriv(static_cast<GLenum>(glTarget),
+                                                 kSwizzlePnames[channel], &value);
+            out->Swizzle[channel] = static_cast<int>(value);
+        }
+
+        // The depth/stencil aspect mode is ES 3.1 and is INVALID_ENUM on a driver without it, so
+        // it is asked for last and its own error decides whether the answer is usable. The queue
+        // is drained first because a stale error from anywhere else would be indistinguishable
+        // from this call's - Espryt drains it the same way at every one of its own sync sites
+        // (DebugImpl::ErrorLopper), and this module's own GL errors are read from the FRONTEND
+        // state (ScenarioTest::FirstGLError), which none of this touches.
+        while (MGB::g_GLESFuncs.glGetError() != GL_NO_ERROR) {
+        }
+        GLint mode = 0;
+        MGB::g_GLESFuncs.glGetTexParameteriv(static_cast<GLenum>(glTarget),
+                                             GL_DEPTH_STENCIL_TEXTURE_MODE, &mode);
+        out->DepthStencilModeIsReadable = MGB::g_GLESFuncs.glGetError() == GL_NO_ERROR;
+        out->DepthStencilMode = static_cast<int>(mode);
+
+        MGB::g_GLESFuncs.glBindTexture(static_cast<GLenum>(glTarget),
+                                       static_cast<GLuint>(previousBinding));
+        while (MGB::g_GLESFuncs.glGetError() != GL_NO_ERROR) {
+        }
+        return true;
+    }
+
+    bool PeekEsprytHasSamplerViewForTexture(unsigned glTextureName, bool* outExists) {
+        if (outExists == nullptr) return false;
+        MobileGL::MG_State::GLState::ITextureObject* const object = FrontendTexture(glTextureName);
+        if (object == nullptr) return false;
+        // Espryt must be the backend running, or "no view" would be true of every texture on
+        // every other backend and the assertion would be vacuous where it is loudest.
+        if (EsprytTwin(glTextureName) == nullptr) return false;
+        // HandleOfSamplerViewForTexture is the monolith glue that derives the view's handle from
+        // the TEXTURE's lifetime id (D-F2: one view per ITextureObject), so this asks Espryt's
+        // own table the same way Espryt asks it - it does not consult the applier record's
+        // ViewCso, which is the client's statement about the same fact and would make one side
+        // of the seam vouch for the other.
+        const MGP::MGPipeHandle view = MGB::SamplerViewImpl::HandleOfSamplerViewForTexture(object);
+        if (MGP::MGPipeHandleIsNull(view)) {
+            *outExists = false;
+            return true;
+        }
+        *outExists = MGB::SamplerViewImpl::FindSamplerViewForHandle(view) != nullptr;
+        return true;
+    }
+
+    bool PeekPipeApplierRefusedNoConsumer(unsigned long long* outCount) {
+        if (outCount == nullptr) return false;
+        *outCount = static_cast<unsigned long long>(MGP::MGPipeApplier().RefusedNoConsumer);
+        return true;
+    }
+#else
+    bool PeekPipeTextureParamsRecord(unsigned, PipeTextureParamsRecordPeek*) { return false; }
+    bool PeekEsprytAppliedTextureParams(unsigned, unsigned, EsprytAppliedTextureParamsPeek*) {
+        return false;
+    }
+    bool PeekEsprytHasSamplerViewForTexture(unsigned, bool*) { return false; }
+    bool PeekPipeApplierRefusedNoConsumer(unsigned long long*) { return false; }
+#endif
+
+} // namespace MGITest

MobileGL/MG_IntegrationTest/Harness/PipeApplyPeek.h

@@ -0,0 +1,110 @@
+// MobileGL - MobileGL/MG_IntegrationTest/Harness/PipeApplyPeek.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
+//
+// The APPLIER's texture-parameter record, ESPRYT's applied value for the same texture, and
+// whether that texture has a sampler view yet. Three readings taken from a scenario, for gate
+// G9's WHITE-BOX half.
+//
+// WHY A WHITE-BOX HALF EXISTS AT ALL (ID-19, brief section F, gates review R1). G9's public-GL
+// cases in TextureParamsWithoutASamplerViewScenario.cpp catch "the parameter never reached the
+// driver". They CANNOT catch "the parameter reached the driver LATE", because a texture
+// parameter's only public-GL observable is a SAMPLE and the sample is itself what repairs an
+// unsynced parameter: it puts the texture on the unit list, and that walk pushes the parameters
+// for anything whose params serial moved. A backend that deferred every attachment-only
+// texture's parameters to the first sampler view would be green on all four of those cases,
+// forever, on every tree. The distinction only exists on the inside, so the reading has to be
+// taken there - while the texture is still attachment-only, before any sample.
+//
+// A SEPARATE TRANSLATION UNIT for PipeSlotPeek.h's reason, verbatim: the scenario sources
+// include the GL headers with prototypes and MobileGL's umbrella header is not meant to meet
+// them in one file. This one goes further than PipeSlotPeek and includes Espryt's own
+// Managers.h, which is exactly why it may not be anywhere near a scenario's GL headers.
+//
+// EVERY ENTRY POINT RETURNS false, TOUCHING NOTHING, WHERE IT CANNOT LOOK, and a caller that
+// gets false has learned NOTHING - "could not look" is not "was applied". Out of reach means:
+// a PULL build (there is no applier: it is `#if MOBILEGL_PIPE_PUSH`); Android, where this module
+// links the shipping libMobileGL.so built -fvisibility=hidden and no internal symbol resolves;
+// a backend that is not DirectGLES (Espryt is the subject; Magma answers the same GL question
+// through P7's own paths); and, for the record peek, a mask whose texture-resource bit is off,
+// where no record exists to find because nothing was ever emitted.
+
+#pragma once
+
+namespace MGITest {
+
+    // ---- the applier's set_texture_params record for one GL texture name ------------------
+    //
+    // ADDRESSED BY GL NAME, and the search key is MGPResourceDesc::GlNameForDiag. That field is
+    // diagnostics-only by contract - never an identity, never a memo key (MGPipeTypes.h) - and
+    // this is a diagnostic: a test harness looking for the record a named GL object produced.
+    // The alternative would be to ask the CLIENT emitter for the texture's handle, and the
+    // review is explicit that this probe must arm on package D's applier/backend state and not
+    // on the emitter markers B and C set: they are different questions, and a shared marker
+    // would re-create the shape review F-M5 was raised about.
+    struct PipeTextureParamsRecordPeek {
+        // The handle the record sits at, so a caller can print it.
+        unsigned Slot;
+        unsigned Gen;
+        // set_texture_params' own serial. 0 means the record exists (the resource was created)
+        // but NO set_texture_params has ever been applied to it - which is a different finding
+        // from "no record", and the two must not be merged.
+        unsigned long long ParamsSerial;
+        // MGPTextureParams::Swizzle[4], translated to the GL enums the application passed to
+        // glTextureParameteri (GL_ZERO / GL_ONE / GL_RED / GL_GREEN / GL_BLUE / GL_ALPHA), so
+        // the scenario compares what it set against what the record carries in ONE vocabulary
+        // and neither side has to know the other's encoding.
+        int Swizzle[4];
+        // MGPTextureParams::DepthStencilMode, translated the same way: GL_DEPTH_COMPONENT or
+        // GL_STENCIL_INDEX.
+        int DepthStencilMode;
+    };
+
+    bool PeekPipeTextureParamsRecord(unsigned glTextureName, PipeTextureParamsRecordPeek* out);
+
+    // ---- Espryt's APPLIED value for the same texture --------------------------------------
+    //
+    // Read from the DRIVER, through the twin's own ES name, because "applied" means the driver
+    // was told - the same thing package D's white-box unit probe asserts against its mocked
+    // driver (esprytobj-v2 (9)). The current binding on the ACTIVE unit is saved and restored
+    // around the query and no unit is switched, so Espryt's binding shadow still describes
+    // reality afterwards: nothing is perturbed for it to be stale about.
+    //
+    // `glTarget` is the texture's GL target (only GL_TEXTURE_2D is supported today; any other
+    // target returns false rather than guessing a binding query).
+    struct EsprytAppliedTextureParamsPeek {
+        // The driver name Espryt minted for this texture, for the caller's message.
+        unsigned BackendTextureId;
+        int Swizzle[4];
+        int DepthStencilMode;
+        // False when the driver rejected the depth/stencil query - a non-depth texture, or an ES
+        // level without GL_DEPTH_STENCIL_TEXTURE_MODE. The swizzle half is still valid.
+        bool DepthStencilModeIsReadable;
+    };
+
+    bool PeekEsprytAppliedTextureParams(unsigned glTextureName, unsigned glTarget,
+                                        EsprytAppliedTextureParamsPeek* out);
+
+    // ---- and the claim that makes the two above mean anything ------------------------------
+    //
+    // Whether Espryt holds a SAMPLER VIEW twin for this texture. This is the assertion the
+    // public-GL cases cannot make, because making it there would create the view. `*outExists`
+    // is written only on true.
+    bool PeekEsprytHasSamplerViewForTexture(unsigned glTextureName, bool* outExists);
+
+    // ---- c0f's belt, for the ObjectSubsystemControl arms -----------------------------------
+    //
+    // MGPipeApplierState::RefusedNoConsumer: the number of P4a-family entry points that were
+    // refused because no backend had registered MGPipeResourceOps. On a backend WITH a consumer
+    // it must never move; on one without (Magma, ID-39/ID-40) the client's own gate is supposed
+    // to stop the emission before the belt is reached, so it must never move there either. A
+    // non-zero delta says the gate and the belt disagreed, which is the whole point of having
+    // both. Reset by MGPipeApplierReset, so a caller reads it as a DELTA and treats a value that
+    // went DOWN as "the applier was reset, count everything since as `after`".
+    bool PeekPipeApplierRefusedNoConsumer(unsigned long long* outCount);
+
+} // namespace MGITest

MobileGL/MG_IntegrationTest/Harness/PipeSlotPeek.cpp

@@ -0,0 +1,82 @@
+// MobileGL - MobileGL/MG_IntegrationTest/Harness/PipeSlotPeek.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 "PipeSlotPeek.h"
+
+#if !defined(__ANDROID__)
+#include <MG_Pipe/MGPipe.h>
+#if MOBILEGL_PIPE_PUSH
+#include <MG_Impl/Pipe/SlotAllocator.h>
+#define MGITEST_PIPE_SLOT_PEEK_LIVE 1
+#endif
+#endif
+
+namespace MGITest {
+
+#if defined(MGITEST_PIPE_SLOT_PEEK_LIVE)
+    namespace {
+        // One arm per member, and NO `default:` on purpose: adding a PipeSlotKind without
+        // deciding which MGPipeKind it names is a compiler warning here (-Wswitch) rather than
+        // a row that silently counts VertexElementsCso and reports "did not leak" about a kind
+        // it never looked at. The trailing return is the unreachable one the compiler needs.
+        MobileGL::MG_Pipe::MGPipeKind Translate(PipeSlotKind kind) {
+            switch (kind) {
+                case PipeSlotKind::Buffer: return MobileGL::MG_Pipe::MGPipeKind::Buffer;
+                case PipeSlotKind::VertexElementsCso:
+                    return MobileGL::MG_Pipe::MGPipeKind::VertexElementsCso;
+                case PipeSlotKind::Texture: return MobileGL::MG_Pipe::MGPipeKind::Texture;
+                case PipeSlotKind::Renderbuffer: return MobileGL::MG_Pipe::MGPipeKind::Renderbuffer;
+                case PipeSlotKind::Framebuffer: return MobileGL::MG_Pipe::MGPipeKind::Framebuffer;
+                case PipeSlotKind::SamplerCso: return MobileGL::MG_Pipe::MGPipeKind::SamplerCso;
+                case PipeSlotKind::SamplerViewCso:
+                    return MobileGL::MG_Pipe::MGPipeKind::SamplerViewCso;
+                case PipeSlotKind::ShaderCso: return MobileGL::MG_Pipe::MGPipeKind::ShaderCso;
+            }
+            return MobileGL::MG_Pipe::MGPipeKind::None;
+        }
+    } // namespace
+
+    bool PeekPipeSlotLiveCount(PipeSlotKind kind, unsigned* outLive) {
+        if (outLive == nullptr) return false;
+        *outLive = static_cast<unsigned>(MobileGL::MG_Pipe::MGPipeSlots().LiveCount(Translate(kind)));
+        return true;
+    }
+
+    bool PeekPipeSlotHighWater(PipeSlotKind kind, unsigned* outHighWater) {
+        if (outHighWater == nullptr) return false;
+        // The ORDINARY space only, for every kind including ShaderCso (contract-v2.md 4.3).
+        *outHighWater = static_cast<unsigned>(MobileGL::MG_Pipe::MGPipeSlots().HighWater(Translate(kind)));
+        return true;
+    }
+
+    bool PeekPipeCompositeSlotLiveCount(unsigned* outLive) {
+        if (outLive == nullptr) return false;
+        *outLive = static_cast<unsigned>(MobileGL::MG_Pipe::MGPipeSlots().CompositeLiveCount());
+        return true;
+    }
+
+    bool PeekPipeCompositeSlotHighWater(unsigned* outHighWater) {
+        if (outHighWater == nullptr) return false;
+        *outHighWater = static_cast<unsigned>(MobileGL::MG_Pipe::MGPipeSlots().CompositeHighWater());
+        return true;
+    }
+
+    bool PeekPipeCompositeSlotBandBase(unsigned* outBandBase) {
+        if (outBandBase == nullptr) return false;
+        *outBandBase = static_cast<unsigned>(MobileGL::MG_Pipe::kMGPipeShaderCsoCompositeSlotBase);
+        return true;
+    }
+#else
+    bool PeekPipeSlotLiveCount(PipeSlotKind, unsigned*) { return false; }
+    bool PeekPipeSlotHighWater(PipeSlotKind, unsigned*) { return false; }
+    bool PeekPipeCompositeSlotLiveCount(unsigned*) { return false; }
+    bool PeekPipeCompositeSlotHighWater(unsigned*) { return false; }
+    bool PeekPipeCompositeSlotBandBase(unsigned*) { return false; }
+#endif
+
+} // namespace MGITest

MobileGL/MG_IntegrationTest/Harness/PipeSlotPeek.h

@@ -0,0 +1,101 @@
+// MobileGL - MobileGL/MG_IntegrationTest/Harness/PipeSlotPeek.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 CLIENT slot allocator's occupancy, read from a scenario.
+//
+// It exists for one assertion, P3a's C-1: a frontend object that dies must return its
+// MGPipeHandle slot WHATEVER BACKEND IS RUNNING. That question has no answer in the GL API -
+// the leak it rules out is entirely inside the library, and it is invisible in pixels, in GL
+// names and in `glGetError` - so the only honest observable is the allocator's own live count
+// and high-water mark. Reading them is what makes the case fail on the backend it actually
+// failed on (DirectVulkan, which installs no StateObjectDeathOps) rather than only on the one
+// where a backend-owned free happened to exist.
+//
+// A separate translation unit for BackendCapsPeek.h's reason, verbatim: 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 {
+
+    // Which client-side object kind to ask about. Mirrors MG_Pipe::MGPipeKind for exactly the
+    // kinds a scenario has a reason to count, so that the enum does not travel through this
+    // header and the GL headers together.
+    enum class PipeSlotKind {
+        Buffer,
+        VertexElementsCso,
+        // P4a's six (G8b). Every one of them is a kind the CLIENT mints and the client alone
+        // frees (BRIEF-P4A.md D-I1: one death helper per kind, called from the frontend
+        // object's own destructor, whatever backend is running), so every one of them can leak
+        // the P3a C-1 way - and the leak is invisible in pixels, in GL names and in
+        // glGetError, exactly as the VertexElementsCso one was.
+        Texture,
+        Renderbuffer,
+        // Framebuffer has a HANDLE but no wire lifetime (D-I2): no create_*, no destroy row in
+        // the catalogue, and its death helper does the notice and the free and emits nothing.
+        // That makes the allocator the ONLY observable of its lifetime, so this row matters
+        // more here than the others rather than less.
+        Framebuffer,
+        SamplerCso,
+        SamplerViewCso,
+        // ShaderCso covers BOTH the ordinary program slots and the program-pipeline COMPOSITES
+        // minted out of the reserved high band (MGPipeHandles.h:86-107, D-H7). One kind, because
+        // that is what the allocator has: the band is a second dense table inside the same kind
+        // and LiveCount counts both.
+        //
+        // THE TWO SPACES' HIGH-WATER MARKS ARE NOT ONE NUMBER, and the correction matters here
+        // more than anywhere else. c0b split them (contract-v2.md 4.3): HighWater(ShaderCso) is
+        // now the ORDINARY space only and the band's own mark is CompositeHighWater(), because
+        // a merged mark is pinned at ~983k from the first composite mint onward and every "the
+        // high-water mark did not move over N churn rounds" assertion about ordinary programs
+        // would be vacuously true for the rest of the process. The composite's leak case is a
+        // separate CASE and reads the BAND'S OWN counters below (PeekPipeCompositeSlot*) - a
+        // composite's slot has TWO independent release paths (the pipeline cache's LRU eviction
+        // and the composite ProgramObject's destructor), and a slot that never comes back to
+        // the band moves neither of the ordinary numbers.
+        ShaderCso,
+    };
+
+    // Live slots of this kind right now, and one past the highest slot ever handed out.
+    // Both return false, touching nothing, where the allocator is out of reach: in a PULL
+    // build there is no allocator at all (it is `#if MOBILEGL_PIPE_PUSH`), and on Android this
+    // module links the shipping libMobileGL.so built -fvisibility=hidden, so no internal symbol
+    // resolves. A caller that gets false must SKIP rather than pass - "could not look" is not
+    // "did not leak".
+    bool PeekPipeSlotLiveCount(PipeSlotKind kind, unsigned* outLive);
+    bool PeekPipeSlotHighWater(PipeSlotKind kind, unsigned* outHighWater);
+
+    // The ShaderCso COMPOSITE BAND's own three numbers, the seventh..ninth members
+    // contract-v2.md 4.3 asks this header for. There is no `kind` argument because the band is
+    // ShaderCso's alone - AllocateComposite is the one door into it and no other kind has one.
+    // All three return false on the same terms as the two above, and a caller that gets false
+    // must SKIP.
+    //
+    //   PeekPipeCompositeSlotLiveCount   = MGPipeSlotAllocator::CompositeLiveCount(), the band's
+    //                                      share of LiveCount(ShaderCso).
+    //   PeekPipeCompositeSlotHighWater   = CompositeHighWater() VERBATIM, i.e. one past the
+    //                                      highest band slot ever handed out. It is an ABSOLUTE
+    //                                      slot number and therefore starts at the band's base,
+    //                                      not at zero - "no composite was ever minted" reads as
+    //                                      `high water == band base`, which is what the third
+    //                                      member is for. It is not returned base-relative
+    //                                      because a peek whose name says HighWater and whose
+    //                                      value is a delta is exactly the kind of quietly
+    //                                      redefined counter this member exists to correct.
+    //   PeekPipeCompositeSlotBandBase    = kMGPipeShaderCsoCompositeSlotBase, the floor the
+    //                                      other two are read against. A constant, but it
+    //                                      reaches a scenario only through this header: the
+    //                                      MG_Pipe headers and the GL headers are not meant to
+    //                                      meet in one translation unit, which is why this
+    //                                      harness exists at all.
+    bool PeekPipeCompositeSlotLiveCount(unsigned* outLive);
+    bool PeekPipeCompositeSlotHighWater(unsigned* outHighWater);
+    bool PeekPipeCompositeSlotBandBase(unsigned* outBandBase);
+
+} // namespace MGITest

MobileGL/MG_IntegrationTest/Harness/PipeStatsWindow.h

@@ -0,0 +1,94 @@
+// MobileGL - MobileGL/MG_IntegrationTest/Harness/PipeStatsWindow.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
+//
+// Reading ONE PipeStats summary window out of the library's own log, for the scenarios whose
+// claim is about a counter rather than about pixels.
+//
+// WHY THROUGH A LOG FILE AT ALL. MG_Util::PipeStats is internal to the library and this module
+// cannot link against it (ScenarioFixture.h has the long version: on Android this binary links
+// the SHIPPING libMobileGL.so, built -fvisibility=hidden). The library's `MGPipe stats:` line is
+// the only channel, so a lane that wants to read a counter sets MOBILEGL_PIPE_STATS=1,
+// MOBILEGL_PIPE_STATS_PERIOD=1 - one line per eglSwapBuffers - and a MOBILEGL_LOG_FILE_PATH of
+// its OWN.
+//
+// THE LOG PATH HAS TO BE PRIVATE TO ONE CTEST ENTRY, and that is not a style rule: the library
+// opens it fopen(path, "w"), so every process launched in a lane TRUNCATES it. Two entries of one
+// lane reading the same path race under `ctest -j`, and the shape of the failure is an empty read
+// that looks exactly like "the counter was never emitted". So a case that reads a window gets a
+// ctest entry whose TEST_FILTER selects that case alone, with a log path nothing else writes -
+// the rule PipeVerifyArmingScenario and CsoContentAddressingScenario already follow.
+//
+// THE WINDOW IS "SINCE THE PREVIOUS LINE" (PipeStats::FormatWindowLine), so the caller closes the
+// setup window with a swap, runs the workload, swaps again, and reads the LAST line - which then
+// covers the workload and nothing else.
+
+#pragma once
+
+#include <cstdlib>
+#include <cstring>
+#include <fstream>
+#include <iterator>
+#include <string>
+
+namespace MGITest::PipeStatsWindow {
+
+    // The lane's private log path, or empty when the lane configured none.
+    inline std::string LibraryLogPath() {
+        const char* path = std::getenv("MOBILEGL_LOG_FILE_PATH");
+        return (path != nullptr && *path != '\0') ? std::string(path) : std::string();
+    }
+
+    inline std::string ReadWholeFile(const std::string& path) {
+        if (path.empty()) return {};
+        std::ifstream file(path, std::ios::binary);
+        if (!file.good()) return {};
+        return std::string((std::istreambuf_iterator<char>(file)), std::istreambuf_iterator<char>());
+    }
+
+    // The last summary line in the log, verbatim. `found` is false when the library never emitted
+    // one, which is a different failure from "the counter read zero" and has to be reported as
+    // one: it means the stats channel never reached the process, not that the workload did
+    // nothing.
+    struct Window {
+        bool found = false;
+        std::string line;
+    };
+
+    inline Window Last(const std::string& log) {
+        Window window;
+        const std::string marker = "MGPipe stats:";
+        const std::size_t at = log.rfind(marker);
+        if (at == std::string::npos) return window;
+        const std::size_t end = log.find('\n', at);
+        window.line = log.substr(at, end == std::string::npos ? std::string::npos : end - at);
+        window.found = true;
+        return window;
+    }
+
+    inline Window LastFromLaneLog() { return Last(ReadWholeFile(LibraryLogPath())); }
+
+    // One counter out of that line, by its short name ("mpr", "draws", "csom"), or -1 when the
+    // line does not carry it. The search includes the SEPARATOR before the name and the `=` after
+    // it, so "draws" cannot match "draws/f=" and "mpr" cannot match a longer name ending in it -
+    // a substring match here would read a neighbouring counter's value and report it as this
+    // one's, which is the one way a counter assertion can be wrong without ever failing.
+    inline long long CounterOrAbsent(const Window& window, const char* shortName) {
+        if (!window.found) return -1;
+        // A counter is preceded either by a space (` mpr=`, ` draws=`) or by its bracket's
+        // opening (`cso[csom=`, `bytes/f[stage-buffer=`); nothing in the line is preceded by
+        // anything else.
+        for (const char* prefix : {" ", "["}) {
+            const std::string key = std::string(prefix) + shortName + "=";
+            const std::size_t at = window.line.find(key);
+            if (at == std::string::npos) continue;
+            return std::strtoll(window.line.c_str() + at + key.size(), nullptr, 10);
+        }
+        return -1;
+    }
+
+} // namespace MGITest::PipeStatsWindow

MobileGL/MG_IntegrationTest/Scenarios/CsoContentAddressingScenario.cpp

@@ -0,0 +1,329 @@
+// MobileGL - MobileGL/MG_IntegrationTest/Scenarios/CsoContentAddressingScenario.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 - THE CSO CONTENT-ADDRESSING NEGATIVE CONTROL (gate G12).
+//
+// P2's render-state CSO is content-addressed: the client hashes the 396 pipeline bytes, probes a
+// 64-entry cache, memcmps a hash hit and reuses the handle. The whole design is measured against
+// a knob that turns that off - kMGPipeBehaviourNoCsoContentAddressing, bit 63 of the runtime
+// MOBILEGL_PIPE_PUSH bitmask - so that "push is slower" can be told apart from "the CSO design is
+// slower" (P2 brief D.4.5). A measurement knob has one characteristic failure mode: it stops
+// steering anything and every later number is quietly taken against a switch that does nothing.
+// This file is the entry that cannot let that happen.
+//
+// WHAT IT ASSERTS, per arm, and why those are the right shapes:
+//
+//   content-addressed (MOBILEGL_PIPE_PUSH=0x7f)
+//       A Blaze3D blend toggle - enable / draw / disable / draw, N times, which is the workload
+//       the CsoCache exists for (ARCHITECTURE.md 5.1: the push happens at validate rather than in
+//       the setter precisely because Blaze3D brackets every batch this way) - visits exactly TWO
+//       distinct pipeline subsets. So the mint count must stay small and BOUNDED while the bind
+//       count grows with the draws: csom << csob.
+//
+//   no content addressing (MOBILEGL_PIPE_PUSH=0x800000000000007f)
+//       Every pipeline-version change mints a fresh CSO and the map is never probed, so mint and
+//       bind must move together: csom == csob. This is the assertion a dead switch fails - with
+//       the bit ignored, this arm would report csom << csob just like the other one.
+//
+//   both arms
+//       The PIXELS must not move. The quad is drawn with alpha 1.0 through
+//       GL_SRC_ALPHA / GL_ONE_MINUS_SRC_ALPHA, so the blended and unblended draws produce the
+//       same colour by construction and the readback is the same image in both arms and after
+//       every toggle. "The counters moved and the picture did not" is the whole claim.
+//
+// HOW THE COUNTERS ARE READ. MG_Util::PipeStats is internal to the library and this module cannot
+// link against it (ScenarioFixture.h explains why: on Android this binary links the SHIPPING
+// libMobileGL.so, built -fvisibility=hidden). The library's own summary line is the only channel,
+// so each lane sets MOBILEGL_PIPE_STATS=1, MOBILEGL_PIPE_STATS_PERIOD=1 - one line per
+// eglSwapBuffers - and a MOBILEGL_LOG_FILE_PATH of its OWN. The log path has to be private: the
+// library opens it fopen(path, "w"), so every process in a lane truncates it, and a whole-file
+// read in a shared lane races a neighbour's bring-up. That is the same rule, and the same
+// remedy, as PipeVerifyArmingScenario's arming lane.
+//
+// The window a summary line reports is "since the previous line" (PipeStats::FormatWindowLine), so
+// the workload runs inside ONE frame: a swap before it closes the setup window, and the swap after
+// it emits a line whose csom / csob cover the toggle loop and nothing else.
+//
+// WHY IT CAN SKIP. The counters are minted by the client-side tracker (P2 package B), and this
+// file is written against the P2 contract commit, before that package lands. Until the tracker
+// exists there is no CSO to mint, csom is structurally 0 and an assertion about its ratio to csob
+// would be a statement about nothing. The build answers the question rather than a hand-maintained
+// list: MG_IntegrationTest/CMakeLists.txt greps every source under MG_Impl/Pipe/ for the two
+// counters' names and passes the answer in as MGITEST_PIPE_TRACKER_PRESENT, with a
+// CONFIGURE_DEPENDS on that directory and on each file it finds so the answer cannot go stale.
+// It is a CONTENT probe, not a filename probe, precisely so that the owning package keeps control
+// of its own file layout - it implements the tracker and the cache header-only today, and a glob
+// for `Tracker.cpp` would have kept this control skipping forever after that package landed, with
+// a reason that had become false. When an emitter lands the arms arm themselves; until then the
+// entries are registered, visible and SKIPPED with the reason - never absent, and never green for
+// having asserted nothing.
+
+#include <cstdlib>
+#include <cstring>
+#include <filesystem>
+#include <fstream>
+#include <iterator>
+#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 {
+
+        // Set by the two CsoContentAddressing. ctest entries and by nothing else; a harness
+        // marker, never read by the library. Its absence means an ambient entry, where neither
+        // the stats channel nor a private log path is configured.
+        constexpr const char* kLaneMarker = "MGITEST_CSO_LANE";
+        constexpr const char* kLaneContentAddressed = "content-addressed";
+        constexpr const char* kLaneNoContentAddressing = "no-content-addressing";
+
+        // Toggle pairs per frame. 8 is small enough to keep the frame cheap and large enough that
+        // "mints stay bounded" and "mints track binds" are different numbers by a wide margin.
+        constexpr int kTogglePairs = 8;
+        constexpr int kDrawsPerFrame = kTogglePairs * 2;
+        // The blend toggle visits two distinct pipeline subsets, so two CSOs. The bound is
+        // deliberately a little looser than 2: a future chunk-table change could legitimately
+        // split one of them, and the claim being pinned here is "bounded, not per-draw".
+        constexpr long long kMaxDistinctCsos = 4;
+
+        constexpr const char* kVS = R"(#version 330 core
+in vec2 aPos;
+void main() { gl_Position = vec4(aPos, 0.0, 1.0); }
+)";
+
+        constexpr const char* kFS = R"(#version 330 core
+out vec4 oColor;
+void main() { oColor = vec4(0.0, 1.0, 0.0, 1.0); }
+)";
+
+        constexpr int kInset = 2;
+
+        bool BuildMarkerIsSet(const char* name) {
+            const char* value = std::getenv(name);
+            return value != nullptr && value[0] == '1' && value[1] == '\0';
+        }
+
+        std::string LaneName() {
+            const char* lane = std::getenv(kLaneMarker);
+            return lane != nullptr ? std::string(lane) : std::string();
+        }
+
+        std::string LibraryLogPath() {
+            const char* path = std::getenv("MOBILEGL_LOG_FILE_PATH");
+            return (path != nullptr && *path != '\0') ? std::string(path) : std::string();
+        }
+
+        std::string ReadWholeFile(const std::string& path) {
+            if (path.empty()) return {};
+            std::ifstream file(path, std::ios::binary);
+            if (!file.good()) return {};
+            return std::string((std::istreambuf_iterator<char>(file)), std::istreambuf_iterator<char>());
+        }
+
+        // One window's CSO counters, as the library printed them.
+        struct CsoWindow {
+            bool found = false;
+            long long mints = -1;
+            long long binds = -1;
+            std::string line;
+        };
+
+        // Parses `... cso[csom=<N> csob=<M>] ...` out of the LAST "MGPipe stats:" line in the log.
+        // The last line, because the window a line reports is "since the previous line" and the
+        // caller closes the setup window with a swap before the workload.
+        CsoWindow LastCsoWindow(const std::string& log) {
+            CsoWindow window;
+            const std::string marker = "MGPipe stats:";
+            std::size_t at = log.rfind(marker);
+            if (at == std::string::npos) return window;
+            const std::size_t end = log.find('\n', at);
+            window.line = log.substr(at, end == std::string::npos ? std::string::npos : end - at);
+
+            const std::string mintKey = "csom=";
+            const std::string bindKey = "csob=";
+            const std::size_t mintAt = window.line.find(mintKey);
+            const std::size_t bindAt = window.line.find(bindKey);
+            if (mintAt == std::string::npos || bindAt == std::string::npos) return window;
+            window.mints = std::strtoll(window.line.c_str() + mintAt + mintKey.size(), nullptr, 10);
+            window.binds = std::strtoll(window.line.c_str() + bindAt + bindKey.size(), nullptr, 10);
+            window.found = true;
+            return window;
+        }
+
+        class CsoContentAddressingScenario : public ScenarioTest {
+        protected:
+            void SetUp() override {
+                ScenarioTest::SetUp();
+                if (!Ready()) return;
+                m_lane = LaneName();
+                std::string error;
+                m_program = CompileProgram(kVS, kFS, &error);
+                ASSERT_NE(m_program, 0u) << error;
+
+                const float quad[12] = {-1.0f, -1.0f, 1.0f, -1.0f, 1.0f, 1.0f,
+                                        -1.0f, -1.0f, 1.0f, 1.0f,  -1.0f, 1.0f};
+                glGenVertexArrays(1, &m_vao);
+                glBindVertexArray(m_vao);
+                glGenBuffers(1, &m_vbo);
+                glBindBuffer(GL_ARRAY_BUFFER, m_vbo);
+                glBufferData(GL_ARRAY_BUFFER, sizeof(quad), quad, GL_STATIC_DRAW);
+                glEnableVertexAttribArray(0);
+                glVertexAttribPointer(0, 2, GL_FLOAT, GL_FALSE, 2 * sizeof(float), nullptr);
+                ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR)) << "scene setup left a GL error behind";
+                RecordProperty("lane", m_lane.empty() ? "ambient" : m_lane.c_str());
+            }
+
+            void TearDown() override {
+                if (!Ready()) return;
+                glBindVertexArray(0);
+                glBindBuffer(GL_ARRAY_BUFFER, 0);
+                if (m_vbo != 0) glDeleteBuffers(1, &m_vbo);
+                if (m_vao != 0) glDeleteVertexArrays(1, &m_vao);
+                if (m_program != 0) glDeleteProgram(m_program);
+            }
+
+            // GTEST_SKIP() returns from the function it is written in, so this cannot report
+            // through a return value; every caller pairs it with `if (IsSkipped()) return;`.
+            void SkipUnlessTheLaneIsAssertableHere() {
+                if (m_lane.empty()) {
+                    GTEST_SKIP() << "runs only in its own lane: the two CsoContentAddressing. ctest entries set "
+                                    "MGITEST_CSO_LANE together with the MOBILEGL_PIPE_PUSH bitmask, "
+                                    "MOBILEGL_PIPE_STATS=1, MOBILEGL_PIPE_STATS_PERIOD=1 and a private "
+                                    "MOBILEGL_LOG_FILE_PATH. None of that is configured in the ambient "
+                                    "entries, and the ambient log is shared, so a read here would race.";
+                    return;
+                }
+                if (!BuildMarkerIsSet("MGITEST_PIPE_PUSH_BUILD")) {
+                    GTEST_SKIP() << "this library was built without MOBILEGL_PIPE_PUSH, so there is no "
+                                    "render-state CSO to mint, no cso[] bracket in the summary line and "
+                                    "nothing for the content-addressing bit to steer. The entry is "
+                                    "registered here anyway so that `ctest -L integration-gpu` names the "
+                                    "same tests in the pull build and the push build (gate G2).";
+                    return;
+                }
+                if (!BuildMarkerIsSet("MGITEST_PIPE_TRACKER_PRESENT")) {
+                    GTEST_SKIP() << "the CSO counters have no emitter in this build: no source under "
+                                    "MobileGL/MG_Impl/Pipe/ names RenderStateCsoMints or "
+                                    "RenderStateCsoBinds, so nothing mints or binds a render-state CSO "
+                                    "and csom / csob are structurally zero. P2 package B owns the tracker "
+                                    "and the CSO cache; this entry arms itself when they land, whatever "
+                                    "files that package chooses to put them in.";
+                    return;
+                }
+                if (LibraryLogPath().empty()) {
+                    GTEST_SKIP() << "the lane configured no MOBILEGL_LOG_FILE_PATH, and the library's summary "
+                                    "line is the only channel this module has for reading PipeStats";
+                    return;
+                }
+            }
+
+            // enable / draw / disable / draw, kTogglePairs times, entirely inside one frame.
+            // Returns the readback taken at the end of that frame, before the swap.
+            Image RunBlendToggleFrame() {
+                BindDefaultFramebuffer();
+                ClearTo(0.0f, 0.0f, 0.0f, 1.0f);
+                glUseProgram(m_program);
+                glBindVertexArray(m_vao);
+                glBlendFuncSeparate(GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA, GL_ONE, GL_ZERO);
+                for (int i = 0; i < kTogglePairs; ++i) {
+                    glEnable(GL_BLEND);
+                    glDrawArrays(GL_TRIANGLES, 0, 6);
+                    glDisable(GL_BLEND);
+                    glDrawArrays(GL_TRIANGLES, 0, 6);
+                }
+                const Image image = ReadPixels(Gl().Width(), Gl().Height());
+                Gl().EndFrame();
+                return image;
+            }
+
+            std::string m_lane;
+            GLuint m_program = 0;
+            GLuint m_vao = 0;
+            GLuint m_vbo = 0;
+        };
+
+        // ONE case per lane, and that is a hard constraint rather than a style choice.
+        //
+        // This case READS the library log, and the log is a per-LANE resource: the library opens it
+        // fopen(path, "w"), so every process in a lane truncates it. A second case in this lane would
+        // therefore race this one under `ctest -j`, and the shape of the failure is a silent, empty
+        // read that looks exactly like "the counters were never emitted". Splitting the plumbing
+        // assertion into its own case would have bought a clearer failure message and paid for it
+        // with a flake in the thing the message is about. The plumbing is asserted first, with its
+        // own message, inside this one process instead.
+        TEST_F(CsoContentAddressingScenario, TheBlendToggleMintsBoundedlyWithContentAddressingAndPerBindWithout) {
+            if (!Ready()) return;
+            SkipUnlessTheLaneIsAssertableHere();
+            if (IsSkipped()) return;
+
+            Gl().EndFrame(); // close the setup window
+            const Image first = RunBlendToggleFrame();
+            const CsoWindow window = LastCsoWindow(ReadWholeFile(LibraryLogPath()));
+            // The plumbing first, with its own message, so a counter-ratio failure below can never
+            // be confused with "the lane never turned the stats channel on".
+            ASSERT_TRUE(window.found)
+                << "no 'MGPipe stats:' line carrying cso[csom= csob=] in " << LibraryLogPath()
+                << ". This IS a push build (the lane checked MGITEST_PIPE_PUSH_BUILD before getting "
+                   "here) and the cso[] bracket is unconditional inside that #if, so it cannot be "
+                   "missing for a build reason: either MOBILEGL_PIPE_STATS / "
+                   "MOBILEGL_PIPE_STATS_PERIOD did not reach the process, or no summary line was "
+                   "emitted at all because nothing reached PipeStats::OnPresent.";
+            RecordProperty("cso_line", window.line.c_str());
+
+            // Every draw in the frame changed the pipeline subset, so every draw is a bind. This
+            // is the denominator both arms are read against; without it, "csom == csob" would also
+            // be satisfied by a frame in which neither happened at all.
+            ASSERT_GE(window.binds, static_cast<long long>(kDrawsPerFrame))
+                << "the toggle frame issued " << kDrawsPerFrame
+                << " draws whose pipeline subset alternates, so it must have issued at least that many "
+                   "render-state binds. It reported: "
+                << window.line;
+
+            if (m_lane == kLaneContentAddressed) {
+                EXPECT_LE(window.mints, kMaxDistinctCsos)
+                    << "with content addressing on, enable/draw/disable/draw x " << kTogglePairs
+                    << " visits two distinct pipeline subsets and must mint a bounded number of CSOs, then "
+                       "reuse them. It reported: "
+                    << window.line;
+                EXPECT_LT(window.mints, window.binds)
+                    << "with content addressing on the cache must be answering binds it did not mint. "
+                    << window.line;
+            } else if (m_lane == kLaneNoContentAddressing) {
+                EXPECT_EQ(window.mints, window.binds)
+                    << "kMGPipeBehaviourNoCsoContentAddressing (bit 63 of MOBILEGL_PIPE_PUSH) must make every "
+                       "bind mint a fresh CSO - the map is never probed and no handle is ever reused. Equal "
+                       "counters are the only reading that proves the bit STEERED anything: if it were "
+                       "ignored, this arm would report the same bounded mint count as the other one. It "
+                       "reported: "
+                    << window.line;
+            } else {
+                FAIL() << "unknown " << kLaneMarker << " value '" << m_lane << "'";
+            }
+
+            // ... and the picture is the same in both arms and after every toggle. The quad is
+            // opaque, so the blended and unblended draws agree by construction.
+            EXPECT_TRUE(RegionIsMostly(first, kInset, first.Width() - kInset, kInset, first.Height() - kInset,
+                                       "green", 0.0, "the blend-toggle frame [" + m_lane + "]"));
+            const Image second = RunBlendToggleFrame();
+            EXPECT_TRUE(second == first)
+                << "the second toggle frame does not match the first: " << second.ByteDiffCount(first)
+                << " bytes differ. The CSO path must not change what is drawn.";
+        }
+
+    } // namespace
+} // namespace MGITest

MobileGL/MG_IntegrationTest/Scenarios/LargeArenaAdoptionScenario.cpp

@@ -25,13 +25,31 @@
 //   * GetBufferSubData reads back the latest CPU write - the shadow IS the map;
 //   * a compute-shader write through an SSBO binding of the same arena is read
 //     back - the GPU-written path for adopted stores (glFinish + direct read).
+//
+// P3a (gate G10, G12) adds a fourth case and two more lanes, and neither of them
+// changes what the three above assert:
+//
+//   * AnAdoptionCostsExactlyOneMapPersistentRoundtrip counts the acquisition.
+//     ARCHITECTURE.md:474 prices the adopted store at one round trip per STORAGE
+//     DEFINITION; `map-persistent-roundtrips` counts every map_persistent
+//     emission, mint or decline (D-B2), so one definition plus a frame of draws
+//     must publish exactly one. It reads the library's summary line, so it needs
+//     a lane with the stats channel and a private log path, and it SKIPS - with
+//     the reason - anywhere else and on any tree that does not emit the counter.
+//   * the three original cases are registered TWICE MORE, with P3a's resource and
+//     vertex-input subsystem bits set and cleared, because this file is where an
+//     adopted store's whole life is exercised: definition, in-flight SubData,
+//     readback and a GPU write. If the handle path and the legacy BufferBackendOps
+//     path disagree about any of it, one of the two arms goes red here.
 
 #include <array>
+#include <cstdlib>
 #include <cstring>
 #include <string>
 #include <vector>
 
 #include "../Harness/HeadlessGL.h"
+#include "../Harness/PipeStatsWindow.h"
 #include "../Harness/ScenarioFixture.h"
 
 #ifdef GLAPI
@@ -73,6 +91,19 @@ layout(std430, binding = 0) buffer Arena { uint word; };
 void main() { word = 0xC0FFEEu; }
 )";
 
+        // Set by the MapPersistentRoundtrips. ctest entry and by nothing else; a harness marker,
+        // never read by the library.
+        constexpr const char* kLaneMarker = "MGITEST_MPR_LANE";
+        // Draws issued against the arena inside the counted window. One definition, many draws:
+        // "one per definition" (1) and "one per draw" (kDrawsInTheWindow) have to be different
+        // numbers or the assertion cannot tell them apart.
+        constexpr int kDrawsInTheWindow = 5;
+
+        bool BuildMarkerIsSet(const char* name) {
+            const char* value = std::getenv(name);
+            return value != nullptr && value[0] == '1' && value[1] == '\0';
+        }
+
         struct Vertex {
             float x, y;
             float r, g, b;
@@ -102,6 +133,12 @@ void main() { word = 0xC0FFEEu; }
                 // The NULL-data definition is the adoption point (and Minecraft's
                 // arena-creation idiom).
                 glBufferData(GL_ARRAY_BUFFER, kArenaBytes, nullptr, GL_DYNAMIC_DRAW);
+                ConfigureVertexArray(m_vao);
+            }
+
+            void ConfigureVertexArray(GLuint vao) {
+                glBindVertexArray(vao);
+                glBindBuffer(GL_ARRAY_BUFFER, m_arena);
                 glVertexAttribPointer(0, 2, GL_FLOAT, GL_FALSE, sizeof(Vertex),
                                       reinterpret_cast<void*>(kVertexOffset));
                 glVertexAttribPointer(1, 3, GL_FLOAT, GL_FALSE, sizeof(Vertex),
@@ -173,15 +210,49 @@ void main() { word = 0xC0FFEEu; }
                                 GLsizeiptr(vertices.size() * sizeof(Vertex)), vertices.data());
             }
 
-            void DrawQuad() {
+            void DrawQuad(GLuint vao = 0) {
                 glViewport(0, 0, Gl().Width(), Gl().Height());
                 glClearColor(0.f, 0.f, 0.f, 1.f);
                 glClear(GL_COLOR_BUFFER_BIT);
                 glUseProgram(m_program);
-                glBindVertexArray(m_vao);
+                glBindVertexArray(vao != 0 ? vao : m_vao);
                 glDrawArrays(GL_TRIANGLES, 0, 6);
             }
 
+            // GTEST_SKIP() returns from the function it is written in, so this cannot report
+            // through a return value; the caller pairs it with `if (IsSkipped()) return;`.
+            void SkipUnlessTheRoundtripCounterIsReadableHere() {
+                if (std::getenv(kLaneMarker) == nullptr) {
+                    GTEST_SKIP() << "runs only in its own lane: the MapPersistentRoundtrips. ctest entry "
+                                    "sets MGITEST_MPR_LANE together with MOBILEGL_PIPE_PUSH's P3a mask, "
+                                    "MOBILEGL_PIPE_STATS=1, MOBILEGL_PIPE_STATS_PERIOD=1 and a private "
+                                    "MOBILEGL_LOG_FILE_PATH. The ambient entries and the two subsystem "
+                                    "arms configure none of that, and their log is shared - a read there "
+                                    "would race a neighbour's bring-up.";
+                    return;
+                }
+                if (!BuildMarkerIsSet("MGITEST_PIPE_PUSH_BUILD")) {
+                    GTEST_SKIP() << "this library was built without MOBILEGL_PIPE_PUSH, so "
+                                    "CallClass::MapPersistentRoundtrips does not exist and the summary "
+                                    "line carries no mpr=. The entry stays registered so that "
+                                    "`ctest -L integration-gpu` names the same tests in both builds (G2).";
+                    return;
+                }
+                if (!BuildMarkerIsSet("MGITEST_PIPE_RESOURCE_EMITTER_PRESENT")) {
+                    GTEST_SKIP() << "subsystem not implemented on this tree: no source under "
+                                    "MobileGL/MG_Impl/Pipe/ names MapPersistentRoundtrips, so nothing "
+                                    "emits map_persistent and mpr= is structurally zero. P3a package B "
+                                    "owns that emitter; this entry arms itself when it lands.";
+                    return;
+                }
+                if (PipeStatsWindow::LibraryLogPath().empty()) {
+                    GTEST_SKIP() << "the lane configured no MOBILEGL_LOG_FILE_PATH, and the library's "
+                                    "summary line is the only channel this module has for reading "
+                                    "PipeStats";
+                    return;
+                }
+            }
+
             std::array<unsigned char, 4> CenterPixel() {
                 std::array<unsigned char, 4> px = {0, 0, 0, 0};
                 glReadPixels(Gl().Width() / 2, Gl().Height() / 2, 1, 1, GL_RGBA, GL_UNSIGNED_BYTE,
@@ -222,6 +293,97 @@ void main() { word = 0xC0FFEEu; }
         EXPECT_LT(px[0], 50) << "the draw still shows the previous frame's bytes";
     }
 
+    // Respecifying a frontend buffer preserves its VAO attachments even when the
+    // backend replaces the adopted store's GL name. Keep every attribute binding
+    // unchanged so a stale backend VAO cannot be repaired by a frontend rebind.
+    TEST_F(LargeArenaAdoptionScenario, RespecifiedVertexArenaKeepsVaoBindings) {
+        if (!Ready() || IsSkipped()) return;
+
+        UploadQuad(1.f, 0.f, 0.f);
+        DrawQuad();
+        ASSERT_GT(CenterPixel()[0], 200);
+        ASSERT_EQ(FirstGLError(), 0u);
+
+        GLuint otherVao = 0;
+        glGenVertexArrays(1, &otherVao);
+        ConfigureVertexArray(otherVao);
+        DrawQuad(otherVao);
+        EXPECT_GT(CenterPixel()[0], 200);
+        EXPECT_EQ(FirstGLError(), 0u);
+
+        constexpr std::array<GLsizeiptr, 3> sizes = {
+            kArenaBytes, kArenaBytes + 4096, kArenaBytes - 4096,
+        };
+        constexpr std::array<std::array<float, 3>, 3> colors = {{
+            {0.f, 1.f, 0.f}, {0.f, 0.f, 1.f}, {1.f, 0.f, 0.f},
+        }};
+        for (std::size_t i = 0; i < sizes.size(); ++i) {
+            SCOPED_TRACE(sizes[i]);
+            glBindBuffer(GL_ARRAY_BUFFER, m_arena);
+            glBufferData(GL_ARRAY_BUFFER, sizes[i], nullptr, GL_DYNAMIC_DRAW);
+            UploadQuad(colors[i][0], colors[i][1], colors[i][2]);
+            // The unbound VAO can retain the deleted store; the current VAO's
+            // attachments can be cleared by deletion. Both must be repaired.
+            for (GLuint vao : {m_vao, otherVao}) {
+                SCOPED_TRACE(vao);
+                DrawQuad(vao);
+                const auto px = CenterPixel();
+                EXPECT_EQ(FirstGLError(), 0u);
+                for (std::size_t channel = 0; channel < 3; ++channel) {
+                    if (colors[i][channel] != 0.f) {
+                        EXPECT_GT(px[channel], 200) << "VAO did not fetch the replacement vertex store";
+                    } else {
+                        EXPECT_LT(px[channel], 50) << "VAO still fetched the previous vertex store";
+                    }
+                }
+            }
+        }
+        glDeleteVertexArrays(1, &otherVao);
+    }
+
+    TEST_F(LargeArenaAdoptionScenario, RespecifiedIndexArenaKeepsVaoBinding) {
+        if (!Ready() || IsSkipped()) return;
+
+        auto vertices = QuadVertices(1.f, 0.f, 0.f);
+        const auto green = QuadVertices(0.f, 1.f, 0.f);
+        vertices.insert(vertices.end(), green.begin(), green.end());
+        glBindBuffer(GL_ARRAY_BUFFER, m_arena);
+        glBufferSubData(GL_ARRAY_BUFFER, kVertexOffset,
+                        GLsizeiptr(vertices.size() * sizeof(Vertex)), vertices.data());
+
+        GLuint indices = 0;
+        glGenBuffers(1, &indices);
+        glBindVertexArray(m_vao);
+        glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, indices);
+        // Redefine through COPY_WRITE_BUFFER so the element binding slot never
+        // changes. The small final store also exercises returning to shadow storage.
+        glBindBuffer(GL_COPY_WRITE_BUFFER, indices);
+        constexpr std::array<GLsizeiptr, 4> sizes = {
+            kArenaBytes, kArenaBytes, kArenaBytes + 4096, 4096,
+        };
+        for (std::size_t i = 0; i < sizes.size(); ++i) {
+            SCOPED_TRACE(sizes[i]);
+            const GLuint first = (i % 2) == 0 ? 0u : 6u;
+            const std::array<GLuint, 6> elements = {
+                first, first + 1, first + 2, first + 3, first + 4, first + 5,
+            };
+            glBufferData(GL_COPY_WRITE_BUFFER, sizes[i], nullptr, GL_DYNAMIC_DRAW);
+            glBufferSubData(GL_COPY_WRITE_BUFFER, 0, sizeof(elements), elements.data());
+            glViewport(0, 0, Gl().Width(), Gl().Height());
+            glClearColor(0.f, 0.f, 0.f, 1.f);
+            glClear(GL_COLOR_BUFFER_BIT);
+            glUseProgram(m_program);
+            glDrawElements(GL_TRIANGLES, 6, GL_UNSIGNED_INT, nullptr);
+            const auto px = CenterPixel();
+            EXPECT_EQ(FirstGLError(), 0u);
+            EXPECT_GT(px[first == 0 ? 0 : 1], 200) << "VAO did not fetch the replacement index store";
+            EXPECT_LT(px[first == 0 ? 1 : 0], 50) << "VAO still fetched the previous index store";
+        }
+        glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, 0);
+        glBindBuffer(GL_COPY_WRITE_BUFFER, 0);
+        glDeleteBuffers(1, &indices);
+    }
+
     // The shadow IS the mapping: a readback straight after a CPU write must hand
     // back exactly those bytes.
     TEST_F(LargeArenaAdoptionScenario, ReadbackSeesTheLatestCpuWrite) {
@@ -275,4 +437,72 @@ void main() { word = 0xC0FFEEu; }
             << "the compute write into the adopted arena did not reach the CPU readback";
     }
 
+    // G10, the per-adoption half: ONE storage definition of an arena costs ONE map_persistent
+    // emission, however many draws read it afterwards.
+    //
+    // The arena SetUp defined is deliberately re-defined inside the counted window rather than
+    // measured from outside it: the window a summary line reports is "since the previous line",
+    // so the definition has to happen between the two swaps that bracket it, and a case that
+    // counted SetUp's definition would be reading a window it did not control.
+    //
+    // ONE reading case per lane, for the reason PipeStatsWindow.h gives: the library truncates the
+    // log per process, so two readers in a lane race under `ctest -j`.
+    TEST_F(LargeArenaAdoptionScenario, AnAdoptionCostsExactlyOneMapPersistentRoundtrip) {
+        if (!Ready() || IsSkipped()) return;
+        SkipUnlessTheRoundtripCounterIsReadableHere();
+        if (IsSkipped()) return;
+
+        Gl().EndFrame(); // close the setup window, SetUp's own definition included
+
+        // One definition of a store past the 16 MiB adoption threshold, taken by RE-SPECIFYING
+        // SetUp's arena while m_vao's attributes are still pointing into it - and the attributes
+        // are deliberately NOT re-declared afterwards, so the draws below can only land if the
+        // backend VAO followed the new store on its own.
+        //
+        // That is the hard shape on purpose. It was routed around in the first cut of this file
+        // because feat/disaggregated did not yet carry `dev`'s d7655247 ("rebind VAOs when an
+        // adopted buffer is respecified - the immediate retire path forgot the buffer-id
+        // generation") and the workload was a hard SIGSEGV inside the vertex fetch on the first
+        // draw after the re-specification. ID-9 merged that fix (feat/disaggregated 5cb826b0) and
+        // requires it to hold in BOTH the legacy and the handle arm of the respecify/retire path,
+        // so this workload counts the path rather than avoiding it: under the
+        // ResourceSubsystemOn./Off. lanes the same body runs on both arms, and a handle arm that
+        // re-implemented the retire without the rebind is a crash here rather than a silent
+        // divergence found on device.
+        glBindBuffer(GL_ARRAY_BUFFER, m_arena);
+        glBufferData(GL_ARRAY_BUFFER, kArenaBytes, nullptr, GL_DYNAMIC_DRAW);
+        ASSERT_EQ(FirstGLError(), 0u) << "re-specifying the arena inside the counted window failed";
+
+        // ... and then a frame's worth of traffic against it, of the shape the arena exists for:
+        // a SubData per draw, every one of which lands in the adopted mapping and none of which
+        // may acquire it again.
+        for (int draw = 0; draw < kDrawsInTheWindow; ++draw) {
+            UploadQuad(0.f, 1.f, 0.f);
+            DrawQuad();
+        }
+        const auto px = CenterPixel();
+        EXPECT_EQ(FirstGLError(), 0u);
+        EXPECT_GT(px[1], 200) << "the draws inside the counted window never landed, so the count below "
+                                 "would be a number about nothing";
+
+        Gl().EndFrame(); // the swap that emits the window covering exactly the work above
+        const PipeStatsWindow::Window window = PipeStatsWindow::LastFromLaneLog();
+        ASSERT_TRUE(window.found) << "no 'MGPipe stats:' line in " << PipeStatsWindow::LibraryLogPath()
+                                  << ": either MOBILEGL_PIPE_STATS / MOBILEGL_PIPE_STATS_PERIOD did not "
+                                     "reach the process, or nothing reached PipeStats::OnPresent.";
+        RecordProperty("stats_line", window.line.c_str());
+
+        const long long roundtrips = PipeStatsWindow::CounterOrAbsent(window, "mpr");
+        ASSERT_GE(roundtrips, 0) << "the summary line carries no mpr= field: " << window.line;
+        EXPECT_EQ(roundtrips, 1)
+            << "one storage definition of an adopted arena is one map_persistent emission "
+               "(ARCHITECTURE.md:474, D-B2: mint OR decline, both need an answer from the resource "
+               "owner). This window defined the arena once and drew from it "
+            << kDrawsInTheWindow << " times, so 1 is the whole cost; " << kDrawsInTheWindow
+            << " would mean the acquisition moved onto the draw path - the ~167 ms/arena hiccup this "
+               "adoption removed, re-introduced - and 0 would mean the emission stopped happening. It "
+               "reported: "
+            << window.line;
+    }
+
 } // namespace MGITest

MobileGL/MG_IntegrationTest/Scenarios/ObjectSubsystemControlScenario.cpp

@@ -0,0 +1,878 @@
+// MobileGL - MobileGL/MG_IntegrationTest/Scenarios/ObjectSubsystemControlScenario.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 - THE P4a SUBSYSTEM A/B IS REAL, AND ITS DEPENDENCY REFUSALS ARE EXERCISED (gate G12).
+//
+// P4a migrates FOUR subsystems (D-K1, MG_Pipe/MGPipe.h):
+//
+//   bit  9  kMGPipeSubsystemFramebuffer       set_framebuffer_state
+//   bit 10  kMGPipeSubsystemTextureResources  texture + renderbuffer resource_*, set_texture_params
+//   bit 11  kMGPipeSubsystemSamplers          sampler CSO, sampler view, the three unit sets
+//   bit 12  kMGPipeSubsystemPrograms          shader CSO, draw/dispatch program, global constants
+//
+// so the push build's default mask becomes kMGPipeSubsystemsMigratedAtP4a = 0x1fff, and P3a's
+// 0x1ff survives as the control that clears exactly those four - MGPipe.h's rule that every phase's
+// constant keeps meaning what it meant, so an operator's recorded mask is still readable a phase
+// later. THE OFF LANE IS 0x1ff AND NOT A HAND-PICKED PATTERN, for that reason.
+//
+// That A/B is what every "push vs pull" number in MEASUREMENTS.md is taken against, and it has one
+// characteristic failure mode: the bits stop steering anything, both arms run the same code, and
+// every later comparison is quietly taken against a switch that does nothing. This file is the
+// entry that cannot let that happen. It is the P4a analogue of ResourceSubsystemControlScenario and
+// deliberately its twin in shape.
+//
+// WHAT IT ASSERTS, per lane:
+//
+//   on  (MOBILEGL_PIPE_PUSH=0x1fff)
+//       The client emits P4a's records for the workload: a framebuffer state per bound target that
+//       moved, the three unit sets, and the client-side texture upload record. The window's
+//       emit[fbe= sve= sse= sie= ctu=] bracket therefore carries a NON-ZERO total.
+//
+//   off (MOBILEGL_PIPE_PUSH=0x1ff, P3a's default = P4a's four subsystems cleared)
+//       The frontend dispatch falls through to the legacy MGB_CTX-reading arms, nothing is emitted
+//       through any of the four families, and every one of those five counters must read ZERO.
+//       This is the reading a dead switch fails: with the bits ignored, this lane would report the
+//       same non-zero counts as the other one.
+//
+//   refused (MOBILEGL_PIPE_PUSH=0x9ff = bits 0..8 plus bit 11, samplers, WITHOUT bit 10)
+//       D-K2's dependency refusal. Every MGPBoundView::Texture and MGPImageView::Res names a
+//       Texture handle and only bit 10 populates the texture slot table, so a sampler subsystem
+//       without it would miss every lookup and walk on without unbinding. The bring-up logs ONE
+//       error naming BOTH bits, refuses bit 11 and runs the legacy sampler arm - modelled on the
+//       bit-8-requires-bit-7 refusal that already ships (Managers.cpp:2393-2410). The assertion is
+//       that the refusal is NAMED and that the run then produces the same pixels as any other
+//       lane: a refusal that half-ran, or that aborted, would both be failures here.
+//
+//   refused-texture (MOBILEGL_PIPE_PUSH=0x5ff = bits 0..8 plus bit 10, texture resources, WITHOUT
+//       bit 11)
+//       D-K2's FOURTH row (ID-15), and the direction the brief originally called harmless.
+//       MGPTextureParams::BuiltinSampler is a SamplerCso HANDLE and only bit 11 mints sampler
+//       CSOs, so with bit 10 alone every set_texture_params would carry a null there and the
+//       applier's Fatal{ProtocolCorruption} is the next thing that happens. Same two assertions
+//       as the lane above, with the two bits' roles swapped.
+//
+//   both refusal lanes
+//       "NAMED" means ONE LINE of the library's log, at ERROR severity, that says it REFUSED and
+//       names both bits. Not a substring anywhere in the file: the word "sampler" appears in
+//       almost any log the sampler path writes to, and an assertion that cannot go red for its
+//       stated reason is worse than no assertion (review F-M6).
+//
+//   every lane
+//       THE PIXELS MUST NOT MOVE. The workload draws one solid-colour quad through a texture, an
+//       explicit sampler object and a user framebuffer, and every lane must read back that colour.
+//       "The counters moved and the picture did not" is the whole claim - a switch that changed
+//       what is drawn would not be an A/B, it would be a bug.
+//
+// WHY IT CAN SKIP. The counters are emitted by the client-side emitters P4a packages B and C own,
+// and this file is written against the P4a contract commit, before either lands. Until then nothing
+// emits, the five counters are structurally zero in BOTH lanes, and an assertion about the
+// difference would be a statement about nothing. The build answers the question rather than a
+// hand-maintained list: MG_IntegrationTest/CMakeLists.txt greps every source under MG_Impl/Pipe/
+// for the counters' names and passes the answer in as MGITEST_PIPE_OBJECT_EMITTER_PRESENT, with a
+// CONFIGURE_DEPENDS on that directory and on each file it finds so the answer cannot go stale. It
+// is a CONTENT probe, not a filename probe, so the owning packages keep control of their own file
+// layout - P4a's new client files are headers (D-P), and a glob for a named .cpp would have kept
+// this control skipping forever with a reason that had become false.
+//
+// DIRECTGLES ONLY, and that is the honest scope: P4a migrates Espryt's framebuffer, texture,
+// sampler and program paths. Magma's are P7 (D-Q) and register nothing here, so a DirectVulkan lane
+// would be measuring the client emitters against a backend nobody asked to change.
+
+#include <cstdint>
+#include <cstdlib>
+#include <cstring>
+#include <iostream>
+#include <string>
+#include <vector>
+
+#include "../Harness/HeadlessGL.h"
+#include "../Harness/PipeApplyPeek.h"
+#include "../Harness/PipeStatsWindow.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 {
+
+        // Set by the three ObjectSubsystemControl. ctest entries and by nothing else; a harness
+        // marker, never read by the library. Its absence means an ambient entry, where neither the
+        // stats channel nor a private log path is configured.
+        constexpr const char* kLaneMarker = "MGITEST_OBJECT_SUBSYSTEM_LANE";
+        constexpr const char* kLaneOn = "on";
+        constexpr const char* kLaneOff = "off";
+        constexpr const char* kLaneRefused = "refused";
+        // D-K2's FOURTH row (ID-15): bit 10 without bit 11. 0x5ff is 0x1ff plus bit 10.
+        constexpr const char* kLaneRefusedTexture = "refused-texture";
+        // c0f's two halves (ID-39/ID-40), run at the phase default on BOTH backends: the client
+        // GATE (a P4a family emits only where a backend registered MGPipeResourceOps) and the
+        // applier's BELT (every P4a entry point refuses and counts RefusedNoConsumer when none
+        // did). One lane per backend, because the interesting one is the backend with NO
+        // consumer - Magma - and the other is the control that says the assertion is not
+        // vacuously true of a tree where nothing emits at all.
+        constexpr const char* kLaneConsumer = "consumer";
+        constexpr const char* kLaneNoConsumer = "no-consumer";
+
+        bool LaneIsARefusalLane(const std::string& lane) {
+            return lane == kLaneRefused || lane == kLaneRefusedTexture;
+        }
+
+        bool LaneIsAConsumerLane(const std::string& lane) {
+            return lane == kLaneConsumer || lane == kLaneNoConsumer;
+        }
+
+        constexpr int kInset = 2;
+        constexpr int kTextureSize = 4;
+        // Enough frames that a per-frame emitter and a per-draw emitter read differently, and few
+        // enough that one summary window covers exactly this.
+        constexpr int kDrawsInTheWindow = 4;
+
+        constexpr const char* kVS = R"(#version 330 core
+in vec2 aPos;
+out vec2 vUv;
+void main() {
+    vUv = aPos * 0.5 + 0.5;
+    gl_Position = vec4(aPos, 0.0, 1.0);
+}
+)";
+
+        constexpr const char* kFS = R"(#version 330 core
+in vec2 vUv;
+uniform sampler2D uTex;
+out vec4 oColor;
+void main() { oColor = texture(uTex, vUv); }
+)";
+
+        struct Vertex {
+            float x, y;
+        };
+
+        bool BuildMarkerIsSet(const char* name) {
+            const char* value = std::getenv(name);
+            return value != nullptr && value[0] == '1' && value[1] == '\0';
+        }
+
+        std::string LaneName() {
+            const char* lane = std::getenv(kLaneMarker);
+            return lane != nullptr ? std::string(lane) : std::string();
+        }
+
+        // ---- reading the refusal out of the library's own log ------------------------------
+        //
+        // THE UNIT IS A LINE, AND THE LINE HAS TO BE THE REFUSAL (review F-M6). The first cut of
+        // this asked whether the WHOLE FILE contained a lowercase "sampler" and whether it
+        // contained "texture resource", anywhere, in any order, at any severity. Both are true of
+        // almost any log the moment the sampler path says anything at all, so the assertion could
+        // not go red for the reason it claims and the one P4a control that is not vacuous before
+        // the emitters land would have been vacuous too.
+        //
+        // What is matched instead is one line that is ALL of:
+        //   * at ERROR severity - the library writes "[<time>] [<os> <thread>/<TAG>]: <message>",
+        //     one record per line (MG_Util/Debug/Log.cpp), and D-K2 asks for an MGLOG_E. A refusal
+        //     that was demoted to a D or a W is a refusal an operator's log will not carry;
+        //   * carrying the helper's own decision clause, verbatim - so a line that merely
+        //     mentions the two bits (a future summary, a comment echoed into the log) is not
+        //     mistaken for the decision;
+        //   * naming the bit that was SET and the bit it NEEDED, on that same line, AND IN THAT
+        //     ORDER - see the direction check below.
+        //
+        // Espryt's text is one MGLOG_E from the helper the three dependent families share
+        // (Managers.cpp, PipeSubsystemDependencyMissing): "MGPipe: <A> (bit N) is set but <B>
+        // (bit M) is clear; <why> - REFUSING the dependent bit and running the legacy arm. Set
+        // both bits, or clear both". Three spellings are accepted per bit - the constant's name,
+        // "(bit N)", and the hexadecimal mask - so the assertion pins the DECISION and the
+        // DIRECTION, and not the family-specific prose in <why>.
+        //
+        // THE DIRECTION IS THE HALF THIS FILE USED TO BE MISSING (review F-v2-m1). The first form
+        // of the matcher asked "does the line name bit A?" AND "does the line name bit B?", which
+        // is a SYMMETRIC conjunction: swapping the two arguments - exactly what separates the
+        // 0x5ff case from the 0x9ff one below, and what each of their comments claims to be
+        // doing - could not change the answer, and both cases went green on either line. A
+        // resolver that refused correctly but printed the MIRROR sentence would have been green
+        // on a refusal that told the operator the wrong dependency, which is the same class of
+        // "the log says something plausible" defect that made the whole-file substring search
+        // (F-M6) worthless one level up. The two resolvers are forty lines apart in one file,
+        // share this helper and differ only in the `what` string, so the copy-paste is one edit
+        // away at all times.
+        //
+        // What makes the direction readable is the sentence's own shape: the SET bit is named
+        // before " is set but " and the NEEDED bit between that and " is clear". So the check is
+        // four offsets in strictly increasing order, and it is the sentence Espryt emits rather
+        // than a re-statement of it.
+        constexpr const char* kSaysItRefused = "REFUSING the dependent bit and running the legacy arm";
+        constexpr const char* kSaysWhichIsSet = " is set but ";
+        constexpr const char* kSaysWhichIsClear = " is clear";
+
+        // The earliest offset at which any accepted spelling of one bit appears, or npos. The
+        // EARLIEST rather than any: a spelling that also occurs later in <why> (Espryt's
+        // bit-10-requires-bit-11 sentence says "only bit 11 mints sampler CSOs" in its reason)
+        // must not be able to satisfy an ordering the first occurrence does not.
+        std::size_t EarliestSpellingOffset(const std::string& line,
+                                           const std::vector<std::string>& spellings) {
+            std::size_t earliest = std::string::npos;
+            for (const std::string& spelling : spellings) {
+                const std::size_t at = line.find(spelling);
+                if (at != std::string::npos && (earliest == std::string::npos || at < earliest)) {
+                    earliest = at;
+                }
+            }
+            return earliest;
+        }
+
+        // The matching line, or an empty string. Returned rather than a bool so the case can print
+        // what it found: a reader of a green refusal lane must be able to see the sentence.
+        std::string FindTheRefusalLine(const std::string& log,
+                                       const std::vector<std::string>& bitThatWasSet,
+                                       const std::vector<std::string>& bitThatWasNeeded) {
+            std::size_t pos = 0;
+            while (pos <= log.size()) {
+                const std::size_t newline = log.find('\n', pos);
+                const std::string line = log.substr(
+                    pos, newline == std::string::npos ? std::string::npos : newline - pos);
+                const bool atErrorSeverity = line.find("/ERROR]") != std::string::npos;
+                const std::size_t refusedAt = line.find(kSaysItRefused);
+                const std::size_t setAt = EarliestSpellingOffset(line, bitThatWasSet);
+                const std::size_t setClauseAt = line.find(kSaysWhichIsSet);
+                const std::size_t neededAt = EarliestSpellingOffset(line, bitThatWasNeeded);
+                const std::size_t clearClauseAt = line.find(kSaysWhichIsClear);
+                const bool everyPartIsThere =
+                    refusedAt != std::string::npos && setAt != std::string::npos &&
+                    setClauseAt != std::string::npos && neededAt != std::string::npos &&
+                    clearClauseAt != std::string::npos;
+                // "<set bit> ... is set but ... <needed bit> ... is clear", strictly in that
+                // order. Swapping the caller's two arguments breaks the chain, which is the
+                // whole of F-v2-m1.
+                const bool inTheRightDirection =
+                    everyPartIsThere && setAt < setClauseAt && setClauseAt < neededAt &&
+                    neededAt < clearClauseAt;
+                if (atErrorSeverity && inTheRightDirection) {
+                    return line;
+                }
+                if (newline == std::string::npos) break;
+                pos = newline + 1;
+            }
+            return std::string();
+        }
+
+        // The three accepted spellings of each of the two P4a bits this file's two refusal lanes
+        // are about. MGPipe.h: bit 10 = kMGPipeSubsystemTextureResources = 0x400,
+        // bit 11 = kMGPipeSubsystemSamplers = 0x800.
+        std::vector<std::string> SamplerBitSpellings() {
+            return {"kMGPipeSubsystemSamplers", "(bit 11)", "0x800"};
+        }
+
+        std::vector<std::string> TextureResourceBitSpellings() {
+            return {"kMGPipeSubsystemTextureResources", "(bit 10)", "0x400"};
+        }
+
+        class ObjectSubsystemControlScenario : public ScenarioTest {
+        protected:
+            void SetUp() override {
+                ScenarioTest::SetUp();
+                if (!Ready()) return;
+                m_lane = LaneName();
+                std::string error;
+                m_program = CompileProgram(kVS, kFS, &error);
+                ASSERT_NE(m_program, 0u) << error;
+
+                static const Vertex quad[6] = {{-1.0f, -1.0f}, {1.0f, -1.0f}, {1.0f, 1.0f},
+                                               {-1.0f, -1.0f}, {1.0f, 1.0f},  {-1.0f, 1.0f}};
+                glGenBuffers(1, &m_quadBuffer);
+                glBindBuffer(GL_ARRAY_BUFFER, m_quadBuffer);
+                glBufferData(GL_ARRAY_BUFFER, sizeof(quad), quad, GL_STATIC_DRAW);
+                glGenVertexArrays(1, &m_vao);
+                glBindVertexArray(m_vao);
+                glEnableVertexAttribArray(0);
+                glVertexAttribPointer(0, 2, GL_FLOAT, GL_FALSE, sizeof(Vertex), nullptr);
+                glBindVertexArray(0);
+                RecordProperty("lane", m_lane.empty() ? "ambient" : m_lane.c_str());
+            }
+
+            void TearDown() override {
+                if (!Ready()) return;
+                glUseProgram(0);
+                glBindVertexArray(0);
+                glBindSampler(0, 0);
+                glBindBuffer(GL_ARRAY_BUFFER, 0);
+                if (m_vao != 0) glDeleteVertexArrays(1, &m_vao);
+                if (m_quadBuffer != 0) glDeleteBuffers(1, &m_quadBuffer);
+                if (m_program != 0) glDeleteProgram(m_program);
+            }
+
+            // GTEST_SKIP() returns from the function it is written in, so this cannot report
+            // through a return value; every caller pairs it with `if (IsSkipped()) return;`.
+            void SkipUnlessTheLaneIsAssertableHere(bool needsTheEmitters) {
+                if (m_lane.empty()) {
+                    GTEST_SKIP() << "runs only in its own lane: the three ObjectSubsystemControl. "
+                                    "ctest entries set " << kLaneMarker
+                                 << " together with the MOBILEGL_PIPE_PUSH bitmask that arm means, "
+                                    "MOBILEGL_PIPE_STATS=1, MOBILEGL_PIPE_STATS_PERIOD=1 and a "
+                                    "private MOBILEGL_LOG_FILE_PATH. None of that is configured in "
+                                    "the ambient entries, and the ambient log is shared, so a read "
+                                    "here would race.";
+                    return;
+                }
+                if (!BuildMarkerIsSet("MGITEST_PIPE_PUSH_BUILD")) {
+                    GTEST_SKIP() << "this library was built without MOBILEGL_PIPE_PUSH: there are no "
+                                    "subsystem bits to clear, P4a's five CallClass members do not "
+                                    "exist and the summary line carries no emit[...] bracket. The "
+                                    "entry is registered here anyway so that `ctest -L "
+                                    "integration-gpu` names the same tests in the pull build and the "
+                                    "push build (gate G2).";
+                    return;
+                }
+                if (needsTheEmitters && !BuildMarkerIsSet("MGITEST_PIPE_OBJECT_EMITTER_PRESENT")) {
+                    GTEST_SKIP() << "subsystem not implemented on this tree: no source under "
+                                    "MobileGL/MG_Impl/Pipe/ emits FramebufferEmissions, so nothing "
+                                    "sends a P4a record, every counter in the emit[] bracket is "
+                                    "structurally zero in BOTH lanes and the difference between them "
+                                    "is not observable yet. P4a packages B (framebuffer, texture) "
+                                    "and C (sampler, image, program) own those emitters; this "
+                                    "control arms itself when they land, whatever files they use.";
+                    return;
+                }
+                if (PipeStatsWindow::LibraryLogPath().empty()) {
+                    GTEST_SKIP() << "the lane configured no MOBILEGL_LOG_FILE_PATH, and the library's "
+                                    "own log is the only channel this module has for reading "
+                                    "PipeStats and the bring-up's refusal line";
+                    return;
+                }
+            }
+
+            // The workload, and every one of P4a's four families is in it exactly once per draw:
+            // a USER FRAMEBUFFER with a texture attachment (bit 9), a TEXTURE with parameters and
+            // an upload (bit 10), an explicit SAMPLER OBJECT on the unit (bit 11) and a PROGRAM
+            // with a default-uniform-block write (bit 12). A lane that steered only one of the four
+            // would move only its own counter, which is why they are counted separately.
+            void RunTheWorkload() {
+                std::vector<std::uint8_t> texels(kTextureSize * kTextureSize * 4);
+                for (std::size_t i = 0; i < texels.size(); i += 4) {
+                    texels[i] = 0;
+                    texels[i + 1] = 255;
+                    texels[i + 2] = 0;
+                    texels[i + 3] = 255;
+                }
+                glGenTextures(1, &m_texture);
+                glBindTexture(GL_TEXTURE_2D, m_texture);
+                glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA8, kTextureSize, kTextureSize, 0, GL_RGBA,
+                             GL_UNSIGNED_BYTE, texels.data());
+                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_BASE_LEVEL, 0);
+                glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, 0);
+
+                glGenSamplers(1, &m_sampler);
+                glSamplerParameteri(m_sampler, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+                glSamplerParameteri(m_sampler, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
+                glSamplerParameteri(m_sampler, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);
+                glSamplerParameteri(m_sampler, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);
+
+                // The user framebuffer, drawn into once per iteration so that the framebuffer
+                // record has a reason to move: the binding alternates between it and the default
+                // framebuffer, which is exactly what a per-target set_framebuffer_state counts.
+                glGenTextures(1, &m_attachment);
+                glBindTexture(GL_TEXTURE_2D, m_attachment);
+                glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA8, kTextureSize, kTextureSize, 0, GL_RGBA,
+                             GL_UNSIGNED_BYTE, nullptr);
+                glBindTexture(GL_TEXTURE_2D, 0);
+                glGenFramebuffers(1, &m_fbo);
+                glBindFramebuffer(GL_FRAMEBUFFER, m_fbo);
+                glFramebufferTexture2D(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D,
+                                       m_attachment, 0);
+                BindDefaultFramebuffer();
+
+                for (int draw = 0; draw < kDrawsInTheWindow; ++draw) {
+                    // Into the user framebuffer...
+                    glBindFramebuffer(GL_FRAMEBUFFER, m_fbo);
+                    glViewport(0, 0, kTextureSize, kTextureSize);
+                    glUseProgram(m_program);
+                    glUniform1i(glGetUniformLocation(m_program, "uTex"), 0);
+                    glActiveTexture(GL_TEXTURE0);
+                    glBindTexture(GL_TEXTURE_2D, m_texture);
+                    glBindSampler(0, m_sampler);
+                    glBindVertexArray(m_vao);
+                    glDrawArrays(GL_TRIANGLES, 0, 6);
+                    // ...and into the default one, which is what the case reads back.
+                    BindDefaultFramebuffer();
+                    glViewport(0, 0, Gl().Width(), Gl().Height());
+                    glDrawArrays(GL_TRIANGLES, 0, 6);
+                    // One sub-region upload per iteration, so the client-side texture upload
+                    // counter (ctu) has something to count and the server's tex[emit=] has the
+                    // same something.
+                    const std::uint8_t green[4] = {0, 255, 0, 255};
+                    glBindTexture(GL_TEXTURE_2D, m_texture);
+                    glTexSubImage2D(GL_TEXTURE_2D, 0, draw % kTextureSize, 0, 1, 1, GL_RGBA,
+                                    GL_UNSIGNED_BYTE, green);
+                }
+            }
+
+            void ReleaseTheWorkload() {
+                glBindSampler(0, 0);
+                glBindTexture(GL_TEXTURE_2D, 0);
+                BindDefaultFramebuffer();
+                if (m_fbo != 0) glDeleteFramebuffers(1, &m_fbo);
+                if (m_sampler != 0) glDeleteSamplers(1, &m_sampler);
+                if (m_texture != 0) glDeleteTextures(1, &m_texture);
+                if (m_attachment != 0) glDeleteTextures(1, &m_attachment);
+                m_fbo = m_sampler = m_texture = m_attachment = 0;
+            }
+
+            std::string m_lane;
+            GLuint m_program = 0;
+            GLuint m_vao = 0;
+            GLuint m_quadBuffer = 0;
+            GLuint m_texture = 0;
+            GLuint m_attachment = 0;
+            GLuint m_sampler = 0;
+            GLuint m_fbo = 0;
+        };
+
+        // ONE case per lane, and it is a constraint rather than a preference: this case READS the
+        // library log, the log is a per-LANE resource (the library opens it fopen(path, "w"), so
+        // every process in a lane truncates it), and a second case in the same lane would race this
+        // one under `ctest -j` with a failure indistinguishable from "the counter was never
+        // emitted". The CMake registration gives each lane a TEST_FILTER naming one case.
+        TEST_F(ObjectSubsystemControlScenario, ClearingTheP4aBitsStopsTheEmissionsAndNotThePixels) {
+            if (!Ready()) return;
+            SkipUnlessTheLaneIsAssertableHere(/*needsTheEmitters=*/true);
+            if (IsSkipped()) return;
+            if (LaneIsAConsumerLane(m_lane)) {
+                GTEST_SKIP() << "the two consumer lanes run their own case instead "
+                                "(TheAppliersNoConsumerBeltNeverFiresBehindTheClientsGate). They "
+                                "are at the phase default on both backends and their subject is "
+                                "c0f's gate/belt pair, not the on/off A/B: on the backend with no "
+                                "consumer the emit[] bracket is structurally zero AT the default "
+                                "mask, which is neither the on-lane's expectation nor the "
+                                "off-lane's.";
+            }
+            if (LaneIsARefusalLane(m_lane)) {
+                GTEST_SKIP() << "the refusal lanes run their own case instead (0x9ff -> "
+                                "ASamplerBitWithoutTheTextureBitIsRefusedAndNamed, 0x5ff -> "
+                                "ATextureBitWithoutTheSamplerBitIsRefusedAndNamed): a refused "
+                                "subsystem's emission counts are neither the on-lane's nor the "
+                                "off-lane's, and asserting either would be reading a third arm as "
+                                "if it were one of the two.";
+            }
+
+            BindDefaultFramebuffer();
+            Gl().EndFrame(); // close the setup window: everything below is one window
+
+            ClearTo(0.0f, 0.0f, 0.0f, 1.0f);
+            RunTheWorkload();
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR)) << "the workload left a GL error behind";
+            const Image image = ReadPixels(Gl().Width(), Gl().Height());
+            Gl().EndFrame(); // the swap that emits the window covering exactly the work above
+
+            const PipeStatsWindow::Window window = PipeStatsWindow::LastFromLaneLog();
+            ASSERT_TRUE(window.found)
+                << "no 'MGPipe stats:' line in " << PipeStatsWindow::LibraryLogPath()
+                << ". This IS a push build (the lane checked MGITEST_PIPE_PUSH_BUILD before getting "
+                   "here), so either MOBILEGL_PIPE_STATS / MOBILEGL_PIPE_STATS_PERIOD did not reach "
+                   "the process, or no summary line was emitted at all because nothing reached "
+                   "PipeStats::OnPresent.";
+            RecordProperty("stats_line", window.line.c_str());
+
+            // The five counters of the emit[] bracket, read individually so that a lane which
+            // steered one family and not another says WHICH.
+            const long long framebuffer = PipeStatsWindow::CounterOrAbsent(window, "fbe");
+            const long long samplerViews = PipeStatsWindow::CounterOrAbsent(window, "sve");
+            const long long samplerStates = PipeStatsWindow::CounterOrAbsent(window, "sse");
+            const long long shaderImages = PipeStatsWindow::CounterOrAbsent(window, "sie");
+            const long long clientUploads = PipeStatsWindow::CounterOrAbsent(window, "ctu");
+            ASSERT_GE(framebuffer, 0)
+                << "the summary line carries no fbe= field, so this build's PipeStats has no P4a "
+                   "emission counters to read: "
+                << window.line;
+            ASSERT_GE(samplerViews, 0) << "no sve= field: " << window.line;
+            ASSERT_GE(samplerStates, 0) << "no sse= field: " << window.line;
+            ASSERT_GE(shaderImages, 0) << "no sie= field: " << window.line;
+            ASSERT_GE(clientUploads, 0) << "no ctu= field: " << window.line;
+            const long long total = framebuffer + samplerViews + samplerStates + shaderImages +
+                                    clientUploads;
+
+            if (m_lane == kLaneOn) {
+                EXPECT_GT(total, 0)
+                    << "with bits 9|10|11|12 SET the four P4a families are the path this workload "
+                       "takes - a user framebuffer bound and unbound "
+                    << kDrawsInTheWindow
+                    << " times, a texture with parameters and a sub-region upload per iteration, an "
+                       "explicit sampler object on the unit and a program with a default-uniform "
+                       "write - so the window's emit[] bracket must carry something. All five "
+                       "reading zero means the emitters never ran on the arm that is supposed to run "
+                       "them. It reported: "
+                    << window.line;
+                // The framebuffer family on its own, because it is the one that would be hidden by
+                // a large upload count: a suppressor that stopped suppressing shows up as fbe
+                // tracking the DRAW count, and a family that never emitted shows up as zero.
+                EXPECT_GT(framebuffer, 0)
+                    << "fbe= is zero on the ON lane: set_framebuffer_state never went out even "
+                       "though the workload bound a user framebuffer and the default framebuffer "
+                    << kDrawsInTheWindow << " times each. " << window.line;
+            } else if (m_lane == kLaneOff) {
+                EXPECT_EQ(total, 0)
+                    << "with bits 9|10|11|12 CLEARED (MOBILEGL_PIPE_PUSH=0x1ff, P3a's default) the "
+                       "frontend dispatch must fall through to the legacy MGB_CTX-reading arms and "
+                       "emit nothing through any of the four P4a families, so every counter in the "
+                       "emit[] bracket must be zero. A non-zero count here is the dead-switch "
+                       "reading: the bits are being ignored, both arms run the same code, and every "
+                       "push-vs-pull number taken against this A/B is measuring one arm twice. It "
+                       "reported: "
+                    << window.line;
+            } else {
+                FAIL() << "unknown " << kLaneMarker << " value '" << m_lane
+                       << "': the arms are on / off / refused / refused-texture / consumer / "
+                          "no-consumer. Reading an unrecognised name as any of them would make "
+                          "this lane assert another arm's expectation while claiming to test "
+                          "this one.";
+            }
+
+            // ... and the picture is the same whichever arm ran.
+            EXPECT_TRUE(RegionIsMostly(image, kInset, image.Width() - kInset, kInset,
+                                       image.Height() - kInset, "green", 0.0,
+                                       "the sampled draw [" + m_lane + "]"))
+                << "the subsystem bits changed what is DRAWN, which is not an A/B - the handle path "
+                   "and the legacy path must produce the same pixels from the same texture, sampler "
+                   "and framebuffer.";
+
+            ReleaseTheWorkload();
+        }
+
+        // ------------------------------------------------------------------------------------
+        // D-K2's dependency refusal, in the direction that has to be refused.
+        //
+        // 0x9ff is bits 0..8 (everything P3a shipped) plus bit 11 (samplers) and WITHOUT bit 10
+        // (texture resources). Every MGPBoundView::Texture and every MGPImageView::Res names a
+        // Texture handle, and only bit 10 populates the texture slot table, so with bit 11 alone
+        // every lookup would miss and the unit walk would `continue` without unbinding - a
+        // half-run subsystem, which ROADMAP.md:7 forbids as loudly as a dead switch. The bring-up
+        // logs ONE error naming both bits, refuses bit 11, and runs the legacy sampler arm.
+        //
+        // TWO ASSERTIONS, and the second is the one that stops this from being a log-scraping test:
+        // the refusal is NAMED in the library's own log, and the run then draws the same picture as
+        // every other lane. A refusal that aborted the process, and a refusal that silently let the
+        // half-configured arm run, are both failures - and they look completely different here.
+        // ------------------------------------------------------------------------------------
+        TEST_F(ObjectSubsystemControlScenario, ASamplerBitWithoutTheTextureBitIsRefusedAndNamed) {
+            if (!Ready()) return;
+            // needsTheEmitters=false: the refusal is a BRING-UP decision made from the bitmask
+            // alone, so it is assertable before any emitter exists - which is exactly what makes it
+            // the one P4a control that is not vacuous on the contract tree.
+            SkipUnlessTheLaneIsAssertableHere(/*needsTheEmitters=*/false);
+            if (IsSkipped()) return;
+            if (m_lane != kLaneRefused) {
+                GTEST_SKIP() << "runs only in the refusal lane (MOBILEGL_PIPE_PUSH=0x9ff): the "
+                                "on/off lanes configure a mask whose dependencies are all satisfied, "
+                                "so there is no refusal there to find and a search for one would "
+                                "report a healthy lane as red.";
+            }
+            // The refusal is decided from the bitmask, but it is a BACKEND's decision: D-K2 puts it
+            // in ResolveSamplersSubsystemArm(), beside the bit-8-requires-bit-7 refusal that
+            // already ships, and that function is package D's (Managers.cpp). A backend that does
+            // not yet honour P4a's mask at all cannot refuse a dependency inside it, so on such a
+            // tree there is nothing here to find and this case SKIPS rather than reporting the
+            // absence of an unimplemented subsystem as a failure. The marker is the same one
+            // HandleRecycle's P4a cases read - "does any source under this backend name one of the
+            // four P4a subsystem constants" - because naming the constant is exactly what honouring
+            // the mask means.
+            {
+                const std::string& backend = Gl().BackendName();
+                const std::string marker =
+                    "MGITEST_HANDLE_REKEY_OBJECTS_" + (backend == "DirectVulkan"
+                                                           ? std::string("DirectVulkan")
+                                                           : std::string("DirectGLES"));
+                if (!BuildMarkerIsSet(marker.c_str())) {
+                    GTEST_SKIP() << "subsystem not implemented on this tree: no source under "
+                                    "MobileGL/MG_Backend/"
+                                 << backend
+                                 << " names any of kMGPipeSubsystem{Framebuffer, TextureResources, "
+                                    "Samplers, Programs}, so this backend does not honour P4a's mask "
+                                    "and cannot refuse a dependency inside it. D-K2's refusal lives "
+                                    "in ResolveSamplersSubsystemArm() beside the bit-8-requires-bit-7 "
+                                    "one that already ships (Managers.cpp:2393-2410), which is P4a "
+                                    "package D's file; this control arms itself when that lands. The "
+                                    "lane itself is not wasted: the library came up under 0x9ff, "
+                                    "which on a tree with no P4a arm is P3a's mask plus one inert "
+                                    "bit, and a mask that aborted a bring-up would have failed this "
+                                    "entry before the skip.";
+                }
+            }
+
+            BindDefaultFramebuffer();
+            ClearTo(0.0f, 0.0f, 0.0f, 1.0f);
+            RunTheWorkload();
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR))
+                << "the workload left a GL error behind on the refused lane, which would mean the "
+                   "refusal did not fall back cleanly to the legacy arm";
+            const Image image = ReadPixels(Gl().Width(), Gl().Height());
+            Gl().EndFrame();
+
+            const std::string log = PipeStatsWindow::ReadWholeFile(PipeStatsWindow::LibraryLogPath());
+            ASSERT_FALSE(log.empty())
+                << "the library wrote nothing to " << PipeStatsWindow::LibraryLogPath()
+                << ", so the refusal cannot be read back. MOBILEGL_LOG_FILE_PATH is the only channel "
+                   "this module has for the library's own report.";
+            // ONE LINE, at ERROR severity, saying it refused and naming BOTH bits. See
+            // FindTheRefusalLine: a substring search over the whole file cannot go red for the
+            // reason this case claims (F-M6).
+            const std::string refusal =
+                FindTheRefusalLine(log, SamplerBitSpellings(), TextureResourceBitSpellings());
+            EXPECT_FALSE(refusal.empty())
+                << "MOBILEGL_PIPE_PUSH=0x9ff sets the sampler subsystem (bit 11) without the texture "
+                   "resource subsystem (bit 10) it depends on, and no single ERROR line of the "
+                   "library's log both says it REFUSED and names the two bits. D-K2 requires ONE "
+                   "MGLOG_E naming both and a fall back to the legacy sampler arm; a mask that is "
+                   "silently half-honoured is the failure this case exists to catch, and it is "
+                   "invisible in the pixels by construction. Accepted spellings per bit are the "
+                   "constant's name, '(bit 11)' / '(bit 10)', and '0x800' / '0x400'. The log was "
+                << log.size() << " bytes and is at " << PipeStatsWindow::LibraryLogPath() << ".";
+            if (!refusal.empty()) {
+                // Printed on the pass as well: a reader of a green refusal lane must be able to
+                // see the sentence the lane went green on.
+                std::cout << "[ ObjectSubsystemControl ] refusal line: " << refusal << std::endl;
+                RecordProperty("refusal_line", refusal.c_str());
+            }
+
+            EXPECT_TRUE(RegionIsMostly(image, kInset, image.Width() - kInset, kInset,
+                                       image.Height() - kInset, "green", 0.0,
+                                       "the sampled draw [refused]"))
+                << "the refused configuration did not draw what every other lane draws. A refusal is "
+                   "supposed to run the LEGACY arm, which is the arm that ships in a pull build - so "
+                   "the pixels are the one thing it may not change.";
+
+            ReleaseTheWorkload();
+        }
+
+        // ------------------------------------------------------------------------------------
+        // D-K2's FOURTH dependency row, in the OTHER direction: bit 10 without bit 11 (ID-15).
+        //
+        // 0x5ff is bits 0..8 plus bit 10 (texture resources) and WITHOUT bit 11 (samplers).
+        //
+        // WHY THIS IS A REFUSAL AND NOT THE "FINE" MIRROR PAIR THE BRIEF ORIGINALLY CALLED IT.
+        // BRIEF-P4A.md's D-K2 says "bit 10 without bit 11 is fine", and that sentence is wrong for
+        // P4a AS BUILT: MGPTextureParams carries a BuiltinSampler, which is a SamplerCso HANDLE,
+        // and only bit 11 mints sampler CSOs - c0b's four unconditional mints deliberately exclude
+        // that kind (contract-v2.md), and package C content-addresses them through its own cache
+        // (ID-14). With bit 10 set and bit 11 clear every set_texture_params would therefore carry
+        // a NULL BuiltinSampler, which the applier treats as Fatal{ProtocolCorruption} (wire H1),
+        // and minting it client-side in the arm that exists to exclude samplers was rejected. So
+        // the dependency is real and it has to be refused at bring-up, exactly like bit 11 without
+        // bit 10 above and bit 8 without bit 7 one phase earlier. ID-15 puts the refusal in the
+        // texture family's Resolve*SubsystemArm - package D's Managers.cpp - and this case is the
+        // pin that says it is there.
+        //
+        // ON A TREE WHOSE BACKEND DOES NOT HONOUR P4a's MASK THIS SKIPS, NAMED, exactly as the
+        // 0x9ff case does and for the same reason: a backend that never reads the four constants
+        // cannot refuse a dependency between two of them, and reporting the absence of an
+        // unimplemented subsystem as a failure is what ID-2 forbids. Once the backend DOES name
+        // them the case is a hard pin, which is the point - if D's texture-family resolver honours
+        // the mask and does not carry this row, this entry is where that shows.
+        //
+        // WHAT THIS ARM DOES ON THE INTEGRATED TREE, corrected (review F-v2-m2). An earlier
+        // round's report told the integrator to expect this lane to go RED between esprytobj's
+        // integration and package D's rework, and to read that red as expected. That window does
+        // not exist: esprytobj v2 already carries D-K2's fourth row - Managers.cpp's
+        // ResolveTextureResourceSubsystemArm refuses bit 10 without bit 11 with the sentence this
+        // case matches - so the arm ARMS AND PASSES. A red here is therefore a real finding about
+        // that resolver (it stopped refusing, refused for the wrong reason, or printed the mirror
+        // sentence, which the direction check above is what catches) and must not be waved
+        // through as a sequencing artefact.
+        // ------------------------------------------------------------------------------------
+        TEST_F(ObjectSubsystemControlScenario, ATextureBitWithoutTheSamplerBitIsRefusedAndNamed) {
+            if (!Ready()) return;
+            // needsTheEmitters=false, for the 0x9ff case's reason: a bring-up decision made from
+            // the bitmask alone is assertable before any emitter exists.
+            SkipUnlessTheLaneIsAssertableHere(/*needsTheEmitters=*/false);
+            if (IsSkipped()) return;
+            if (m_lane != kLaneRefusedTexture) {
+                GTEST_SKIP() << "runs only in the texture-side refusal lane "
+                                "(MOBILEGL_PIPE_PUSH=0x5ff): every other lane configures a mask "
+                                "whose dependencies are satisfied or a different refusal, so there "
+                                "is nothing here to find and a search for one would report a "
+                                "healthy lane as red.";
+            }
+            {
+                const std::string& backend = Gl().BackendName();
+                const std::string marker =
+                    "MGITEST_HANDLE_REKEY_OBJECTS_" + (backend == "DirectVulkan"
+                                                           ? std::string("DirectVulkan")
+                                                           : std::string("DirectGLES"));
+                if (!BuildMarkerIsSet(marker.c_str())) {
+                    GTEST_SKIP() << "subsystem not implemented on this tree: no source under "
+                                    "MobileGL/MG_Backend/"
+                                 << backend
+                                 << " names any of kMGPipeSubsystem{Framebuffer, TextureResources, "
+                                    "Samplers, Programs}, so this backend does not honour P4a's mask "
+                                    "and cannot refuse a dependency inside it. D-K2's fourth row "
+                                    "(bit 10 requires bit 11, ID-15) lives in the texture family's "
+                                    "Resolve*SubsystemArm beside the bit-11-requires-bit-10 and "
+                                    "bit-8-requires-bit-7 refusals, which is P4a package D's file; "
+                                    "this control arms itself when that lands. The lane itself is "
+                                    "not wasted: the library came up under 0x5ff, which on a tree "
+                                    "with no P4a arm is P3a's mask plus one inert bit, and a mask "
+                                    "that aborted a bring-up would have failed this entry before "
+                                    "the skip.";
+                }
+            }
+
+            BindDefaultFramebuffer();
+            ClearTo(0.0f, 0.0f, 0.0f, 1.0f);
+            RunTheWorkload();
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR))
+                << "the workload left a GL error behind on the refused lane, which would mean the "
+                   "refusal did not fall back cleanly to the legacy arm";
+            const Image image = ReadPixels(Gl().Width(), Gl().Height());
+            Gl().EndFrame();
+
+            const std::string log = PipeStatsWindow::ReadWholeFile(PipeStatsWindow::LibraryLogPath());
+            ASSERT_FALSE(log.empty())
+                << "the library wrote nothing to " << PipeStatsWindow::LibraryLogPath()
+                << ", so the refusal cannot be read back. MOBILEGL_LOG_FILE_PATH is the only channel "
+                   "this module has for the library's own report.";
+            // The same line shape as the 0x9ff arm, with the two bits' roles swapped: the bit that
+            // was SET is the texture-resource one and the bit it NEEDED is the sampler one. The
+            // swap is now a REAL difference between the two cases: FindTheRefusalLine requires the
+            // set bit to be named before " is set but " and the needed bit after it (F-v2-m1), so
+            // this call and the 0x9ff one above accept disjoint sentences. Espryt's is
+            // "kMGPipeSubsystemTextureResources (bit 10) is set but kMGPipeSubsystemSamplers
+            // (bit 11) is clear; MGPTextureParams::BuiltinSampler is a SamplerCso handle, only
+            // bit 11 mints sampler CSOs, and the applier's verdict for a null one is
+            // Fatal{ProtocolCorruption} - REFUSING the dependent bit and running the legacy arm.
+            // Set both bits, or clear both" (Managers.cpp, ResolveTextureResourceSubsystemArm).
+            const std::string refusal =
+                FindTheRefusalLine(log, TextureResourceBitSpellings(), SamplerBitSpellings());
+            EXPECT_FALSE(refusal.empty())
+                << "MOBILEGL_PIPE_PUSH=0x5ff sets the texture resource subsystem (bit 10) without "
+                   "the sampler subsystem (bit 11) that MGPTextureParams::BuiltinSampler depends on, "
+                   "and no single ERROR line of the library's log both says it REFUSED and names the "
+                   "two bits. Only bit 11 mints sampler CSOs, so every set_texture_params emitted "
+                   "under this mask would carry a null BuiltinSampler and the applier's Fatal is the "
+                   "next thing that happens - which is why this pair is a refusal at bring-up and "
+                   "not the harmless mirror of the 0x9ff one. Accepted spellings per bit are the "
+                   "constant's name, '(bit 10)' / '(bit 11)', and '0x400' / '0x800'. The log was "
+                << log.size() << " bytes and is at " << PipeStatsWindow::LibraryLogPath() << ".";
+            if (!refusal.empty()) {
+                std::cout << "[ ObjectSubsystemControl ] refusal line: " << refusal << std::endl;
+                RecordProperty("refusal_line", refusal.c_str());
+            }
+
+            EXPECT_TRUE(RegionIsMostly(image, kInset, image.Width() - kInset, kInset,
+                                       image.Height() - kInset, "green", 0.0,
+                                       "the sampled draw [refused-texture]"))
+                << "the refused configuration did not draw what every other lane draws. A refusal is "
+                   "supposed to run the LEGACY arm, which is the arm that ships in a pull build - so "
+                   "the pixels are the one thing it may not change.";
+
+            ReleaseTheWorkload();
+        }
+
+        // ------------------------------------------------------------------------------------
+        // c0f's GATE AND BELT, MEASURED TOGETHER (ID-39, ID-40).
+        //
+        // WHAT WENT WRONG AND WHY IT NEEDS A LANE. P4a's four families were wired without the
+        // gate P3a's buffers have had since PipeFill.cpp ~656: emission required the family's bit
+        // and nothing else. On Magma, which registers no MGPipeResourceOps, the client therefore
+        // emitted, THE APPLIER ACCEPTED, the client cleared its dirty flags on that acceptance -
+        // and Magma's legacy path then found nothing to upload. Sixty-six DirectVulkan cases went
+        // red at once, all texture-upload-shaped, and every one of them was green at 0x1ff. The
+        // fix has two halves that are deliberately independent: the client's gate (bit N AND
+        // wired AND a backend registered the ops) and the applier's belt (every P4a entry point
+        // returns accepted = false and counts RefusedNoConsumer when none did).
+        //
+        // THE ASSERTION IS THAT THE BELT NEVER FIRES, and it is the same assertion on both
+        // backends, which is what makes it worth having twice:
+        //
+        //   no-consumer (DirectVulkan, 0x1fff): the belt is the SAFETY NET. A non-zero count here
+        //     means a record reached the applier on a backend with no consumer - i.e. the client
+        //     gate leaked and only the belt stopped the dirty flag from being cleared. That is
+        //     ID-39's bug caught one layer later, and it is invisible in these pixels because the
+        //     belt does its job; the 66 red cases were in another suite entirely.
+        //   consumer (DirectGLES, 0x1fff): the CONTROL. Espryt registers the ops, so no entry
+        //     point may take the no-consumer arm at all. Without this lane a green above could
+        //     also mean "nothing is ever emitted anywhere", which is exactly what a gate that was
+        //     accidentally always-false would look like.
+        //
+        // A DELTA, not an absolute: the counter is process-global and other cases in this binary
+        // run before this one. MGPipeApplierReset also zeroes it, so a count that went DOWN is
+        // read as "the applier was reset and everything since is `after`" rather than as an
+        // underflow.
+        // ------------------------------------------------------------------------------------
+        TEST_F(ObjectSubsystemControlScenario, TheAppliersNoConsumerBeltNeverFiresBehindTheClientsGate) {
+            if (!Ready()) return;
+            // needsTheEmitters=false: the assertion is that a counter did NOT move, which is
+            // meaningful before the emitters land as well as after - and on the no-consumer lane
+            // it is meaningful precisely BECAUSE nothing may be emitted there.
+            SkipUnlessTheLaneIsAssertableHere(/*needsTheEmitters=*/false);
+            if (IsSkipped()) return;
+            if (!LaneIsAConsumerLane(m_lane)) {
+                GTEST_SKIP() << "runs only in the two consumer lanes (MGITEST_OBJECT_SUBSYSTEM_LANE="
+                             << kLaneConsumer << " / " << kLaneNoConsumer
+                             << "), which pin MOBILEGL_PIPE_PUSH at the phase default on the two "
+                                "backends. Every other lane configures a mask or a backend whose "
+                                "emission shape is a different question.";
+            }
+
+            unsigned long long before = 0;
+            if (!PeekPipeApplierRefusedNoConsumer(&before)) {
+                GTEST_SKIP() << "MGPipeApplierState::RefusedNoConsumer is out of reach here: there "
+                                "is no applier in a PULL build (it is #if MOBILEGL_PIPE_PUSH), and "
+                                "on Android this module links the shipping libMobileGL.so built "
+                                "-fvisibility=hidden. 'Could not look' is not 'did not fire'.";
+            }
+
+            BindDefaultFramebuffer();
+            ClearTo(0.0f, 0.0f, 0.0f, 1.0f);
+            RunTheWorkload();
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR))
+                << "the workload left a GL error behind on the " << m_lane << " lane";
+            const Image image = ReadPixels(Gl().Width(), Gl().Height());
+            Gl().EndFrame();
+
+            unsigned long long after = 0;
+            ASSERT_TRUE(PeekPipeApplierRefusedNoConsumer(&after))
+                << "the counter could be read before the workload and not after it";
+            // Down means MGPipeApplierReset ran inside the window, so everything still counted is
+            // what happened since - which is the number this case is about either way.
+            const unsigned long long fired = after >= before ? after - before : after;
+
+            std::cout << "[ ObjectSubsystemControl ] " << m_lane
+                      << " lane: applier RefusedNoConsumer " << before << " -> " << after
+                      << " over the workload (delta " << fired << ")" << std::endl;
+            RecordProperty("refused_no_consumer_delta", static_cast<int>(fired));
+
+            EXPECT_EQ(fired, 0u)
+                << "the applier's no-consumer BELT fired " << fired
+                << " time(s) during this workload on the " << m_lane
+                << " lane. The belt exists so that a P4a record arriving on a backend that "
+                   "registered no MGPipeResourceOps is refused rather than accepted - and an "
+                   "accepted record is what makes the client clear the dirty flags whose texels "
+                   "nobody then uploads (ID-39: sixty-six DirectVulkan cases, all texture-upload "
+                   "shaped). A non-zero count means the CLIENT'S GATE let an emission through and "
+                   "only the belt caught it: the two are supposed to agree, and PipeFill's "
+                   "P4aFamilyHasItsConsumer() is where they stopped.";
+
+            // The pixels, on both lanes, for the reason every arm of this file asserts them: a
+            // backend running its legacy path because no consumer is registered must draw exactly
+            // what a backend running the handle arm draws.
+            EXPECT_TRUE(RegionIsMostly(image, kInset, image.Width() - kInset, kInset,
+                                       image.Height() - kInset, "green", 0.0,
+                                       "the sampled draw [" + m_lane + "]"))
+                << "the " << m_lane
+                << " lane did not draw what every other lane draws, so whatever the counter says, "
+                   "this configuration is not running the workload correctly.";
+
+            ReleaseTheWorkload();
+        }
+
+    } // namespace
+} // namespace MGITest

MobileGL/MG_IntegrationTest/Scenarios/P4aFinalFixScenario.cpp

@@ -0,0 +1,658 @@
+// MobileGL - MobileGL/MG_IntegrationTest/Scenarios/P4aFinalFixScenario.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 - THE THREE FINDINGS OF THE P4a FINAL WHOLE-DIFF REVIEW (final-review-v1.md C-1, C-2,
+// M-A), each pinned by the public-GL sequence that was red on the tree the review read and is
+// green with its fix. Every sequence here is legal GL and none of the 80-odd scenarios before
+// this file drove it, which is how two criticals shipped through a green gate.
+//
+//   C-1  The client never passed the applier the LEVEL a respecify redefines, so every per-level
+//        glTexImage*D / glGenerateMipmap grow took the applier's whole-resource arm and dropped
+//        EVERY pending upload of the texture - including a level the applier had already
+//        accepted and whose client-side dirty flag was therefore already clear (D-D5 step 1).
+//        Nobody owed those texels any more. The window is "accepted but not yet consumed":
+//        a verb the texture is not reached by (a draw with another texture) drains the level
+//        into the applier, Espryt does not sync the texture, and the next level definition eats
+//        the entry. Two hazard cases (a level-1 definition, a glGenerateMipmap) read a black
+//        level 0 on the handle arm; the three controls beside them (no verb between, level 0
+//        consumed first, an immediate generate) are red on every arm, which is what pins the
+//        window rather than the mip path.
+//   C-2  A dead-but-not-recycled texture handle still resolved to the freed ITextureObject*
+//        inside the client's drain: the death helper freed the slot without telling the emitter,
+//        the drain list kept the level, and the next verb's drain called virtual
+//        GetStorageType() on freed memory - `glTexImage2D; glDeleteTextures; <any verb>` was a
+//        SIGABRT ("pure virtual method called") at the shipping default mask. The same
+//        delete-then-use shape is driven for every kind P4a mints (renderbuffer, sampler object,
+//        program, framebuffer) and for a slot recycled straight after the death (ABA), on both
+//        backends: the death path is backend-neutral by ruling (ID-8) and the DirectVulkan lane
+//        must see it too.
+//   M-A  Nothing produced kMGPipeBindSampler / kMGPipeBindShaderImage, so ImageBindableHint was
+//        dead: the applier never saw a texture become image-bound, the metadata respecify
+//        (ID-18 M4) had no live trigger, and the remint pull the hint exists to prevent was
+//        neither prevented nor counted. The case here reads the applier's record around a
+//        glBindImageTexture: the hint arrives as a metadata update that keeps the pending upload
+//        standing beside it, and the picture after the transition is the texels that upload
+//        carried.
+//
+// A WHITE-BOX READING THAT CANNOT BE TAKEN IS DECLINED BY NAME AND THE CASE CONTINUES with its
+// public-GL half (P4aSeamAuditScenario.cpp's shape): a pull build or a backend with no P4a
+// consumer holds no record to read, and skipping the whole case there would delete the verdict
+// those lanes carry. The C-1 and M-A cases assert their pictures on DirectGLES only - Espryt is
+// the one consumer of the texture records this phase wires, so on any other backend the handle
+// arm is inert by design and the picture proves nothing about it.
+
+#include <cstdint>
+#include <cstdlib>
+#include <iostream>
+#include <string>
+#include <vector>
+
+#include "../Harness/HeadlessGL.h"
+#include "../Harness/P4aFinalFixPeek.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 kInset = 2;
+
+        constexpr const char* kVS = R"(#version 330 core
+in vec2 aPos;
+out vec2 vUv;
+void main() {
+    vUv = aPos * 0.5 + 0.5;
+    gl_Position = vec4(aPos, 0.0, 1.0);
+}
+)";
+
+        constexpr const char* kFS = R"(#version 330 core
+in vec2 vUv;
+uniform sampler2D uTex;
+out vec4 oColor;
+void main() { oColor = texture(uTex, vUv); }
+)";
+
+        struct Vertex {
+            float x, y;
+        };
+
+        class P4aFinalFixScenario : public ScenarioTest {
+        protected:
+            void SetUp() override {
+                ScenarioTest::SetUp();
+                if (!Ready()) return;
+                std::string error;
+                m_program = CompileProgram(kVS, kFS, &error);
+                ASSERT_NE(m_program, 0u) << error;
+
+                static const Vertex quad[6] = {{-1.0f, -1.0f}, {1.0f, -1.0f}, {1.0f, 1.0f},
+                                               {-1.0f, -1.0f}, {1.0f, 1.0f},  {-1.0f, 1.0f}};
+                glGenBuffers(1, &m_quadBuffer);
+                glBindBuffer(GL_ARRAY_BUFFER, m_quadBuffer);
+                glBufferData(GL_ARRAY_BUFFER, sizeof(quad), quad, GL_STATIC_DRAW);
+                glGenVertexArrays(1, &m_vao);
+                glBindVertexArray(m_vao);
+                glEnableVertexAttribArray(0);
+                glVertexAttribPointer(0, 2, GL_FLOAT, GL_FALSE, sizeof(Vertex), nullptr);
+                glBindVertexArray(0);
+                glDisable(GL_BLEND);
+                glDisable(GL_SCISSOR_TEST);
+                glDisable(GL_DEPTH_TEST);
+                glColorMask(GL_TRUE, GL_TRUE, GL_TRUE, GL_TRUE);
+
+                // The "other" texture: a complete, single-level white texture, so a draw that
+                // samples it is a verb the texture under test is not reached by.
+                m_other = MakeLevel0(255, 255, 255, /*maxLevel=*/0);
+                while (glGetError() != GL_NO_ERROR) {
+                }
+            }
+
+            void TearDown() override {
+                if (!Ready()) return;
+                glUseProgram(0);
+                glBindVertexArray(0);
+                glBindTexture(GL_TEXTURE_2D, 0);
+                glBindBuffer(GL_ARRAY_BUFFER, 0);
+                glBindFramebuffer(GL_FRAMEBUFFER, 0);
+                if (m_other != 0) glDeleteTextures(1, &m_other);
+                if (m_vao != 0) glDeleteVertexArrays(1, &m_vao);
+                if (m_quadBuffer != 0) glDeleteBuffers(1, &m_quadBuffer);
+                if (m_program != 0) glDeleteProgram(m_program);
+                while (glGetError() != GL_NO_ERROR) {
+                }
+            }
+
+            // The C-1 and M-A pictures are about Espryt's consumption of the texture records;
+            // Magma registers no consumer for the P4a families (c0f), so the handle arm is inert
+            // there by design and a green picture proves nothing about the finding. Marks the
+            // case skipped; the caller tests IsSkipped() and returns.
+            void SkipUnlessEspryt(const char* what) {
+                if (Gl().BackendName() == "DirectGLES") return;
+                GTEST_SKIP() << what << " is consumed by DirectGLES only; backend is " << Gl().BackendName();
+            }
+
+            static std::vector<std::uint8_t> Solid(int size, std::uint8_t r, std::uint8_t g, std::uint8_t b) {
+                std::vector<std::uint8_t> texels(static_cast<std::size_t>(size) * size * 4);
+                for (std::size_t i = 0; i < texels.size(); i += 4) {
+                    texels[i] = r;
+                    texels[i + 1] = g;
+                    texels[i + 2] = b;
+                    texels[i + 3] = 255;
+                }
+                return texels;
+            }
+
+            // A 4x4 level 0 of one colour, NEAREST_MIPMAP_NEAREST with the level range clamped
+            // to `maxLevel`, so a single-level texture is complete and a chain is complete once
+            // its levels exist.
+            static GLuint MakeLevel0(std::uint8_t r, std::uint8_t g, std::uint8_t b, int maxLevel, int size = 4) {
+                const std::vector<std::uint8_t> texels = Solid(size, r, g, b);
+                GLuint texture = 0;
+                glGenTextures(1, &texture);
+                glBindTexture(GL_TEXTURE_2D, texture);
+                glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA8, size, size, 0, GL_RGBA, GL_UNSIGNED_BYTE, texels.data());
+                glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST_MIPMAP_NEAREST);
+                glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
+                glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_BASE_LEVEL, 0);
+                glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, maxLevel);
+                glBindTexture(GL_TEXTURE_2D, 0);
+                return texture;
+            }
+
+            static void DefineLevel1(GLuint texture, std::uint8_t r, std::uint8_t g, std::uint8_t b) {
+                const std::vector<std::uint8_t> texels = Solid(2, r, g, b);
+                glBindTexture(GL_TEXTURE_2D, texture);
+                glTexImage2D(GL_TEXTURE_2D, 1, GL_RGBA8, 2, 2, 0, GL_RGBA, GL_UNSIGNED_BYTE, texels.data());
+                glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, 1);
+                glBindTexture(GL_TEXTURE_2D, 0);
+            }
+
+            // A full-viewport draw sampling `texture` on unit 0 through `program` (the fixture's
+            // by default). The viewport is far larger than the 4x4 base level, so this is
+            // MAGNIFICATION and reads LEVEL 0 whatever the chain holds above it.
+            Image DrawSampled(GLuint texture, GLuint program = 0) {
+                if (program == 0) program = m_program;
+                BindDefaultFramebuffer();
+                glViewport(0, 0, Gl().Width(), Gl().Height());
+                glUseProgram(program);
+                glUniform1i(glGetUniformLocation(program, "uTex"), 0);
+                glActiveTexture(GL_TEXTURE0);
+                glBindTexture(GL_TEXTURE_2D, texture);
+                glBindVertexArray(m_vao);
+                glDrawArrays(GL_TRIANGLES, 0, 6);
+                Image image = ReadPixels(Gl().Width(), Gl().Height());
+                glBindTexture(GL_TEXTURE_2D, 0);
+                glBindVertexArray(0);
+                Gl().EndFrame();
+                return image;
+            }
+
+            ::testing::AssertionResult Mostly(const Image& image, const char* color, const std::string& when) {
+                return RegionIsMostly(image, kInset, image.Width() - kInset, kInset, image.Height() - kInset, color,
+                                      0.0, when);
+            }
+
+            void Report(const char* caseName, const Image& image) {
+                const char* mask = std::getenv("MOBILEGL_PIPE_PUSH");
+                const int cx = image.Width() / 2;
+                const int cy = image.Height() / 2;
+                std::cout << "[ P4aFinalFix ] case=" << caseName << " backend=" << Gl().BackendName()
+                          << " MOBILEGL_PIPE_PUSH=" << (mask ? mask : "(unset)") << " centre=" << image.At(cx, cy)
+                          << " (" << image.ColorName(cx, cy) << ")" << std::endl;
+            }
+
+            // The white-box gate of the M-A case: true when the applier holds a record for the
+            // texture in this process. Prints the decline.
+            bool RecordIsReadable(unsigned glTextureName, const char* what, PipeTextureResourceRecordPeek* out) {
+                if (PeekPipeTextureResourceRecord(glTextureName, out)) return true;
+                std::cout << "[ P4aFinalFix ] white-box reading DECLINED for " << what
+                          << ": the applier holds no record for texture " << glTextureName
+                          << " (a pull build, or a backend with no P4a consumer); the public-GL half of "
+                             "the case still runs"
+                          << std::endl;
+                RecordProperty("p4a_finalfix_white_box", "declined");
+                return false;
+            }
+
+            GLuint m_program = 0;
+            GLuint m_vao = 0;
+            GLuint m_quadBuffer = 0;
+            GLuint m_other = 0;
+        };
+
+        // ======================================================================================
+        // C-1: a per-level definition around a verb the texture is not reached by
+        // ======================================================================================
+
+        // THE HAZARD. L0's upload is accepted at the unrelated draw's validate point (the client
+        // clears its flag), Espryt never syncs T there (it is bound nowhere), then the level-1
+        // definition respecifies the resource. Before the fix that respecify carried no level and
+        // the applier dropped every pending upload; level 0 was allocated undefined.
+        TEST_F(P4aFinalFixScenario, PerLevelDefinitionAcrossAnUnrelatedDraw) {
+            if (!Ready()) return;
+            SkipUnlessEspryt("C-1's per-level respecify");
+            if (IsSkipped()) return;
+
+            const GLuint texture = MakeLevel0(255, 0, 0, /*maxLevel=*/0);
+            const Image unrelated = DrawSampled(m_other);
+            EXPECT_TRUE(Mostly(unrelated, "white", "the unrelated draw"));
+            DefineLevel1(texture, 255, 0, 0);
+            const Image image = DrawSampled(texture);
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+            Report("PerLevelDefinitionAcrossAnUnrelatedDraw", image);
+            EXPECT_TRUE(Mostly(image, "red",
+                               "level 0 after a level-1 definition that followed a draw the texture was not "
+                               "reached by - its accepted-but-unconsumed upload was dropped by the whole-"
+                               "resource arm"));
+            GLuint cleanup = texture;
+            glDeleteTextures(1, &cleanup);
+        }
+
+        // CONTROL: both levels defined before any verb; both are pending at the first sync.
+        TEST_F(P4aFinalFixScenario, ConsecutiveDefinitionsNoVerbBetween) {
+            if (!Ready()) return;
+            SkipUnlessEspryt("C-1's per-level respecify");
+            if (IsSkipped()) return;
+
+            const GLuint texture = MakeLevel0(255, 0, 0, /*maxLevel=*/0);
+            DefineLevel1(texture, 255, 0, 0);
+            const Image image = DrawSampled(texture);
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+            Report("ConsecutiveDefinitionsNoVerbBetween", image);
+            EXPECT_TRUE(Mostly(image, "red", "level 0 with both levels defined back to back"));
+            GLuint cleanup = texture;
+            glDeleteTextures(1, &cleanup);
+        }
+
+        // CONTROL: level 0 is consumed by Espryt (T is sampled) before level 1 is defined.
+        TEST_F(P4aFinalFixScenario, LevelZeroConsumedBeforeLevelOne) {
+            if (!Ready()) return;
+            SkipUnlessEspryt("C-1's per-level respecify");
+            if (IsSkipped()) return;
+
+            const GLuint texture = MakeLevel0(255, 0, 0, /*maxLevel=*/0);
+            const Image first = DrawSampled(texture);
+            EXPECT_TRUE(Mostly(first, "red", "level 0 alone"));
+            DefineLevel1(texture, 255, 0, 0);
+            const Image image = DrawSampled(texture);
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+            Report("LevelZeroConsumedBeforeLevelOne", image);
+            EXPECT_TRUE(Mostly(image, "red", "level 0 after level 1 was added to a synced texture"));
+            GLuint cleanup = texture;
+            glDeleteTextures(1, &cleanup);
+        }
+
+        // THE HAZARD, glGenerateMipmap flavour: the frontend grows the level chain (one
+        // AllocateStorage -> respecify per level) BEFORE the backend generate runs, with level 0
+        // accepted-but-unconsumed. The driver then built the chain from an undefined level 0.
+        TEST_F(P4aFinalFixScenario, GenerateMipmapAcrossAnUnrelatedDraw) {
+            if (!Ready()) return;
+            SkipUnlessEspryt("C-1's per-level respecify");
+            if (IsSkipped()) return;
+
+            const GLuint texture = MakeLevel0(255, 0, 0, /*maxLevel=*/1000);
+            const Image unrelated = DrawSampled(m_other);
+            EXPECT_TRUE(Mostly(unrelated, "white", "the unrelated draw"));
+            glBindTexture(GL_TEXTURE_2D, texture);
+            glGenerateMipmap(GL_TEXTURE_2D);
+            glBindTexture(GL_TEXTURE_2D, 0);
+            const Image image = DrawSampled(texture);
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+            Report("GenerateMipmapAcrossAnUnrelatedDraw", image);
+            EXPECT_TRUE(Mostly(image, "red",
+                               "level 0 after a glGenerateMipmap that followed a draw the texture was not "
+                               "reached by"));
+            GLuint cleanup = texture;
+            glDeleteTextures(1, &cleanup);
+        }
+
+        // CONTROL for the generate: no verb between the upload and the generate.
+        TEST_F(P4aFinalFixScenario, GenerateMipmapImmediately) {
+            if (!Ready()) return;
+            SkipUnlessEspryt("C-1's per-level respecify");
+            if (IsSkipped()) return;
+
+            const GLuint texture = MakeLevel0(255, 0, 0, /*maxLevel=*/1000);
+            glBindTexture(GL_TEXTURE_2D, texture);
+            glGenerateMipmap(GL_TEXTURE_2D);
+            glBindTexture(GL_TEXTURE_2D, 0);
+            const Image image = DrawSampled(texture);
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+            Report("GenerateMipmapImmediately", image);
+            EXPECT_TRUE(Mostly(image, "red", "level 0 after an immediate glGenerateMipmap"));
+            GLuint cleanup = texture;
+            glDeleteTextures(1, &cleanup);
+        }
+
+        // ======================================================================================
+        // C-2: delete-then-use, for every kind P4a mints, on both backends
+        // ======================================================================================
+
+        // A level goes dirty, the texture dies before any verb, and the next verb's drain walks
+        // the entry. Before the fix the emitter resolved the dead handle to the freed object and
+        // the drain called a virtual on it: SIGABRT in the first round. Eight rounds, and the
+        // lane registered with MALLOC_PERTURB_ scribbles every freed block so a resolved-but-
+        // dead pointer faults rather than reads the object's ghost.
+        TEST_F(P4aFinalFixScenario, ADirtyTextureDeletedBeforeAnyVerbIsWalkedByTheNextDrain) {
+            if (!Ready()) return;
+            for (int round = 0; round < 8; ++round) {
+                GLuint texture = 0;
+                glGenTextures(1, &texture);
+                glBindTexture(GL_TEXTURE_2D, texture);
+                const std::vector<std::uint8_t> texels = Solid(4, 255, 0, 0);
+                glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA8, 4, 4, 0, GL_RGBA, GL_UNSIGNED_BYTE, texels.data());
+                glBindTexture(GL_TEXTURE_2D, 0);
+                glDeleteTextures(1, &texture); // the last reference: the frontend object dies here
+                // Something else is allocated between the death and the drain, so the freed
+                // storage is not simply re-handed to the next object.
+                std::vector<std::uint8_t> churn(4096 + round * 1024, static_cast<std::uint8_t>(round));
+                (void)churn;
+                const Image image = DrawSampled(m_other); // the validate point: the drain runs here
+                ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+                if (round == 0) Report("ADirtyTextureDeletedBeforeAnyVerbIsWalkedByTheNextDrain", image);
+                EXPECT_TRUE(Mostly(image, "white", "the draw after a dirty texture died"));
+            }
+        }
+
+        // ABA: the slot the dead texture held is handed straight to the next texture (the free
+        // list is LIFO). The new texture's picture must be its own, and the dead one's drain
+        // entry must not be replayed onto it.
+        TEST_F(P4aFinalFixScenario, ATextureRecycledOntoTheDeadSlotDoesNotInheritItsDrainEntry) {
+            if (!Ready()) return;
+            {
+                GLuint dead = 0;
+                glGenTextures(1, &dead);
+                glBindTexture(GL_TEXTURE_2D, dead);
+                const std::vector<std::uint8_t> texels = Solid(4, 255, 0, 0);
+                glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA8, 4, 4, 0, GL_RGBA, GL_UNSIGNED_BYTE, texels.data());
+                glBindTexture(GL_TEXTURE_2D, 0);
+                glDeleteTextures(1, &dead); // dirty, dead, no verb between
+            }
+            const GLuint successor = MakeLevel0(0, 0, 255, /*maxLevel=*/0, /*size=*/8);
+            const Image image = DrawSampled(successor);
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+            Report("ATextureRecycledOntoTheDeadSlotDoesNotInheritItsDrainEntry", image);
+            EXPECT_TRUE(Mostly(image, "blue", "the successor of a dead dirty texture on the recycled slot"));
+            const Image other = DrawSampled(m_other);
+            EXPECT_TRUE(Mostly(other, "white", "an unrelated draw after the recycled slot was used"));
+            GLuint cleanup = successor;
+            glDeleteTextures(1, &cleanup);
+        }
+
+        // A renderbuffer with defined storage, attached, cleared through its framebuffer, then
+        // both die before the next verb.
+        TEST_F(P4aFinalFixScenario, ARenderbufferAndItsFramebufferDeletedAfterAClearLeaveTheNextDrawIntact) {
+            if (!Ready()) return;
+            GLuint renderbuffer = 0;
+            glGenRenderbuffers(1, &renderbuffer);
+            glBindRenderbuffer(GL_RENDERBUFFER, renderbuffer);
+            glRenderbufferStorage(GL_RENDERBUFFER, GL_RGBA8, 8, 8);
+            GLuint fbo = 0;
+            glGenFramebuffers(1, &fbo);
+            glBindFramebuffer(GL_FRAMEBUFFER, fbo);
+            glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_RENDERBUFFER, renderbuffer);
+            ASSERT_EQ(glCheckFramebufferStatus(GL_FRAMEBUFFER), GLenum(GL_FRAMEBUFFER_COMPLETE));
+            glViewport(0, 0, 8, 8);
+            glClearColor(0.0f, 1.0f, 0.0f, 1.0f);
+            glClear(GL_COLOR_BUFFER_BIT);
+            const Image cleared = ReadPixels(8, 8);
+            EXPECT_TRUE(RegionIsMostly(cleared, 0, 8, 0, 8, "green", 0.0, "the renderbuffer after the clear"));
+            glBindFramebuffer(GL_FRAMEBUFFER, 0);
+            glBindRenderbuffer(GL_RENDERBUFFER, 0);
+            glDeleteFramebuffers(1, &fbo);
+            glDeleteRenderbuffers(1, &renderbuffer); // the attachment's last reference went with the FBO
+            const Image image = DrawSampled(m_other);
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+            Report("ARenderbufferAndItsFramebufferDeletedAfterAClearLeaveTheNextDrawIntact", image);
+            EXPECT_TRUE(Mostly(image, "white", "the draw after a renderbuffer and its framebuffer died"));
+        }
+
+        // A sampler object bound to the unit the draw samples through, deleted while bound: GL
+        // unbinds it from every unit at glDeleteSamplers, and the texture's own parameters apply
+        // again. Both draws must be the texture's colour.
+        TEST_F(P4aFinalFixScenario, ASamplerObjectDeletedWhileBoundLeavesTheNextDrawIntact) {
+            if (!Ready()) return;
+            const GLuint texture = MakeLevel0(255, 0, 0, /*maxLevel=*/0);
+            GLuint sampler = 0;
+            glGenSamplers(1, &sampler);
+            glSamplerParameteri(sampler, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+            glSamplerParameteri(sampler, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
+            glBindSampler(0, sampler);
+            const Image withSampler = DrawSampled(texture);
+            EXPECT_TRUE(Mostly(withSampler, "red", "the draw through the bound sampler object"));
+            glDeleteSamplers(1, &sampler); // bound: unbound by the delete, then dies
+            const Image image = DrawSampled(texture);
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+            Report("ASamplerObjectDeletedWhileBoundLeavesTheNextDrawIntact", image);
+            EXPECT_TRUE(Mostly(image, "red", "the draw after the bound sampler object died"));
+            glBindSampler(0, 0);
+            GLuint cleanup = texture;
+            glDeleteTextures(1, &cleanup);
+        }
+
+        // A second program, in use when it is deleted (GL keeps it alive until it is no longer
+        // current), then released by a glUseProgram of the fixture's program: it dies there, and
+        // the draw that follows runs through the survivor.
+        TEST_F(P4aFinalFixScenario, AProgramDeletedWhileInUseLeavesTheNextDrawIntact) {
+            if (!Ready()) return;
+            std::string error;
+            const GLuint second = CompileProgram(kVS, kFS, &error);
+            ASSERT_NE(second, 0u) << error;
+            const GLuint texture = MakeLevel0(255, 0, 0, /*maxLevel=*/0);
+            const Image throughSecond = DrawSampled(texture, second);
+            EXPECT_TRUE(Mostly(throughSecond, "red", "the draw through the second program"));
+            glDeleteProgram(second); // current: flagged for deletion, still very much alive
+            const Image stillCurrent = DrawSampled(texture, second);
+            EXPECT_TRUE(Mostly(stillCurrent, "red", "the draw through a program flagged for deletion"));
+            const Image image = DrawSampled(texture); // glUseProgram(m_program): the second dies here
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+            Report("AProgramDeletedWhileInUseLeavesTheNextDrawIntact", image);
+            EXPECT_TRUE(Mostly(image, "red", "the draw after the deleted program was released"));
+            GLuint cleanup = texture;
+            glDeleteTextures(1, &cleanup);
+        }
+
+        // A framebuffer handed to the server BY NAME (a DSA clear emits a Named record, ID-19(c))
+        // and deleted before the next verb; its attachment lives on and carries the clear.
+        TEST_F(P4aFinalFixScenario, AFramebufferDeletedAfterADsaClearLeavesItsAttachmentIntact) {
+            if (!Ready()) return;
+            const GLuint texture = MakeLevel0(255, 0, 0, /*maxLevel=*/0);
+            GLuint fbo = 0;
+            glGenFramebuffers(1, &fbo);
+            glBindFramebuffer(GL_FRAMEBUFFER, fbo);
+            glFramebufferTexture2D(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, texture, 0);
+            ASSERT_EQ(glCheckFramebufferStatus(GL_FRAMEBUFFER), GLenum(GL_FRAMEBUFFER_COMPLETE));
+            glBindFramebuffer(GL_FRAMEBUFFER, 0);
+            const GLfloat green[4] = {0.0f, 1.0f, 0.0f, 1.0f};
+            glClearNamedFramebufferfv(fbo, GL_COLOR, 0, green);
+            glDeleteFramebuffers(1, &fbo); // unbound and named: dies here
+            const Image image = DrawSampled(texture);
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+            Report("AFramebufferDeletedAfterADsaClearLeavesItsAttachmentIntact", image);
+            EXPECT_TRUE(Mostly(image, "green", "the attachment of a framebuffer that died after a DSA clear"));
+            GLuint cleanup = texture;
+            glDeleteTextures(1, &cleanup);
+        }
+
+        // ======================================================================================
+        // M-A: an image bind after the allocation is a metadata respecify with the hint set
+        // ======================================================================================
+
+        // glTexStorage2D (immutable: no later respecify to ride), a red upload consumed by a draw,
+        // then a blue upload drained by a verb the texture is not reached by (accepted, standing
+        // in the applier's pending set), then glBindImageTexture. The bind must reach the record
+        // as a metadata update - ImageBindableHint 1, the pending upload still standing - and the
+        // draw after it must show the blue that upload carried through the widened carrier the
+        // hint schedules.
+        TEST_F(P4aFinalFixScenario, AnImageBindAfterAllocationReachesTheApplierAsAMetadataRespecify) {
+            if (!Ready()) return;
+            SkipUnlessEspryt("M-A's image-bindable hint");
+            if (IsSkipped()) return;
+            GLint maxImageUnits = 0;
+            glGetIntegerv(GL_MAX_IMAGE_UNITS, &maxImageUnits);
+            while (glGetError() != GL_NO_ERROR) {
+            }
+            if (maxImageUnits < 1) {
+                GTEST_SKIP() << "no image units";
+                return;
+            }
+
+            // THE NUMBER ROADMAP OPEN QUESTION 2 ASKS FOR: a texture Espryt allocated BEFORE the
+            // hint reached it is re-minted image-bindable at the bind and its levels replayed
+            // from the client's shadow - one remint pull, counted. Arming the counter here is
+            // what makes it readable without a stats-enabled lane.
+            unsigned long long pullsBefore = 0;
+            const bool pullsReadable = PeekPipeStatsTextureRemintPulls(&pullsBefore);
+
+            GLuint texture = 0;
+            glGenTextures(1, &texture);
+            glBindTexture(GL_TEXTURE_2D, texture);
+            glTexStorage2D(GL_TEXTURE_2D, 1, GL_RGBA8, 4, 4);
+            const std::vector<std::uint8_t> red = Solid(4, 255, 0, 0);
+            glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, 4, 4, GL_RGBA, GL_UNSIGNED_BYTE, red.data());
+            glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+            glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
+            glBindTexture(GL_TEXTURE_2D, 0);
+            const Image before = DrawSampled(texture); // allocated and consumed, NOT image-bindable
+            EXPECT_TRUE(Mostly(before, "red", "the immutable texture before the image bind"));
+
+            PipeTextureResourceRecordPeek record{};
+            const bool readable = RecordIsReadable(texture, "M-A's image-bindable hint", &record);
+            if (readable) {
+                EXPECT_EQ(record.ImageBindableHint, 0u) << "nothing has image-bound this texture yet";
+                EXPECT_EQ(record.PendingUploads, 0u) << "the red upload was consumed by the draw";
+            }
+
+            // A blue upload, drained by a verb that does not reach T: accepted, unconsumed.
+            const std::vector<std::uint8_t> blue = Solid(4, 0, 0, 255);
+            glBindTexture(GL_TEXTURE_2D, texture);
+            glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, 4, 4, GL_RGBA, GL_UNSIGNED_BYTE, blue.data());
+            glBindTexture(GL_TEXTURE_2D, 0);
+            const Image unrelated = DrawSampled(m_other);
+            EXPECT_TRUE(Mostly(unrelated, "white", "the unrelated draw"));
+            if (readable) {
+                ASSERT_TRUE(PeekPipeTextureResourceRecord(texture, &record));
+                EXPECT_EQ(record.PendingUploads, 1u) << "the blue upload was not drained into the applier";
+            }
+            const unsigned long long serialBeforeBind = record.Serial;
+            unsigned long long uploadsBeforeBind = 0;
+            const bool uploadsReadable = PeekPipeStatsTextureUploadEmissions(&uploadsBeforeBind);
+
+            // THE TRANSITION. An immutable texture has no storage-defining respecify left, so the
+            // hint can only arrive as a metadata update (ID-18 M4). Espryt syncs the texture
+            // eagerly inside glBindImageTexture and the widening re-mints its storage, replaying
+            // every defined level from the shadow (the remint pull the counter below counts), so
+            // the standing upload is consumed by that regeneration here and the picture that
+            // follows is blue whatever the metadata respecify did to the record - the KEPT
+            // property is proved further down, on a texture no remint stands in front of.
+            (void)uploadsBeforeBind;
+            (void)uploadsReadable;
+            glBindImageTexture(0, texture, 0, GL_FALSE, 0, GL_READ_ONLY, GL_RGBA8);
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+            if (readable) {
+                ASSERT_TRUE(PeekPipeTextureResourceRecord(texture, &record));
+                EXPECT_EQ(record.ImageBindableHint, 1u)
+                    << "glBindImageTexture did not reach the applier's record as ImageBindableHint";
+                EXPECT_NE(record.BindMask & (1u << 6), 0u) << "kMGPipeBindShaderImage was not produced";
+                EXPECT_GT(record.Serial, serialBeforeBind) << "the metadata respecify moved no serial";
+            }
+
+            const Image image = DrawSampled(texture);
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+            Report("AnImageBindAfterAllocationReachesTheApplierAsAMetadataRespecify", image);
+            EXPECT_TRUE(Mostly(image, "blue", "the texture after the image bind that followed an unconsumed upload"));
+            glBindImageTexture(0, 0, 0, GL_FALSE, 0, GL_READ_ONLY, GL_RGBA8);
+            unsigned long long pullsAfter = 0;
+            if (pullsReadable && readable && PeekPipeStatsTextureRemintPulls(&pullsAfter)) {
+                EXPECT_EQ(pullsAfter, pullsBefore + 1)
+                    << "the re-mint of a texture allocated before its hint was not counted as a remint pull "
+                       "(trp= on the stats line is ROADMAP open question 2's number)";
+            }
+
+            // THE PREVENTION HALF, measured the other way round: a texture whose hint arrives at
+            // the bind, BEFORE its first sync, is allocated image-bindable up front and pulls
+            // nothing - the counter does not move.
+            GLuint early = 0;
+            glGenTextures(1, &early);
+            glBindTexture(GL_TEXTURE_2D, early);
+            glTexStorage2D(GL_TEXTURE_2D, 1, GL_RGBA8, 4, 4);
+            glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, 4, 4, GL_RGBA, GL_UNSIGNED_BYTE, red.data());
+            glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+            glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
+            glBindTexture(GL_TEXTURE_2D, 0);
+            glBindImageTexture(0, early, 0, GL_FALSE, 0, GL_READ_ONLY, GL_RGBA8); // before any sync
+            const Image earlyImage = DrawSampled(early);
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+            EXPECT_TRUE(Mostly(earlyImage, "red", "a texture image-bound before its first sync"));
+            glBindImageTexture(0, 0, 0, GL_FALSE, 0, GL_READ_ONLY, GL_RGBA8);
+            unsigned long long pullsEarly = 0;
+            if (pullsReadable && readable && PeekPipeStatsTextureRemintPulls(&pullsEarly)) {
+                EXPECT_EQ(pullsEarly, pullsAfter)
+                    << "a texture whose hint preceded its first sync was still re-minted (the prevention "
+                       "half of the hint did not fire)";
+            }
+
+            // THE METADATA RESPECIFY KEEPS A STANDING UPLOAD, end to end and with no remint in the
+            // way: `early` is image-bindable already, so a NEW sticky bit reaching it - the
+            // RENDER_TARGET bit a DSA attachment produces at its setter (a Named record, ID-19(c)),
+            // with no sync of the texture in between - is a pure metadata update. The blue upload
+            // drained before it must still stand in the record afterwards (or, if a sync did run,
+            // have been uploaded rather than dropped) and reach the driver at the next draw.
+            glBindTexture(GL_TEXTURE_2D, early);
+            glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, 4, 4, GL_RGBA, GL_UNSIGNED_BYTE, blue.data());
+            glBindTexture(GL_TEXTURE_2D, 0);
+            const Image unrelatedAgain = DrawSampled(m_other);
+            EXPECT_TRUE(Mostly(unrelatedAgain, "white", "the unrelated draw"));
+            PipeTextureResourceRecordPeek earlyRecord{};
+            const bool earlyReadable = PeekPipeTextureResourceRecord(early, &earlyRecord);
+            if (earlyReadable) {
+                EXPECT_EQ(earlyRecord.PendingUploads, 1u) << "the blue upload was not drained into the applier";
+            }
+            const unsigned long long earlySerialBefore = earlyRecord.Serial;
+            unsigned long long uploadsBeforeAttach = 0;
+            const bool uploadsCounted = PeekPipeStatsTextureUploadEmissions(&uploadsBeforeAttach);
+            GLuint namedFbo = 0;
+            glCreateFramebuffers(1, &namedFbo);
+            glNamedFramebufferTexture(namedFbo, GL_COLOR_ATTACHMENT0, early, 0);
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+            if (earlyReadable) {
+                ASSERT_TRUE(PeekPipeTextureResourceRecord(early, &earlyRecord));
+                EXPECT_NE(earlyRecord.BindMask & (1u << 7), 0u)
+                    << "the DSA attachment did not produce kMGPipeBindRenderTarget";
+                EXPECT_GT(earlyRecord.Serial, earlySerialBefore) << "the mask move reached the record as no respecify";
+                unsigned long long uploadsAfterAttach = 0;
+                if (earlyRecord.PendingUploads == 0 && uploadsCounted &&
+                    PeekPipeStatsTextureUploadEmissions(&uploadsAfterAttach)) {
+                    EXPECT_GT(uploadsAfterAttach, uploadsBeforeAttach)
+                        << "the standing upload vanished from the record without Espryt uploading anything: "
+                           "the metadata respecify dropped it";
+                } else {
+                    EXPECT_EQ(earlyRecord.PendingUploads, 1u)
+                        << "the metadata respecify dropped the pending upload standing beside it";
+                }
+            }
+            const Image earlyAfter = DrawSampled(early);
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR));
+            EXPECT_TRUE(Mostly(earlyAfter, "blue", "the upload that stood across a metadata respecify"));
+            glDeleteFramebuffers(1, &namedFbo);
+            GLuint cleanup = texture;
+            glDeleteTextures(1, &cleanup);
+            GLuint cleanupEarly = early;
+            glDeleteTextures(1, &cleanupEarly);
+        }
+
+    } // namespace
+} // namespace MGITest

MobileGL/MG_IntegrationTest/Scenarios/P4aSeamAuditScenario.cpp

@@ -0,0 +1,803 @@
+// MobileGL - MobileGL/MG_IntegrationTest/Scenarios/P4aSeamAuditScenario.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 - THE FOUR SEAMS THE P4a FABLE SEAM AUDIT PROVED, each pinned by the public-GL sequence
+// (or the white-box reading) that was red on the tree the audit read and is green with its fix.
+//
+// The audit's rule, which every case here is an instance of: EVERY FIELD OF EVERY EMITTED RECORD
+// NAMES THE FRONTEND SETTER THAT CHANGES IT, AND THAT SETTER MOVES A COUNTER THE EMITTING BIT'S
+// SHUTTER READS. A record whose field has a setter no shutter sees is a stale record with nothing
+// to refuse - no census line, no Fatal, a wrong picture or a permanent silent fallback - which is
+// why none of the 80 scenarios before this file caught any of the four (Tracker.h carries the
+// record-field -> setter -> shutter table this file is the gate for).
+//
+//   F-3  set_framebuffer_state INLINES an attachment's format (D-C1) and a storage redefinition of
+//        an ATTACHED texture or renderbuffer moved nothing bit 11 read: Espryt's handle arm then
+//        answered its alpha-widening / snorm-clamp / integer masks from the stale copy. AND THE
+//        PRE-HANDLE ARM WAS NOT FRESH EITHER, which these cases found on the 0x1ff / 0 / pull
+//        lanes: a redefinition that keeps the driver id (mutable texture storage regenerated in
+//        place, a renderbuffer re-storaged in place) moves neither the framebuffer's frontend
+//        versions nor the backend-id generation the FBO memo reads, so SyncToBackend never
+//        re-ran and the masks stayed on both arms. The texture half is fixed on both arms OF A
+//        PUSH BUILD (an in-place regeneration now takes the same generation a re-mint takes -
+//        compiled under MOBILEGL_PIPE_PUSH because G1 keeps the pull library byte-identical to
+//        the P4a baseline, so the pull build keeps the pre-P4a hole until the fix lands on dev on
+//        its own and the texture cases decline by name there); the renderbuffer half only on the
+//        handle arm, where the resource record carries the re-storage - on the pre-handle arm a
+//        renderbuffer's twin is only ever reached from inside the FBO walk the memo skips (D-D2's
+//        documented hole, pre-P4a code), so that case asserts on the handle arm and declines by
+//        name elsewhere. Three cases, both directions, texture and renderbuffer. DirectGLES only:
+//        the masks are Espryt's substitution machinery.
+//   F-1  set_sampler_views is resolved for the PROGRAM IN USE and bit 12's shutter read no program
+//        input, so a glUseProgram alone never re-emitted it; E's record epoch (the two set serials)
+//        then kept the program-independent texture sync list from ever rebuilding, and a texture
+//        bound to an EMPTY slot under one program was never synced for the next. One case, both
+//        backends, red as a black quad.
+//   F-2  bit 14's plain-program arm mixed a per-program COUNTER two programs routinely share, so a
+//        program switch never re-emitted set_shader_images and the window stayed the previous
+//        program's - and E's SD-4 (a buffer image never reaching the record at all) is the same
+//        bit through the null -> program transition. One case, white-box, both backends run it.
+//   F-4  BindCurrentUnitSamplers' record arm looked a CONTENT-addressed CSO handle up in the
+//        IDENTITY-keyed twin registry: a miss on every draw, hidden because the pre-handle program
+//        pass bound the same values. One case, white-box: the unit's driver sampler must be the
+//        CSO's own twin.
+//
+// A WHITE-BOX READING THAT CANNOT BE TAKEN IS DECLINED BY NAME AND THE CASE CONTINUES with its
+// public-GL half (the shape TextureParamsWithoutASamplerViewScenario.cpp argues for): a pull
+// build, Magma, or a lane whose mask leaves Espryt's sampler family on its legacy arm has no
+// record arm to assert about, and skipping the whole case there would delete the public-GL
+// verdict those lanes carry. Every decline is printed and RecordProperty'd.
+
+#include <cstdint>
+#include <cstring>
+#include <iostream>
+#include <string>
+#include <vector>
+
+#include "../Harness/HeadlessGL.h"
+#include "../Harness/P4aSeamPeek.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 kSize = 16;
+        constexpr int kInset = 2;
+
+        // No attributes: the quad's corners come from gl_VertexID, so a bare VAO is all a draw
+        // needs and no vertex-input state can enter any of the sequences below.
+        constexpr const char* kQuadVS = R"(#version 330 core
+void main() {
+    vec2 corner = vec2((gl_VertexID & 1) == 0 ? -1.0 : 1.0,
+                       (gl_VertexID & 2) == 0 ? -1.0 : 1.0);
+    gl_Position = vec4(corner, 0.0, 1.0);
+}
+)";
+
+        constexpr const char* kColorFS = R"(#version 330 core
+uniform vec4 uColor;
+out vec4 oColor;
+void main() { oColor = uColor; }
+)";
+
+        // texelFetch, so WHICH image the unit holds is the whole answer and no filter, wrap or
+        // completeness rule can explain a colour away.
+        constexpr const char* kFetchFS = R"(#version 330 core
+uniform sampler2D uTex;
+out vec4 oColor;
+void main() { oColor = texelFetch(uTex, ivec2(0, 0), 0); }
+)";
+
+        // texture() at (1.5, 1.5): outside the image on both axes, so the WRAP mode of whichever
+        // sampler applies - the unit's sampler object or the texture's built-in one - decides
+        // whether the texel or the border colour comes back.
+        constexpr const char* kOutsideSampleFS = R"(#version 330 core
+uniform sampler2D uTex;
+out vec4 oColor;
+void main() { oColor = texture(uTex, vec2(1.5, 1.5)); }
+)";
+
+        // F-2 / SD-4: two compute programs over BUFFER images (the SD-4 shape - the kind E's I2
+        // flip found never reached the record at all), the second naming one unit more than the
+        // first, and both image-unit counters equal (layout(binding) assigns the unit at link, so
+        // neither program ever moves it through glUniform1i).
+        constexpr const char* kOneBufferImageCS = R"(#version 430 core
+layout(local_size_x = 1) in;
+layout(binding = 0, r32ui) writeonly uniform uimageBuffer i0;
+void main() { imageStore(i0, 0, uvec4(7u, 0u, 0u, 0u)); }
+)";
+
+        constexpr const char* kTwoBufferImagesCS = R"(#version 430 core
+layout(local_size_x = 1) in;
+layout(binding = 0, r32ui) readonly uniform uimageBuffer i0;
+layout(binding = 1, r32ui) writeonly uniform uimageBuffer i1;
+void main() { imageStore(i1, 0, imageLoad(i0, 0) + uvec4(2u, 0u, 0u, 0u)); }
+)";
+
+        class P4aSeamAuditScenario : public ScenarioTest {
+        protected:
+            void SetUp() override {
+                ScenarioTest::SetUp();
+                if (!Ready()) return;
+                glGenVertexArrays(1, &m_vao);
+                glBindVertexArray(m_vao);
+                glDisable(GL_BLEND);
+                glDisable(GL_SCISSOR_TEST);
+                glDisable(GL_DEPTH_TEST);
+                glColorMask(GL_TRUE, GL_TRUE, GL_TRUE, GL_TRUE);
+                while (glGetError() != GL_NO_ERROR) {
+                }
+            }
+
+            void TearDown() override {
+                if (!Ready()) return;
+                glUseProgram(0);
+                glBindVertexArray(0);
+                if (m_vao != 0) glDeleteVertexArrays(1, &m_vao);
+                glBindFramebuffer(GL_FRAMEBUFFER, 0);
+                while (glGetError() != GL_NO_ERROR) {
+                }
+            }
+
+            // The F-3 cases are about Espryt's four cross-object masks, which are its own
+            // substitution machinery (three-channel widening, SNORM/UNORM clamp, integer outputs);
+            // Magma answers the same GL questions on its own terms, so a verdict there would pin
+            // a coincidence - the same reason SnormAttachment and ThreeChannelAttachment skip.
+            // Marks the case skipped; the caller tests IsSkipped() and returns (GTEST_SKIP is a
+            // void statement, so it cannot return the verdict itself).
+            void SkipUnlessEspryt(const char* what) {
+                if (Gl().BackendName() == "DirectGLES") return;
+                GTEST_SKIP() << what << " is a DirectGLES handle-arm seam; backend is " << Gl().BackendName();
+            }
+
+            static void DrawQuad() { glDrawArrays(GL_TRIANGLE_STRIP, 0, 4); }
+
+            // One pixel's RGBA as floats, from the currently bound READ framebuffer.
+            static void ReadPixelFloat(int x, int y, float out[4]) {
+                out[0] = out[1] = out[2] = out[3] = -1.0f;
+                glReadPixels(x, y, 1, 1, GL_RGBA, GL_FLOAT, out);
+            }
+
+            // The white-box gate shared by F-2 and F-4: true when Espryt's sampler family is on
+            // its handle arm in this process, so the applier's unit sets are consumed and an
+            // assertion about them can only be red for its own reason. Prints the decline.
+            bool SamplerHandleArmIsLive(const char* what) {
+                bool live = false;
+                std::string why;
+                if (!PeekEsprytSamplerHandleArmIsLive(&live)) {
+                    why = "the reading cannot be taken here (a pull build, Android, or a backend that "
+                          "is not DirectGLES)";
+                } else if (!live) {
+                    why = "Espryt's sampler family runs its legacy arm in this process "
+                          "(MOBILEGL_PIPE_PUSH leaves bit 11 clear or refuses it)";
+                }
+                if (why.empty()) return true;
+                std::cout << "[ P4aSeamAudit ] white-box reading DECLINED for " << what << ": " << why
+                          << "; the public-GL half of the case still runs" << std::endl;
+                RecordProperty("p4a_seam_white_box", "declined");
+                RecordProperty("p4a_seam_white_box_reason", why);
+                return false;
+            }
+
+            // A 2x2 RGBA8 texture filled with one colour, NEAREST, single level - complete under
+            // every rule, so nothing about completeness can enter the F-1 and F-4 sequences.
+            static GLuint MakeSolidTexture2D(std::uint8_t r, std::uint8_t g, std::uint8_t b) {
+                std::uint8_t texels[2 * 2 * 4];
+                for (int i = 0; i < 4; ++i) {
+                    texels[i * 4 + 0] = r;
+                    texels[i * 4 + 1] = g;
+                    texels[i * 4 + 2] = b;
+                    texels[i * 4 + 3] = 255;
+                }
+                GLuint texture = 0;
+                glGenTextures(1, &texture);
+                glBindTexture(GL_TEXTURE_2D, texture);
+                glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA8, 2, 2, 0, GL_RGBA, GL_UNSIGNED_BYTE, texels);
+                glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+                glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
+                return texture;
+            }
+
+            bool ComputeImagesAreUsable() const {
+                GLint maxImageUnits = 0;
+                glGetIntegerv(GL_MAX_IMAGE_UNITS, &maxImageUnits);
+                GLint maxComputeImageUniforms = 0;
+                glGetIntegerv(GL_MAX_COMPUTE_IMAGE_UNIFORMS, &maxComputeImageUniforms);
+                GLint maxBufferSize = 0;
+                glGetIntegerv(GL_MAX_TEXTURE_BUFFER_SIZE, &maxBufferSize);
+                while (glGetError() != GL_NO_ERROR) {
+                }
+                return maxImageUnits >= 2 && maxComputeImageUniforms >= 2 && maxBufferSize >= 4;
+            }
+
+            static GLuint MakeComputeProgram(const char* source, std::string* outError) {
+                const GLuint shader = glCreateShader(GL_COMPUTE_SHADER);
+                glShaderSource(shader, 1, &source, nullptr);
+                glCompileShader(shader);
+                GLint compiled = GL_FALSE;
+                glGetShaderiv(shader, GL_COMPILE_STATUS, &compiled);
+                if (compiled == GL_FALSE) {
+                    char log[2048] = {};
+                    glGetShaderInfoLog(shader, sizeof(log) - 1, nullptr, log);
+                    *outError = 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 = GL_FALSE;
+                glGetProgramiv(program, GL_LINK_STATUS, &linked);
+                if (linked == GL_FALSE) {
+                    char log[2048] = {};
+                    glGetProgramInfoLog(program, sizeof(log) - 1, nullptr, log);
+                    *outError = std::string("compute program did not link: ") + log;
+                    glDeleteProgram(program);
+                    return 0;
+                }
+                return program;
+            }
+
+            // An R32UI buffer texture over a fresh 4-texel buffer, every texel `fill`.
+            static GLuint MakeBufferTexture(GLuint* outBuffer, GLuint fill) {
+                const GLuint texels[4] = {fill, fill, fill, fill};
+                glGenBuffers(1, outBuffer);
+                glBindBuffer(GL_TEXTURE_BUFFER, *outBuffer);
+                glBufferData(GL_TEXTURE_BUFFER, sizeof(texels), texels, GL_DYNAMIC_COPY);
+                GLuint texture = 0;
+                glGenTextures(1, &texture);
+                glBindTexture(GL_TEXTURE_BUFFER, texture);
+                glTexBuffer(GL_TEXTURE_BUFFER, GL_R32UI, *outBuffer);
+                return texture;
+            }
+
+            static GLuint ReadBufferTexel0(GLuint buffer) {
+                GLuint value = 0xFFFFFFFFu;
+                glBindBuffer(GL_TEXTURE_BUFFER, buffer);
+                glGetBufferSubData(GL_TEXTURE_BUFFER, 0, sizeof(value), &value);
+                return value;
+            }
+
+            GLuint m_vao = 0;
+        };
+
+        // -----------------------------------------------------------------------------------
+        // F-3: a storage redefinition WHILE ATTACHED reaches the framebuffer record
+        // -----------------------------------------------------------------------------------
+        //
+        // GL_SRGB8 is a format Espryt can only render into through its three-channel widening
+        // (llvmpipe reports INCOMPLETE_ATTACHMENT for it natively - ThreeChannelAttachmentScenario
+        // measured the table), so its draw buffer carries the alpha-widened mask: every draw has its
+        // alpha masked off so the stored alpha stays at the 1.0 a three-channel format implies.
+        // Redefine the same attached texture as GL_SRGB8_ALPHA8 and the application owns alpha
+        // again - the mask must clear. On the tree the audit read the record still said SRGB8, the
+        // handle arm kept masking, and the 0.25 this case draws never reached the storage; on the
+        // pre-handle arm the twin regenerated the (mutable) storage on the same driver id, nothing
+        // the FBO memo reads moved, and the masks stayed the same way.
+
+        TEST_F(P4aSeamAuditScenario, ATextureRespecifiedWhileAttachedReachesTheFramebufferRecord) {
+            if (!Ready()) return;
+            SkipUnlessEspryt("F-3");
+            if (IsSkipped()) return;
+
+            std::string error;
+            const GLuint program = CompileProgram(kQuadVS, kColorFS, &error);
+            ASSERT_NE(program, 0u) << error;
+            const GLint colorLocation = glGetUniformLocation(program, "uColor");
+            ASSERT_GE(colorLocation, 0);
+
+            GLuint texture = 0;
+            glGenTextures(1, &texture);
+            glBindTexture(GL_TEXTURE_2D, texture);
+            glTexImage2D(GL_TEXTURE_2D, 0, GL_SRGB8, kSize, kSize, 0, GL_RGB, GL_UNSIGNED_BYTE, nullptr);
+            glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+            glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
+            ASSERT_EQ(FirstGLError(), 0u) << "the SRGB8 texture was refused";
+
+            GLuint fbo = 0;
+            glGenFramebuffers(1, &fbo);
+            glBindFramebuffer(GL_FRAMEBUFFER, fbo);
+            glFramebufferTexture2D(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, texture, 0);
+            ASSERT_EQ(glCheckFramebufferStatus(GL_FRAMEBUFFER), static_cast<GLenum>(GL_FRAMEBUFFER_COMPLETE))
+                << "an SRGB8 colour attachment must be complete (natively or through the widening)";
+            glViewport(0, 0, kSize, kSize);
+            glUseProgram(program);
+            glUniform4f(colorLocation, 0.0f, 1.0f, 0.0f, 0.25f);
+
+            // Phase 1: the three-channel format. Whatever the draw writes, GL reports alpha 1.0.
+            glClearColor(0.0f, 0.0f, 0.0f, 0.0f);
+            glClear(GL_COLOR_BUFFER_BIT);
+            DrawQuad();
+            float pixel[4];
+            glReadBuffer(GL_COLOR_ATTACHMENT0);
+            ReadPixelFloat(kSize / 2, kSize / 2, pixel);
+            EXPECT_NEAR(pixel[3], 1.0f, 0.02f) << "a three-channel attachment reports alpha 1.0";
+
+            // Phase 2: THE RESPECIFY, while attached, with no re-attach and no rebind of the FBO.
+            // The only thing that moves between the two draws is the texture's storage.
+            glBindTexture(GL_TEXTURE_2D, texture);
+            glTexImage2D(GL_TEXTURE_2D, 0, GL_SRGB8_ALPHA8, kSize, kSize, 0, GL_RGBA, GL_UNSIGNED_BYTE, nullptr);
+            ASSERT_EQ(FirstGLError(), 0u) << "the respecify to SRGB8_ALPHA8 was refused";
+            ASSERT_EQ(glCheckFramebufferStatus(GL_FRAMEBUFFER), static_cast<GLenum>(GL_FRAMEBUFFER_COMPLETE));
+
+            glClearColor(0.0f, 0.0f, 0.0f, 0.0f);
+            glClear(GL_COLOR_BUFFER_BIT);
+            DrawQuad();
+            ReadPixelFloat(kSize / 2, kSize / 2, pixel);
+            EXPECT_NEAR(pixel[1], 1.0f, 0.05f) << "the draw did not land at all";
+            // PUSH BUILDS ONLY, EVERY ARM OF THEM. The pre-handle half of the fix (an in-place
+            // regeneration takes the backend-id generation a re-mint takes) is Espryt code the
+            // pull build would share, and G1 keeps the pull library byte-identical to the P4a
+            // baseline - so it is compiled under MOBILEGL_PIPE_PUSH and the pull build keeps the
+            // pre-P4a hole until the same lines land on dev on their own. The peek returns true
+            // exactly where it could look, which for a case that already skipped off Espryt means
+            // "a push build"; what it writes (is the handle arm live) does not matter here.
+            bool framebufferArmLive = false;
+            if (PeekEsprytFramebufferHandleArmIsLive(&framebufferArmLive)) {
+                EXPECT_NEAR(pixel[3], 0.25f, 0.02f)
+                    << "the draw's alpha never reached a four-channel attachment: the framebuffer record "
+                       "(handle arm) or the FBO twin's memo (pre-handle arm) still describes the "
+                       "three-channel storage the texture was attached with, so alpha stayed masked off (F-3)";
+            } else {
+                std::cout << "[ P4aSeamAudit ] texture respecify verdict DECLINED on the pull build (the "
+                             "in-place regeneration bump is push-only by G1); alpha read "
+                          << pixel[3] << std::endl;
+                RecordProperty("p4a_seam_white_box", "declined");
+                RecordProperty("p4a_seam_white_box_reason", "texture respecify: pull build (G1)");
+            }
+
+            glBindFramebuffer(GL_FRAMEBUFFER, 0);
+            glDeleteFramebuffers(1, &fbo);
+            glDeleteTextures(1, &texture);
+            glDeleteProgram(program);
+            EXPECT_EQ(FirstGLError(), 0u) << GLErrorName(FirstGLError());
+        }
+
+        // The renderbuffer twin. A renderbuffer's three storage setters bump no version at all:
+        // D-D2 closed the RESOURCE record by emitting from the entry point and left the framebuffer
+        // record - and with it the masks - describing the storage it was attached with.
+        TEST_F(P4aSeamAuditScenario, ARenderbufferRestoragedWhileAttachedReachesTheFramebufferRecord) {
+            if (!Ready()) return;
+            SkipUnlessEspryt("F-3");
+            if (IsSkipped()) return;
+
+            std::string error;
+            const GLuint program = CompileProgram(kQuadVS, kColorFS, &error);
+            ASSERT_NE(program, 0u) << error;
+            const GLint colorLocation = glGetUniformLocation(program, "uColor");
+            ASSERT_GE(colorLocation, 0);
+
+            GLuint renderbuffer = 0;
+            glGenRenderbuffers(1, &renderbuffer);
+            glBindRenderbuffer(GL_RENDERBUFFER, renderbuffer);
+            glRenderbufferStorage(GL_RENDERBUFFER, GL_SRGB8, kSize, kSize);
+            ASSERT_EQ(FirstGLError(), 0u) << "the SRGB8 renderbuffer was refused";
+
+            GLuint fbo = 0;
+            glGenFramebuffers(1, &fbo);
+            glBindFramebuffer(GL_FRAMEBUFFER, fbo);
+            glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_RENDERBUFFER, renderbuffer);
+            ASSERT_EQ(glCheckFramebufferStatus(GL_FRAMEBUFFER), static_cast<GLenum>(GL_FRAMEBUFFER_COMPLETE))
+                << "an SRGB8 renderbuffer attachment must be complete (natively or through the widening)";
+            glViewport(0, 0, kSize, kSize);
+            glUseProgram(program);
+            glUniform4f(colorLocation, 0.0f, 1.0f, 0.0f, 0.25f);
+
+            glClearColor(0.0f, 0.0f, 0.0f, 0.0f);
+            glClear(GL_COLOR_BUFFER_BIT);
+            DrawQuad();
+            float pixel[4];
+            glReadBuffer(GL_COLOR_ATTACHMENT0);
+            ReadPixelFloat(kSize / 2, kSize / 2, pixel);
+            EXPECT_NEAR(pixel[3], 1.0f, 0.02f) << "a three-channel attachment reports alpha 1.0";
+
+            // THE RE-STORAGE, while attached.
+            glBindRenderbuffer(GL_RENDERBUFFER, renderbuffer);
+            glRenderbufferStorage(GL_RENDERBUFFER, GL_SRGB8_ALPHA8, kSize, kSize);
+            ASSERT_EQ(FirstGLError(), 0u) << "the re-storage to SRGB8_ALPHA8 was refused";
+            ASSERT_EQ(glCheckFramebufferStatus(GL_FRAMEBUFFER), static_cast<GLenum>(GL_FRAMEBUFFER_COMPLETE));
+
+            glClearColor(0.0f, 0.0f, 0.0f, 0.0f);
+            glClear(GL_COLOR_BUFFER_BIT);
+            DrawQuad();
+            ReadPixelFloat(kSize / 2, kSize / 2, pixel);
+            EXPECT_NEAR(pixel[1], 1.0f, 0.05f) << "the draw did not land at all";
+            // THE HANDLE ARM ONLY. On the pre-handle arm a renderbuffer's twin is reached only from
+            // inside the FBO walk, and nothing that walk's memo reads moves on glRenderbufferStorage
+            // - the frontend setters bump no version (D-D2), no framebuffer version sees them, and
+            // the twin that would bump the backend generation is exactly what the memo skips. That
+            // is pre-P4a code and D-D2's documented hole; the resource record is what closes it,
+            // so the verdict is taken where the record is consumed and declined by name elsewhere
+            // (measured: alpha 1.0 on the pull build and at 0x1ff / 0, the mask of the storage the
+            // renderbuffer was attached with).
+            bool framebufferArmLive = false;
+            if (PeekEsprytFramebufferHandleArmIsLive(&framebufferArmLive) && framebufferArmLive) {
+                EXPECT_NEAR(pixel[3], 0.25f, 0.02f)
+                    << "the draw's alpha never reached the four-channel renderbuffer: the framebuffer "
+                       "record still describes the storage it was attached with (F-3)";
+            } else {
+                std::cout << "[ P4aSeamAudit ] renderbuffer re-storage verdict DECLINED on the pre-handle arm "
+                             "(D-D2's documented hole: no frontend version and no backend generation moves on a "
+                             "renderbuffer re-storage until the FBO walk the memo skips); alpha read "
+                          << pixel[3] << std::endl;
+                RecordProperty("p4a_seam_white_box", "declined");
+                RecordProperty("p4a_seam_white_box_reason", "renderbuffer re-storage: pre-handle arm (D-D2)");
+            }
+
+            glBindFramebuffer(GL_FRAMEBUFFER, 0);
+            glDeleteFramebuffers(1, &fbo);
+            glDeleteRenderbuffers(1, &renderbuffer);
+            glDeleteProgram(program);
+            EXPECT_EQ(FirstGLError(), 0u) << GLErrorName(FirstGLError());
+        }
+
+        // The mirror direction, four channels -> three, and it needs the driver to READ the stored
+        // alpha because the readback fixup (which consults the frontend) would hide it: after the
+        // respecify to SRGB8 the widening discipline has to hold - the clear puts 1.0 into the
+        // carrier's alpha and the draw is masked away from it - so a GL_DST_ALPHA blend of white sees
+        // 1.0. On a stale record the draw wrote its 0.25 into the carrier and the blend saw that.
+        TEST_F(P4aSeamAuditScenario, ATextureRespecifiedToThreeChannelsWhileAttachedReachesTheFramebufferRecord) {
+            if (!Ready()) return;
+            SkipUnlessEspryt("F-3");
+            if (IsSkipped()) return;
+
+            std::string error;
+            const GLuint program = CompileProgram(kQuadVS, kColorFS, &error);
+            ASSERT_NE(program, 0u) << error;
+            const GLint colorLocation = glGetUniformLocation(program, "uColor");
+            ASSERT_GE(colorLocation, 0);
+
+            GLuint texture = 0;
+            glGenTextures(1, &texture);
+            glBindTexture(GL_TEXTURE_2D, texture);
+            glTexImage2D(GL_TEXTURE_2D, 0, GL_SRGB8_ALPHA8, kSize, kSize, 0, GL_RGBA, GL_UNSIGNED_BYTE, nullptr);
+            glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+            glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
+            ASSERT_EQ(FirstGLError(), 0u) << "the SRGB8_ALPHA8 texture was refused";
+
+            GLuint fbo = 0;
+            glGenFramebuffers(1, &fbo);
+            glBindFramebuffer(GL_FRAMEBUFFER, fbo);
+            glFramebufferTexture2D(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, texture, 0);
+            ASSERT_EQ(glCheckFramebufferStatus(GL_FRAMEBUFFER), static_cast<GLenum>(GL_FRAMEBUFFER_COMPLETE));
+            glViewport(0, 0, kSize, kSize);
+            glUseProgram(program);
+
+            // Phase 1: four channels, the application owns alpha.
+            glClearColor(0.0f, 0.0f, 0.0f, 0.0f);
+            glClear(GL_COLOR_BUFFER_BIT);
+            glUniform4f(colorLocation, 0.0f, 1.0f, 0.0f, 0.25f);
+            DrawQuad();
+            float pixel[4];
+            glReadBuffer(GL_COLOR_ATTACHMENT0);
+            ReadPixelFloat(kSize / 2, kSize / 2, pixel);
+            EXPECT_NEAR(pixel[3], 0.25f, 0.02f) << "a four-channel attachment stores the draw's alpha";
+
+            // Phase 2: THE RESPECIFY to three channels, while attached.
+            glBindTexture(GL_TEXTURE_2D, texture);
+            glTexImage2D(GL_TEXTURE_2D, 0, GL_SRGB8, kSize, kSize, 0, GL_RGB, GL_UNSIGNED_BYTE, nullptr);
+            ASSERT_EQ(FirstGLError(), 0u) << "the respecify to SRGB8 was refused";
+            ASSERT_EQ(glCheckFramebufferStatus(GL_FRAMEBUFFER), static_cast<GLenum>(GL_FRAMEBUFFER_COMPLETE));
+
+            glDisable(GL_BLEND);
+            glClearColor(0.0f, 0.0f, 0.0f, 0.0f);
+            glClear(GL_COLOR_BUFFER_BIT);
+            glUniform4f(colorLocation, 0.0f, 1.0f, 0.0f, 0.25f);
+            DrawQuad();
+            // dst = stored alpha; src factor GL_DST_ALPHA, dst factor GL_ZERO, source white =>
+            // the colour becomes (storedAlpha, storedAlpha, storedAlpha) - ThreeChannelAttachment's
+            // own probe, which nothing on the readback path can doctor.
+            glEnable(GL_BLEND);
+            glBlendFunc(GL_DST_ALPHA, GL_ZERO);
+            glUniform4f(colorLocation, 1.0f, 1.0f, 1.0f, 1.0f);
+            DrawQuad();
+            glDisable(GL_BLEND);
+            ReadPixelFloat(kSize / 2, kSize / 2, pixel);
+            // PUSH BUILDS ONLY, EVERY ARM OF THEM (the mirror). The pre-handle half of the fix (an in-place
+            // regeneration takes the backend-id generation a re-mint takes) is Espryt code the
+            // pull build would share, and G1 keeps the pull library byte-identical to the P4a
+            // baseline - so it is compiled under MOBILEGL_PIPE_PUSH and the pull build keeps the
+            // pre-P4a hole until the same lines land on dev on their own. The peek returns true
+            // exactly where it could look, which for a case that already skipped off Espryt means
+            // "a push build"; what it writes (is the handle arm live) does not matter here.
+            bool framebufferArmLive = false;
+            if (PeekEsprytFramebufferHandleArmIsLive(&framebufferArmLive)) {
+                EXPECT_NEAR(pixel[0], 1.0f, 0.05f)
+                    << "GL_DST_ALPHA read the stored alpha of a three-channel attachment and it was not "
+                       "1.0: the framebuffer record (handle arm) or the FBO twin's memo (pre-handle arm) "
+                       "still describes the four-channel storage the texture was attached with, so the "
+                       "draw was let write alpha (F-3, mirror)";
+            } else {
+                std::cout << "[ P4aSeamAudit ] three-channel respecify verdict DECLINED on the pull build (the "
+                             "in-place regeneration bump is push-only by G1); red read "
+                          << pixel[0] << std::endl;
+                RecordProperty("p4a_seam_white_box", "declined");
+                RecordProperty("p4a_seam_white_box_reason", "three-channel respecify: pull build (G1)");
+            }
+
+            glBindFramebuffer(GL_FRAMEBUFFER, 0);
+            glDeleteFramebuffers(1, &fbo);
+            glDeleteTextures(1, &texture);
+            glDeleteProgram(program);
+            EXPECT_EQ(FirstGLError(), 0u) << GLErrorName(FirstGLError());
+        }
+
+        // -----------------------------------------------------------------------------------
+        // F-1 / F-1b: a program switch re-resolves the view set, and the texture sync list with it
+        // -----------------------------------------------------------------------------------
+        //
+        // The sequence the audit named, and every step of it is ordinary: two programs sampling two
+        // different units, a texture bound to a unit's EMPTY 2D slot - the unit was already touched
+        // through another target, so the high-water mark does not move - while a program that does
+        // not sample it is in use, then the switch to the one that does. Nothing between the two
+        // draws touches a parameter, a level or a populated slot, which is exactly what leaves the
+        // record epoch - and the program-independent texture sync list keyed on it - unmoved on
+        // the tree the audit read: the second program sampled an unbound unit and drew black.
+        TEST_F(P4aSeamAuditScenario, ATextureBoundToAnEmptySlotUnderOneProgramIsSampledByTheNext) {
+            if (!Ready()) return;
+
+            std::string error;
+            const GLuint first = CompileProgram(kQuadVS, kFetchFS, &error);
+            ASSERT_NE(first, 0u) << error;
+            const GLuint second = CompileProgram(kQuadVS, kFetchFS, &error);
+            ASSERT_NE(second, 0u) << error;
+            glUseProgram(first);
+            glUniform1i(glGetUniformLocation(first, "uTex"), 0);
+            glUseProgram(second);
+            glUniform1i(glGetUniformLocation(second, "uTex"), 1);
+            glUseProgram(0);
+
+            // Every texture exists, complete, with its parameters set, BEFORE the first draw: a
+            // parameter or a level defined between the two draws would move the sampling-resolution
+            // generation and rescue the list by accident.
+            const GLuint red = MakeSolidTexture2D(255, 0, 0);
+            const GLuint green = MakeSolidTexture2D(0, 255, 0);
+            GLuint touch3D = 0;
+            glGenTextures(1, &touch3D);
+            glBindTexture(GL_TEXTURE_3D, touch3D);
+            const std::uint8_t blue[2 * 2 * 2 * 4] = {0, 0, 255, 255, 0, 0, 255, 255, 0, 0, 255, 255, 0, 0, 255, 255,
+                                                     0, 0, 255, 255, 0, 0, 255, 255, 0, 0, 255, 255, 0, 0, 255, 255};
+            glTexImage3D(GL_TEXTURE_3D, 0, GL_RGBA8, 2, 2, 2, 0, GL_RGBA, GL_UNSIGNED_BYTE, blue);
+            glTexParameteri(GL_TEXTURE_3D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+            glTexParameteri(GL_TEXTURE_3D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
+            glBindTexture(GL_TEXTURE_3D, 0);
+            glBindTexture(GL_TEXTURE_2D, 0);
+            ASSERT_EQ(FirstGLError(), 0u) << "texture setup left a GL error behind";
+
+            ColorFbo target = MakeColorFbo(kSize, kSize);
+            ASSERT_NE(target.fbo, 0u);
+            BindFbo(target);
+            glBindVertexArray(m_vao);
+
+            // Unit 1 is TOUCHED through its 3D slot; its 2D slot stays empty. Unit 0 holds red.
+            // The first program is in use BEFORE the first verb (the clear), so the very first
+            // view set that goes out is already resolved for it - measured: with no program in
+            // use at the clear the first set is [null, null], and the bind below then re-resolves
+            // to [red, null], a DIFFERENT set that moves the serial and rescues the case by
+            // accident.
+            glActiveTexture(GL_TEXTURE1);
+            glBindTexture(GL_TEXTURE_3D, touch3D);
+            glActiveTexture(GL_TEXTURE0);
+            glBindTexture(GL_TEXTURE_2D, red);
+            glUseProgram(first);
+            ClearTo(0.0f, 0.0f, 1.0f, 1.0f);
+            DrawQuad();
+
+            // THE BIND ONTO THE EMPTY SLOT, under a program that does not sample unit 1, and a
+            // draw with THAT program so the bind's own re-resolution of the view set happens under
+            // it (the bind generation fires bit 12 at the next verb; a switch inside the same verb
+            // gap would let that fire resolve under the second program by accident) ...
+            glActiveTexture(GL_TEXTURE1);
+            glBindTexture(GL_TEXTURE_2D, green);
+            glActiveTexture(GL_TEXTURE0);
+            DrawQuad();
+            // ... and THE SWITCH to the one that does sample it. No other state moves.
+            glUseProgram(second);
+            DrawQuad();
+            EXPECT_EQ(FirstGLError(), 0u) << "the two draws left a GL error behind";
+
+            const Image image = ReadPixels(kSize, kSize);
+            ASSERT_FALSE(image.Empty());
+            EXPECT_TRUE(RegionIsMostly(image, kInset, kSize - 1 - kInset, kInset, kSize - 1 - kInset, "green", 0.0,
+                                       "the draw after the program switch"))
+                << "black means the second program sampled an unbound unit: the texture bound to the "
+                   "empty slot was never synced because the view set - and E's record epoch with it - "
+                   "did not move on the program switch (F-1 / F-1b); red means the first program's "
+                   "set was still in force";
+
+            glActiveTexture(GL_TEXTURE1);
+            glBindTexture(GL_TEXTURE_2D, 0);
+            glBindTexture(GL_TEXTURE_3D, 0);
+            glActiveTexture(GL_TEXTURE0);
+            glBindTexture(GL_TEXTURE_2D, 0);
+            glUseProgram(0);
+            DestroyColorFbo(target);
+            glDeleteTextures(1, &red);
+            glDeleteTextures(1, &green);
+            glDeleteTextures(1, &touch3D);
+            glDeleteProgram(first);
+            glDeleteProgram(second);
+            EXPECT_EQ(FirstGLError(), 0u) << GLErrorName(FirstGLError());
+        }
+
+        // -----------------------------------------------------------------------------------
+        // F-2 / SD-4: the image window follows the program, through buffer images
+        // -----------------------------------------------------------------------------------
+        //
+        // Public-GL half: both dispatches store what they should (every arm passes this - the
+        // server's window/high-water union takes the pre-handle bind for a unit the record does
+        // not cover, which is exactly why the seam was silent). White-box half, on Espryt's handle
+        // arm: after the first dispatch set_shader_images must have arrived with a window of ONE
+        // unit (SD-4: on the tree the audit read a buffer image never reached the record at all -
+        // the null -> program transition moved nothing bit 14 read), and after the switch to the
+        // program naming two units the window must be TWO (F-2: the two programs' image-unit
+        // counters are equal, so the switch alone moved nothing either).
+        TEST_F(P4aSeamAuditScenario, AProgramSwitchWithEqualImageUnitCountersMovesTheImageWindow) {
+            if (!Ready()) return;
+            if (!ComputeImagesAreUsable()) GTEST_SKIP() << "no compute image units / buffer textures on this host";
+
+            std::string error;
+            const GLuint one = MakeComputeProgram(kOneBufferImageCS, &error);
+            ASSERT_NE(one, 0u) << error;
+            const GLuint two = MakeComputeProgram(kTwoBufferImagesCS, &error);
+            ASSERT_NE(two, 0u) << error;
+
+            GLuint buffer0 = 0;
+            GLuint buffer1 = 0;
+            const GLuint image0 = MakeBufferTexture(&buffer0, 0u);
+            const GLuint image1 = MakeBufferTexture(&buffer1, 0u);
+            ASSERT_EQ(FirstGLError(), 0u) << "buffer texture setup left a GL error behind";
+
+            // Both units bound BEFORE any dispatch, so the bind generation does not move between
+            // the two dispatches and the only thing that changes is the program in use.
+            glBindImageTexture(0, image0, 0, GL_FALSE, 0, GL_READ_WRITE, GL_R32UI);
+            glBindImageTexture(1, image1, 0, GL_FALSE, 0, GL_READ_WRITE, GL_R32UI);
+            ASSERT_EQ(FirstGLError(), 0u) << "binding the buffer images left a GL error behind";
+
+            const bool whiteBox = SamplerHandleArmIsLive("F-2 / SD-4");
+
+            glUseProgram(one);
+            glDispatchCompute(1, 1, 1);
+            glMemoryBarrier(GL_ALL_BARRIER_BITS);
+            EXPECT_EQ(FirstGLError(), 0u) << "the first dispatch leaked a GL error";
+            if (whiteBox) {
+                PipeShaderImageWindowPeek window{};
+                ASSERT_TRUE(PeekPipeShaderImageWindow(&window));
+                EXPECT_EQ(window.Start, 0u);
+                EXPECT_EQ(window.Count, 1u)
+                    << "set_shader_images never arrived for a program whose only image is a BUFFER "
+                       "image (SD-4): the null -> program transition moved nothing bit 14 read";
+            }
+
+            glUseProgram(two);
+            glDispatchCompute(1, 1, 1);
+            glMemoryBarrier(GL_ALL_BARRIER_BITS);
+            EXPECT_EQ(FirstGLError(), 0u) << "the second dispatch leaked a GL error";
+            if (whiteBox) {
+                PipeShaderImageWindowPeek window{};
+                ASSERT_TRUE(PeekPipeShaderImageWindow(&window));
+                EXPECT_EQ(window.Start, 0u);
+                EXPECT_EQ(window.Count, 2u)
+                    << "the image window did not follow the program switch: two programs with equal "
+                       "image-unit counters, and bit 14 mixed only the counter (F-2)";
+            }
+
+            EXPECT_EQ(ReadBufferTexel0(buffer0), 7u) << "the first program's store did not land";
+            EXPECT_EQ(ReadBufferTexel0(buffer1), 9u) << "the second program's store did not land";
+
+            glBindImageTexture(0, 0, 0, GL_FALSE, 0, GL_READ_ONLY, GL_R32UI);
+            glBindImageTexture(1, 0, 0, GL_FALSE, 0, GL_READ_ONLY, GL_R32UI);
+            glBindBuffer(GL_TEXTURE_BUFFER, 0);
+            glBindTexture(GL_TEXTURE_BUFFER, 0);
+            glUseProgram(0);
+            glDeleteProgram(one);
+            glDeleteProgram(two);
+            glDeleteTextures(1, &image0);
+            glDeleteTextures(1, &image1);
+            glDeleteBuffers(1, &buffer0);
+            glDeleteBuffers(1, &buffer1);
+            EXPECT_EQ(FirstGLError(), 0u) << GLErrorName(FirstGLError());
+        }
+
+        // -----------------------------------------------------------------------------------
+        // F-4: the unit's driver sampler is the CSO's own twin on the handle arm
+        // -----------------------------------------------------------------------------------
+        //
+        // Public-GL half: a glBindSampler'd object whose wrap differs from the texture's built-in
+        // sampler wins (GL 4.6 core 8.10) - every arm passes this, because the pre-handle program
+        // pass bound the object through its identity twin. White-box half, on Espryt's handle arm:
+        // the sampler the unit carries on the driver must be the twin Espryt holds AT THE CSO
+        // HANDLE bind_sampler_states named for the unit. On the tree the audit read that twin did
+        // not exist - the handle is content-addressed, the registry's twins were minted off
+        // lifetime ids - so the record arm bound nothing on every draw.
+        TEST_F(P4aSeamAuditScenario, ABoundSamplerObjectIsDrivenThroughItsCsoTwinOnTheHandleArm) {
+            if (!Ready()) return;
+
+            std::string error;
+            const GLuint program = CompileProgram(kQuadVS, kOutsideSampleFS, &error);
+            ASSERT_NE(program, 0u) << error;
+
+            // The texture's built-in sampler REPEATS, so (1.5, 1.5) reads the red texel through it;
+            // the sampler object CLAMPS TO A WHITE BORDER, so the same coordinate reads white
+            // through it. White is a Vulkan palette border colour, so Magma needs no extension.
+            const GLuint red = MakeSolidTexture2D(255, 0, 0);
+            glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_REPEAT);
+            glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_REPEAT);
+            GLuint sampler = 0;
+            glGenSamplers(1, &sampler);
+            glSamplerParameteri(sampler, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+            glSamplerParameteri(sampler, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
+            glSamplerParameteri(sampler, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_BORDER);
+            glSamplerParameteri(sampler, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_BORDER);
+            const GLfloat white[4] = {1.0f, 1.0f, 1.0f, 1.0f};
+            glSamplerParameterfv(sampler, GL_TEXTURE_BORDER_COLOR, white);
+            ASSERT_EQ(FirstGLError(), 0u) << "sampler setup left a GL error behind";
+
+            ColorFbo target = MakeColorFbo(kSize, kSize);
+            ASSERT_NE(target.fbo, 0u);
+            BindFbo(target);
+            glBindVertexArray(m_vao);
+            glActiveTexture(GL_TEXTURE0);
+            glBindTexture(GL_TEXTURE_2D, red);
+            glBindSampler(0, sampler);
+            glUseProgram(program);
+            glUniform1i(glGetUniformLocation(program, "uTex"), 0);
+            ClearTo(0.0f, 0.0f, 1.0f, 1.0f);
+            DrawQuad();
+            EXPECT_EQ(FirstGLError(), 0u) << "the draw left a GL error behind";
+
+            const Image image = ReadPixels(kSize, kSize);
+            ASSERT_FALSE(image.Empty());
+            EXPECT_TRUE(RegionIsMostly(image, kInset, kSize - 1 - kInset, kInset, kSize - 1 - kInset, "white", 0.0,
+                                       "the draw through the bound sampler object"))
+                << "red means the texture's own REPEAT sampler applied instead of the bound object's "
+                   "CLAMP_TO_BORDER";
+
+            if (SamplerHandleArmIsLive("F-4")) {
+                EsprytUnitSamplerPeek peek{};
+                ASSERT_TRUE(PeekEsprytUnitSampler(0, sampler, &peek));
+                std::cout << "[ P4aSeamAudit ] white-box: unit 0 driver sampler " << peek.BoundSamplerId
+                          << ", bind_sampler_states handle {" << peek.CsoHandleSlot << ", " << peek.CsoHandleGen
+                          << "} inside window " << (peek.UnitInsideWindow ? "yes" : "no") << ", CSO twin "
+                          << peek.CsoTwinSamplerId << ", identity twin " << peek.IdentityTwinSamplerId << std::endl;
+                EXPECT_TRUE(peek.UnitInsideWindow) << "bind_sampler_states did not describe unit 0";
+                EXPECT_NE(peek.CsoHandleSlot, 0u) << "bind_sampler_states names no CSO for a unit that carries "
+                                                     "a sampler object";
+                EXPECT_NE(peek.CsoTwinSamplerId, 0u)
+                    << "Espryt holds no twin at the CSO handle bind_sampler_states named: the record arm's "
+                       "lookup went to the identity-keyed registry with a content-addressed handle and "
+                       "could never hit (F-4)";
+                EXPECT_EQ(peek.BoundSamplerId, peek.CsoTwinSamplerId)
+                    << "the driver sampler on unit 0 is not the CSO's twin, so it was put there by the "
+                       "pre-handle program pass and not by the record arm (F-4)";
+                EXPECT_EQ(peek.IdentityTwinSamplerId, 0u)
+                    << "an identity-keyed twin was minted for the sampler object on the handle arm: the "
+                       "pre-handle pass is still the one doing the binding";
+            }
+
+            glBindSampler(0, 0);
+            glBindTexture(GL_TEXTURE_2D, 0);
+            glUseProgram(0);
+            DestroyColorFbo(target);
+            glDeleteSamplers(1, &sampler);
+            glDeleteTextures(1, &red);
+            glDeleteProgram(program);
+            EXPECT_EQ(FirstGLError(), 0u) << GLErrorName(FirstGLError());
+        }
+
+    } // namespace
+} // namespace MGITest

MobileGL/MG_IntegrationTest/Scenarios/ResourceSubsystemControlScenario.cpp

@@ -0,0 +1,327 @@
+// MobileGL - MobileGL/MG_IntegrationTest/Scenarios/ResourceSubsystemControlScenario.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 - THE P3a SUBSYSTEM A/B IS REAL (gate G12).
+//
+// P3a migrates two subsystems: kMGPipeSubsystemResources (bit 7, the resource_* family) and
+// kMGPipeSubsystemVertexInput (bit 8, vertex elements / buffers / index). The push build's default
+// mask becomes kMGPipeSubsystemsMigratedAtP3a = 0x1ff, and P2's 0x7f survives as the control that
+// clears exactly those two bits - MGPipe.h:79's rule that every phase's constant keeps meaning what
+// it meant, so an operator's recorded mask is still readable a phase later.
+//
+// That A/B is what every "push vs pull" number in MEASUREMENTS.md is taken against, and it has one
+// characteristic failure mode: the bits stop steering anything, both arms run the same code, and
+// every later comparison is quietly taken against a switch that does nothing. This file is the
+// entry that cannot let that happen.
+//
+// WHAT IT ASSERTS, per arm:
+//
+//   on  (MOBILEGL_PIPE_PUSH=0x1ff)
+//       The client emits map_persistent for every definition of a store past
+//       BufferObject::TryAdoptLargeStorage's 16 MiB threshold, so the window's
+//       map-persistent-roundtrips (`mpr=`) equals the number of definitions in it - one per
+//       storage definition, mint or decline (D-B2).
+//
+//   off (MOBILEGL_PIPE_PUSH=0x7f, P2's default = P3a's subsystems cleared)
+//       The frontend dispatch falls through to the legacy BufferBackendOps arm, nothing is emitted
+//       through the resource family, and mpr= must read ZERO. This is the reading a dead switch
+//       fails: with bit 7 ignored, this arm would report the same non-zero count as the other one.
+//
+//   both arms
+//       THE PIXELS MUST NOT MOVE. The arena is filled with one solid-colour quad and drawn, and
+//       both arms must read back that colour. "The counters moved and the picture did not" is the
+//       whole claim - a switch that changed what is drawn would not be an A/B, it would be a bug.
+//
+// WHY IT CAN SKIP. The counter is emitted by the client-side resource tracker (P3a package B), and
+// this file is written against the P3a contract commit, before that package lands. Until then
+// nothing emits map_persistent, mpr= is structurally zero in BOTH arms, and an assertion about the
+// difference would be a statement about nothing. The build answers the question rather than a
+// hand-maintained list: MG_IntegrationTest/CMakeLists.txt greps every source under MG_Impl/Pipe/
+// for the counter's name and passes the answer in as MGITEST_PIPE_RESOURCE_EMITTER_PRESENT, with a
+// CONFIGURE_DEPENDS on that directory and on each file it finds so the answer cannot go stale. It
+// is a CONTENT probe, not a filename probe, so the owning package keeps control of its own file
+// layout - P3a's new client files are headers (D-N), and a glob for `ResourceTracker.cpp` would
+// have kept this control skipping forever with a reason that had become false.
+//
+// DirectGLES ONLY, and that is the honest scope: P3a migrates Espryt's buffer and VAO paths.
+// Magma's buffer path is P7 and registers no MGPipeResourceOps, so a DirectVulkan lane here would
+// be measuring the client emitter against a backend that has not been asked to change - which is
+// a real question, but it is P7's, not this control's.
+
+#include <cstdlib>
+#include <cstring>
+#include <string>
+#include <vector>
+
+#include "../Harness/HeadlessGL.h"
+#include "../Harness/PipeStatsWindow.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 {
+
+        // Set by the two ResourceSubsystemControl. ctest entries and by nothing else; a harness
+        // marker, never read by the library. Its absence means an ambient entry, where neither the
+        // stats channel nor a private log path is configured.
+        constexpr const char* kLaneMarker = "MGITEST_RESOURCE_SUBSYSTEM_LANE";
+        constexpr const char* kLaneOn = "on";
+        constexpr const char* kLaneOff = "off";
+
+        // Past BufferObject::TryAdoptLargeStorage's 16 MiB threshold, so the store is offered for
+        // adoption at all; the vertex payload sits deep inside it so a clamped or aliased adopted
+        // range would miss it. Same shape as LargeArenaAdoptionScenario, deliberately: this
+        // control's workload has to be one the buffer path really takes.
+        constexpr GLsizeiptr kArenaBytes = GLsizeiptr(20) * 1024 * 1024;
+        constexpr GLintptr kVertexOffset = GLintptr(16) * 1024 * 1024;
+        // Two definitions and several draws each, so "one per definition", "one per draw" and
+        // "none at all" are three different numbers.
+        //
+        // ONE ARENA DEFINED TWICE, not two arenas defined once each: the second definition
+        // RE-SPECIFIES a store whose bytes the VAO's attributes are already pointing into, and
+        // the attributes are not re-declared afterwards. That makes this control also the place
+        // where the respecify/retire path is exercised on BOTH arms of the A/B, which is what
+        // ID-9 asks for: `dev`'s d7655247 ("rebind VAOs when an adopted buffer is respecified -
+        // the immediate retire path forgot the buffer-id generation") arrived in
+        // feat/disaggregated with the 5cb826b0 merge, and the handle arm duplicates that retire
+        // core, so an arm that forgot the rebind must be visible somewhere. Here it is a dead
+        // draw or a fault, not a silent divergence. The first cut of this file routed around the
+        // path because the fix was not yet in this branch's history and the workload reproduced
+        // as a hard SIGSEGV in the vertex fetch; that detour is what ID-9 supersedes.
+        //
+        // The COUNT is unaffected by the change: two storage definitions either way, which is
+        // what ARCHITECTURE.md:474 prices.
+        constexpr int kDefinitionsInTheWindow = 2;
+        constexpr int kDrawsPerDefinition = 3;
+        constexpr int kInset = 2;
+
+        constexpr const char* kVS = R"(#version 330 core
+in vec2 aPos;
+in vec3 aColor;
+out vec3 vColor;
+void main() {
+    vColor = aColor;
+    gl_Position = vec4(aPos, 0.0, 1.0);
+}
+)";
+
+        constexpr const char* kFS = R"(#version 330 core
+in vec3 vColor;
+out vec4 oColor;
+void main() { oColor = vec4(vColor, 1.0); }
+)";
+
+        struct Vertex {
+            float x, y;
+            float r, g, b;
+        };
+
+        std::vector<Vertex> Quad(float r, float g, float b) {
+            return {
+                {-1.0f, -1.0f, r, g, b}, {1.0f, -1.0f, r, g, b}, {1.0f, 1.0f, r, g, b},
+                {-1.0f, -1.0f, r, g, b}, {1.0f, 1.0f, r, g, b},  {-1.0f, 1.0f, r, g, b},
+            };
+        }
+
+        bool BuildMarkerIsSet(const char* name) {
+            const char* value = std::getenv(name);
+            return value != nullptr && value[0] == '1' && value[1] == '\0';
+        }
+
+        std::string LaneName() {
+            const char* lane = std::getenv(kLaneMarker);
+            return lane != nullptr ? std::string(lane) : std::string();
+        }
+
+        class ResourceSubsystemControlScenario : public ScenarioTest {
+        protected:
+            void SetUp() override {
+                ScenarioTest::SetUp();
+                if (!Ready()) return;
+                m_lane = LaneName();
+                std::string error;
+                m_program = CompileProgram(kVS, kFS, &error);
+                ASSERT_NE(m_program, 0u) << error;
+
+                // The VAO only. The arena is created and defined inside the counted window - the
+                // window a summary line reports is "since the previous line", so a definition
+                // taken in SetUp would be counted in a window this case does not control - and
+                // its attribute pointers are declared only once the store exists, because an
+                // attribute whose offset is 16 MiB into a store that has not been defined yet is
+                // a range no driver has to accept.
+                glGenVertexArrays(1, &m_vao);
+                glBindVertexArray(m_vao);
+                RecordProperty("lane", m_lane.empty() ? "ambient" : m_lane.c_str());
+            }
+
+            void TearDown() override {
+                if (!Ready()) return;
+                glUseProgram(0);
+                glBindVertexArray(0);
+                glBindBuffer(GL_ARRAY_BUFFER, 0);
+                if (m_arena != 0) glDeleteBuffers(1, &m_arena);
+                m_arena = 0;
+                if (m_vao != 0) glDeleteVertexArrays(1, &m_vao);
+                if (m_program != 0) glDeleteProgram(m_program);
+            }
+
+            // GTEST_SKIP() returns from the function it is written in, so this cannot report
+            // through a return value; the caller pairs it with `if (IsSkipped()) return;`.
+            void SkipUnlessTheLaneIsAssertableHere() {
+                if (m_lane.empty()) {
+                    GTEST_SKIP() << "runs only in its own lane: the two ResourceSubsystemControl. ctest "
+                                    "entries set MGITEST_RESOURCE_SUBSYSTEM_LANE together with the "
+                                    "MOBILEGL_PIPE_PUSH bitmask that arm means, MOBILEGL_PIPE_STATS=1, "
+                                    "MOBILEGL_PIPE_STATS_PERIOD=1 and a private MOBILEGL_LOG_FILE_PATH. "
+                                    "None of that is configured in the ambient entries, and the ambient "
+                                    "log is shared, so a read here would race.";
+                    return;
+                }
+                if (!BuildMarkerIsSet("MGITEST_PIPE_PUSH_BUILD")) {
+                    GTEST_SKIP() << "this library was built without MOBILEGL_PIPE_PUSH: there are no "
+                                    "subsystem bits to clear, CallClass::MapPersistentRoundtrips does "
+                                    "not exist and the summary line carries no mpr=. The entry is "
+                                    "registered here anyway so that `ctest -L integration-gpu` names the "
+                                    "same tests in the pull build and the push build (gate G2).";
+                    return;
+                }
+                if (!BuildMarkerIsSet("MGITEST_PIPE_RESOURCE_EMITTER_PRESENT")) {
+                    GTEST_SKIP() << "subsystem not implemented on this tree: no source under "
+                                    "MobileGL/MG_Impl/Pipe/ names MapPersistentRoundtrips, so nothing "
+                                    "emits map_persistent, mpr= is structurally zero in BOTH arms and "
+                                    "the difference between them is not observable yet. P3a package B "
+                                    "owns the client-side resource tracker; this control arms itself "
+                                    "when it lands, whatever file that package puts the emitter in.";
+                    return;
+                }
+                if (PipeStatsWindow::LibraryLogPath().empty()) {
+                    GTEST_SKIP() << "the lane configured no MOBILEGL_LOG_FILE_PATH, and the library's "
+                                    "summary line is the only channel this module has for reading "
+                                    "PipeStats";
+                    return;
+                }
+            }
+
+            // ONE storage definition - the NULL-data glBufferData past the adoption threshold,
+            // which is Minecraft's arena-creation idiom and the adoption point - then a few
+            // draws. Entirely inside one frame, so one summary window covers exactly this.
+            //
+            // The attribute pointers are declared ONCE, on the first definition, and never again:
+            // definition 0 creates the store, every later index RE-SPECIFIES it under the live
+            // VAO. Re-declaring them afterwards would re-sync the VAO by hand and hide the thing
+            // the second definition is here to exercise (see kDefinitionsInTheWindow above).
+            void DefineTheArenaAndDrawFromIt(int index, float r, float g, float b) {
+                if (index == 0) glGenBuffers(1, &m_arena);
+                glBindBuffer(GL_ARRAY_BUFFER, m_arena);
+                glBufferData(GL_ARRAY_BUFFER, kArenaBytes, nullptr, GL_DYNAMIC_DRAW);
+                const std::vector<Vertex> vertices = Quad(r, g, b);
+                glBufferSubData(GL_ARRAY_BUFFER, kVertexOffset,
+                                GLsizeiptr(vertices.size() * sizeof(Vertex)), vertices.data());
+                if (index == 0) {
+                    glBindVertexArray(m_vao);
+                    glVertexAttribPointer(0, 2, GL_FLOAT, GL_FALSE, sizeof(Vertex),
+                                          reinterpret_cast<void*>(kVertexOffset));
+                    glVertexAttribPointer(1, 3, GL_FLOAT, GL_FALSE, sizeof(Vertex),
+                                          reinterpret_cast<void*>(kVertexOffset + 2 * sizeof(float)));
+                    glEnableVertexAttribArray(0);
+                    glEnableVertexAttribArray(1);
+                }
+                glUseProgram(m_program);
+                for (int draw = 0; draw < kDrawsPerDefinition; ++draw) {
+                    glDrawArrays(GL_TRIANGLES, 0, 6);
+                }
+            }
+
+            std::string m_lane;
+            GLuint m_program = 0;
+            GLuint m_vao = 0;
+            GLuint m_arena = 0;
+        };
+
+        // ONE case per lane, and it is a constraint rather than a preference: this case READS the
+        // library log, the log is a per-LANE resource (the library opens it fopen(path, "w"), so
+        // every process in a lane truncates it), and a second case here would race this one under
+        // `ctest -j` with a failure indistinguishable from "the counter was never emitted". The
+        // plumbing is asserted first, with its own message, inside this one process.
+        TEST_F(ResourceSubsystemControlScenario, ClearingTheP3aBitsStopsTheEmissionsAndNotThePixels) {
+            if (!Ready()) return;
+            SkipUnlessTheLaneIsAssertableHere();
+            if (IsSkipped()) return;
+
+            BindDefaultFramebuffer();
+            Gl().EndFrame(); // close the setup window: everything below is one window
+
+            ClearTo(0.0f, 0.0f, 0.0f, 1.0f);
+            for (int definition = 0; definition < kDefinitionsInTheWindow; ++definition) {
+                DefineTheArenaAndDrawFromIt(definition, 0.0f, 1.0f, 0.0f);
+                ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR))
+                    << "arena definition " << definition
+                    << " left a GL error behind (definition 0 creates the store, every later one "
+                       "re-specifies it under the live VAO)";
+            }
+            const Image image = ReadPixels(Gl().Width(), Gl().Height());
+            Gl().EndFrame(); // the swap that emits the window covering exactly the work above
+
+            const PipeStatsWindow::Window window = PipeStatsWindow::LastFromLaneLog();
+            ASSERT_TRUE(window.found)
+                << "no 'MGPipe stats:' line in " << PipeStatsWindow::LibraryLogPath()
+                << ". This IS a push build (the lane checked MGITEST_PIPE_PUSH_BUILD before getting "
+                   "here), so either MOBILEGL_PIPE_STATS / MOBILEGL_PIPE_STATS_PERIOD did not reach the "
+                   "process, or no summary line was emitted at all because nothing reached "
+                   "PipeStats::OnPresent.";
+            RecordProperty("stats_line", window.line.c_str());
+
+            const long long roundtrips = PipeStatsWindow::CounterOrAbsent(window, "mpr");
+            ASSERT_GE(roundtrips, 0)
+                << "the summary line carries no mpr= field, so this build's PipeStats has no "
+                   "map-persistent-roundtrips counter to read: "
+                << window.line;
+
+            if (m_lane == kLaneOn) {
+                EXPECT_EQ(roundtrips, static_cast<long long>(kDefinitionsInTheWindow))
+                    << "with bits 7|8 SET the resource family is the path a store definition takes, so "
+                       "each of the " << kDefinitionsInTheWindow
+                    << " definitions in this window is one map_persistent emission (mint or decline - "
+                       "both need an answer from the resource owner, D-B2). "
+                    << (kDefinitionsInTheWindow * kDrawsPerDefinition)
+                    << " would mean an acquisition per draw, and 0 would mean the emission never "
+                       "happened on the arm that is supposed to do it. It reported: "
+                    << window.line;
+            } else if (m_lane == kLaneOff) {
+                EXPECT_EQ(roundtrips, 0)
+                    << "with bits 7|8 CLEARED (MOBILEGL_PIPE_PUSH=0x7f, P2's default) the frontend "
+                       "dispatch must fall through to the legacy BufferBackendOps arm and emit nothing "
+                       "through the resource family, so mpr= must be zero. A non-zero count here is the "
+                       "dead-switch reading: the bits are being ignored, both arms run the same code, "
+                       "and every push-vs-pull number taken against this A/B is measuring one arm twice. "
+                       "It reported: "
+                    << window.line;
+            } else {
+                FAIL() << "unknown " << kLaneMarker << " value '" << m_lane
+                       << "': the arms are on / off. Reading an unrecognised name as either would make "
+                          "this lane assert the other arm's expectation while claiming to test this one.";
+            }
+
+            // ... and the picture is the same whichever arm ran. The arena is drawn with one solid
+            // colour, so both arms must read back exactly that.
+            EXPECT_TRUE(RegionIsMostly(image, kInset, image.Width() - kInset, kInset,
+                                       image.Height() - kInset, "green", 0.0,
+                                       "the arena draw [" + m_lane + "]"))
+                << "the subsystem bits changed what is DRAWN, which is not an A/B - the handle path and "
+                   "the legacy path must produce the same pixels from the same arena.";
+        }
+
+    } // namespace
+} // namespace MGITest

MobileGL/MG_IntegrationTest/Scenarios/StorageBufferRegrowScenario.cpp

@@ -19,11 +19,26 @@
 //
 // The assertion is deliberately on the WHOLE grown range, so a partial write names the byte the
 // stale extent stopped at.
+//
+// P3a ADDS THE COST OF THAT REGROWTH (gate G10). ARCHITECTURE.md:474 prices a persistently mapped
+// store at "one round trip per STORAGE DEFINITION, not one per store" - and, emphatically, not one
+// per draw. `map-persistent-roundtrips` (`mpr=` in the summary line) counts every map_persistent
+// EMISSION, mint or decline (D-B2), so the claim is directly countable: N definitions of an
+// adopted store must publish exactly N, whatever the workload does between them. A regression that
+// re-acquires per dispatch reports N x dispatches, which is the failure this case exists to name;
+// a regression that stops emitting reports 0.
+//
+// The second case therefore respecifies a store LARGE ENOUGH TO BE ADOPTED
+// (BufferObject::TryAdoptLargeStorage's 16 MiB threshold), several times, with several dispatches
+// between the definitions, and reads the one window that covers exactly that workload. It skips -
+// visibly, with the reason - on a tree where nothing emits the counter yet.
 
+#include <cstdlib>
 #include <string>
 #include <vector>
 
 #include "../Harness/HeadlessGL.h"
+#include "../Harness/PipeStatsWindow.h"
 #include "../Harness/ScenarioFixture.h"
 
 #ifdef GLAPI
@@ -50,6 +65,31 @@ void main() {
         constexpr int kSmallElements = 6;  // 24 bytes - the first iteration's size
         constexpr int kLargeElements = 24; // 96 bytes - what the second iteration grows to
 
+        // The G10 case's sizes. Every one of them is past BufferObject::TryAdoptLargeStorage's
+        // 16 MiB threshold, because a store below it is never offered for adoption at all and the
+        // window would then be asserting that nothing happened. They GROW, which is the scenario's
+        // subject: each glBufferData is a new storage definition, so each is one acquisition.
+        constexpr int kAdoptedDefinitions = 3;
+        constexpr int kAdoptedBaseElements = 5 * 1024 * 1024;  // 20 MiB of uint
+        constexpr int kAdoptedGrowthElements = 1024 * 1024;    // + 4 MiB per definition
+        // Enough dispatches per definition that "one per definition" and "one per dispatch" are
+        // different numbers by a wide margin (3 vs 12), and few enough to stay cheap.
+        constexpr int kDispatchesPerDefinition = 4;
+        // Only the first elements are dispatched over: the point of the large store is the
+        // ADOPTION, not the compute cost.
+        constexpr int kDispatchedElements = 6;
+
+        // Set by the MapPersistentRoundtrips. ctest entry and by nothing else; a harness marker,
+        // never read by the library. Its absence means an ambient entry, where neither the stats
+        // channel nor a private log path is configured - and where the shared log makes a read
+        // race a neighbour's bring-up.
+        constexpr const char* kLaneMarker = "MGITEST_MPR_LANE";
+
+        bool BuildMarkerIsSet(const char* name) {
+            const char* value = std::getenv(name);
+            return value != nullptr && value[0] == '1' && value[1] == '\0';
+        }
+
         class StorageBufferRegrowScenario : public ScenarioTest {
         protected:
             void SetUp() override {
@@ -114,6 +154,56 @@ void main() {
                 return values;
             }
 
+            // The Minecraft arena idiom, and the adoption point: a NULL-data definition of a
+            // store past the threshold. No host-side vector, so a 28 MiB definition costs
+            // nothing on this side of the API.
+            void DefineAdoptedStore(int elements) {
+                glBindBuffer(GL_SHADER_STORAGE_BUFFER, m_buffer);
+                glBufferData(GL_SHADER_STORAGE_BUFFER,
+                             static_cast<GLsizeiptr>(static_cast<GLsizeiptr>(elements) *
+                                                     static_cast<GLsizeiptr>(sizeof(unsigned int))),
+                             nullptr, GL_DYNAMIC_DRAW);
+            }
+
+            // GTEST_SKIP() returns from the function it is written in, so this cannot report
+            // through a return value; the caller pairs it with `if (IsSkipped()) return;`.
+            void SkipUnlessTheRoundtripCounterIsReadableHere() {
+                if (std::getenv(kLaneMarker) == nullptr) {
+                    GTEST_SKIP() << "runs only in its own lane: the MapPersistentRoundtrips. ctest entry "
+                                    "sets MGITEST_MPR_LANE together with MOBILEGL_PIPE_PUSH's P3a mask, "
+                                    "MOBILEGL_PIPE_STATS=1, MOBILEGL_PIPE_STATS_PERIOD=1 and a private "
+                                    "MOBILEGL_LOG_FILE_PATH. None of that is configured in the ambient "
+                                    "entries, and the ambient log is shared, so a read here would race a "
+                                    "neighbour's bring-up.";
+                    return;
+                }
+                if (!BuildMarkerIsSet("MGITEST_PIPE_PUSH_BUILD")) {
+                    GTEST_SKIP() << "this library was built without MOBILEGL_PIPE_PUSH, so "
+                                    "CallClass::MapPersistentRoundtrips does not exist (PipeStats.h "
+                                    "declares it inside the push guard, because growing the enum in a "
+                                    "pull build resizes the counter arrays and the name table - a G1 "
+                                    "break for a counter that could never leave zero) and the summary "
+                                    "line carries no mpr=. The entry is registered here anyway so that "
+                                    "`ctest -L integration-gpu` names the same tests in the pull build "
+                                    "and the push build (gate G2).";
+                    return;
+                }
+                if (!BuildMarkerIsSet("MGITEST_PIPE_RESOURCE_EMITTER_PRESENT")) {
+                    GTEST_SKIP() << "subsystem not implemented on this tree: no source under "
+                                    "MobileGL/MG_Impl/Pipe/ names MapPersistentRoundtrips, so nothing "
+                                    "emits map_persistent and mpr= is structurally zero. P3a package B "
+                                    "owns the client-side resource tracker; this entry arms itself when "
+                                    "it lands, whatever file that package puts the emitter in.";
+                    return;
+                }
+                if (PipeStatsWindow::LibraryLogPath().empty()) {
+                    GTEST_SKIP() << "the lane configured no MOBILEGL_LOG_FILE_PATH, and the library's "
+                                    "summary line is the only channel this module has for reading "
+                                    "PipeStats";
+                    return;
+                }
+            }
+
             unsigned int m_program = 0;
             GLuint m_buffer = 0;
             std::string m_buildLog;
@@ -153,4 +243,68 @@ void main() {
 
         glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 0, 0);
     }
+
+    // G10. N storage definitions of an adopted store cost N map-persistent round trips - not one
+    // per draw, and not zero.
+    //
+    // ONE case in this lane, and that is a constraint rather than a preference: it READS the
+    // library log, the log is a per-lane resource (the library opens it fopen(path, "w"), so every
+    // process in a lane truncates it), and a second reading entry in the same lane would race this
+    // one under `ctest -j` with a failure that looks exactly like "the counter was never emitted".
+    // The plumbing is therefore asserted first, with its own message, inside this one process.
+    TEST_F(StorageBufferRegrowScenario, NStorageDefinitionsCostNMapPersistentRoundtripsNotOnePerDraw) {
+        if (!Ready()) return;
+        SkipUnlessTheRoundtripCounterIsReadableHere();
+        if (IsSkipped()) return;
+
+        glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 0, m_buffer);
+        Gl().EndFrame(); // close the setup window: everything below is one window
+
+        for (int definition = 0; definition < kAdoptedDefinitions; ++definition) {
+            DefineAdoptedStore(kAdoptedBaseElements + definition * kAdoptedGrowthElements);
+            ASSERT_EQ(FirstGLError(), 0u) << "definition " << definition << " of the adopted store failed";
+            for (int dispatch = 0; dispatch < kDispatchesPerDefinition; ++dispatch) {
+                glUseProgram(m_program);
+                glDispatchCompute(static_cast<GLuint>(kDispatchedElements), 1, 1);
+                glMemoryBarrier(GL_BUFFER_UPDATE_BARRIER_BIT);
+            }
+        }
+
+        // The store is still the one the last definition made, and it still works: a counter
+        // assertion over a workload that silently stopped functioning would be measuring nothing.
+        std::vector<unsigned int> values(static_cast<std::size_t>(kDispatchedElements), 0xDEADBEEFu);
+        glBindBuffer(GL_SHADER_STORAGE_BUFFER, m_buffer);
+        glGetBufferSubData(GL_SHADER_STORAGE_BUFFER, 0,
+                           static_cast<GLsizeiptr>(values.size() * sizeof(unsigned int)), values.data());
+        EXPECT_EQ(FirstGLError(), 0u);
+        for (int i = 0; i < kDispatchedElements; ++i) {
+            EXPECT_EQ(values[static_cast<std::size_t>(i)], static_cast<unsigned int>(i + 1))
+                << "the adopted store's own dispatch did not write element " << i;
+        }
+
+        Gl().EndFrame(); // the swap that emits the window covering exactly the loop above
+        const PipeStatsWindow::Window window = PipeStatsWindow::LastFromLaneLog();
+        ASSERT_TRUE(window.found) << "no 'MGPipe stats:' line in " << PipeStatsWindow::LibraryLogPath()
+                                  << ". This IS a push build (the lane checked MGITEST_PIPE_PUSH_BUILD "
+                                     "before getting here), so either MOBILEGL_PIPE_STATS / "
+                                     "MOBILEGL_PIPE_STATS_PERIOD did not reach the process or no summary "
+                                     "line was emitted at all because nothing reached PipeStats::OnPresent.";
+        RecordProperty("stats_line", window.line.c_str());
+
+        const long long roundtrips = PipeStatsWindow::CounterOrAbsent(window, "mpr");
+        ASSERT_GE(roundtrips, 0)
+            << "the summary line carries no mpr= field, so this build's PipeStats has no "
+               "map-persistent-roundtrips counter to read: " << window.line;
+        EXPECT_EQ(roundtrips, static_cast<long long>(kAdoptedDefinitions))
+            << "an adopted store costs ONE map_persistent per STORAGE DEFINITION "
+               "(ARCHITECTURE.md:474). This window defined the store " << kAdoptedDefinitions
+            << " times and dispatched " << kDispatchesPerDefinition << " times against each of them, so "
+            << kAdoptedDefinitions << " is the whole cost. "
+            << (kAdoptedDefinitions * kDispatchesPerDefinition)
+            << " would mean an acquisition per DRAW - the regression this counter exists to catch - and 0 "
+               "would mean nothing emitted map_persistent at all. It reported: "
+            << window.line;
+
+        glBindBufferBase(GL_SHADER_STORAGE_BUFFER, 0, 0);
+    }
 } // namespace MGITest

MobileGL/MG_IntegrationTest/Scenarios/TextureParamsWithoutASamplerViewScenario.cpp

@@ -0,0 +1,775 @@
+// MobileGL - MobileGL/MG_IntegrationTest/Scenarios/TextureParamsWithoutASamplerViewScenario.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 - A TEXTURE'S PARAMETERS TAKE EFFECT EVEN WHEN IT HAS NO SAMPLER VIEW (gate G9), the
+// one scenario ROADMAP.md:20 names by hand and requires to be RED before P4a lands.
+//
+// THE DESIGN STATEMENT IT TESTS, ARCHITECTURE.md:100 (D10), verbatim: "SetTextureParams 按资源寻址、
+// 与 sampler view 分开（D10）：只作 FBO attachment / image 单元 / glCopyImageSubData 端点的纹理没有
+// sampler view，但 Espryt 对 attachment 也同步纹理参数，且 RequireImageBindableStorage 需要在前端参数
+// 版本不动时强制重同步." A texture that is only an attachment, only an image-unit binding or only a
+// copy endpoint has NO sampler view at all - so a design that carried texture parameters on the
+// view would silently drop them for exactly those textures. P4a addresses set_texture_params by
+// RESOURCE, independently of any binding, which is what makes the record exist for every texture
+// the moment its parameters move.
+//
+// THE FOUR CASES, and BRIEF-P4A.md D-E3 is where their expected verdicts come from:
+//
+//   AnAttachmentOnlyTexturesSwizzleReachesTheDriver                        green today, green after
+//   AReadAttachmentOnlyTexturesDepthStencilModeReachesTheDriver            D-E3 says RED today
+//   AnImageUnitOnlyTexturesSwizzleSurvivesARequireImageBindableStorageRemint green today, green after
+//   ACopyImageEndpointOnlyTexturesParamsReachTheDriver                     green today, green after
+//
+// [MEASURED, AND IT DOES NOT MATCH D-E3] All four are GREEN on the P4a contract commit (08192d72),
+// on llvmpipe, in the push build. The second one is green FOR A REASON THAT IS ITSELF THE FINDING,
+// and it is written here rather than in a review comment because the next person to try to make it
+// red needs to know why they cannot:
+//
+//   A texture parameter's only public-GL observable is a SAMPLE - nothing about an attachment, an
+//   image binding or a copy endpoint reads a swizzle or an aspect mode - and a sample puts the
+//   texture on SyncNeccessaryTextures' UNIT list. That walk calls SyncTextureParamsToBackend for
+//   every entry whose IsDrawSyncClean is false (DirectGLES.cpp:1896-1898), and IsDrawSyncClean is
+//   false whenever the frontend's parameter version has moved since the last sync
+//   (Managers.h:1399-1416, `m_syncedTextureParamsVersion != paramsVersion`). So the very act of
+//   observing the parameter repairs the state the observation was meant to catch: the gap between
+//   "the parameter moved on a read-attachment-only texture" and "the driver was told" is REAL, and
+//   it is closed by the next sampler binding, which is also the only thing that can see it.
+//
+// What that means for the gate, stated plainly so nobody reads a green here as evidence of
+// anything it is not:
+//
+//   * the four cases are a REGRESSION NET around D10, not the evidence for the change. They pin
+//     the design statement: a texture's parameters take effect however the texture is reached, and
+//     in particular they would go RED if any of the four reachability paths were ever made to
+//     depend on the texture having a sampler VIEW - which is exactly the coupling P4a's
+//     resource-addressed set_texture_params removes and the thing a later phase could reintroduce;
+//   * the "落地前必须红" artefact ROADMAP.md:20 asks for is NOT produced by the public-GL half
+//     of this file, and no public-GL integration scenario on a monolith tree can produce it.
+//     Producing it needs an observation of the DRIVER's texture object taken while the texture is
+//     still read-attachment-only. ID-19 rules that G9 is therefore a WHITE-BOX assertion, and this
+//     file now carries the SCENARIO half of it (the unit half is package D's,
+//     MG_Test/SanityTest.cpp's DirectGLESTextureSync.AnAttachmentOnlyTexturesParametersReachThe
+//     DriverWithNoSamplerView).
+//
+// THE WHITE-BOX HALF, and what it adds to the four cases below. Each case, at the point where its
+// texture is reachable ONLY its own way and BEFORE the observing sample, takes three readings
+// through MG_IntegrationTest/Harness/PipeApplyPeek.h and asserts all three:
+//
+//   (a) the APPLIER holds a set_texture_params record for this texture, at a non-zero ParamsSerial,
+//       carrying the field the case moved;
+//   (b) ESPRYT's applied value for the same texture - read back from the DRIVER, through the twin's
+//       own ES name - is that value ALREADY, not after the first sampler view;
+//   (c) Espryt holds NO SAMPLER VIEW for this texture yet, which is what turns (b) from "applied"
+//       into "applied WITHOUT one" and is the whole claim D10 makes.
+//
+// (c) is the assertion the public-GL half structurally cannot make: making it there would create
+// the view. (b) is the half that goes red on a backend that DEFERS - a tree where the parameter
+// push is gated on a sampler view existing is green on all four public-GL cases forever, because
+// the sample that observes the parameter is also what mints the view and repairs the state.
+//
+// WHAT THE THREE READINGS DO **NOT** COVER, so the next reader does not over-trust them
+// (esprytobj re-review N-9, carried here by request). D's unit probe drives
+// SyncTextureParamsToBackend directly, so the only deferral shape IT can see is one INSIDE that
+// function. These three run through the real per-frame paths and therefore also see a deferral
+// introduced ABOVE it - in SyncNeccessaryTextures, in the attachment walk, or in E's per-unit walk.
+// Between them the two halves cover both, and neither covers both alone.
+//
+// A READING THAT CANNOT BE TAKEN IS DECLINED BY NAME AND THE CASE CONTINUES - it is not a
+// GTEST_SKIP, and that is a deliberate departure from the shape the review sketched. These four
+// cases are dual-purpose: they are also the END-TO-END regression net around D10, and that net is
+// the ONLY thing measuring D10 on exactly the arms where the peek cannot look (the pull build,
+// which has no applier at all; the 0x1ff and 0 lanes, where the texture family is switched off;
+// Magma, which has no Espryt twin). Skipping the case there would delete the one verdict those
+// lanes carry in order to report the absence of a second one. The decline is printed, recorded as
+// a test property and named, so a lane that silently stopped taking the reading is visible in the
+// log rather than in a count.
+//
+// WHY THE SECOND ONE IS THE RED, mechanically (scout-espryt-framebuffer.md 2.6, re-opened at the
+// base ref). Today a texture's parameters ride on the UNIT BINDING and on the DRAW attachment set:
+//
+//   bound to a sampler unit <= the high-water mark  SyncNeccessaryTextures' unit list   -> synced
+//   attachment of the DRAW framebuffer              SyncNeccessaryTextures' FBO list    -> synced
+//   attachment of the READ framebuffer ONLY         SyncCurrentFBO -> SyncToBackend ->
+//                                                   SyncAttachmentObject, which calls
+//                                                   SyncMipmapsToBackend at Managers.cpp:7161
+//                                                   and NOTHING ELSE                    -> NOT synced
+//   bound to an image unit                          SyncImageTextureBinding ->
+//                                                   SyncTextureObjectToBackend, and
+//                                                   RequireImageBindableStorage additionally
+//                                                   forces m_forceTextureParamsResync    -> synced
+//   a glCopyImageSubData endpoint                   MakeGLESCopyImageEndpoint ->
+//                                                   SyncTextureObjectToBackend           -> synced
+//
+// SyncNeccessaryTextures' attachment list reads GetFramebufferBindingSlotChecked(Draw) only
+// (DirectGLES.cpp:1944), so a texture that is exclusively a READ attachment gets its STORAGE synced
+// and its PARAMETERS never. P4a closes that gap deliberately (D-E3): the record is addressed by
+// resource, and Espryt's SyncAttachmentObject applies parameters for ANY attachment, draw or read.
+// It is a behaviour change and it is the deliverable, not a drive-by dev fix (ROADMAP.md:98).
+//
+// HOW EACH CASE OBSERVES "REACHED THE DRIVER", and why the observation is always a LATER SAMPLE.
+// A texture parameter is by definition a sampling parameter: nothing about an attachment, an image
+// binding or a copy endpoint reads a swizzle or a depth/stencil aspect mode, so the only thing that
+// can see one is a sample. Each case therefore does the same three things -
+//
+//   1. put the texture through ONE of the five reachability paths above, and only that one,
+//   2. move a parameter while it is reachable ONLY that way (through the DSA entry points
+//      glTextureParameteri / glTextureSubImage2D, so no step of the setup ever binds the texture to
+//      a sampler unit - a bind would put it on the unit list and answer the question by accident),
+//   3. sample it once, at the end, and read the colour back.
+//
+// - and the difference between them is step 1 alone. A case that is red says: the parameter set
+// while the texture was reachable only that way did not survive to the sample.
+//
+// DIRECTGLES ONLY. The gap is Espryt's - it is a statement about SyncAttachmentObject and
+// SyncNeccessaryTextures - and P4a does not touch MG_Backend/DirectVulkan (D-Q). Magma answers the
+// same GL question through an entirely different path, so a red or a green there would be evidence
+// about P7's work rather than about this gate; the cases SKIP on any other backend, naming that.
+
+#include <cstdint>
+#include <cstdlib>
+#include <iostream>
+#include <string>
+#include <vector>
+
+#include "../Harness/HeadlessGL.h"
+#include "../Harness/PipeApplyPeek.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 kInset = 2;
+        constexpr int kTextureSize = 4;
+
+        constexpr const char* kQuadVS = R"(#version 330 core
+in vec2 aPos;
+out vec2 vUv;
+void main() {
+    vUv = aPos * 0.5 + 0.5;
+    gl_Position = vec4(aPos, 0.0, 1.0);
+}
+)";
+
+        // Samples texel (0,0) with an explicit fetch: no filtering, no derivatives, no wrap - so
+        // the colour that comes back is the texel as the driver's swizzle presents it and nothing
+        // else can move it.
+        constexpr const char* kFetchFS = R"(#version 330 core
+uniform sampler2D uTex;
+out vec4 oColor;
+void main() { oColor = texelFetch(uTex, ivec2(0, 0), 0); }
+)";
+
+        // The stencil aspect of a packed depth/stencil texture is an UNSIGNED INTEGER texture, so
+        // it needs a usampler2D. The case that uses it turns "the stencil value is what was
+        // cleared" into a colour, because a colour is the only thing this harness can read back.
+        constexpr const char* kStencilFetchFS = R"(#version 330 core
+uniform usampler2D uTex;
+uniform uint uExpected;
+out vec4 oColor;
+void main() {
+    uint value = texelFetch(uTex, ivec2(0, 0), 0).r;
+    oColor = (value == uExpected) ? vec4(0.0, 1.0, 0.0, 1.0) : vec4(1.0, 0.0, 0.0, 1.0);
+}
+)";
+
+        struct Vertex {
+            float x, y;
+        };
+
+        class TextureParamsWithoutASamplerViewScenario : public ScenarioTest {
+        protected:
+            void SetUp() override {
+                ScenarioTest::SetUp();
+                if (!Ready()) return;
+                if (Gl().BackendName() != "DirectGLES") {
+                    GTEST_SKIP() << "DirectGLES only: this scenario is about Espryt's own reachability "
+                                    "table - SyncNeccessaryTextures' attachment list walks the DRAW "
+                                    "slot only (DirectGLES.cpp:1944) and SyncAttachmentObject syncs "
+                                    "storage and not parameters (Managers.cpp:7161). "
+                                 << Gl().BackendName()
+                                 << " answers the same GL question through a different path, so a "
+                                    "verdict here would be evidence about that backend rather than "
+                                    "about this gate (P4a touches no DirectVulkan source but "
+                                    "MagmaPipeArms.h, D-Q).";
+                }
+                std::string error;
+                m_fetchProgram = CompileProgram(kQuadVS, kFetchFS, &error);
+                ASSERT_NE(m_fetchProgram, 0u) << error;
+
+                static const Vertex quad[6] = {{-1.0f, -1.0f}, {1.0f, -1.0f}, {1.0f, 1.0f},
+                                               {-1.0f, -1.0f}, {1.0f, 1.0f},  {-1.0f, 1.0f}};
+                glGenBuffers(1, &m_quadBuffer);
+                glBindBuffer(GL_ARRAY_BUFFER, m_quadBuffer);
+                glBufferData(GL_ARRAY_BUFFER, sizeof(quad), quad, GL_STATIC_DRAW);
+                glGenVertexArrays(1, &m_vao);
+                glBindVertexArray(m_vao);
+                glEnableVertexAttribArray(0);
+                glVertexAttribPointer(0, 2, GL_FLOAT, GL_FALSE, sizeof(Vertex), nullptr);
+                glBindVertexArray(0);
+                ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR)) << "the scene setup left a GL error behind";
+            }
+
+            void TearDown() override {
+                if (!Ready() || IsSkipped()) return;
+                glUseProgram(0);
+                glBindVertexArray(0);
+                if (m_vao != 0) glDeleteVertexArrays(1, &m_vao);
+                if (m_quadBuffer != 0) glDeleteBuffers(1, &m_quadBuffer);
+                if (m_fetchProgram != 0) glDeleteProgram(m_fetchProgram);
+            }
+
+            // Direct State Access is how every setup step below touches a texture, and it is the
+            // whole reason the cases can claim "this texture never had a sampler view": the classic
+            // entry points (glTexImage2D, glTexParameteri) all require the texture to be BOUND to a
+            // unit first, and a bind is exactly what puts it on SyncNeccessaryTextures' unit list.
+            // A case that used them would sync the parameters through the path it is trying to
+            // exclude and would be green for the wrong reason, on every tree, forever.
+            bool DirectStateAccessIsAvailable() {
+                GLuint probe = 0;
+                glCreateTextures(GL_TEXTURE_2D, 1, &probe);
+                const bool ok = FirstGLError() == GLenum(GL_NO_ERROR) && probe != 0;
+                if (probe != 0) glDeleteTextures(1, &probe);
+                return ok;
+            }
+
+            void SkipWithoutDirectStateAccess() {
+                if (!DirectStateAccessIsAvailable()) {
+                    GTEST_SKIP() << "glCreateTextures is not usable here, and every case in this file "
+                                    "needs the DSA entry points: the classic ones bind the texture to "
+                                    "a unit, which is the reachability path these cases exist to "
+                                    "exclude. A case that fell back to them would be green for the "
+                                    "wrong reason rather than measuring anything.";
+                }
+            }
+
+            // A 4x4 RGBA8 texture, one solid colour, created and filled WITHOUT EVER BINDING IT.
+            GLuint MakeSolidTextureWithoutBinding(std::uint8_t r, std::uint8_t g, std::uint8_t b) {
+                std::vector<std::uint8_t> texels(kTextureSize * kTextureSize * 4);
+                for (std::size_t i = 0; i < texels.size(); i += 4) {
+                    texels[i] = r;
+                    texels[i + 1] = g;
+                    texels[i + 2] = b;
+                    texels[i + 3] = 255;
+                }
+                GLuint texture = 0;
+                glCreateTextures(GL_TEXTURE_2D, 1, &texture);
+                glTextureStorage2D(texture, 1, GL_RGBA8, kTextureSize, kTextureSize);
+                glTextureSubImage2D(texture, 0, 0, 0, kTextureSize, kTextureSize, GL_RGBA,
+                                    GL_UNSIGNED_BYTE, texels.data());
+                glTextureParameteri(texture, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+                glTextureParameteri(texture, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
+                return texture;
+            }
+
+            // THE OBSERVATION, shared by every case: bind the texture to a unit for the first time
+            // in its life and read one texel back through the default framebuffer.
+            Image SampleAndRead(GLuint program, GLuint texture) {
+                BindDefaultFramebuffer();
+                glViewport(0, 0, Gl().Width(), Gl().Height());
+                ClearTo(0.0f, 0.0f, 0.0f, 1.0f);
+                glUseProgram(program);
+                glUniform1i(glGetUniformLocation(program, "uTex"), 0);
+                glActiveTexture(GL_TEXTURE0);
+                glBindTexture(GL_TEXTURE_2D, texture);
+                glBindVertexArray(m_vao);
+                glDrawArrays(GL_TRIANGLES, 0, 6);
+                const Image image = ReadPixels(Gl().Width(), Gl().Height());
+                Gl().EndFrame();
+                glBindTexture(GL_TEXTURE_2D, 0);
+                return image;
+            }
+
+            ::testing::AssertionResult WholeViewportIs(const Image& image, const char* expected,
+                                                       const std::string& when) {
+                return RegionIsMostly(image, kInset, image.Width() - kInset, kInset,
+                                      image.Height() - kInset, expected, 0.0, when);
+            }
+
+            // R -> ZERO and G -> ONE, so a RED texel samples as GREEN if and only if the swizzle
+            // reached the driver, and as RED if it did not. Two colours the harness can name, from
+            // one parameter change, with no third outcome that could be mistaken for either.
+            void SwizzleRedIntoGreen(GLuint texture) {
+                glTextureParameteri(texture, GL_TEXTURE_SWIZZLE_R, GL_ZERO);
+                glTextureParameteri(texture, GL_TEXTURE_SWIZZLE_G, GL_ONE);
+                glTextureParameteri(texture, GL_TEXTURE_SWIZZLE_B, GL_ZERO);
+            }
+
+            // Which of the two parameters a case moved, and therefore which one the white-box
+            // reading has to find on both sides of the seam. Two, because they are the two the
+            // four cases use and because a peek that reported "some parameter" would be green for
+            // a backend that applied the wrong one.
+            enum class MovedParameter { Swizzle, DepthStencilMode };
+
+            // ------------------------------------------------------------------------------
+            // G9's WHITE-BOX READING (ID-19). Called by every case at the point where its
+            // texture is reachable only its own way and BEFORE the observing sample - which is
+            // the whole of the design, because the sample repairs what it observes.
+            //
+            // `expectedSwizzle` is the four GL enums the case set (or left at their defaults);
+            // `expectedDepthStencilMode` is GL_DEPTH_COMPONENT or GL_STENCIL_INDEX. Both are
+            // always passed and `moved` says which one is the case's subject, so a reader of a
+            // failure can see the untouched half beside the moved one.
+            void TakeTheWhiteBoxReadingBeforeAnySample(GLuint texture, GLenum target,
+                                                       MovedParameter moved,
+                                                       const GLint expectedSwizzle[4],
+                                                       GLint expectedDepthStencilMode,
+                                                       const char* whatMadeItReachable) {
+                const char* const movedName =
+                    moved == MovedParameter::Swizzle ? "GL_TEXTURE_SWIZZLE_*"
+                                                     : "GL_DEPTH_STENCIL_TEXTURE_MODE";
+
+                PipeTextureParamsRecordPeek record{};
+                if (!PeekPipeTextureParamsRecord(static_cast<unsigned>(texture), &record)) {
+                    DeclineTheWhiteBoxReading(
+                        "no set_texture_params record for this texture in the applier. Either "
+                        "there is no applier here (a PULL build: MGPipeApplierState is "
+                        "#if MOBILEGL_PIPE_PUSH), or this lane's MOBILEGL_PIPE_PUSH leaves "
+                        "kMGPipeSubsystemTextureResources (bit 10) clear, or no backend "
+                        "registered MGPipeResourceOps so the client never emitted (c0f). The "
+                        "end-to-end half of this case below is unaffected and still decides it.");
+                    return;
+                }
+
+                // From here the reading WAS taken, so everything is a hard assertion: a record
+                // that exists and does not carry the parameter is exactly the finding.
+                EXPECT_NE(record.ParamsSerial, 0u)
+                    << "the applier holds a resource record for texture " << texture
+                    << " at handle {" << record.Slot << ", " << record.Gen
+                    << "} but its ParamsSerial is 0, i.e. NO set_texture_params has ever been "
+                       "applied to it - and this case moved " << movedName << " while the texture "
+                       "was " << whatMadeItReachable
+                    << ". A parameter change on a texture with no sampler view has to produce a "
+                       "record addressed BY RESOURCE (D10, D-E1); a zero here means the client "
+                       "never emitted one, which is the coupling P4a exists to remove reappearing "
+                       "on the emitter's side of the seam.";
+
+                if (moved == MovedParameter::Swizzle) {
+                    for (int channel = 0; channel < 4; ++channel) {
+                        EXPECT_EQ(record.Swizzle[channel], static_cast<int>(expectedSwizzle[channel]))
+                            << "the applier's set_texture_params record for texture " << texture
+                            << " carries the wrong swizzle in channel " << channel
+                            << " (record 0x" << std::hex << record.Swizzle[channel] << ", expected 0x"
+                            << expectedSwizzle[channel] << std::dec
+                            << "). The record is what Espryt reads, so a wrong value here is a "
+                               "wrong value everywhere downstream of it.";
+                    }
+                } else {
+                    EXPECT_EQ(record.DepthStencilMode, static_cast<int>(expectedDepthStencilMode))
+                        << "the applier's set_texture_params record for texture " << texture
+                        << " carries GL_DEPTH_STENCIL_TEXTURE_MODE 0x" << std::hex
+                        << record.DepthStencilMode << ", expected 0x" << expectedDepthStencilMode
+                        << std::dec << ".";
+                }
+
+                // (c) - and it is checked BEFORE (b) is read, because (b) reads the driver and a
+                // reader of a failure needs to know the view question was answered on the state
+                // this case built rather than on anything the peek did.
+                bool hasSamplerView = true;
+                if (!PeekEsprytHasSamplerViewForTexture(static_cast<unsigned>(texture),
+                                                        &hasSamplerView)) {
+                    DeclineTheWhiteBoxReading(
+                        "Espryt holds no twin for this texture, so neither the sampler-view "
+                        "question nor the applied-value one can be asked here. On a backend other "
+                        "than DirectGLES that is the designed state (P4a touches no DirectVulkan "
+                        "source but MagmaPipeArms.h, D-Q).");
+                    return;
+                }
+                EXPECT_FALSE(hasSamplerView)
+                    << "Espryt already holds a SAMPLER VIEW for texture " << texture
+                    << ", which was " << whatMadeItReachable
+                    << " and has never been bound to a sampler unit in this case. The whole claim "
+                       "of D10 is that a texture reached this way has no view, so if one exists "
+                       "the reading below cannot separate 'applied by resource' from 'applied "
+                       "through the view' and this case has stopped measuring G9.";
+
+                EsprytAppliedTextureParamsPeek applied{};
+                if (!PeekEsprytAppliedTextureParams(static_cast<unsigned>(texture),
+                                                    static_cast<unsigned>(target), &applied)) {
+                    DeclineTheWhiteBoxReading(
+                        "Espryt's applied value could not be read back from the driver (no twin, "
+                        "no ES name yet, or a target this peek has no binding query for).");
+                    return;
+                }
+
+                std::cout << "[ TextureParamsWithoutASamplerView ] white-box: texture " << texture
+                          << " -> applier handle {" << record.Slot << ", " << record.Gen
+                          << "} paramsSerial " << record.ParamsSerial << ", Espryt ES name "
+                          << applied.BackendTextureId << ", sampler view: none, " << movedName
+                          << " applied before any sample" << std::endl;
+
+                if (moved == MovedParameter::Swizzle) {
+                    for (int channel = 0; channel < 4; ++channel) {
+                        EXPECT_EQ(applied.Swizzle[channel], static_cast<int>(expectedSwizzle[channel]))
+                            << "ESPRYT HAS NOT APPLIED THE SWIZZLE YET. Channel " << channel
+                            << " of the driver texture (ES name " << applied.BackendTextureId
+                            << ") reads 0x" << std::hex << applied.Swizzle[channel] << ", the "
+                            << "application set 0x" << expectedSwizzle[channel] << std::dec
+                            << ", and the applier's record already carries the right value - so "
+                               "the record reached the server and the server has not pushed it. "
+                               "The texture was " << whatMadeItReachable
+                            << " and has NO sampler view (asserted above), which makes this "
+                               "exactly the deferred-to-first-view shape G9 exists to catch: the "
+                               "sample at the end of this case would repair it, and the "
+                               "end-to-end assertion below would then pass on a driver that was "
+                               "told late. That is the half no public-GL case can see.";
+                    }
+                } else {
+                    if (!applied.DepthStencilModeIsReadable) {
+                        DeclineTheWhiteBoxReading(
+                            "this driver would not answer glGetTexParameteriv("
+                            "GL_DEPTH_STENCIL_TEXTURE_MODE), so the applied aspect mode cannot be "
+                            "read back. The record half above was still asserted.");
+                        return;
+                    }
+                    EXPECT_EQ(applied.DepthStencilMode, static_cast<int>(expectedDepthStencilMode))
+                        << "ESPRYT HAS NOT APPLIED THE DEPTH/STENCIL ASPECT MODE YET. The driver "
+                           "texture (ES name " << applied.BackendTextureId << ") reads 0x"
+                        << std::hex << applied.DepthStencilMode << ", the application set 0x"
+                        << expectedDepthStencilMode << std::dec
+                        << ", and the applier's record already carries the right value. The "
+                           "texture was " << whatMadeItReachable
+                        << " and has no sampler view, so this is D-E3's gap measured directly "
+                           "rather than through a sample that would repair it: a driver left at "
+                           "GL_DEPTH_COMPONENT samples the DEPTH bits where the application asked "
+                           "for stencil.";
+                }
+            }
+
+            // Printed, recorded and named, never silent - a lane that stopped taking the reading
+            // must be visible in the log. See this file's header for why it is not a GTEST_SKIP.
+            void DeclineTheWhiteBoxReading(const std::string& why) {
+                std::cout << "[ TextureParamsWithoutASamplerView ] white-box reading DECLINED: "
+                          << why << std::endl;
+                RecordProperty("g9_white_box", "declined");
+                RecordProperty("g9_white_box_reason", why.c_str());
+            }
+
+            GLuint m_fetchProgram = 0;
+            GLuint m_vao = 0;
+            GLuint m_quadBuffer = 0;
+        };
+
+        // The swizzle SwizzleRedIntoGreen leaves behind, as GL enums: R -> ZERO, G -> ONE,
+        // B -> ZERO and A untouched at its GL default. Written once here because both the
+        // applier record and the driver read-back are compared against it.
+        constexpr GLint kRedIntoGreenSwizzle[4] = {GL_ZERO, GL_ONE, GL_ZERO, GL_ALPHA};
+        // A texture whose aspect mode was never touched, i.e. the GL initial value - which is
+        // also what a zeroed MGPTextureParams::DepthStencilMode decodes to (MGPipeTypes.h).
+        constexpr GLint kUntouchedDepthStencilMode = GL_DEPTH_COMPONENT;
+        // ...and the identity swizzle, for the case whose subject is the aspect mode: the moved
+        // half is asserted, and the untouched half is carried so a failure prints both.
+        constexpr GLint kUntouchedSwizzle[4] = {GL_RED, GL_GREEN, GL_BLUE, GL_ALPHA};
+
+        // ------------------------------------------------------------------------------------
+        // 1. DRAW ATTACHMENT ONLY. Green today (SyncNeccessaryTextures' FBO list walks the draw
+        //    slot and calls SyncTextureObjectToBackend, which syncs parameters) and green after.
+        //    It is the regression net for the half of D-E3 that already works: P4a moves the
+        //    parameter sync from the sync list onto the record, and this case is what says the
+        //    move did not lose the case that used to be covered.
+        // ------------------------------------------------------------------------------------
+        TEST_F(TextureParamsWithoutASamplerViewScenario, AnAttachmentOnlyTexturesSwizzleReachesTheDriver) {
+            if (!Ready()) return;
+            SkipWithoutDirectStateAccess();
+            if (IsSkipped()) return;
+
+            const GLuint texture = MakeSolidTextureWithoutBinding(255, 0, 0);
+            GLuint fbo = 0;
+            glGenFramebuffers(1, &fbo);
+            glBindFramebuffer(GL_DRAW_FRAMEBUFFER, fbo);
+            glFramebufferTexture2D(GL_DRAW_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, texture, 0);
+            ASSERT_EQ(glCheckFramebufferStatus(GL_DRAW_FRAMEBUFFER), GLenum(GL_FRAMEBUFFER_COMPLETE));
+
+            // The parameter moves while the texture is reachable ONLY as a draw attachment...
+            SwizzleRedIntoGreen(texture);
+            // ...and a frame runs with it bound that way, so whatever the draw path syncs, syncs.
+            glViewport(0, 0, kTextureSize, kTextureSize);
+            ClearTo(1.0f, 0.0f, 0.0f, 1.0f);
+            BindDefaultFramebuffer();
+            Gl().EndFrame();
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR)) << "the draw-attachment frame left a GL error";
+
+            glBindFramebuffer(GL_FRAMEBUFFER, 0);
+            glDeleteFramebuffers(1, &fbo);
+
+            // G9's white-box reading, taken here: the texture has been a draw attachment and
+            // nothing else, and the sample below has not happened yet.
+            TakeTheWhiteBoxReadingBeforeAnySample(texture, GL_TEXTURE_2D, MovedParameter::Swizzle,
+                                                  kRedIntoGreenSwizzle, kUntouchedDepthStencilMode,
+                                                  "an attachment of the DRAW framebuffer and "
+                                                  "nothing else");
+
+            const Image image = SampleAndRead(m_fetchProgram, texture);
+            EXPECT_TRUE(WholeViewportIs(image, "green",
+                                        "a texture that was only ever a DRAW attachment, sampled "
+                                        "after its swizzle moved"))
+                << "the swizzle set while this texture was reachable only as a draw-framebuffer "
+                   "attachment did not reach the driver: a red texel with R->ZERO, G->ONE must "
+                   "sample as green.";
+
+            GLuint cleanup = texture;
+            glDeleteTextures(1, &cleanup);
+        }
+
+        // ------------------------------------------------------------------------------------
+        // 2. READ ATTACHMENT ONLY - THE MANDATORY RED (ROADMAP.md:20, D-E3, G9).
+        //
+        //    The parameter is GL_DEPTH_STENCIL_TEXTURE_MODE and the texture is a packed
+        //    depth/stencil one, because that is the parameter whose absence is not a mis-filtered
+        //    picture but the WRONG ASPECT: a driver left at the GL default samples the depth bits
+        //    where the application asked for stencil, and the value that comes back is not the
+        //    stencil that was written. RecreateBackendTexture's own comment says exactly this -
+        //    "the mode makes it visible because falling back to the default silently samples the
+        //    wrong aspect rather than merely mis-filtering" (Managers.cpp:4826-4833).
+        //
+        //    The framebuffer is bound to GL_READ_FRAMEBUFFER and the DRAW target is left on the
+        //    default framebuffer for the whole window, which is the ONE thing that separates this
+        //    case from the one above.
+        // ------------------------------------------------------------------------------------
+        TEST_F(TextureParamsWithoutASamplerViewScenario,
+               AReadAttachmentOnlyTexturesDepthStencilModeReachesTheDriver) {
+            if (!Ready()) return;
+            SkipWithoutDirectStateAccess();
+            if (IsSkipped()) return;
+
+            std::string error;
+            const GLuint stencilProgram = CompileProgram(kQuadVS, kStencilFetchFS, &error);
+            ASSERT_NE(stencilProgram, 0u) << error;
+
+            GLuint texture = 0;
+            glCreateTextures(GL_TEXTURE_2D, 1, &texture);
+            glTextureStorage2D(texture, 1, GL_DEPTH24_STENCIL8, kTextureSize, kTextureSize);
+            glTextureParameteri(texture, GL_TEXTURE_MIN_FILTER, GL_NEAREST);
+            glTextureParameteri(texture, GL_TEXTURE_MAG_FILTER, GL_NEAREST);
+            if (FirstGLError() != GLenum(GL_NO_ERROR)) {
+                glDeleteTextures(1, &texture);
+                glDeleteProgram(stencilProgram);
+                GTEST_SKIP() << "this driver would not create an immutable DEPTH24_STENCIL8 texture, "
+                                "so there is no packed depth/stencil aspect here to sample and the "
+                                "case cannot answer";
+            }
+
+            // Write a stencil value through the texture AS A DRAW ATTACHMENT once, so that there is
+            // something in the stencil aspect to read. This is setup, not the window: the window
+            // below never makes it a draw attachment again.
+            GLuint fbo = 0;
+            glGenFramebuffers(1, &fbo);
+            glBindFramebuffer(GL_FRAMEBUFFER, fbo);
+            glFramebufferTexture2D(GL_FRAMEBUFFER, GL_DEPTH_STENCIL_ATTACHMENT, GL_TEXTURE_2D, texture, 0);
+            glDrawBuffer(GL_NONE);
+            glReadBuffer(GL_NONE);
+            if (glCheckFramebufferStatus(GL_FRAMEBUFFER) != GLenum(GL_FRAMEBUFFER_COMPLETE)) {
+                BindDefaultFramebuffer();
+                glDeleteFramebuffers(1, &fbo);
+                glDeleteTextures(1, &texture);
+                glDeleteProgram(stencilProgram);
+                GTEST_SKIP() << "a depth-stencil-only framebuffer is incomplete on this driver, so "
+                                "the stencil aspect cannot be written and the case cannot answer";
+            }
+            constexpr GLint kStencil = 42;
+            glViewport(0, 0, kTextureSize, kTextureSize);
+            glStencilMask(0xFFu);
+            glClearBufferfi(GL_DEPTH_STENCIL, 0, 0.5f, kStencil);
+            BindDefaultFramebuffer();
+            Gl().EndFrame();
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR)) << "clearing the stencil aspect left a GL error";
+
+            // ---- the window: the texture is reachable ONLY as a READ attachment ----
+            //
+            // The DRAW binding is the default framebuffer throughout, so SyncNeccessaryTextures'
+            // attachment list - which walks the DRAW slot only - never sees this texture, and
+            // SyncCurrentFBO's read path reaches it through SyncAttachmentObject, which syncs
+            // storage and not parameters. That is the gap.
+            glBindFramebuffer(GL_READ_FRAMEBUFFER, fbo);
+            glBindFramebuffer(GL_DRAW_FRAMEBUFFER, 0);
+            glTextureParameteri(texture, GL_DEPTH_STENCIL_TEXTURE_MODE, GL_STENCIL_INDEX);
+            // A frame with the read binding live, and it has to contain a REAL DRAW: SyncCurrentFBO
+            // and the whole sync-list walk run at the validate point, so a frame that only cleared
+            // and swapped would never reach the read-side path this case is about.
+            glViewport(0, 0, Gl().Width(), Gl().Height());
+            ClearTo(0.0f, 0.0f, 0.0f, 1.0f);
+            glUseProgram(m_fetchProgram);
+            glUniform1i(glGetUniformLocation(m_fetchProgram, "uTex"), 0);
+            glActiveTexture(GL_TEXTURE0);
+            glBindTexture(GL_TEXTURE_2D, 0);
+            glBindVertexArray(m_vao);
+            glDrawArrays(GL_TRIANGLES, 0, 6);
+            Gl().EndFrame();
+            glBindFramebuffer(GL_READ_FRAMEBUFFER, 0);
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR)) << "the read-attachment frame left a GL error";
+
+            // G9's white-box reading, and this is the case it matters most for: the aspect
+            // mode was set while the texture was reachable ONLY as a read attachment, and the
+            // observation below is a sample that would repair an unsynced parameter on its way
+            // to reporting it.
+            TakeTheWhiteBoxReadingBeforeAnySample(texture, GL_TEXTURE_2D,
+                                                  MovedParameter::DepthStencilMode,
+                                                  kUntouchedSwizzle, GL_STENCIL_INDEX,
+                                                  "an attachment of the READ framebuffer and "
+                                                  "nothing else");
+
+            // ---- the observation ----
+            BindDefaultFramebuffer();
+            glViewport(0, 0, Gl().Width(), Gl().Height());
+            ClearTo(0.0f, 0.0f, 0.0f, 1.0f);
+            glUseProgram(stencilProgram);
+            glUniform1i(glGetUniformLocation(stencilProgram, "uTex"), 0);
+            glUniform1ui(glGetUniformLocation(stencilProgram, "uExpected"),
+                         static_cast<GLuint>(kStencil));
+            glActiveTexture(GL_TEXTURE0);
+            glBindTexture(GL_TEXTURE_2D, texture);
+            glBindVertexArray(m_vao);
+            glDrawArrays(GL_TRIANGLES, 0, 6);
+            const Image image = ReadPixels(Gl().Width(), Gl().Height());
+            Gl().EndFrame();
+            glBindTexture(GL_TEXTURE_2D, 0);
+
+            const GLenum sampleError = FirstGLError();
+            std::cout << "[ TextureParamsWithoutASamplerView ] read-attachment-only "
+                         "GL_DEPTH_STENCIL_TEXTURE_MODE=GL_STENCIL_INDEX, expecting stencil "
+                      << kStencil << "; sample GL error 0x" << std::hex << sampleError << std::dec
+                      << std::endl;
+
+            EXPECT_TRUE(WholeViewportIs(image, "green",
+                                        "a texture that was only ever a READ attachment, sampled "
+                                        "through its stencil aspect"))
+                << "GL_DEPTH_STENCIL_TEXTURE_MODE = GL_STENCIL_INDEX was set while this texture was "
+                   "reachable ONLY as an attachment of the READ framebuffer, and the sample did not "
+                   "come back as the stencil value that was cleared into it. The parameter did not "
+                   "reach the driver, and - because the sample itself would have repaired an "
+                   "unsynced parameter (see this file's header: the unit sync list calls "
+                   "SyncTextureParamsToBackend whenever the params version moved) - a red here means "
+                   "something stronger than the D-E3 gap: a reachability path that does not sync "
+                   "parameters AT ALL, i.e. the sampler-view coupling ARCHITECTURE.md:100 (D10) "
+                   "exists to remove. Read the case's own stdout line for the GL error the sample "
+                   "raised before assuming an aspect-mode bug.";
+
+            BindDefaultFramebuffer();
+            glDeleteFramebuffers(1, &fbo);
+            glDeleteTextures(1, &texture);
+            glDeleteProgram(stencilProgram);
+        }
+
+        // ------------------------------------------------------------------------------------
+        // 3. IMAGE UNIT ONLY, across a storage RE-MINT. Green today and green after, and the
+        //    reason it is green is the thing P4a must not lose: glBindImageTexture drives
+        //    RequireImageBindableStorage, which re-mints the storage in a possibly WIDENED carrier
+        //    and sets m_forceTextureParamsResync (Managers.cpp:4787) precisely because the
+        //    frontend's parameter version does not move across that transition. A parameter sync
+        //    gated only on the frontend version would leave the driver at its defaults and sample
+        //    the carrier's surplus channels raw.
+        // ------------------------------------------------------------------------------------
+        TEST_F(TextureParamsWithoutASamplerViewScenario,
+               AnImageUnitOnlyTexturesSwizzleSurvivesARequireImageBindableStorageRemint) {
+            if (!Ready()) return;
+            SkipWithoutDirectStateAccess();
+            if (IsSkipped()) return;
+
+            const GLuint texture = MakeSolidTextureWithoutBinding(255, 0, 0);
+
+            // The parameter moves while the texture is reachable only as an image-unit binding...
+            SwizzleRedIntoGreen(texture);
+            glBindImageTexture(0, texture, 0, GL_FALSE, 0, GL_READ_ONLY, GL_RGBA8);
+            const GLenum bindError = FirstGLError();
+            if (bindError != GLenum(GL_NO_ERROR)) {
+                glBindImageTexture(0, 0, 0, GL_FALSE, 0, GL_READ_ONLY, GL_RGBA8);
+                GLuint cleanup = texture;
+                glDeleteTextures(1, &cleanup);
+                GTEST_SKIP() << "glBindImageTexture is not usable here (GL error 0x" << std::hex
+                             << bindError << std::dec
+                             << "), so the RequireImageBindableStorage transition this case is about "
+                                "cannot be reached";
+            }
+            // ...and a frame runs with the image binding live, which is what drives the re-mint.
+            glViewport(0, 0, Gl().Width(), Gl().Height());
+            ClearTo(0.0f, 0.0f, 0.0f, 1.0f);
+            Gl().EndFrame();
+            glBindImageTexture(0, 0, 0, GL_FALSE, 0, GL_READ_ONLY, GL_RGBA8);
+
+            // G9's white-box reading. RGBA8 is a core image format, so no widening carrier is
+            // minted and the driver's swizzle is the application's own - the composition
+            // RecreateBackendTexture applies for a NON-core format would show up here as a
+            // legitimate difference and this case deliberately does not use one.
+            TakeTheWhiteBoxReadingBeforeAnySample(texture, GL_TEXTURE_2D, MovedParameter::Swizzle,
+                                                  kRedIntoGreenSwizzle, kUntouchedDepthStencilMode,
+                                                  "an image-unit binding and nothing else, across "
+                                                  "a RequireImageBindableStorage re-mint");
+
+            const Image image = SampleAndRead(m_fetchProgram, texture);
+            EXPECT_TRUE(WholeViewportIs(image, "green",
+                                        "a texture that was only ever an image-unit binding, sampled "
+                                        "after its swizzle moved and its storage was re-minted"))
+                << "the swizzle did not survive the RequireImageBindableStorage re-mint. The re-mint "
+                   "creates a new driver texture at the ES defaults without moving the frontend's "
+                   "parameter version, so the forced resync (Managers.cpp:4787) is the only thing "
+                   "that puts the application's parameters back.";
+
+            GLuint cleanup = texture;
+            glDeleteTextures(1, &cleanup);
+        }
+
+        // ------------------------------------------------------------------------------------
+        // 4. glCopyImageSubData ENDPOINT ONLY. Green today (MakeGLESCopyImageEndpoint calls
+        //    SyncTextureObjectToBackend, DirectGLES.cpp:7588) and green after. The endpoint is the
+        //    DESTINATION, so the case also proves the copy itself still lands: a swizzled read of
+        //    the copied texel is only meaningful if the texel arrived.
+        // ------------------------------------------------------------------------------------
+        TEST_F(TextureParamsWithoutASamplerViewScenario,
+               ACopyImageEndpointOnlyTexturesParamsReachTheDriver) {
+            if (!Ready()) return;
+            SkipWithoutDirectStateAccess();
+            if (IsSkipped()) return;
+
+            const GLuint source = MakeSolidTextureWithoutBinding(255, 0, 0);
+            const GLuint destination = MakeSolidTextureWithoutBinding(0, 0, 255);
+
+            // The parameter moves while the destination is reachable only as a copy endpoint...
+            SwizzleRedIntoGreen(destination);
+            glCopyImageSubData(source, GL_TEXTURE_2D, 0, 0, 0, 0, destination, GL_TEXTURE_2D, 0, 0, 0,
+                               0, kTextureSize, kTextureSize, 1);
+            const GLenum copyError = FirstGLError();
+            if (copyError != GLenum(GL_NO_ERROR)) {
+                GLuint cleanup[2] = {source, destination};
+                glDeleteTextures(2, cleanup);
+                GTEST_SKIP() << "glCopyImageSubData is not usable here (GL error 0x" << std::hex
+                             << copyError << std::dec << "), so there is no copy endpoint to be";
+            }
+            Gl().EndFrame();
+
+            // G9's white-box reading, on the DESTINATION - the endpoint whose parameters moved.
+            TakeTheWhiteBoxReadingBeforeAnySample(destination, GL_TEXTURE_2D,
+                                                  MovedParameter::Swizzle, kRedIntoGreenSwizzle,
+                                                  kUntouchedDepthStencilMode,
+                                                  "a glCopyImageSubData destination and nothing "
+                                                  "else");
+
+            // ...and the destination now holds the source's RED texel, which the swizzle must turn
+            // into GREEN when it is finally sampled.
+            const Image image = SampleAndRead(m_fetchProgram, destination);
+            EXPECT_TRUE(WholeViewportIs(image, "green",
+                                        "a texture that was only ever a glCopyImageSubData endpoint, "
+                                        "sampled after its swizzle moved"))
+                << "the three readings are distinct and each names its own cause: GREEN is the pass "
+                   "(the swizzle reached the driver); RED means the copy landed and the swizzle did "
+                   "not; BLUE means neither happened, i.e. the destination is still its own original "
+                   "texel and glCopyImageSubData wrote nothing. The swizzle is what turns any texel "
+                   "into (0,1,0), so the colour separates the two failures rather than merging them.";
+
+            GLuint cleanup[2] = {source, destination};
+            glDeleteTextures(2, cleanup);
+        }
+
+    } // namespace
+} // namespace MGITest

MobileGL/MG_IntegrationTest/Scenarios/TextureUploadShapeScenario.cpp

@@ -0,0 +1,396 @@
+// MobileGL - MobileGL/MG_IntegrationTest/Scenarios/TextureUploadShapeScenario.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 - THE TEXTURE UPLOAD SHAPE, RECORDED (BRIEF-P4A.md D-D4). NOT A GATE IN P4a.
+//
+// WHAT IT IS FOR. SSIM is completely blind to the difference between "one union box" and "N
+// separate rects", and that difference is the Mali upload cliff: Mali prices an upload by the
+// number of JOBS, and ~100 one-rect sprite jobs against one union box measured +6 ms/frame
+// (ARCHITECTURE.md:249-251). Every P4a gate can be green while the emission shape has silently
+// inverted, so the shape needs a number - and there are TWO numbers, deliberately:
+//
+//   tex[emit= box= rect= jobs=]   the SERVER's count, Espryt's own, which has existed since P2
+//   emit[ctu=]                    the CLIENT's count of the same records (CallClass::
+//                                 ClientTextureUploadEmissions, minted by the P4a contract commit)
+//
+// The two agreeing is the whole reason both are printed (D-L). An emission-shape divergence between
+// the client that decides the rect model and the server that pays the GPU cost is then a difference
+// of two published numbers rather than something only a device can see.
+//
+// WHY IT IS RECORDED AND NOT GATED, and this is a scope decision rather than a hedge. ROADMAP.md:23
+// puts "dirty 归属反转（按存储属主键控的发射游标）" in the P3b/P4b cell: P4a lands the flat drain
+// list, and the per-storage-owner emission cursor with view/owner index remapping - which is what
+// actually decides the shape for a texture uploaded through a VIEW - is the next phase's. Gating a
+// shape the phase has not finished deciding would either pin today's shape as if it were the
+// answer, or fail on a change that is the point of the next phase. So P4a BUILDS the scenario (it
+// is meaningless without P4a's records) and runs it as a RECORDED comparison; P3b/P4b turns it into
+// a gate with the Mali frame-time delta published beside it (D-D4).
+//
+// WHAT IT THEREFORE ASSERTS, and it is not nothing:
+//
+//   1. the numbers could be READ AT ALL - the counters exist, the window covers the workload, and
+//      the workload really uploaded (a run that uploaded nothing would record four zeroes and look
+//      exactly like a healthy run whose emitter had been switched off);
+//   2. the internal ARITHMETIC of the server's own bracket holds: emit == box + rect, and
+//      jobs >= emit, because a box emission is one job and a rect-list emission is N;
+//   3. the client and the server agree on the RECORD COUNT when both are non-zero (ctu == emit).
+//      Not "when the client is non-zero": on the P4a contract tree the client emits nothing and
+//      that is a SKIP-shaped observation, not a divergence.
+//
+// Everything else - which shape each texture took, and whether that is the right shape - is
+// RECORDED with RecordProperty and printed, for MEASUREMENTS.md and for the P3b/P4b gate to be
+// written against.
+//
+// THE WORKLOAD is the shape the decision is about: one texture receiving MANY SMALL SCATTERED
+// SUB-REGIONS per frame (the sprite-atlas / chunk-renderer shape), and one receiving a single large
+// contiguous one. The first is where the box-versus-rect choice is made - MipmapStorage's 96-rect
+// cascade and the summedArea*4 >= unionArea*3 union-box fallback - and the second is the control
+// that must always be one box whatever the policy is.
+//
+// DIRECTGLES ONLY. The server-side counters are Espryt's (Managers.cpp:6360-6395); Magma's upload
+// path is P7 and contributes nothing to them, so a DirectVulkan lane would record a bracket of
+// zeroes and call it a shape.
+
+#include <cstdint>
+#include <cstdlib>
+#include <iostream>
+#include <string>
+#include <vector>
+
+#include "../Harness/HeadlessGL.h"
+#include "../Harness/PipeStatsWindow.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 {
+
+        // Set by the TextureUploadShape. ctest entry and by nothing else; a harness marker, never
+        // read by the library.
+        constexpr const char* kLaneMarker = "MGITEST_TEXTURE_UPLOAD_SHAPE_LANE";
+
+        constexpr int kInset = 2;
+        constexpr int kAtlasSize = 64;
+        // Enough scattered rects that the box-versus-rect policy has a real decision to make: the
+        // rect cascade caps at MipmapStorage::kMaxDirtyRects = 96, and the union-box fallback fires
+        // on summedArea*4 >= unionArea*3, so a handful of rects would take neither branch
+        // interestingly.
+        constexpr int kScatteredRects = 40;
+        constexpr int kRectSize = 2;
+        constexpr int kFrames = 3;
+
+        constexpr const char* kVS = R"(#version 330 core
+in vec2 aPos;
+out vec2 vUv;
+void main() {
+    vUv = aPos * 0.5 + 0.5;
+    gl_Position = vec4(aPos, 0.0, 1.0);
+}
+)";
+
+        constexpr const char* kFS = R"(#version 330 core
+in vec2 vUv;
+uniform sampler2D uTex;
+out vec4 oColor;
+void main() { oColor = texture(uTex, vUv); }
+)";
+
+        struct Vertex {
+            float x, y;
+        };
+
+        bool BuildMarkerIsSet(const char* name) {
+            const char* value = std::getenv(name);
+            return value != nullptr && value[0] == '1' && value[1] == '\0';
+        }
+
+        class TextureUploadShapeScenario : public ScenarioTest {
+        protected:
+            void SetUp() override {
+                ScenarioTest::SetUp();
+                if (!Ready()) return;
+                std::string error;
+                m_program = CompileProgram(kVS, kFS, &error);
+                ASSERT_NE(m_program, 0u) << error;
+
+                static const Vertex quad[6] = {{-1.0f, -1.0f}, {1.0f, -1.0f}, {1.0f, 1.0f},
+                                               {-1.0f, -1.0f}, {1.0f, 1.0f},  {-1.0f, 1.0f}};
+                glGenBuffers(1, &m_quadBuffer);
+                glBindBuffer(GL_ARRAY_BUFFER, m_quadBuffer);
+                glBufferData(GL_ARRAY_BUFFER, sizeof(quad), quad, GL_STATIC_DRAW);
+                glGenVertexArrays(1, &m_vao);
+                glBindVertexArray(m_vao);
+                glEnableVertexAttribArray(0);
+                glVertexAttribPointer(0, 2, GL_FLOAT, GL_FALSE, sizeof(Vertex), nullptr);
+                glBindVertexArray(0);
+            }
+
+            void TearDown() override {
+                if (!Ready()) return;
+                glUseProgram(0);
+                glBindVertexArray(0);
+                glBindTexture(GL_TEXTURE_2D, 0);
+                glBindBuffer(GL_ARRAY_BUFFER, 0);
+                if (m_scattered != 0) glDeleteTextures(1, &m_scattered);
+                if (m_contiguous != 0) glDeleteTextures(1, &m_contiguous);
+                if (m_vao != 0) glDeleteVertexArrays(1, &m_vao);
+                if (m_quadBuffer != 0) glDeleteBuffers(1, &m_quadBuffer);
+                if (m_program != 0) glDeleteProgram(m_program);
+            }
+
+            void SkipUnlessTheLaneIsAssertableHere() {
+                if (std::getenv(kLaneMarker) == nullptr) {
+                    GTEST_SKIP() << "runs only in its own lane: the TextureUploadShape. ctest entry "
+                                    "sets " << kLaneMarker
+                                 << " together with MOBILEGL_PIPE_STATS=1, "
+                                    "MOBILEGL_PIPE_STATS_PERIOD=1 and a private "
+                                    "MOBILEGL_LOG_FILE_PATH. None of that is configured in the "
+                                    "ambient entries, and the ambient log is shared, so a read here "
+                                    "would race.";
+                    return;
+                }
+                if (Gl().BackendName() != "DirectGLES") {
+                    GTEST_SKIP() << "DirectGLES only: the upload-shape counters are Espryt's "
+                                    "(Managers.cpp:6360-6395) and " << Gl().BackendName()
+                                 << " contributes nothing to them, so this lane would record a "
+                                    "bracket of zeroes and call it a shape.";
+                    return;
+                }
+                if (!BuildMarkerIsSet("MGITEST_PIPE_PUSH_BUILD")) {
+                    GTEST_SKIP() << "this library was built without MOBILEGL_PIPE_PUSH: the client's "
+                                    "half of the comparison (CallClass::ClientTextureUploadEmissions, "
+                                    "the ctu= field) does not exist there, and a one-sided reading is "
+                                    "not the comparison this scenario is for. The entry is registered "
+                                    "in every build so that `ctest -L integration-gpu` names the same "
+                                    "tests in the pull build and the push build (gate G2).";
+                    return;
+                }
+                if (PipeStatsWindow::LibraryLogPath().empty()) {
+                    GTEST_SKIP() << "the lane configured no MOBILEGL_LOG_FILE_PATH, and the library's "
+                                    "summary line is the only channel this module has for reading "
+                                    "PipeStats";
+                    return;
+                }
+            }
+
+            GLuint MakeAtlas(std::uint8_t r, std::uint8_t g, std::uint8_t b) {
+                std::vector<std::uint8_t> texels(static_cast<std::size_t>(kAtlasSize) * kAtlasSize * 4);
+                for (std::size_t i = 0; i < texels.size(); i += 4) {
+                    texels[i] = r;
+                    texels[i + 1] = g;
+                    texels[i + 2] = b;
+                    texels[i + 3] = 255;
+                }
+                GLuint texture = 0;
+                glGenTextures(1, &texture);
+                glBindTexture(GL_TEXTURE_2D, texture);
+                glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA8, kAtlasSize, kAtlasSize, 0, GL_RGBA,
+                             GL_UNSIGNED_BYTE, texels.data());
+                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_BASE_LEVEL, 0);
+                glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAX_LEVEL, 0);
+                return texture;
+            }
+
+            Image DrawSampled(GLuint texture) {
+                BindDefaultFramebuffer();
+                glViewport(0, 0, Gl().Width(), Gl().Height());
+                glUseProgram(m_program);
+                glUniform1i(glGetUniformLocation(m_program, "uTex"), 0);
+                glActiveTexture(GL_TEXTURE0);
+                glBindTexture(GL_TEXTURE_2D, texture);
+                glBindVertexArray(m_vao);
+                glDrawArrays(GL_TRIANGLES, 0, 6);
+                return ReadPixels(Gl().Width(), Gl().Height());
+            }
+
+            GLuint m_program = 0;
+            GLuint m_vao = 0;
+            GLuint m_quadBuffer = 0;
+            GLuint m_scattered = 0;
+            GLuint m_contiguous = 0;
+        };
+
+        TEST_F(TextureUploadShapeScenario, TheEmittedUploadShapeIsRecordedAndTheTwoSidesAgree) {
+            if (!Ready()) return;
+            SkipUnlessTheLaneIsAssertableHere();
+            if (IsSkipped()) return;
+
+            m_scattered = MakeAtlas(0, 255, 0);
+            m_contiguous = MakeAtlas(0, 255, 0);
+            // The first draw of each texture uploads its whole level, which is not the shape this
+            // scenario is about; it happens in the SETUP window, before the one that is read.
+            DrawSampled(m_scattered);
+            DrawSampled(m_contiguous);
+            BindDefaultFramebuffer();
+            Gl().EndFrame();
+
+            // ---- the counted window ----
+            Image lastScattered;
+            Image lastContiguous;
+            for (int frame = 0; frame < kFrames; ++frame) {
+                // MANY SMALL SCATTERED RECTS: the shape whose box-versus-rect decision is the whole
+                // subject. They are spread over the atlas on a coarse stride so that their union
+                // box is most of the texture and their summed area is a small fraction of it -
+                // which is the input the summedArea*4 >= unionArea*3 fallback is written for.
+                glBindTexture(GL_TEXTURE_2D, m_scattered);
+                const std::uint8_t patch[kRectSize * kRectSize * 4] = {
+                    0, 255, 0, 255, 0, 255, 0, 255, 0, 255, 0, 255, 0, 255, 0, 255};
+                for (int rect = 0; rect < kScatteredRects; ++rect) {
+                    const int x = ((rect * 7) % (kAtlasSize / kRectSize)) * kRectSize;
+                    const int y = ((rect * 5) % (kAtlasSize / kRectSize)) * kRectSize;
+                    glTexSubImage2D(GL_TEXTURE_2D, 0, x, y, kRectSize, kRectSize, GL_RGBA,
+                                    GL_UNSIGNED_BYTE, patch);
+                }
+                lastScattered = DrawSampled(m_scattered);
+
+                // ONE LARGE CONTIGUOUS REGION: the control. Whatever the policy is, this is one
+                // box and one job, and a reading where it is not says the policy has stopped
+                // looking at the region at all.
+                glBindTexture(GL_TEXTURE_2D, m_contiguous);
+                std::vector<std::uint8_t> band(static_cast<std::size_t>(kAtlasSize) * 8 * 4);
+                for (std::size_t i = 0; i < band.size(); i += 4) {
+                    band[i] = 0;
+                    band[i + 1] = 255;
+                    band[i + 2] = 0;
+                    band[i + 3] = 255;
+                }
+                glTexSubImage2D(GL_TEXTURE_2D, 0, 0, 0, kAtlasSize, 8, GL_RGBA, GL_UNSIGNED_BYTE,
+                                band.data());
+                lastContiguous = DrawSampled(m_contiguous);
+            }
+            ASSERT_EQ(FirstGLError(), GLenum(GL_NO_ERROR)) << "the upload workload left a GL error behind";
+            Gl().EndFrame(); // the swap that emits the window covering exactly the work above
+
+            const PipeStatsWindow::Window window = PipeStatsWindow::LastFromLaneLog();
+            ASSERT_TRUE(window.found)
+                << "no 'MGPipe stats:' line in " << PipeStatsWindow::LibraryLogPath()
+                << ", so the shape could not be read at all";
+            RecordProperty("stats_line", window.line.c_str());
+
+            const long long emissions = PipeStatsWindow::CounterOrAbsent(window, "emit");
+            const long long box = PipeStatsWindow::CounterOrAbsent(window, "box");
+            const long long rect = PipeStatsWindow::CounterOrAbsent(window, "rect");
+            const long long jobs = PipeStatsWindow::CounterOrAbsent(window, "jobs");
+            const long long clientEmissions = PipeStatsWindow::CounterOrAbsent(window, "ctu");
+            ASSERT_GE(emissions, 0) << "the summary line carries no tex[emit=]: " << window.line;
+            ASSERT_GE(box, 0) << "no box=: " << window.line;
+            ASSERT_GE(rect, 0) << "no rect=: " << window.line;
+            ASSERT_GE(jobs, 0) << "no jobs=: " << window.line;
+
+            // THE RECORD. This is the deliverable of this scenario in P4a: a number, in the ctest
+            // XML and in the log, for MEASUREMENTS.md and for the P3b/P4b gate to be written
+            // against. Printed as well as recorded, because a RecordProperty is invisible in a
+            // console run.
+            std::cout << "[ TextureUploadShape ] backend=" << Gl().BackendName() << " frames=" << kFrames
+                      << " scattered_rects_per_frame=" << kScatteredRects
+                      << " server[emit=" << emissions << " box=" << box << " rect=" << rect
+                      << " jobs=" << jobs << "] client[ctu=" << clientEmissions << "]" << std::endl;
+            RecordProperty("server_emissions", static_cast<int>(emissions));
+            RecordProperty("server_box_emissions", static_cast<int>(box));
+            RecordProperty("server_rect_emissions", static_cast<int>(rect));
+            RecordProperty("server_upload_jobs", static_cast<int>(jobs));
+            RecordProperty("client_emissions", static_cast<int>(clientEmissions));
+
+            // 1. the workload really uploaded. Without this the three assertions below are all
+            //    0 == 0 and a run whose emitter was switched off records the same "healthy" shape
+            //    as one that worked.
+            ASSERT_GT(emissions, 0)
+                << "the server counted no texture upload emission at all over " << kFrames
+                << " frames of " << kScatteredRects
+                << " sub-regions each plus a contiguous band. Either the uploads never reached the "
+                   "backend or the counter stopped counting; in both cases every shape number below "
+                   "would be a zero that means nothing. "
+                << window.line;
+
+            // 2. the server bracket's own arithmetic.
+            EXPECT_EQ(box + rect, emissions)
+                << "tex[box=] + tex[rect=] must be tex[emit=]: every emission takes exactly one of "
+                   "the two shapes. " << window.line;
+            EXPECT_GE(jobs, emissions)
+                << "tex[jobs=] must be at least tex[emit=]: a box emission is one driver upload job "
+                   "and a rect-list emission is N. " << window.line;
+
+            // 3. the two sides agree, WHEN THERE ARE TWO SIDES - and "there are two sides" is
+            //    answered by the BUILD, not by the number (review F-m7).
+            //
+            //    ctu= IS ALWAYS PRESENT IN A PUSH BUILD: PipeStats.cpp writes the field whether or
+            //    not anything ever incremented the counter, so `clientEmissions > 0` conflated
+            //    three different trees - "no client emitter exists", "the emitter exists and
+            //    emitted nothing", and "the counter was not published at all" - into one branch
+            //    that asserts nothing and prints a sentence that is only true of the first. Once
+            //    package B's texture emitter lands, an emitter that STOPPED emitting would read
+            //    exactly like no emitter at all and this case would have gone green over it, which
+            //    is the failure mode the whole scenario exists to make impossible.
+            //
+            //    So the discriminator is MGITEST_PIPE_CLIENT_TEXTURE_UPLOAD_EMITTER_PRESENT, the
+            //    build's own content probe for a MG_Impl/Pipe source that emits
+            //    CallClass::ClientTextureUploadEmissions - the same mechanism as every other arming
+            //    decision in this package - and each side of it asserts something real.
+            const bool clientEmitterExists =
+                BuildMarkerIsSet("MGITEST_PIPE_CLIENT_TEXTURE_UPLOAD_EMITTER_PRESENT");
+            ASSERT_GE(clientEmissions, 0)
+                << "the summary line carries no ctu= field at all, in a push build, where PipeStats "
+                   "publishes it unconditionally. The client half of the comparison cannot be read: "
+                << window.line;
+            RecordProperty("client_emitter_present", clientEmitterExists ? 1 : 0);
+            if (clientEmitterExists) {
+                EXPECT_GT(clientEmissions, 0)
+                    << "a MG_Impl/Pipe source emits CallClass::ClientTextureUploadEmissions on this "
+                       "tree, and the SERVER counted " << emissions
+                    << " texture upload emissions for this workload, but the client counted NONE. An "
+                       "emitter that has stopped emitting reads exactly like no emitter at all in "
+                       "this field, which is why this case asks the build rather than the number. "
+                    << window.line;
+                EXPECT_EQ(clientEmissions, emissions)
+                    << "the CLIENT counted " << clientEmissions
+                    << " texture upload records and the SERVER counted " << emissions
+                    << " for the same workload in the same window. The two counting the same records "
+                       "is the entire reason both are published (D-L): a divergence here is an "
+                       "emission-shape divergence that SSIM is blind to and that costs ~+6 ms/frame "
+                       "on Mali when it goes the wrong way. "
+                    << window.line;
+            } else {
+                // Not merely "not asserted": on a tree with no client emitter the counter must be
+                // ZERO, and a non-zero one would mean the probe is looking for the wrong symbol -
+                // i.e. that the arming decision above is wrong and every future run of this case
+                // is mis-armed.
+                EXPECT_EQ(clientEmissions, 0)
+                    << "no MG_Impl/Pipe source emits CallClass::ClientTextureUploadEmissions on this "
+                       "tree, yet the client counted " << clientEmissions
+                    << " of them. Something is incrementing that counter which this build's probe "
+                       "cannot see, so the probe is looking for the wrong symbol and this case's "
+                       "arming decision is unreliable in both directions. " << window.line;
+                std::cout << "[ TextureUploadShape ] no P4a client emitter has landed on this tree "
+                             "(the build's ClientTextureUploadEmissions probe found none), so this "
+                             "run records the SERVER shape only and pins ctu=0. That is the expected "
+                             "reading on the contract tree and it is not a divergence."
+                          << std::endl;
+                RecordProperty("client_side", "absent");
+            }
+
+            // The pixels, so that a recorded shape cannot be the shape of a workload that drew
+            // nothing.
+            EXPECT_TRUE(RegionIsMostly(lastScattered, kInset, lastScattered.Width() - kInset, kInset,
+                                       lastScattered.Height() - kInset, "green", 0.0,
+                                       "the scattered-rect atlas"));
+            EXPECT_TRUE(RegionIsMostly(lastContiguous, kInset, lastContiguous.Width() - kInset, kInset,
+                                       lastContiguous.Height() - kInset, "green", 0.0,
+                                       "the contiguous-band atlas"));
+        }
+
+    } // namespace
+} // namespace MGITest

MobileGL/MG_Pipe/Coverage.def

@@ -34,11 +34,39 @@
     /* dead: no backend reads it since D21; kept for inventory row 594 */  \
     X(GetBoundTransformFeedbackName,            SetStreamOutputTargets)  \
     X(GetBoundVertexArray,                      BindVertexElements)      \
-    /* Polymorphic over BufferTarget: its rows split across set_vertex_buffers,     */  \
-    /* set_index_buffer, set_indirect_buffers and set_shader_buffers when the       */  \
-    /* inventory is re-vendored carrying the target argument (deferred out of P1:   */  \
-    /* the extractor lives in MobileGL-CS). Named for the plan's explicit           */  \
-    /* replacement of the DrawIndirect/Parameter pair.                              */  \
+    /* Polymorphic over BufferTarget, and P3a SPLITS it - not by re-vendoring the   */  \
+    /* inventory (the extractor lives in MobileGL-CS and still does not carry the   */  \
+    /* target argument), but by supplying the target from the EMISSION SITE, which  */  \
+    /* knows it exactly. The split, target by target, in BufferTarget's OWN         */  \
+    /* spelling (BufferObject.h:15-33) so the list cross-checks mechanically:       */  \
+    /*   Vertex, and the per-attribute buffer of a VAO     -> set_vertex_buffers    */  \
+    /*   Index                                             -> set_index_buffer      */  \
+    /*   DrawIndirect, Parameter                           -> set_indirect_buffers  */  \
+    /*   Uniform, ShaderStorage, AtomicCounter,                                     */  \
+    /*   TransformFeedback                                 -> set_shader_buffers    */  \
+    /*   CopyRead, CopyWrite, PixelPack, PixelUnpack,                               */  \
+    /*   Texture                                           -> still pulled: the     */  \
+    /*     transfer and pixel-store targets have no call of their own yet;          */  \
+    /*   DispatchIndirect                                  -> still pulled, and     */  \
+    /*     it is the one a reader most plausibly assumes set_indirect_buffers       */  \
+    /*     covers: BOTH backends read it (SyncBoundBuffer at every                  */  \
+    /*     glDispatchComputeIndirect) and NO call carries it - MGPIndirectBuffers   */  \
+    /*     is the DrawIndirect + Parameter pair only - so the field stays the       */  \
+    /*     fill loop's for this target;                                             */  \
+    /*   Query                                             -> still pulled: no      */  \
+    /*     backend reads it at all and no call names it.                            */  \
+    /* That is 15 of 15; a target missing from this list is a target a later        */  \
+    /* phase would retire the pull for on the strength of a split that never        */  \
+    /* covered it.                                                                  */  \
+    /* THE ROW STAYS ONE ROW, and that is structural rather than a shortcut: this   */  \
+    /* list IS the MGPipeInputField enum and the PipeInputs field set, and the      */  \
+    /* field is ONE array (m_bufferBindingSlot[kBufferTargetCount]) that a second   */  \
+    /* row of the same name could only duplicate. So the split lives here and in    */  \
+    /* the emitters, and the row keeps naming set_indirect_buffers for the plan's   */  \
+    /* explicit replacement of the DrawIndirect/Parameter pair. It is deliberately  */  \
+    /* NOT in the EMITTED list below: seven targets above are still pulled, and a   */  \
+    /* row there says "the whole field is supplied", which for this field would be  */  \
+    /* the same half-truth GetPixelStoreParameters is kept out for.                 */  \
     X(GetBufferBindingSlot,                     SetIndirectBuffers)      \
     X(GetBufferBindingPoint,                    SetShaderBuffers)        \
     X(GetBufferBindingPointCount,               SetShaderBuffers)        \
@@ -71,9 +99,10 @@
     X(GetProgramForDispatch,                    SetDispatchProgram)      \
     X(GetProgramForDraw,                        SetDrawProgram)          \
     X(GetProgramObject,                         CreateShaderState)       \
-    /* Not in ComputePipelineStateHash today even though Vulkan makes it pipeline   */  \
-    /* state; recorded here so the G7 chunk table has to answer for it before it    */  \
-    /* freezes (section 10.3-5).                                                    */  \
+    /* ANSWERED by P2: it is in the pipeline half. SetProvokingVertexMode calls      */  \
+    /* BumpVersions(), and the chunk table's rule is exactly that, so it rides       */  \
+    /* pipeline chunk P4 - a strict superset of what ComputePipelineStateHash used   */  \
+    /* to hash (MGPipeRenderStateSpans.cpp records the provenance).                  */  \
     X(GetProvokingVertexMode,                   CreateRenderState)       \
     X(GetRenderStateParameters,                 CreateRenderState)       \
     X(GetRenderStateParametersVersion,          BindRenderState)         \
@@ -128,4 +157,124 @@
     X(handle-ify (wire handle),                 kStructuralHandle)       \
     X(Buffer ops delta,                         ResourceRespecify)
 
+// X(Accessor, PipeCall) - the EMITTED list (P2 brief D5): which P2 call now SUPPLIES this
+// PipeInputs field, so the per-verb residual fill loop no longer has to pull it out of
+// GLContext. gen_pipe.py turns it into kMGPipeFieldEmittedBy[] (generated/PipeFilled.inc);
+// a field with no row here keeps going through the fill loop, which is what makes the
+// MOBILEGL_PIPE_PUSH bitmask a true per-subsystem A/B rather than an all-or-nothing switch.
+//
+// Every name must be an accessor in MGP_COVERAGE_ACCESSOR_LIST and every call must be a
+// real call in PipeCalls.def; gen_pipe.py refuses anything else.
+//
+// The one row whose call differs from the accessor list's is GetPrimitiveRestartIndex:
+// coverage maps it onto draw_vbo because that is where a backend reads it, but the VALUE
+// travels in dynamic chunk D6, so set_dynamic_state is what supplies it.
+//
+// GetPixelStoreParameters is DELIBERATELY ABSENT, and the reason is the shape of the field
+// rather than of the call. The field is PipeInputs::m_pixelStore[2] - pack AND unpack - and
+// set_pixel_pack_state carries the PACK half only, deliberately and permanently (D10,
+// ARCHITECTURE.md 4.6 D5: nothing on the far side of the boundary reads unpack state). A row
+// here says "this field is supplied, the fill loop may skip it", which would be a half-truth:
+// the moment the render-state bitmask has its bit set, the unpack half would be written by
+// nothing while its poison stamp said it was published, so neither the poison nor the verify
+// comparator could see it. Until the field is split, the whole of it keeps going through the
+// fill loop and the pack half is simply written twice.
+//
+// P3a ADDS ONE ROW, GetBoundVertexArray -> BindVertexElements, and it is the vertex-input
+// family's only candidate: GetBufferBindingSlot is polymorphic over a target set P3a covers
+// only part of (see its comment above) and GetCurrentVertexAttribute has been here since P2.
+// The row is INERT until the vertex-input subsystem is wired - MG_Impl/Pipe/PipeFill.cpp's
+// kMGPipeWiredSubsystems does not carry that bit at the contract commit, because the emitters
+// beside it are still stubs - which is exactly the guard that lets a row land before the call
+// that carries it exists.
+//
+// A NOTE FOR THE COMMIT THAT WIRES IT, because it is not visible from this file: the field is
+// a shared pointer to the frontend VAO, and the vertex-input calls supply the CONFIGURATION
+// (the applier's MGPipeVertexElementsRecord), not the object. So the row is shape-only in the
+// same sense GetCurrentVertexAttribute's is, and it stays so until the backend's twin
+// resolution reads the applier's BoundVertexElements instead of the object - at which point
+// PipeFill.cpp's EmittedCallSuppliesTheWholeField arm is where that is decided, deliberately
+// rather than silently by this row's presence.
+//
+// P4a ADDS SIX ROWS, and the same note applies to every one of them: each is SHAPE-ONLY, each
+// lands in PipeFill.cpp's EmittedCallSuppliesTheWholeField FALSE arm, and the decision is
+// taken THERE rather than inherited from a row's presence here. The rows and their calls:
+//
+//   GetFramebufferBindingSlot -> SetFramebufferState   GetProgramForDraw     -> SetDrawProgram
+//   GetImageTextureBinding    -> SetShaderImages       GetProgramForDispatch -> SetDispatchProgram
+//   GetTextureUnitObject      -> SetSamplerViews       GetMaxTouchedTextureUnit -> SetSamplerViews
+//
+// Five of the six are the pointer-storage case GetBoundVertexArray already documents: the
+// field is a BindingSlot<FramebufferObject>, an ImageTextureBinding, a TextureUnit or a
+// SharedPtr<ProgramObject> - frontend heap references - and the calls carry eight-byte
+// {slot, gen} handles and resolved descriptors. The applier has no way to produce a pointer
+// and P4a deliberately does not give it one; skipping the pull would leave every one of those
+// mirrors null on every draw of every push build. What retires those pulls is not a better
+// applier, it is the phase where the backend stops reading a frontend object at all.
+//
+// THE SIXTH IS A DIFFERENT ARGUMENT AND IT IS WORTH WRITING DOWN, because it looks like the
+// easy one. GetMaxTouchedTextureUnit is a plain Int, and set_sampler_views' Count IS that
+// value plus one (the second merge rule: a high-water mark is directly the count argument).
+// But the set is SUPPRESSED on an unchanged content hash and is emitted only when bit 12 fires,
+// and bit 12's shutter is Mix(textureContent, GetTextureBindGeneration()) - which does NOT
+// move on a redundant re-bind of the object a unit already holds, while the high-water mark
+// DOES (see NoteUnitTouched in DirtySurface.def). So the applier's Count can lag the frontend's
+// high-water mark by exactly the case the suppressor exists to swallow, and the field keeps
+// being pulled. Narrowing that is P3b/P4b's, with the backend debounce it takes over.
+//
+// FOUR ACCESSORS THAT MAP TO A P4a CALL ARE DELIBERATELY NOT HERE, for GetPixelStoreParameters'
+// reason - a row here says "this field is supplied", and for these it would be a half-truth:
+//   GetActiveTextureUnit  - glActiveTexture's selector. set_sampler_views carries the RESOLVED
+//     per-unit set and no active-unit selector at all; nothing on the wire carries it.
+//   GetTextureContextId   - a context identity the backend keys its own tables on. No call
+//     carries it and none should: it is the server's question about the client, not state.
+//   GetTextureBindGeneration / GetSamplingResolutionGeneration - frontend SHUTTERS. What
+//     replaces them server-side is the applier's own Serial, which is a different value with a
+//     different owner; claiming the sets supply the generations would make the fill loop skip
+//     two counters no record carries.
+// And GetTextureObject / GetProgramObject are STICKY (see MGP_COVERAGE_STICKY_LIST): they are
+// keyed by GL name, they are object lookups rather than verb state, and a forwarded field has
+// no storage for an emitted call to supply.
+#define MGP_COVERAGE_EMITTED_LIST(X)                                     \
+    X(GetBlendColor,                            SetDynamicState)         \
+    X(GetBlendEquationIndexed,                  CreateRenderState)       \
+    X(GetBlendFuncIndexed,                      CreateRenderState)       \
+    X(GetBoundVertexArray,                      BindVertexElements)      \
+    X(GetClampReadColor,                        SetDynamicState)         \
+    X(GetClearColor,                            SetDynamicState)         \
+    X(GetClearDepth,                            SetDynamicState)         \
+    X(GetClearStencil,                          SetDynamicState)         \
+    X(GetColorMaskIndexed,                      CreateRenderState)       \
+    X(GetCullFaceMode,                          CreateRenderState)       \
+    X(GetCurrentVertexAttribute,                SetVertexAttribDefaults) \
+    X(GetDepthFunc,                             CreateRenderState)       \
+    X(GetDepthMask,                             CreateRenderState)       \
+    X(GetDepthRangeIndexed,                     SetDynamicState)         \
+    X(GetFramebufferBindingSlot,                SetFramebufferState)     \
+    X(GetImageTextureBinding,                   SetShaderImages)         \
+    X(GetLineWidth,                             SetDynamicState)         \
+    X(GetLogicOp,                               CreateRenderState)       \
+    X(GetMaxTouchedTextureUnit,                 SetSamplerViews)         \
+    X(GetMinSampleShadingValue,                 CreateRenderState)       \
+    X(GetPatchDefaultInnerLevel,                SetPatchState)           \
+    X(GetPatchDefaultOuterLevel,                SetPatchState)           \
+    X(GetPatchVertices,                         SetPatchState)           \
+    X(GetPipelineStateVersion,                  BindRenderState)         \
+    X(GetPolygonModeFront,                      CreateRenderState)       \
+    X(GetPolygonOffsetFactor,                   SetDynamicState)         \
+    X(GetPolygonOffsetUnits,                    SetDynamicState)         \
+    X(GetPrimitiveRestartIndex,                 SetDynamicState)         \
+    X(GetProgramForDispatch,                    SetDispatchProgram)      \
+    X(GetProgramForDraw,                        SetDrawProgram)          \
+    X(GetProvokingVertexMode,                   CreateRenderState)       \
+    X(GetRenderStateParameters,                 CreateRenderState)       \
+    X(GetRenderStateParametersVersion,          BindRenderState)         \
+    X(GetScissorBox,                            SetDynamicState)         \
+    X(GetStencilState,                          CreateRenderState)       \
+    X(GetTextureUnitObject,                     SetSamplerViews)         \
+    X(GetViewport,                              SetDynamicState)         \
+    X(GetViewportIndexed,                       SetDynamicState)         \
+    X(IsCapabilityEnabled,                      CreateRenderState)       \
+    X(IsCapabilityEnabledIndexed,               CreateRenderState)
+
 // clang-format on

MobileGL/MG_Pipe/DirtySurface.def

@@ -0,0 +1,424 @@
+// MobileGL - MobileGL/MG_Pipe/DirtySurface.def
+// 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 dirty-surface mapping (ARCHITECTURE.md 5.2 corollary 4, P2 brief D16).
+//
+// MGPipe replaces "the backend rediscovers what changed" with "the frontend says what
+// changed", which only works if EVERY frontend mutation a backend can observe has an answer
+// to "what publishes this". The failure mode is silent and one-directional: a mutation that
+// forgets to publish renders stale, and no purity gate can see it.
+//
+// So the surface is enumerated MECHANICALLY. scripts/gen_pipe_dirty_surface.py scans
+// MG_Impl/GLImpl AND MG_State/GLState for every pGLContext-> mutator call and every
+// MGP_NOTE_MUTATION site, and, with --check, fails if a scanned mutator has no row here or a
+// row here names a mutator the scan no longer finds. Both directions, so a deleted mutator
+// cannot leave a stale row behind either.
+//
+// THE SECOND ROOT AND THE SECOND MECHANISM ARE P3a's, and they were not cosmetic: reading
+// MG_State/GLState found BumpSamplingResolutionGeneration, a mutator with no row at all
+// whose every caller lives in that directory, and reading MGP_NOTE_MUTATION brought
+// NoteUnitTouched - which no `pGLContext->` pattern can see, because every texture and
+// sampler bind routes through it rather than calling it through the context - inside the
+// gate. A notice site carries a FIELD name, so the mutator a row names is the ENCLOSING
+// function.
+//
+// ANSWERS. A row lists EVERY publisher that fires on EVERY path through that mutator,
+// and only those; several are joined with '|'. A publisher that fires on some paths but
+// not all must not appear, because a shutter built from this file would then UNDER-fire,
+// and ARCHITECTURE.md 13.2 names under-firing as the dangerous direction. The one row that
+// carries a bit which fires on only some paths says so in its answer - kPulledPartialShutter
+// joined with that bit - because the alternative, dropping the bit, tells a reader of this
+// file that a bit P2 already emits a call for has no shutter at all.
+//
+// "EVERY PATH" MEANS EVERY PATH THAT MUTATES. A setter that returns early because the value
+// did not change publishes nothing and needs to publish nothing - there is no mutation to
+// carry - so a redundant-write guard (SetColorMask's `if (changed) BumpVersions();`, the
+// BitwiseEqual guards on the patch levels) does not make its publisher conditional in the
+// sense this rule cares about. A publisher reached on only SOME of the paths that DO mutate
+// - SetCapability's ClipDistance arms, SetStencilFunc's reference-only call - is the thing
+// that must not be named.
+//
+// EVERY BIT ANSWER IN THIS FILE IS DERIVED AND CHECKED, in two families and one
+// direction. The RenderState family (45 rows) is checked both ways against RenderState.cpp,
+// below. Every other NEW_* answer is checked against the shutter Tracker.h builds for that
+// bit: gen_pipe_dirty_surface.py resolves what the shutter READS and what each mutator
+// transitively WRITES (through MGP_NOTE_AGGREGATE too, whose hop it reads out of
+// MGPipeNoteAggregate's own switch, and through the function-like macros of MG_State, which
+// it EXPANDS - sixteen of RenderState.cpp's writes exist only after the preprocessor has
+// pasted them together) to the SAME two-level token, MEM:<member> plus FIELD:<member>.<leaf>,
+// and fails a row that names a bit whose shutter its mutator moves on no path at all. That
+// half is one-directional on purpose - "it does write something the shutter reads" cannot
+// prove it does so on EVERY path - so it catches under-firing and not over-claiming.
+//
+// AN ABSENCE CLAIM IS ONLY WORTH THE READING BEHIND IT, and this gate learned that twice:
+// its write analysis first recorded a write through a member's field as the field alone
+// and never the member, so --check printed, as a fact about RenderState.cpp, that
+// SetPixelStoreParam "writes nothing NEW_PIXEL_PACK's shutter reads" about a setter whose
+// whole body is sixteen writes to exactly that member; then, once it read the member, it
+// still could not see a write through a REFERENCE (SetBlendEquation's `for (auto&
+// blendState : m_parameters.BlendStates)`) and said the same false thing about seven more
+// setters - while its reader side resolved `render.PatchVertices` to the WHOLE of
+// m_parameters, so every setter that touched any byte of it "supported" NEW_PATCH_STATE and
+// a row saying glClearColor publishes the patch state was green. So now: a reference or
+// pointer bound to a member-rooted lvalue is followed, and its writes are credited to the
+// member and the field it was bound to; a write whose root the analysis cannot place (a
+// reference parameter, a call result, a member without the m_ prefix, a token it could not
+// expand) TAINTS the function, and every answer that depends on a tainted function is
+// UNDECIDED - printed with its reason, never a verdict; a writer supports a bit only when
+// the two sides share a member AND, both resolved to fields, their field sets intersect (a
+// whole-member write or read is every field); a member in common with no field information
+// on one side is COARSE, reported and never counted. --check counts only the supported
+// answers as derived, prints the COARSE and UNDECIDED tallies, and FAILS on an UNDECIDED
+// row unless MGP_DIRTY_SURFACE_UNDECIDED_LIST at the bottom of this file marks it - a
+// mark that outlives its reason is a red gate too. What it still cannot claim: a shutter
+// member written outside MG_State/GLState + MG_Impl/Pipe is undecided in the absence
+// direction, a call is resolved by NAME to every body of that name, and a FIELD token is
+// not scoped to a type - all three only widen what a mutator is credited with, and the
+// second is also how a taint spreads.
+//
+// The prose answers (kImmediate, kExplicitDestroy, kUnpublishedDestroy, kNoBackendRead,
+// kPulledEveryVerb, kPulledPartialShutter, kReverseChannel) are statements no derivation
+// checks - except the bits a kPulledPartialShutter row names, which are checked like any
+// other bit answer. --check prints how many rows carry a prose answer, so "all mapped" can
+// never be read as "all verified".
+//
+// For the RenderState family that answer is not a matter of taste and it is CHECKED
+// rather than asserted: scripts/gen_pipe_dirty_surface.py reads RenderState.cpp and
+// derives, per setter, which of NEW_RENDER_STATE / NEW_PIPELINE_STATE moves on every
+// path - BumpVersions() moves both, a bare ++m_version moves only NEW_RENDER_STATE, and
+// a setter with both kinds of path therefore always-fires only NEW_RENDER_STATE - and
+// --check fails when a row disagrees, in either direction. That check exists because
+// this file got exactly two rows wrong: SetCapability, whose ClipDistance0..7 arms move
+// only m_version, and SetStencilFunc, whose pipeline bump is conditional on Func moving.
+// Both named NEW_PIPELINE_STATE, which does not fire for glEnable(GL_CLIP_DISTANCE0) or
+// for a reference-only glStencilFunc.
+//
+//   NEW_*            a MGPipeDirty bit (MG_Impl/Pipe/Tracker.h). The tracker's shutter for
+//                    that bit moves when this mutator runs, so the next verb publishes it.
+//   kImmediate       the mutating function also reaches the backend in the same body, so the
+//                    mutation is published inline and needs no shutter at all.
+//   kReverseChannel  not state: a write INTO the frontend from the backend's side.
+//   kNoBackendRead   no backend read point observes this state at all.
+//   kExplicitDestroy published by the delete_* / resource_destroy call the Track H slice
+//                    emits when the object's last reference drops - an object's DEATH,
+//                    which no generation shutters because there is no longer an object
+//                    to carry one. Only for a kind that HAS an identity on the wire to
+//                    destroy: the resources and CSOs of PipeCalls.def, which is what P2
+//                    brief D13 scopes Espryt 0b's explicit destroy to.
+//   kUnpublishedDestroy
+//                    the same event for a kind NOTHING publishes: a program, a program
+//                    pipeline and a shader have no per-object handle on the wire at all
+//                    in P2 - resource_destroy and the delete_* family name resources and
+//                    CSOs - so their DirectGLES twins are still reclaimed by the
+//                    backend's own registry teardown and no frontend call says they
+//                    died. Recorded as a hole rather than dressed up as a mechanism that
+//                    exists; naming kExplicitDestroy here would be the same defect the
+//                    RenderState derivation above exists to stop, one class down in
+//                    stakes. (D13's prose says 'six kinds' while the Core.cpp ranges it
+//                    cites also cover MarkProgram/MarkShaderForDeletion; the tree
+//                    decides, and the tree has no wire object for those three.)
+//   kPulledEveryVerb no shutter exists at all - no MGPipeDirty bit moves on any path through
+//                    this mutator - and none is needed yet: the PipeInputs field it writes is
+//                    in its verb class's may-read mask, so the residual fill copies it at
+//                    EVERY verb of that class. A shutter here is a P3/P4 optimisation, not a
+//                    correctness gap.
+//   kPulledPartialShutter
+//                    the same pull, but a bit DOES move - on some of the paths that mutate,
+//                    not all of them - so this row must never be read as "no shutter exists".
+//                    The bits that move are named after the '|', which is the one place this
+//                    file joins a prose answer with a bit, and the reason is exactly that a
+//                    P3a shutter builder has to be able to tell "no bit covers this" from "a
+//                    bit covers half of it". The named bits are checked the same way every
+//                    other bit answer is - a dead one is a red gate - but they are NOT a
+//                    licence to narrow: what holds on every mutating path is the pull.
+//                    Which rows need this answer is a human judgement and stays one: the
+//                    derivation's "it does move that shutter" direction over-approximates
+//                    (a call name resolves to every body of that name, a write inside an
+//                    `if` counts), so it can refute a named bit but cannot find the rows
+//                    that should have named one.
+//
+// KNOWN BLIND SPOTS OF THE SCANNER, recorded here rather than left implicit
+// (gen_pipe_dirty_surface.py's own notes plus its scan roots):
+//   1. it matches braced function bodies textually, so a mutator inside a LAMBDA is
+//      attributed to the enclosing function;
+//   2. a mutation published through a HELPER the entry point calls reads as deferred here.
+//   3. CLOSED AT P3a. The scan root was MG_Impl/GLImpl only, so the four MGP_NOTE_MUTATION
+//      sites in MG_State/GLState/TextureState/TextureState.h were outside it entirely. The
+//      root is now both directories and the notice is a recognised publish mechanism; what
+//      that found is the two rows marked "P3a" below.
+// The gate is therefore a COMPLETENESS gate over what the scanner does see. The semantic
+// proof stays the MOBILEGL_PIPE_VERIFY lane, which is blind to none of them.
+//
+// THE MUTATOR PREFIX SET WIDENS AT P4a, and what it does NOT gain is the more interesting
+// half. `pGLContext->` + Add|Set|Mark|Bump|Allocate|Truncate|Record|Notify|Begin|End could
+// not see `UseProgram`, `BindVertexArray`, `BindProgramPipelineObject` or
+// `BindTransformFeedbackObject` - four mutators that each move a field P3a or P4a pushes -
+// because none of them starts with one of those words. `Use` and `Bind` are added, and the
+// complete set the widening surfaces was enumerated by grep at the phase's base ref so it
+// cannot surprise anybody: exactly those four names, on seven call sites.
+//
+// `Create*` and `Pop*` are DELIBERATELY NOT ADDED. They create or destroy objects rather than
+// move a pushed field, and each object class's creation and destruction is already answered
+// twice over - by its own Mark*ForDeletion row below and by the constructor-time
+// resource_create - so adding them would produce rows that restate an answer this file already
+// gives, and every one of them would have to be maintained against a mechanism that is not
+// theirs. A gate whose rows do not each carry their own question is a gate nobody reads.
+//
+// clang-format off
+
+// X(Mutator, Answer)
+#define MGP_DIRTY_SURFACE_LIST(X)                                                               \
+    /* ---- the reverse channel: 836 of the 926 calls, 90% of the surface ---- */               \
+    X(RecordError,                        kReverseChannel)                                      \
+    /* ---- immediate publish points: the same body reaches the backend ---- */                 \
+    X(SetActiveTextureUnit,               kImmediate)                                           \
+    X(BeginTransformFeedback,             kImmediate)                                           \
+    X(EndTransformFeedback,               kImmediate)                                           \
+    X(SetTransformFeedbackPaused,         kImmediate)                                           \
+    X(MarkTransformFeedbackObjectForDeletion, kImmediate)                                       \
+    /* ---- the render state. Derived from RenderState.cpp and gated by --check:  */            \
+    /* a setter that calls BumpVersions() on every path publishes BOTH counters;  */            \
+    /* one that also has a bare ++m_version path publishes only NEW_RENDER_STATE. */            \
+    X(SetBlendEquation,                   NEW_RENDER_STATE|NEW_PIPELINE_STATE)                  \
+    X(SetBlendEquationIndexed,            NEW_RENDER_STATE|NEW_PIPELINE_STATE)                  \
+    X(SetBlendFunc,                       NEW_RENDER_STATE|NEW_PIPELINE_STATE)                  \
+    X(SetBlendFuncIndexed,                NEW_RENDER_STATE|NEW_PIPELINE_STATE)                  \
+    /* SetCapability's ClipDistance0..7 arms write ClipDistanceEnabledMask (dynamic */          \
+    /* chunk D7) and deliberately do NOT BumpVersions, so NEW_PIPELINE_STATE does   */          \
+    /* not fire at all for glEnable(GL_CLIP_DISTANCE0): set_dynamic_state publishes  */         \
+    /* it, and NEW_RENDER_STATE is the only answer that holds on every arm.         */          \
+    X(SetCapability,                      NEW_RENDER_STATE)                                     \
+    X(SetCapabilityIndexed,               NEW_RENDER_STATE|NEW_PIPELINE_STATE)                  \
+    X(SetColorMask,                       NEW_RENDER_STATE|NEW_PIPELINE_STATE)                  \
+    X(SetColorMaskIndexed,                NEW_RENDER_STATE|NEW_PIPELINE_STATE)                  \
+    X(SetCullFaceMode,                    NEW_RENDER_STATE|NEW_PIPELINE_STATE)                  \
+    X(SetDepthFunc,                       NEW_RENDER_STATE|NEW_PIPELINE_STATE)                  \
+    X(SetDepthMask,                       NEW_RENDER_STATE|NEW_PIPELINE_STATE)                  \
+    X(SetFrontFaceMode,                   NEW_RENDER_STATE|NEW_PIPELINE_STATE)                  \
+    X(SetLogicOp,                         NEW_RENDER_STATE|NEW_PIPELINE_STATE)                  \
+    X(SetMinSampleShadingValue,           NEW_RENDER_STATE|NEW_PIPELINE_STATE)                  \
+    X(SetPolygonMode,                     NEW_RENDER_STATE|NEW_PIPELINE_STATE)                  \
+    X(SetProvokingVertexMode,             NEW_RENDER_STATE|NEW_PIPELINE_STATE)                  \
+    X(SetSampleCoverage,                  NEW_RENDER_STATE|NEW_PIPELINE_STATE)                  \
+    X(SetSampleMaskValue,                 NEW_RENDER_STATE|NEW_PIPELINE_STATE)                  \
+    /* SetStencilFunc writes Func (pipeline chunk P2/P3) AND Ref/ValueMask (dynamic */          \
+    /* D3/D4), and ++m_pipelineStateVersion is CONDITIONAL on Func moving - which is */         \
+    /* what keeps a glStencilFunc that moves only the reference from evicting a     */          \
+    /* cached pipeline, and is why only NEW_RENDER_STATE fires on every call.       */          \
+    /* SetStencilOp is wholly pipeline, SetStencilMask wholly dynamic.              */          \
+    X(SetStencilFunc,                     NEW_RENDER_STATE)                                     \
+    X(SetStencilOp,                       NEW_RENDER_STATE|NEW_PIPELINE_STATE)                  \
+    X(SetStencilMask,                     NEW_RENDER_STATE)                                     \
+    X(SetBlendColor,                      NEW_RENDER_STATE)                                     \
+    X(SetClampReadColor,                  NEW_RENDER_STATE)                                     \
+    X(SetClearColor,                      NEW_RENDER_STATE)                                     \
+    X(SetClearDepth,                      NEW_RENDER_STATE)                                     \
+    X(SetClearStencil,                    NEW_RENDER_STATE)                                     \
+    X(SetClipControl,                     NEW_RENDER_STATE)                                     \
+    X(SetDepthRange,                      NEW_RENDER_STATE)                                     \
+    X(SetDepthRangeIndexed,               NEW_RENDER_STATE)                                     \
+    X(SetHint,                            NEW_RENDER_STATE)                                     \
+    X(SetLineWidth,                       NEW_RENDER_STATE)                                     \
+    X(SetPointFadeThresholdSize,          NEW_RENDER_STATE)                                     \
+    X(SetPointSize,                       NEW_RENDER_STATE)                                     \
+    X(SetPointSpriteCoordOrigin,          NEW_RENDER_STATE)                                     \
+    X(SetPolygonOffset,                   NEW_RENDER_STATE)                                     \
+    X(SetPolygonOffsetClamped,            NEW_RENDER_STATE)                                     \
+    X(SetPrimitiveRestartIndex,           NEW_RENDER_STATE)                                     \
+    X(SetScissorBox,                      NEW_RENDER_STATE)                                     \
+    X(SetScissorBoxIndexed,               NEW_RENDER_STATE)                                     \
+    X(SetViewport,                        NEW_RENDER_STATE)                                     \
+    X(SetViewportIndexed,                 NEW_RENDER_STATE)                                     \
+    /* ---- the other value-class bits ---- */                                                  \
+    /* kPulledPartialShutter, NOT kPulledEveryVerb, and NOT a bare NEW_PIXEL_PACK:          */  \
+    /* RenderState::SetPixelStoreParam writes BOTH halves - eight Pack arms and eight       */  \
+    /* Unpack arms - while the tracker's bit 2 is a byte compare of the PACK half alone     */  \
+    /* (Tracker.h), because set_pixel_pack_state deliberately has no unpack counterpart     */  \
+    /* (ARCHITECTURE.md 4.6). So glPixelStorei(GL_PACK_ALIGNMENT, 8) DOES move bit 2 and    */  \
+    /* glPixelStorei(GL_UNPACK_ALIGNMENT, 8) moves nothing at all, and a shutter narrowed   */  \
+    /* to bit 2 would under-fire for eight of the sixteen arms. What is true on every path  */  \
+    /* is the pull: GetPixelStoreParameters is one of the two Coverage.def rows an emitted  */  \
+    /* call does not supply completely (PipeFill.cpp), so the residual fill copies both     */  \
+    /* halves at every verb of the class. The bit is named anyway because P2 already EMITS  */  \
+    /* set_pixel_pack_state off it: a row that said "no shutter exists" about the only      */  \
+    /* mutator behind a shipped call would be a false answer to the one question D16 hands  */  \
+    /* P3a. Splitting this setter into a pack half and an unpack half is what would let the */  \
+    /* pack half answer NEW_PIXEL_PACK outright; that is P3's move, not P2's.               */  \
+    X(SetPixelStoreParam,                 kPulledPartialShutter|NEW_PIXEL_PACK)                 \
+    X(SetPatchDefaultInnerLevel,          NEW_PATCH_STATE|NEW_RENDER_STATE|NEW_PIPELINE_STATE)  \
+    X(SetPatchDefaultOuterLevel,          NEW_PATCH_STATE|NEW_RENDER_STATE|NEW_PIPELINE_STATE)  \
+    /* Also an immediate publish point, but it has a real bit and the bit is   */               \
+    /* the more useful answer: set_patch_state carries it whatever the caller  */               \
+    /* does next.                                                              */               \
+    X(SetPatchVertices,                   NEW_PATCH_STATE|NEW_RENDER_STATE|NEW_PIPELINE_STATE)  \
+    X(SetCurrentVertexAttributeFloat,     NEW_VERTEX_ATTRIB_DEFAULTS)                           \
+    X(SetCurrentVertexAttributeInt,       NEW_VERTEX_ATTRIB_DEFAULTS)                           \
+    X(SetCurrentVertexAttributeUint,      NEW_VERTEX_ATTRIB_DEFAULTS)                           \
+    /* ---- object class ---- */                                                                \
+    X(BumpTextureBindGeneration,          NEW_SAMPLER_VIEWS)                                    \
+    /* P3a, FOUND BY THE WIDENED SCAN ROOT and not by anything before it: every caller of  */   \
+    /* this one is inside MG_State/GLState (SamplerObject::BumpVersion for any sampler     */   \
+    /* parameter, TextureObjectBase's shape bump), which the scan did not read, so it had  */   \
+    /* no row at all while its sibling above did. NEW_SAMPLERS is the bit whose shutter    */   \
+    /* mixes the texture-params aggregate with exactly this generation, and the bump is    */   \
+    /* unconditional on every path that reaches it - the early-outs are in the setters     */   \
+    /* above it, which is the redundant-write guard this file's rule explicitly allows.    */   \
+    X(BumpSamplingResolutionGeneration,   NEW_SAMPLERS)                                         \
+    /* P3a, and the reason the scan now reads MGP_NOTE_MUTATION as a publish mechanism of  */   \
+    /* its own: this function is not a pGLContext-> mutator at all - every texture and     */   \
+    /* sampler bind entry point routes THROUGH it - and it moves two pushed PipeInputs     */   \
+    /* fields with two different answers.                                                   */  \
+    /*   GetTextureBindGeneration moves only on the bindingChanged arm, so NEW_SAMPLER_    */   \
+    /*     VIEWS fires on SOME of the paths that mutate and not all: a redundant re-bind   */   \
+    /*     of the object a unit already holds advances the high-water mark alone. That is  */   \
+    /*     precisely what kPulledPartialShutter is for.                                     */  \
+    /*   GetMaxTouchedTextureUnit has no shutter at all; it is in its verb class's may-read */  \
+    /*     mask and the residual fill copies it at every verb.                              */  \
+    /* Both are ALSO published inline by MGP_NOTE_MUTATION when the write happens inside a */   \
+    /* verb already in flight - a backend binding its own synthesised fallback texture -    */  \
+    /* which is the window no shutter and no pull can cover, and the reason those sites     */  \
+    /* had to come inside this gate rather than stay a recorded blind spot.                 */  \
+    X(NoteUnitTouched,                    kPulledPartialShutter|NEW_SAMPLER_VIEWS)              \
+    /* NOT NEW_SO_TARGETS, and this one was false on EVERY path: GLContext::SetNamed        */  \
+    /* TransformFeedbackBinding either binds a BufferState binding point (index == the      */  \
+    /* bound XFB object) or writes a saved-bindings entry, and NEW_SO_TARGETS mixes the     */  \
+    /* buffer-CONTENT aggregate with the transform-feedback generation - the first moves    */  \
+    /* only at BufferObject.cpp's content sites, the second only in BeginTransformFeedback. */  \
+    /* A binding moves neither. It reaches the backend the same way every other buffer      */  \
+    /* binding point does, through GetBufferBindingPoint in the verb class's may-read mask, */  \
+    /* so the honest answer is the pull. Narrowing it is P3b's, when it takes the subsystem */  \
+    /* over and the binding points get a generation of their own.                           */ \
+    X(SetNamedTransformFeedbackBinding,   kPulledEveryVerb)                                     \
+    /* ---- P4a, THE FOUR THE WIDENED PREFIX SET SURFACES. Every one of them moves a field */  \
+    /* P3a or P4a pushes and none of them was visible to the scan before, because none     */  \
+    /* begins with one of the ten words the pattern matched.                                */ \
+    /*   UseProgram is bit 6's whole subject: the shutter is                                */ \
+    /*     Mix(GetCurrentProgram()->GetLifetimeId(), GetLinkVersion()) and glUseProgram is  */  \
+    /*     what moves the object it reads through. Two call sites. AND SINCE THE FABLE     */  \
+    /*     SEAM ROUND (F-1 / F-2) IT IS BITS 12 AND 14's TOO: set_sampler_views is         */  \
+    /*     resolved for the program in use and set_shader_images' window is the highest    */  \
+    /*     image unit the program in use names, so both shutters mix the same identity     */  \
+    /*     bit 6 reads, and a glUseProgram alone moves all three. Undecided for the same    */  \
+    /*     reason as bit 6 (the taint below), marked the same way.                          */  \
+    /*   BindVertexArray is bit 5's, for the same reason one level down: the shutter mixes  */  \
+    /*     the bound VAO's identity with its configuration version, and this is the bind.   */  \
+    /*     Three call sites.                                                                */ \
+    X(UseProgram,                         NEW_SHADER|NEW_SAMPLER_VIEWS|NEW_SHADER_IMAGES)      \
+    X(BindVertexArray,                    NEW_VERTEX_ELEMENTS)                                 \
+    /* NOT NEW_SHADER, and the derivation refutes it outright rather than leaving it a      */  \
+    /* judgement: this mutator writes m_boundProgramPipeline (plus the pipeline name table) */  \
+    /* and bit 6's shutter reads m_currentProgram's lifetime id and link version - disjoint */  \
+    /* sets, on every path. That is not an oversight in the shutter either: it reads        */  \
+    /* GetCurrentProgram() and DELIBERATELY NOT GetProgramForDraw(), because the tracker    */  \
+    /* must not force a compile just to answer "did the shader move", and flattening a      */  \
+    /* pipeline into its composite is exactly the compile it would force. What a bind moves */  \
+    /* is which program the validate point will flatten, and that field -                   */  \
+    /* GetProgramForDraw - is in the may-read mask of every class that draws and is copied  */  \
+    /* by the residual fill at every verb of those classes, EMITTED-AND-STILL-PULLED like   */  \
+    /* GetBoundVertexArray. So the pull is what holds on every path, and it is the answer.  */ \
+    X(BindProgramPipelineObject,          kPulledEveryVerb)                                    \
+    /* No shutter at all, and none is needed: the transform-feedback binding reaches the    */  \
+    /* backend through GetBoundTransformFeedbackLifetimeId and its siblings, which are in   */  \
+    /* the kDraw and kXfbSpan may-read masks, so the residual fill copies them at every     */  \
+    /* verb of those classes. Narrowing it is P4b's, with set_stream_output_targets.        */ \
+    X(BindTransformFeedbackObject,        kPulledEveryVerb)                                    \
+    /* ---- an object's death: no generation, because there is no longer an object  */          \
+    /* to carry one. Espryt 0b's delete_* / resource_destroy publishes the kinds     */         \
+    /* that have a handle on the wire; programs, program pipelines and shaders have  */         \
+    /* none in P2, so nothing publishes theirs - kUnpublishedDestroy, a known hole.  */         \
+    /* TWO OF THESE ROWS STOPPED BEING ASPIRATIONAL AT P3a, and the call that makes  */         \
+    /* each true is named rather than implied:                                       */        \
+    /*   MarkBufferObjectForDeletion -> resource_destroy (PipeCalls.def), emitted    */         \
+    /*     from ~BufferObject the moment the last reference drops - which is the     */        \
+    /*     glDelete* that only marks the name, followed by whatever unbind actually  */        \
+    /*     releases it - and followed IN THAT ORDER by the client freeing the slot.  */        \
+    /*   MarkVertexArrayForDeletion -> delete_vertex_elements, published through the */        \
+    /*     death notice ~VertexArrayObject already raises for the VertexElementsCso  */        \
+    /*     kind; the CSO handle is minted per frontend VAO off its lifetime id.      */        \
+    /*     THE PUBLISHER IS THE BACKEND'S (Managers.cpp's OnFrontendStateObject-     */        \
+    /*     Destroyed consumer, package espryt), not the client's: the client mints   */        \
+    /*     the CSO handle and emits create/bind, and the free rides with that        */        \
+    /*     consumer. Until it lands the row states the design, not the tree.         */        \
+    /* P4a CLOSES ONE OF THE THREE HOLES ABOVE AND STATES WHY THE OTHER TWO ARE NOT HOLES.  */ \
+    /*   MarkProgramForDeletion -> kExplicitDestroy. delete_shader_state exists now and     */  \
+    /*     ~ProgramObject emits it through the client-side death helper, in the fixed       */  \
+    /*     order: the wire delete first, the backend notice second, the slot free last. A   */  \
+    /*     program pipeline COMPOSITE takes the same call on the same helper - the server   */  \
+    /*     never learns it is a composite.                                                  */ \
+    /*   MarkProgramPipelineForDeletion stays kUnpublishedDestroy, and it is NOT waiting    */  \
+    /*     for a later phase: a ProgramPipelineObject has no lifetime id and no wire object */  \
+    /*     at all (its only identity is m_everBound). It never gets a handle, so there is   */  \
+    /*     nothing for a delete to name. What its cache's eviction DOES publish is the      */  \
+    /*     composite's delete_shader_state, which is the row above.                         */ \
+    /*   MarkShaderForDeletion stays kUnpublishedDestroy for the same kind of reason: a     */  \
+    /*     ShaderObject has no lifetime id and never crosses the boundary - the payload is  */  \
+    /*     per-stage SPIR-V plus the reflection archive, not source, and glslang lives      */  \
+    /*     entirely on the client.                                                          */ \
+    X(MarkBufferObjectForDeletion,        kExplicitDestroy)                                     \
+    X(MarkFramebufferObjectForDeletion,   kExplicitDestroy)                                     \
+    X(MarkProgramForDeletion,             kExplicitDestroy)                                     \
+    X(MarkProgramPipelineForDeletion,     kUnpublishedDestroy)                                  \
+    X(MarkRenderbufferObjectForDeletion,  kExplicitDestroy)                                     \
+    X(MarkSamplerObjectForDeletion,       kExplicitDestroy)                                     \
+    X(MarkShaderForDeletion,              kUnpublishedDestroy)                                  \
+    X(MarkTextureObjectForDeletion,       kExplicitDestroy)                                     \
+    X(MarkVertexArrayForDeletion,         kExplicitDestroy)                                     \
+    /* ---- no backend read point observes these at all ---- */                                 \
+    /* GL_ANY_SAMPLES_PASSED conditional rendering is resolved wholly in the   */               \
+    /* frontend: IsConditionalRenderActive / GetConditionalRenderQuery have no */               \
+    /* reader under MG_Backend and no Coverage.def row.                        */               \
+    X(BeginConditionalRender,             kNoBackendRead)                                       \
+    X(EndConditionalRender,               kNoBackendRead)                                       \
+    /* ---- pulled at every verb of the class, so the next verb publishes them */               \
+    /* unconditionally. The transform-feedback accounting counters reach the   */               \
+    /* backend through GetTransformFeedbackCapturedVertices and friends, which */               \
+    /* are in the kDraw and kXfbSpan may-read masks.                           */               \
+    X(AddTransformFeedbackAccountedCaptureDraw, kPulledEveryVerb)                               \
+    X(AddTransformFeedbackCapturedVertices, kPulledEveryVerb)                                   \
+    X(AddTransformFeedbackGeometryCaptureDraw, kPulledEveryVerb)                                \
+    X(AddTransformFeedbackInputPrimitives, kPulledEveryVerb)                                    \
+    X(AddTransformFeedbackPausedPrimitives, kPulledEveryVerb)                                   \
+    X(AddTransformFeedbackPrimitives,     kPulledEveryVerb)
+
+// X(Mutator, Bit) - the (row, bit) pairs above whose derivation is KNOWN to come out
+// UNDECIDED, each with the reason --check prints for it. Every bit answer NOT listed here
+// is marked derived: --check fails when the derivation cannot decide it, and fails again
+// when a mark here names a pair the derivation now decides, so this list can neither hide a
+// row nor outlive its reason.
+//
+// IT WAS EMPTY UNTIL P4a, and it stops being empty for a reason that is a property of the
+// SCANNER rather than of the rows. Every entry below is a bit answer that is plainly true -
+// glUseProgram is what moves the object bits 6, 12 and 14's shutters read through (the
+// program in use; bits 12 and 14 since the fable seam round, F-1 / F-2), and glBindVertexArray
+// is what moves the object bit 5's shutter reads through - and the write analysis cannot say
+// so, because each of the two mutators reaches, BY NAME, a body that writes a member with no
+// m_ prefix:
+//
+//   UseProgram      -> DestroyProgramSlot() writes `attachedShaders`
+//   BindVertexArray -> a call spelled `Bind(` resolves to every body of that name, one of
+//                      which (ImageTextureBinding::Bind) writes `Access`
+//
+// A call resolved by name to every body of that name is one of the three over-approximations
+// this analysis documents about itself, and an unplaceable write TAINTS the body it is in -
+// which is the right default, because "it does not write anything the shutter reads" must
+// never be claimed about code the script could not read. Widening the taint rule to ignore
+// non-m_ writes would weaken the one mechanism that catches a genuine under-fire, so the rows
+// are MARKED, with the tool's own reason, rather than the tool being made more permissive.
+// Control 21 is what proves every marked row still needs its mark (it reads this list, so a
+// row that gains a bit here is counted rather than assumed), and control 18 is what fails the
+// moment any of them becomes decidable and the mark outlives its reason.
+//
+// The ten mutators that reach a tainted body (--check prints the count) all carry a prose
+// answer, which no derivation checks; these two mutators are the first that carry a bit answer.
+#define MGP_DIRTY_SURFACE_UNDECIDED_LIST(X)                                                     \
+    X(UseProgram,      NEW_SHADER)                                                              \
+    X(UseProgram,      NEW_SAMPLER_VIEWS)                                                       \
+    X(UseProgram,      NEW_SHADER_IMAGES)                                                       \
+    X(BindVertexArray, NEW_VERTEX_ELEMENTS)
+
+// clang-format on

MobileGL/MG_Pipe/FillPoints.def

@@ -26,6 +26,50 @@
 // Fatal{UnmigratedPipeInput, "Field@Verb"} found there is fixed by adding the (class, field)
 // row, never by marking the field sticky.
 //
+// ---------------------------------------------------------------------------------------
+// THE VERDICT ON THE EIGHT STATICALLY OVER-APPROXIMATED ROWS (P2 brief C.1, MEASUREMENTS.md
+// section 4). Every one of them is KEPT, and the reason is the same in all three groups: the
+// row is not a guess, it names a concrete backend path, and the only evidence that could
+// retire it is DYNAMIC - a corpus that never reaches the path proves nothing, because a row
+// removed on that basis turns a rare path into Fatal{UnmigratedPipeInput} in a shipped build.
+//
+//   kReadback + IsTransformFeedbackActive / IsTransformFeedbackPaused
+//     KEPT. The depth/stencil read emulation draws (ScopedEmulationDrawState, DirectGLES.cpp)
+//     and pauses an active capture around its own draw, so a glReadPixels of a depth or
+//     stencil attachment reads the transform-feedback state exactly as a draw does. Reached
+//     only when the emulation is armed, which is a driver-shaped decision, so no desktop
+//     corpus can decide it.
+//
+//   kTextureOp + IsCapabilityEnabled, kDispatch + IsCapabilityEnabled
+//     KEPT. Magma's GenerateMipmap materialises a texture's queued clear before it blits and
+//     PrepareStorageImageTextures does the same for every storage image a dispatch writes;
+//     both go through VkClearManager::PreCompensateSrgbClearColor, which reads
+//     GL_FRAMEBUFFER_SRGB. The P2 contract gave that capability real storage for the first
+//     time, so this row went from reading a compile-time constant to reading real state -
+//     which is the opposite of a row that could be dropped.
+//
+//   kBlitOrCopy / kTextureOp + the shader blit's viewport and vertex/buffer bindings
+//     (GetViewportIndexed, GetDepthRangeIndexed, GetProvokingVertexMode, GetBufferBindingPoint)
+//     KEPT. TryBlitToDefaultFramebufferWithShader is a real draw of a backend-owned helper
+//     program: ApplyGLViewportState -> ComputeGLViewport reads viewport 0 and its depth range,
+//     GetOrCreateBlitPipeline -> SelectProvokingVertexMode reads the provoking vertex, and
+//     BindProgramUniformBuffers' block resolvers read the frontend binding points. It is taken
+//     when a blit's destination is the default framebuffer and the driver cannot do it
+//     natively - again a driver-shaped decision.
+//
+// What WOULD retire a row: the poison build already answers "was this field read at this
+// verb" exactly (MOBILEGL_PIPE_POISON_OMIT withholds one field's stamp for one verb and a
+// read of it aborts naming the pair). Turning that into a retirement gate means running the
+// omission across the full CTS caselist on both devices, not the desktop corpus.
+//
+// P3a STATUS: still not done, and deliberately not done here. P3a is one of the five
+// architecture boundaries that owe a full gl44to46 caselist run on both devices, so the
+// omission sweep rides that run rather than duplicating it - the verdict lands with the
+// caselist result at the phase exit, off the critical path, and every row above stays in
+// place until it does. A row retired on desktop evidence would be retired on evidence that
+// cannot support it, which is the reason this item exists rather than a scheduling excuse.
+// ---------------------------------------------------------------------------------------
+//
 // gen_pipe.py's block regexes end at a blank line: keep the empty line after each macro.
 //
 // clang-format off

MobileGL/MG_Pipe/MGPipe.h

@@ -55,10 +55,75 @@ namespace MobileGL::MG_Pipe {
     };
 
     // The pipeline/dynamic split of RenderStateParameters, defined exactly once (section
-    // 4.5.2). Generated by G7 from the field list ComputePipelineStateHash already hashes;
-    // MGPipeRenderStateSpans.cpp and the setter-consistency test land with P2, which is
-    // when the chunk table can be filled with real offsets.
-    struct MGPipeRenderStateSpans;
+    // 4.5.2): MG_Pipe/MGPipeRenderStateSpans.{h,cpp}, which landed with P2 and computes
+    // every chunk boundary with offsetof. Include that header to use it; what stays here
+    // is the generated member list at the bottom of this file, which is what the chunk
+    // table was derived from.
+
+    // ---- MOBILEGL_PIPE_PUSH's runtime bitmask (Config.h Features.PipePush) ----
+    //
+    // One bit per SUBSYSTEM, so an A/B is per subsystem rather than all-or-nothing, and
+    // bit 63 for the one BEHAVIOUR the design has to be measured against. Bits are
+    // allocated in ROADMAP order and never reused: an operator's recorded 0x7f has to keep
+    // meaning what it meant.
+    //
+    // A clear subsystem bit means "keep pulling", which after P2 is only a valid control
+    // while MOBILEGL_PIPE_LEGACY_MEMOS compiles the pre-handle arm beside it.
+    inline constexpr Uint64 kMGPipeSubsystemRenderState = 1ull << 0;
+    inline constexpr Uint64 kMGPipeSubsystemPixelPack = 1ull << 1;
+    inline constexpr Uint64 kMGPipeSubsystemPatchState = 1ull << 2;
+    inline constexpr Uint64 kMGPipeSubsystemVertexAttribDefaults = 1ull << 3;
+    inline constexpr Uint64 kMGPipeSubsystemResidualValues = 1ull << 4;
+    inline constexpr Uint64 kMGPipeSubsystemEsprytSlots = 1ull << 5;      // Track H, Espryt 0b
+    inline constexpr Uint64 kMGPipeSubsystemMagmaVertexInput = 1ull << 6; // Track H, Magma subsystem 4
+    // P3a's two. Resources is the seven BufferBackendOps hooks turned into the handle-shaped
+    // resource_* family; VertexInput is vertex elements, vertex buffers and the index buffer.
+    // They are separate bits because they are separate A/Bs: a buffer path that regressed and
+    // a vertex path that regressed are different findings, and clearing one must not disarm
+    // the other.
+    inline constexpr Uint64 kMGPipeSubsystemResources = 1ull << 7;
+    inline constexpr Uint64 kMGPipeSubsystemVertexInput = 1ull << 8;
+    // P4a's four. FOUR AND NOT ONE, for P3a's reason one level out: a framebuffer path that
+    // regressed, a texture path that regressed, a sampler path that regressed and a program
+    // path that regressed are four different findings, and clearing one must not disarm the
+    // other three.
+    //
+    // THREE OF THEM HAVE A DEPENDENCY and it is diagnosed at the first use, never half-run -
+    // one Resolve<Family>SubsystemArm per family beside the backend's existing
+    // ResolveResourceSubsystemArm, modelled on the bit-8-requires-bit-7 refusal it already
+    // ships, and lazy rather than at bring-up because a pre-flight child dying on a signal
+    // makes a whole lane SKIP green: bit 11 requires bit 10 because
+    // every MGPBoundView::Texture and MGPImageView::Res names a Texture handle and only bit 10
+    // puts one in the slot table; bit 9 requires bit 10 because MGPSurface::Res does; and bit
+    // 10 requires bit 7 because a buffer texture's BufferForTexBuffer names a Buffer handle.
+    // The mirror pairs (10 without 11, 10 without 9, 7 without 10) are all fine, and are
+    // stated as such because an unreachable branch that says something different is how the
+    // reachable one drifts. Bit 12 depends on nothing.
+    inline constexpr Uint64 kMGPipeSubsystemFramebuffer = 1ull << 9;       // set_framebuffer_state
+    inline constexpr Uint64 kMGPipeSubsystemTextureResources = 1ull << 10; // texture + renderbuffer
+                                                                          // resource_*, set_texture_params
+    inline constexpr Uint64 kMGPipeSubsystemSamplers = 1ull << 11;         // sampler CSO, sampler view,
+                                                                          // the three unit sets
+    inline constexpr Uint64 kMGPipeSubsystemPrograms = 1ull << 12;         // shader CSO, draw/dispatch
+                                                                          // program, global constants
+    // bits 13..62 reserved for the later phases, allocated in ROADMAP order.
+    // NOT a subsystem, a BEHAVIOUR: turn OFF client-side content addressing of CSOs, so
+    // every pipeline-version change mints a fresh CSO and the map is never probed. This is
+    // the negative control the whole CSO design is measured against (ROADMAP.md P2).
+    inline constexpr Uint64 kMGPipeBehaviourNoCsoContentAddressing = 1ull << 63;
+    // The default of a push build with the knob unset (ConfigLoader.cpp). Each phase's
+    // constant STAYS, because it is the A/B control for the phase after it: P3a's
+    // "everything P2 had and nothing of mine" arm is spelled MOBILEGL_PIPE_PUSH=0x7f.
+    inline constexpr Uint64 kMGPipeSubsystemsMigratedAtP2 = 0x7full;   // bits 0..6
+    inline constexpr Uint64 kMGPipeSubsystemsMigratedAtP3a = 0x1ffull; // bits 0..8
+    // P4a's, and the two above are NOT edited: 0x1ff is P4a's T2 arm and its "everything P3a
+    // had and nothing of mine" control, exactly as 0x7f was P3a's.
+    inline constexpr Uint64 kMGPipeSubsystemsMigratedAtP4a = 0x1fffull; // bits 0..12
+    static_assert(kMGPipeSubsystemsMigratedAtP4a ==
+                      (kMGPipeSubsystemsMigratedAtP3a | kMGPipeSubsystemFramebuffer |
+                       kMGPipeSubsystemTextureResources | kMGPipeSubsystemSamplers |
+                       kMGPipeSubsystemPrograms),
+                  "the P4a phase constant and P4a's four subsystem bits have drifted");
 
     // The catalogue itself. Only macros, so it is safe to expand inside the namespace, and
     // consumers (the unit test, later the transport) get MGP_CALL_LIST from this header.

MobileGL/MG_Pipe/MGPipeHandles.h

@@ -85,6 +85,17 @@ namespace MobileGL::MG_Pipe {
     // a pipeline object, and the server never learns it is a composite - it is just another
     // ShaderCso. Reserving a band rather than a flag keeps the composite resolver's
     // lifetime bookkeeping out of the ordinary program slot allocator.
+    //
+    // THE ONE ENTRY POINT INTO THE BAND is MGPipeSlotAllocator::AllocateComposite(lifetimeId)
+    // (MG_Impl/Pipe/SlotAllocator.h, P4a D-H7). MGPipeSlotAllocator::Allocate REFUSES the band
+    // for kind ShaderCso, which is what makes "an ordinary program can never be handed a
+    // composite slot" a property of the allocator rather than of its callers; the band carries
+    // its own exhaustion assert, so exhausting it is a named Fatal rather than silent slot
+    // theft from ordinary programs. A composite's slot has TWO independent release paths - the
+    // pipeline cache's LRU eviction and the composite ProgramObject's own destructor - and
+    // both go through one client-side death helper (MG_Pipe/PipeMutation.h's
+    // MGPipeEmitShaderCsoDestroyAndFree), whose second call is a proven no-op because Free
+    // refuses a slot that is not live at that generation.
     inline constexpr Uint32 kMGPipeShaderCsoSlotLimit = 1u << 20;
     inline constexpr Uint32 kMGPipeShaderCsoCompositeSlotBase =
         kMGPipeShaderCsoSlotLimit - (kMGPipeShaderCsoSlotLimit >> 4);

MobileGL/MG_Pipe/MGPipeRenderStateSpans.cpp

@@ -0,0 +1,196 @@
+// MobileGL - MobileGL/MG_Pipe/MGPipeRenderStateSpans.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 definitions behind MGPipeRenderStateSpans.h and behind the two arrays
+// generated/PipeSpanTable.inc has declared since P0. Compiled ONLY under
+// MOBILEGL_PIPE_PUSH (CMakeLists.txt appends it to SOURCE_FILES there), which is how the
+// pull build gains no symbol from the split - a declaration emits nothing.
+//
+// PROVENANCE OF THE PIPELINE HALF. It began as the enumeration
+// VulkanRenderer::ComputePipelineStateHash carried above itself, which was the contract
+// that function had without being able to say so; it moves here because this file is now
+// that contract. Verbatim, from VulkanRenderer.cpp at feat/disaggregated@48268068:
+//
+//   Value hash over every fixed-function GL state the pipeline payload reads that
+//   the memo key's other fields (mode, program hash, vertex-input hash, render-pass
+//   hash, transform flags) do not already pin down. Enumerated against the payload
+//   build in GetOrCreatePipeline - any new GL-state read there must be added here:
+//     - capability bits: CullFace, DepthTest, PolygonOffsetFill (mode gating rides
+//       the memo's mode key), RasterizerDiscard, ColorLogicOp, StencilTest,
+//       PrimitiveRestart(+FixedIndex), SampleShading, SampleMask, plus the depth write mask
+//     - patch vertices, polygon mode, cull face mode, depth func, logic op,
+//       min sample shading, the glSampleMaski word
+//     - front/back stencil ops + compare funcs (ref/mask are dynamic state)
+//     - per draw buffer up to the render pass's colour span: indexed blend enable,
+//       blend factors/equations, indexed colour write mask (broadcast from index 0
+//       when the device lacks independentBlend - the same read the payload does)
+//   FBO-derived payload inputs (attachment presence/formats/draw-buffer gating) are
+//   pinned by the render-pass hash key, exactly as the version-keyed memo relied on.
+//
+// P2's pipeline half is a strict SUPERSET of that list. It adds SampleCoverageValue,
+// SampleCoverageInvert, FrontFaceModeSetting, ProvokingVertexModeSetting,
+// ScissorTestEnabledMask, PolygonModeBack, the eleven capability bools the hash never read
+// (DebugOutput, DebugOutputSynchronous, Dither, LineSmooth, PolygonOffsetLine,
+// PolygonOffsetPoint, PolygonSmooth, SampleAlphaToCoverage, SampleAlphaToOne, SampleCoverage,
+// ProgramPointSize) and the three capabilities P2 gave storage to (FramebufferSrgb,
+// DepthClamp, TextureCubeMapSeamless). All of them are written by a setter that calls
+// BumpVersions(), so under the header's rule they are pipeline. The alternative - demoting
+// those setters to ++m_version - would change MG_State semantics in the PULL build for the
+// sake of the push path. Growing the subset costs nothing measurable: the hash runs only
+// when m_pipelineStateVersion moves, which is exactly when Magma recomputed
+// ComputePipelineStateHash before.
+//
+// The render-pass facts are deliberately NOT here. ComputePipelineStateHash's signature is
+// (colorAttachmentCount, rasterizationSamples) and it folds ResolveEffectiveSampleMask, so
+// it was never a pure function of RenderStateParameters; a CSO handle cannot replace it on
+// its own and Magma keeps renderPassHash as a separate memo-key component.
+#include <MG_Pipe/MGPipe.h>
+#include <MG_Pipe/MGPipeRenderStateSpans.h>
+
+#include <cstring>
+
+namespace MobileGL::MG_Pipe {
+    namespace {
+        // Half-local chunk index -> global chunk index. The halves alternate, so this is
+        // arithmetic rather than a table.
+        constexpr SizeT GlobalPipelineChunk(SizeT halfIndex) { return halfIndex * 2 + 1; }
+        constexpr SizeT GlobalDynamicChunk(SizeT halfIndex) { return halfIndex * 2; }
+
+        const Uint8* BytesOf(const RenderStateParameters& params) {
+            return reinterpret_cast<const Uint8*>(&params);
+        }
+        Uint8* BytesOf(RenderStateParameters& params) { return reinterpret_cast<Uint8*>(&params); }
+
+        SizeT BlobBytes(Uint32 chunkMask, SizeT halfCount, SizeT (*toGlobal)(SizeT)) {
+            SizeT total = 0;
+            for (SizeT i = 0; i < halfCount; ++i) {
+                if ((chunkMask & (1u << i)) == 0) continue;
+                total += MGPipeRenderStateChunkAt(toGlobal(i)).Length;
+            }
+            return total;
+        }
+
+        void Gather(const RenderStateParameters& params, Uint32 chunkMask, void* dst, SizeT halfCount,
+                    SizeT (*toGlobal)(SizeT)) {
+            Uint8* out = static_cast<Uint8*>(dst);
+            const Uint8* src = BytesOf(params);
+            for (SizeT i = 0; i < halfCount; ++i) {
+                if ((chunkMask & (1u << i)) == 0) continue;
+                const MGPStateChunk chunk = MGPipeRenderStateChunkAt(toGlobal(i));
+                std::memcpy(out, src + chunk.Offset, chunk.Length);
+                out += chunk.Length;
+            }
+        }
+
+        void Scatter(const void* src, Uint32 chunkMask, RenderStateParameters& dst, SizeT halfCount,
+                     SizeT (*toGlobal)(SizeT)) {
+            const Uint8* in = static_cast<const Uint8*>(src);
+            Uint8* out = BytesOf(dst);
+            for (SizeT i = 0; i < halfCount; ++i) {
+                if ((chunkMask & (1u << i)) == 0) continue;
+                const MGPStateChunk chunk = MGPipeRenderStateChunkAt(toGlobal(i));
+                std::memcpy(out + chunk.Offset, in, chunk.Length);
+                in += chunk.Length;
+            }
+        }
+
+        Uint32 ChunksThatMoved(const RenderStateParameters& a, const RenderStateParameters& b,
+                               SizeT halfCount, SizeT (*toGlobal)(SizeT)) {
+            const Uint8* left = BytesOf(a);
+            const Uint8* right = BytesOf(b);
+            Uint32 mask = 0;
+            for (SizeT i = 0; i < halfCount; ++i) {
+                const MGPStateChunk chunk = MGPipeRenderStateChunkAt(toGlobal(i));
+                if (std::memcmp(left + chunk.Offset, right + chunk.Offset, chunk.Length) != 0) {
+                    mask |= 1u << i;
+                }
+            }
+            return mask;
+        }
+
+        constexpr Uint32 AllChunks(SizeT halfCount) {
+            return halfCount >= 32 ? ~Uint32{0} : static_cast<Uint32>((Uint64{1} << halfCount) - 1);
+        }
+    } // namespace
+
+    // The two arrays generated/PipeSpanTable.inc declares. Every entry is
+    // MGPipeRenderStateChunkAt(), so a boundary can only be written once.
+    const MGPStateChunk kMGPipePipelineChunks[kMGPipePipelineChunkCount] = {
+        MGPipeRenderStateChunkAt(GlobalPipelineChunk(0)), MGPipeRenderStateChunkAt(GlobalPipelineChunk(1)),
+        MGPipeRenderStateChunkAt(GlobalPipelineChunk(2)), MGPipeRenderStateChunkAt(GlobalPipelineChunk(3)),
+        MGPipeRenderStateChunkAt(GlobalPipelineChunk(4)), MGPipeRenderStateChunkAt(GlobalPipelineChunk(5)),
+        MGPipeRenderStateChunkAt(GlobalPipelineChunk(6)),
+    };
+    static_assert(sizeof(kMGPipePipelineChunks) / sizeof(kMGPipePipelineChunks[0]) == kMGPipePipelineChunkCount,
+                  "kMGPipePipelineChunks lost an entry");
+
+    const MGPStateChunk kMGPipeDynamicChunks[kMGPipeDynamicChunkCount] = {
+        MGPipeRenderStateChunkAt(GlobalDynamicChunk(0)), MGPipeRenderStateChunkAt(GlobalDynamicChunk(1)),
+        MGPipeRenderStateChunkAt(GlobalDynamicChunk(2)), MGPipeRenderStateChunkAt(GlobalDynamicChunk(3)),
+        MGPipeRenderStateChunkAt(GlobalDynamicChunk(4)), MGPipeRenderStateChunkAt(GlobalDynamicChunk(5)),
+        MGPipeRenderStateChunkAt(GlobalDynamicChunk(6)), MGPipeRenderStateChunkAt(GlobalDynamicChunk(7)),
+    };
+    static_assert(sizeof(kMGPipeDynamicChunks) / sizeof(kMGPipeDynamicChunks[0]) == kMGPipeDynamicChunkCount,
+                  "kMGPipeDynamicChunks lost an entry");
+
+    void MGPipeGatherPipelineBytes(const RenderStateParameters& params, void* dst) {
+        Gather(params, AllChunks(kMGPipePipelineChunkCount), dst, kMGPipePipelineChunkCount,
+               GlobalPipelineChunk);
+    }
+
+    void MGPipeScatterPipelineBytes(const void* src, RenderStateParameters& dst) {
+        Scatter(src, AllChunks(kMGPipePipelineChunkCount), dst, kMGPipePipelineChunkCount,
+                GlobalPipelineChunk);
+    }
+
+    SizeT MGPipePipelineChunkBlobBytes(Uint32 chunkMask) {
+        return BlobBytes(chunkMask, kMGPipePipelineChunkCount, GlobalPipelineChunk);
+    }
+
+    void MGPipeGatherPipelineChunks(const RenderStateParameters& params, Uint32 chunkMask, void* dst) {
+        Gather(params, chunkMask, dst, kMGPipePipelineChunkCount, GlobalPipelineChunk);
+    }
+
+    void MGPipeScatterPipelineChunks(const void* src, Uint32 chunkMask, RenderStateParameters& dst) {
+        Scatter(src, chunkMask, dst, kMGPipePipelineChunkCount, GlobalPipelineChunk);
+    }
+
+    SizeT MGPipeDynamicChunkBlobBytes(Uint32 chunkMask) {
+        return BlobBytes(chunkMask, kMGPipeDynamicChunkCount, GlobalDynamicChunk);
+    }
+
+    void MGPipeGatherDynamicChunks(const RenderStateParameters& params, Uint32 chunkMask, void* dst) {
+        Gather(params, chunkMask, dst, kMGPipeDynamicChunkCount, GlobalDynamicChunk);
+    }
+
+    void MGPipeScatterDynamicChunks(const void* src, Uint32 chunkMask, RenderStateParameters& dst) {
+        Scatter(src, chunkMask, dst, kMGPipeDynamicChunkCount, GlobalDynamicChunk);
+    }
+
+    Uint32 MGPipeDynamicChunksThatMoved(const RenderStateParameters& a, const RenderStateParameters& b) {
+        return ChunksThatMoved(a, b, kMGPipeDynamicChunkCount, GlobalDynamicChunk);
+    }
+
+    Uint32 MGPipePipelineChunksThatMoved(const RenderStateParameters& a, const RenderStateParameters& b) {
+        return ChunksThatMoved(a, b, kMGPipePipelineChunkCount, GlobalPipelineChunk);
+    }
+
+    Uint64 MGPipeHashPipelineBytes(const void* bytes) {
+        return static_cast<Uint64>(
+            XXH64(bytes, kMGPipePipelineChunkBytes, kMGPipeRenderStateChunkTableSeed));
+    }
+
+    Uint64 MGPipeComputePipelineSubsetHash(const RenderStateParameters& params) {
+        // 396 bytes on the stack. A streaming XXH64_state_t would allocate; gathering first
+        // is also what CsoCache wants, because the same bytes are what a hash hit memcmps
+        // against before the handle is reused.
+        Uint8 gathered[kMGPipePipelineChunkBytes];
+        MGPipeGatherPipelineBytes(params, gathered);
+        return MGPipeHashPipelineBytes(gathered);
+    }
+} // namespace MobileGL::MG_Pipe

MobileGL/MG_Pipe/MGPipeRenderStateSpans.h

@@ -0,0 +1,277 @@
+// MobileGL - MobileGL/MG_Pipe/MGPipeRenderStateSpans.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>
+
+#include "MGPipeTypes.h"
+#include "MGPipeValueTypes.h"
+
+// G7: the pipeline/dynamic split of RenderStateParameters, written in EXACTLY ONE PLACE
+// (ARCHITECTURE.md 5.3, D-B1).
+//
+// The rule that decides the split, and it is the only rule:
+//
+//     A byte of RenderStateParameters is in the PIPELINE half if and only if some public
+//     RenderState setter that calls BumpVersions() writes it. Every other byte is in the
+//     DYNAMIC half. There is no third set.
+//
+// That makes the G7 invariant - the pipeline-subset hash moves IF AND ONLY IF
+// m_pipelineStateVersion moves - true by CONSTRUCTION rather than by inspection, and it is
+// what MG_Test/Pipe/RenderStateSpansTest.cpp walks every setter to confirm.
+//
+// The chunks alternate: chunk 0 is dynamic, chunk 1 is pipeline, and so on, so the whole
+// table is 16 BOUNDARIES rather than 15 hand-written ranges. Every boundary is an offsetof
+// or a sizeof - never a literal - because a python guess at a layout it cannot see is
+// exactly the drift the setter-consistency test exists to catch. 8 dynamic chunks + 7
+// pipeline chunks = 15, and both counts fit the Uint32 ChunkMask of MGPRenderStateDesc and
+// MGPDynamicState with room to spare.
+//
+// Note the two splits are ORTHOGONAL and coexist (ARCHITECTURE.md 5.3): DirectGLES'
+// head [0, 312) / blend [312, 536) / tail [536, 1168) spans cut ACROSS this table, and
+// nothing about them changes. StencilFaceState is deliberately NOT reordered - reordering
+// would move Espryt's shadow bytes for no gain.
+namespace MobileGL::MG_Pipe {
+
+    namespace MGPipeRenderStateChunkDetail {
+        using RSP = RenderStateParameters;
+        using SFS = StencilFaceState;
+
+        inline constexpr SizeT kStencilFace0 = offsetof(RSP, StencilStates);
+        inline constexpr SizeT kStencilFace1 = kStencilFace0 + sizeof(SFS);
+        // The pipeline half of one stencil face is [Func, Ref) + [FailOp, end); the dynamic
+        // half is [Ref, FailOp) - Ref and ValueMask are VK_DYNAMIC_STATE_STENCIL_REFERENCE /
+        // _COMPARE_MASK and WriteMask is _WRITE_MASK, which is why glStencilFunc changing only
+        // the reference must not evict a cached pipeline (RenderState.cpp SetStencilFunc).
+        inline constexpr SizeT kFaceDynamicBegin = offsetof(SFS, Ref);
+        inline constexpr SizeT kFaceDynamicEnd = offsetof(SFS, FailOp);
+    } // namespace MGPipeRenderStateChunkDetail
+
+    // 15 chunks, 16 boundaries, strictly ascending, [0, sizeof(RenderStateParameters)).
+    inline constexpr SizeT kMGPipeRenderStateChunkCount = 15;
+
+    inline constexpr Array<SizeT, kMGPipeRenderStateChunkCount + 1> kMGPipeRenderStateChunkBoundaries = {
+        // D0 dynamic: Viewports[16], LineWidth, PointSize
+        SizeT{0},
+        // P0 pipeline: PatchVertices, PatchDefaultOuterLevel, PatchDefaultInnerLevel
+        offsetof(RenderStateParameters, PatchVertices),
+        // D1 dynamic: PolygonOffsetFactor/Units/Clamp, ClipOrigin, ClipDepthMode
+        offsetof(RenderStateParameters, PolygonOffsetFactor),
+        // P1 pipeline: BlendStates[8], LogicOp, DepthTestEnabled, DepthFunc, DepthMask,
+        //              ColorMasks[8], FramebufferSrgbEnabled, DepthClampEnabled,
+        //              TextureCubeMapSeamlessEnabled
+        offsetof(RenderStateParameters, BlendStates),
+        // D2 dynamic: ClearColor, ClearDepth, ClearStencil, BlendColor, DepthRanges[16]
+        offsetof(RenderStateParameters, ClearColor),
+        // P2 pipeline: SampleCoverageValue, SampleCoverageInvert, SampleMaskValue,
+        //              MinSampleShadingValue, StencilStates[0].Func
+        offsetof(RenderStateParameters, SampleCoverageValue),
+        // D3 dynamic: StencilStates[0].{Ref, ValueMask, WriteMask}
+        MGPipeRenderStateChunkDetail::kStencilFace0 + MGPipeRenderStateChunkDetail::kFaceDynamicBegin,
+        // P3 pipeline: StencilStates[0].{FailOp, PassDepthFailOp, PassDepthPassOp},
+        //              StencilStates[1].Func
+        MGPipeRenderStateChunkDetail::kStencilFace0 + MGPipeRenderStateChunkDetail::kFaceDynamicEnd,
+        // D4 dynamic: StencilStates[1].{Ref, ValueMask, WriteMask}
+        MGPipeRenderStateChunkDetail::kStencilFace1 + MGPipeRenderStateChunkDetail::kFaceDynamicBegin,
+        // P4 pipeline: StencilStates[1].{FailOp, PassDepthFailOp, PassDepthPassOp},
+        //              CullFaceEnabled, CullFaceModeSetting, FrontFaceModeSetting,
+        //              ProvokingVertexModeSetting
+        MGPipeRenderStateChunkDetail::kStencilFace1 + MGPipeRenderStateChunkDetail::kFaceDynamicEnd,
+        // D5 dynamic: the four hints, PointFadeThresholdSize, PointSpriteCoordOrigin,
+        //             ClampReadColor
+        offsetof(RenderStateParameters, LineSmoothHint),
+        // P5 pipeline: PolygonModeFront, PolygonModeBack
+        offsetof(RenderStateParameters, PolygonModeFront),
+        // D6 dynamic: PrimitiveRestartIndex
+        offsetof(RenderStateParameters, PrimitiveRestartIndex),
+        // P6 pipeline: the 20 capability bools ColorLogicOpEnabled..ProgramPointSizeEnabled,
+        //              ScissorTestEnabledMask
+        offsetof(RenderStateParameters, ColorLogicOpEnabled),
+        // D7 dynamic: ScissorBoxes[16], ScissorBoxWrittenMask, ClipDistanceEnabledMask
+        offsetof(RenderStateParameters, ScissorBoxes),
+        sizeof(RenderStateParameters),
+    };
+
+    // Chunk 0 is dynamic and they alternate, which is not a coincidence: every boundary above
+    // is a transition between a run of BumpVersions()-written members and a run of
+    // ++m_version-only members, so two adjacent chunks of the same half would mean a boundary
+    // that separates nothing.
+    constexpr Bool MGPipeRenderStateChunkIsPipeline(SizeT index) { return (index % 2) == 1; }
+
+    constexpr MGPStateChunk MGPipeRenderStateChunkAt(SizeT index) {
+        return MGPStateChunk{static_cast<Uint16>(kMGPipeRenderStateChunkBoundaries[index]),
+                             static_cast<Uint16>(kMGPipeRenderStateChunkBoundaries[index + 1] -
+                                                 kMGPipeRenderStateChunkBoundaries[index])};
+    }
+
+    namespace MGPipeRenderStateChunkDetail {
+        constexpr SizeT CountHalf(Bool pipeline) {
+            SizeT count = 0;
+            for (SizeT i = 0; i < kMGPipeRenderStateChunkCount; ++i) {
+                if (MGPipeRenderStateChunkIsPipeline(i) == pipeline) ++count;
+            }
+            return count;
+        }
+        constexpr SizeT BytesOfHalf(Bool pipeline) {
+            SizeT bytes = 0;
+            for (SizeT i = 0; i < kMGPipeRenderStateChunkCount; ++i) {
+                if (MGPipeRenderStateChunkIsPipeline(i) == pipeline) {
+                    bytes += MGPipeRenderStateChunkAt(i).Length;
+                }
+            }
+            return bytes;
+        }
+    } // namespace MGPipeRenderStateChunkDetail
+
+    inline constexpr SizeT kMGPipePipelineChunkCount = MGPipeRenderStateChunkDetail::CountHalf(true);
+    inline constexpr SizeT kMGPipeDynamicChunkCount = MGPipeRenderStateChunkDetail::CountHalf(false);
+    // The CSO's content-addressed identity is exactly this many bytes; CsoCache stores them
+    // per entry and memcmps them on a hash hit.
+    inline constexpr SizeT kMGPipePipelineChunkBytes = MGPipeRenderStateChunkDetail::BytesOfHalf(true);
+    inline constexpr SizeT kMGPipeDynamicChunkBytes = MGPipeRenderStateChunkDetail::BytesOfHalf(false);
+
+    // Bumped by hand when something about the table changes that its BYTES do not show -
+    // the halves' membership, the meaning of a chunk, the gather order.
+    inline constexpr Uint64 kMGPipeRenderStateChunkTableVersion = 1;
+
+    // What actually seeds MGPipeComputePipelineSubsetHash. The version above is a promise a
+    // reader has to keep; this is the part that keeps itself. Folding the boundary table into
+    // the seed means a moved boundary invalidates every persisted key whether or not anyone
+    // remembered to bump the version - and it does so WITHOUT a static_assert on the
+    // boundaries, which would turn G7's negative control (which moves a boundary on purpose
+    // and must still compile) into a build break.
+    namespace MGPipeRenderStateChunkDetail {
+        constexpr Uint64 BoundaryChecksum() {
+            Uint64 hash = 0xcbf29ce484222325ull; // FNV-1a, 64-bit
+            for (SizeT i = 0; i <= kMGPipeRenderStateChunkCount; ++i) {
+                hash = (hash ^ static_cast<Uint64>(kMGPipeRenderStateChunkBoundaries[i])) * 0x100000001b3ull;
+            }
+            return hash;
+        }
+    } // namespace MGPipeRenderStateChunkDetail
+    inline constexpr Uint64 kMGPipeRenderStateChunkTableSeed =
+        kMGPipeRenderStateChunkTableVersion ^ MGPipeRenderStateChunkDetail::BoundaryChecksum();
+
+    // ---- the trip wires. A mistake in the table is a build break, here. ----
+    static_assert(kMGPipeRenderStateChunkBoundaries[0] == 0,
+                  "the chunk table must start at byte 0 of RenderStateParameters");
+    static_assert(kMGPipeRenderStateChunkBoundaries[kMGPipeRenderStateChunkCount] ==
+                      sizeof(RenderStateParameters),
+                  "the chunk table must cover RenderStateParameters to its last byte");
+    static_assert(kMGPipePipelineChunkCount == 7);
+    static_assert(kMGPipeDynamicChunkCount == 8);
+    static_assert(kMGPipePipelineChunkCount + kMGPipeDynamicChunkCount == kMGPipeRenderStateChunkCount);
+    static_assert(kMGPipePipelineChunkBytes + kMGPipeDynamicChunkBytes == sizeof(RenderStateParameters),
+                  "the two halves must partition the block exactly - no gap, no overlap");
+    static_assert(kMGPipeRenderStateChunkCount <= 32,
+                  "a chunk index has to fit the Uint32 ChunkMask of MGPRenderStateDesc/MGPDynamicState");
+
+    // Sorted, non-overlapping and complete: because every chunk is [b[i], b[i+1]) the only
+    // way to violate that is a non-ascending boundary, so this is the whole check.
+    constexpr Bool MGPipeRenderStateChunkBoundariesAscend() {
+        for (SizeT i = 0; i < kMGPipeRenderStateChunkCount; ++i) {
+            if (!(kMGPipeRenderStateChunkBoundaries[i] < kMGPipeRenderStateChunkBoundaries[i + 1])) {
+                return false;
+            }
+            if (kMGPipeRenderStateChunkBoundaries[i + 1] > 0xffffu) return false;
+        }
+        return true;
+    }
+    static_assert(MGPipeRenderStateChunkBoundariesAscend(),
+                  "the chunk boundaries must strictly ascend and fit MGPStateChunk's Uint16 fields");
+
+    // The measured sizes. They are DERIVED above; these two assertions only pin what the P2
+    // brief and MEASUREMENTS.md quote, so a table change that moves them is loud.
+    static_assert(kMGPipePipelineChunkBytes == 396, "the pipeline subset is 396 bytes");
+    static_assert(kMGPipeDynamicChunkBytes == 772, "the dynamic subset is 772 bytes");
+
+    // ---- global chunk bits, so nothing downstream hand-maintains a second table ----
+
+    // The GLOBAL chunk indices (bit i is chunk i of the 15) whose byte range overlaps
+    // [offset, offset + size). It falls straight out of the boundary table, which is the
+    // whole point: the applier scopes its derivation by the chunks a scatter actually moved
+    // (D5/D8), and a hand-written member -> chunk mapping is exactly the second table that
+    // would go stale the first time a boundary moves.
+    constexpr Uint32 MGPipeRenderStateChunkBitsCovering(SizeT offset, SizeT size) {
+        Uint32 bits = 0;
+        for (SizeT i = 0; i < kMGPipeRenderStateChunkCount; ++i) {
+            const SizeT begin = kMGPipeRenderStateChunkBoundaries[i];
+            const SizeT end = kMGPipeRenderStateChunkBoundaries[i + 1];
+            if (offset < end && begin < offset + size) bits |= Uint32{1} << i;
+        }
+        return bits;
+    }
+
+    // The wire masks are HALF-LOCAL (bit i of MGPRenderStateDesc::ChunkMask is pipeline chunk
+    // i); these widen them to the global indices the boundary table is written in. The
+    // halves alternate with chunk 0 dynamic, so the two conversions are arithmetic.
+    constexpr Uint32 MGPipeGlobalChunkBitsOfPipelineMask(Uint32 pipelineMask) {
+        Uint32 bits = 0;
+        for (SizeT i = 0; i < kMGPipePipelineChunkCount; ++i) {
+            if (((pipelineMask >> i) & 1u) != 0) bits |= Uint32{1} << (i * 2 + 1);
+        }
+        return bits;
+    }
+    constexpr Uint32 MGPipeGlobalChunkBitsOfDynamicMask(Uint32 dynamicMask) {
+        Uint32 bits = 0;
+        for (SizeT i = 0; i < kMGPipeDynamicChunkCount; ++i) {
+            if (((dynamicMask >> i) & 1u) != 0) bits |= Uint32{1} << (i * 2);
+        }
+        return bits;
+    }
+    inline constexpr Uint32 kMGPipeAllGlobalChunks =
+        static_cast<Uint32>((Uint64{1} << kMGPipeRenderStateChunkCount) - 1);
+
+    // The two conversions must agree with MGPipeRenderStateChunkIsPipeline, and together they
+    // must cover the table exactly - a widening that dropped or doubled a chunk would make
+    // the applier's scoping silently wrong rather than loud.
+    namespace MGPipeRenderStateChunkDetail {
+        inline constexpr Uint32 kAllPipelineHalfBits =
+            static_cast<Uint32>((Uint64{1} << kMGPipePipelineChunkCount) - 1);
+        inline constexpr Uint32 kAllDynamicHalfBits =
+            static_cast<Uint32>((Uint64{1} << kMGPipeDynamicChunkCount) - 1);
+        inline constexpr Uint32 kWidenedPipeline = MGPipeGlobalChunkBitsOfPipelineMask(kAllPipelineHalfBits);
+        inline constexpr Uint32 kWidenedDynamic = MGPipeGlobalChunkBitsOfDynamicMask(kAllDynamicHalfBits);
+    } // namespace MGPipeRenderStateChunkDetail
+    static_assert((MGPipeRenderStateChunkDetail::kWidenedPipeline &
+                   MGPipeRenderStateChunkDetail::kWidenedDynamic) == 0,
+                  "the two half-local -> global widenings must not overlap");
+    static_assert((MGPipeRenderStateChunkDetail::kWidenedPipeline |
+                   MGPipeRenderStateChunkDetail::kWidenedDynamic) == kMGPipeAllGlobalChunks,
+                  "the two half-local -> global widenings must cover the whole chunk table");
+    static_assert(MGPipeRenderStateChunkBitsCovering(0, sizeof(RenderStateParameters)) == kMGPipeAllGlobalChunks,
+                  "every chunk must be covered by the whole block");
+
+    // ---- the operations everything else is written against ----
+
+    // The 396 pipeline bytes of `params`, in ascending chunk order, into `dst`.
+    void MGPipeGatherPipelineBytes(const RenderStateParameters& params, void* dst);
+    // The inverse: `src` is kMGPipePipelineChunkBytes bytes in the same order.
+    void MGPipeScatterPipelineBytes(const void* src, RenderStateParameters& dst);
+    // Incremental create_render_state: only the pipeline chunks named by `chunkMask` (bit i
+    // is pipeline chunk i, 0-based within the pipeline half), concatenated ascending.
+    SizeT MGPipePipelineChunkBlobBytes(Uint32 chunkMask);
+    void MGPipeGatherPipelineChunks(const RenderStateParameters& params, Uint32 chunkMask, void* dst);
+    void MGPipeScatterPipelineChunks(const void* src, Uint32 chunkMask, RenderStateParameters& dst);
+
+    // set_dynamic_state: bit i of `chunkMask` is dynamic chunk i, 0-based within the dynamic
+    // half; the blob is those chunks concatenated in ascending order.
+    SizeT MGPipeDynamicChunkBlobBytes(Uint32 chunkMask);
+    void MGPipeGatherDynamicChunks(const RenderStateParameters& params, Uint32 chunkMask, void* dst);
+    void MGPipeScatterDynamicChunks(const void* src, Uint32 chunkMask, RenderStateParameters& dst);
+    // Which dynamic chunks differ between two blocks - the chunk-level suppressor's answer.
+    Uint32 MGPipeDynamicChunksThatMoved(const RenderStateParameters& a, const RenderStateParameters& b);
+    // Which pipeline chunks differ - the incremental-create mask against a base CSO.
+    Uint32 MGPipePipelineChunksThatMoved(const RenderStateParameters& a, const RenderStateParameters& b);
+
+    // XXH64 over the seven pipeline chunks in ascending order, seeded with the table version.
+    // Runs ONLY when m_pipelineStateVersion moved, i.e. never in the steady state.
+    Uint64 MGPipeComputePipelineSubsetHash(const RenderStateParameters& params);
+    // The same hash over already-gathered bytes (CsoCache holds them, so it does not re-gather).
+    Uint64 MGPipeHashPipelineBytes(const void* bytes);
+} // namespace MobileGL::MG_Pipe

MobileGL/MG_Pipe/MGPipeTypes.h

@@ -145,15 +145,152 @@ namespace MobileGL::MG_Pipe {
     static_assert(sizeof(MGPCaps) == sizeof(DynamicBackendParameters) + 8 + 24 + 24,
                   "MGPCaps gained padding or a member; update the wire format");
 
+    // ---------------------------------------------------------------------------------
+    // MGPResourceDesc's two discriminators (P4a, D-A3 / D-A4)
+    // ---------------------------------------------------------------------------------
+
+    // MGPResourceDesc::Target. P3a minted no enum for this list because it had exactly one
+    // producer and used the leading member's value (0) for it; P4a's texture family is the
+    // second producer, so the list is written out here, beside the field, in the order the
+    // field's own comment already wrote it.
+    //
+    // TexRect IS A THIRTEENTH ENUMERATOR AND THE BRIEF'S LIST HAS TWELVE. MobileGL's
+    // TextureTarget has TextureRectangle (MG_State/GLState/TextureState/TextureEnum.h), the
+    // table below may not have a `default:` arm, and folding rectangle onto Tex2D would erase
+    // a distinction the frontend keeps and both backends switch on (Espryt's
+    // MapToBackendTextureTarget lowers Tex1D the same way and Tex1D still has its own
+    // enumerator here). It is appended AFTER TexBuffer so every value the design document
+    // names keeps the number it was given.
+    enum class MGPipeResourceTarget : Uint8 {
+        Buffer = 0,
+        Tex1D,
+        Tex2D,
+        Tex3D,
+        Tex1DArray,
+        Tex2DArray,
+        TexCube,
+        TexCubeArray,
+        Tex2DMS,
+        Tex2DMSArray,
+        Renderbuffer,
+        TexBuffer,
+        TexRect,
+        Count,
+    };
+
+    // P3a's constant, moved here from MG_Impl/Pipe/ResourceTracker.h with the enum: the ack
+    // predicate at the bottom of this header now names the buffer target explicitly (D-A2) and
+    // may not reach into MG_Impl to do it. The static_assert is what keeps the two spellings
+    // from drifting; nothing may open-code either.
+    inline constexpr Uint8 kMGPipeResourceTargetBuffer =
+        static_cast<Uint8>(MGPipeResourceTarget::Buffer);
+    static_assert(kMGPipeResourceTargetBuffer == static_cast<Uint8>(MGPipeResourceTarget::Buffer),
+                  "P3a's kMGPipeResourceTargetBuffer and MGPipeResourceTarget::Buffer have drifted");
+
+    // A sentinel the table below returns for a TextureTarget enumerator it does not name. It
+    // is NOT a legal Target value - it does not fit the field's Uint8 - so an unmapped
+    // enumerator is a build break at the static_assert rather than a descriptor that quietly
+    // describes the wrong kind of storage. Exactly kMGPipeBindUnmapped's shape.
+    inline constexpr Uint32 kMGPipeResourceTargetUnmapped = 0x100u;
+
+    // The one table. No `default:` arm on purpose - that is what makes the static_assert
+    // below able to see an unnamed enumerator, and it is the shape
+    // MGPipeBindMaskForBufferTarget already uses for BufferTarget.
+    constexpr Uint32 MGPipeResourceTargetForTextureTarget(MobileGL::TextureTarget target) {
+        switch (target) {
+        case MobileGL::TextureTarget::Texture1D:
+            return static_cast<Uint32>(MGPipeResourceTarget::Tex1D);
+        case MobileGL::TextureTarget::Texture2D:
+            return static_cast<Uint32>(MGPipeResourceTarget::Tex2D);
+        case MobileGL::TextureTarget::Texture3D:
+            return static_cast<Uint32>(MGPipeResourceTarget::Tex3D);
+        case MobileGL::TextureTarget::TextureCubeMap:
+            return static_cast<Uint32>(MGPipeResourceTarget::TexCube);
+        // Its own enumerator rather than Tex2D: see the enum's comment.
+        case MobileGL::TextureTarget::TextureRectangle:
+            return static_cast<Uint32>(MGPipeResourceTarget::TexRect);
+        case MobileGL::TextureTarget::Texture2DMultisample:
+            return static_cast<Uint32>(MGPipeResourceTarget::Tex2DMS);
+        case MobileGL::TextureTarget::TextureBuffer:
+            return static_cast<Uint32>(MGPipeResourceTarget::TexBuffer);
+        case MobileGL::TextureTarget::Texture1DArray:
+            return static_cast<Uint32>(MGPipeResourceTarget::Tex1DArray);
+        case MobileGL::TextureTarget::Texture2DArray:
+            return static_cast<Uint32>(MGPipeResourceTarget::Tex2DArray);
+        case MobileGL::TextureTarget::TextureCubeMapArray:
+            return static_cast<Uint32>(MGPipeResourceTarget::TexCubeArray);
+        case MobileGL::TextureTarget::Texture2DMultisampleArray:
+            return static_cast<Uint32>(MGPipeResourceTarget::Tex2DMSArray);
+        // NOT TEXTURE TARGETS. Listed rather than defaulted so the completeness assert still
+        // sees them, and mapped to the sentinel because no descriptor may carry either: the
+        // count is the enum's bound and Unknown is what an unresolved GL enum becomes.
+        case MobileGL::TextureTarget::TextureTargetCount:
+        case MobileGL::TextureTarget::Unknown:
+            return kMGPipeResourceTargetUnmapped;
+        }
+        return kMGPipeResourceTargetUnmapped;
+    }
+
+    constexpr Bool MGPipeEveryTextureTargetIsMapped() {
+        for (SizeT i = 0; i < static_cast<SizeT>(MobileGL::TextureTarget::TextureTargetCount); ++i) {
+            if (MGPipeResourceTargetForTextureTarget(static_cast<MobileGL::TextureTarget>(i)) ==
+                kMGPipeResourceTargetUnmapped) {
+                return false;
+            }
+        }
+        return true;
+    }
+    static_assert(MGPipeEveryTextureTargetIsMapped(),
+                  "a TextureTarget enumerator has no MGPResourceDesc::Target row: add it to "
+                  "MGPipeResourceTargetForTextureTarget, and give it an enumerator of its own "
+                  "rather than folding it onto a neighbour (D-A3)");
+    static_assert(MGPipeResourceTargetForTextureTarget(MobileGL::TextureTarget::Texture2D) !=
+                      MGPipeResourceTargetForTextureTarget(MobileGL::TextureTarget::TextureRectangle),
+                  "a rectangle texture is not a 2D texture on the wire; both backends switch on "
+                  "the difference");
+
+    // MGPResourceDesc::BindMask's twelve bits, in the order the field's comment names them.
+    //
+    // THEY LIVED IN MG_Impl/Pipe/ResourceTracker.h THROUGH P3a, with that file's own note
+    // saying "the integrator moves them beside the field when a second producer appears
+    // (P4a's texture family)". P4a is that producer: a texture sets kMGPipeBindSampler,
+    // kMGPipeBindShaderImage, kMGPipeBindRenderTarget and kMGPipeBindDepthStencil, which are
+    // the four bits nothing set before. The mask is STICKY - ORed, never cleared - and is
+    // emitted on both resource_create and every resource_respecify.
+    enum MGPipeBindBit : Uint16 {
+        kMGPipeBindNone = 0,
+        kMGPipeBindVertex = 1u << 0,
+        kMGPipeBindIndex = 1u << 1,
+        kMGPipeBindConstant = 1u << 2,
+        kMGPipeBindShaderBuffer = 1u << 3,
+        kMGPipeBindIndirect = 1u << 4,
+        kMGPipeBindSampler = 1u << 5,
+        kMGPipeBindShaderImage = 1u << 6,
+        kMGPipeBindRenderTarget = 1u << 7,
+        kMGPipeBindDepthStencil = 1u << 8,
+        kMGPipeBindStreamOutput = 1u << 9,
+        kMGPipeBindAtomic = 1u << 10,
+        // THE D-B7 SWITCH. With kCapNeedsHostIndexBytes set the server mirrors this
+        // resource's bytes so it can rewrite restart indices and flatten multi-draws
+        // (ARCHITECTURE.md 10.3). Getting it wrong is invisible in monolith and silently
+        // disables both under split, which is why it is set from a table rather than from a
+        // special case at the emission site.
+        kMGPipeBindElementArray = 1u << 11,
+    };
+
     // Discriminated resource descriptor: buffers, every texture target and renderbuffers
     // share one create/respecify shape (section 4.5.1).
     struct MGPResourceDesc {
         MGPipeHandle Resource;
-        Uint8 Target;      // Buffer | Tex1D..TexCubeArray | Tex2DMS.. | Renderbuffer | TexBuffer
+        // MGPipeResourceTarget: Buffer | Tex1D..TexCubeArray | Tex2DMS.. | Renderbuffer |
+        // TexBuffer | TexRect. Never open-coded; the texture half comes from
+        // MGPipeResourceTargetForTextureTarget above.
+        Uint8 Target;
         Uint8 StorageKind; // == TextureStorageType (Mipmap | Buffer)
-        // VERTEX|INDEX|CONSTANT|SHADER_BUFFER|INDIRECT|SAMPLER|SHADER_IMAGE|RENDER_TARGET|
-        // DEPTH_STENCIL|STREAM_OUTPUT|ATOMIC|ELEMENT_ARRAY. The ELEMENT_ARRAY bit is the
-        // D-B7 switch: with kCapNeedsHostIndexBytes set the server mirrors this resource.
+        // MGPipeBindBit, above: VERTEX|INDEX|CONSTANT|SHADER_BUFFER|INDIRECT|SAMPLER|
+        // SHADER_IMAGE|RENDER_TARGET|DEPTH_STENCIL|STREAM_OUTPUT|ATOMIC|ELEMENT_ARRAY. The
+        // ELEMENT_ARRAY bit is the D-B7 switch: with kCapNeedsHostIndexBytes set the server
+        // mirrors this resource.
         Uint16 BindMask;
         Uint32 InternalFormat; // already resolved to an uncompressed fallback by the client
         Uint32 Width, Height, Depth;
@@ -161,9 +298,16 @@ namespace MobileGL::MG_Pipe {
         Uint8 FixedSampleLocations, Immutable;
         Uint32 Usage;        // BufferUsage
         Uint32 StorageFlags; // glBufferStorage flags
-        Uint8 HasDefinedContent;  // false after a NULL-data respecify
+        Uint8 HasDefinedContent;  // false after a NULL-data respecify - STORAGE-DEFINING
         Uint8 ImageBindableHint;  // client-side everImageBound; pre-emptive allocation
         Uint16 Pad0;
+        // ImageBindableHint and BindMask above are the two METADATA fields the rule exists
+        // for: a respecify that moves only them - every storage-defining field equal to the
+        // stored descriptor - is a metadata update, with no reallocation ack and no
+        // PendingUploads clear. The rule, the full storage-defining field set and the third
+        // field allowed to differ (GlNameForDiag) are stated beside
+        // MGPipeResourceRespecifyNeedsAck below (P4a, ID-18 M4), which is where the applier
+        // and both twins read them from.
         // Diagnostics only. A GL name is NEVER an identity, never a memo key and never part
         // of a content hash (section 4.2.1). Widened from the plan's two bytes, which
         // cannot hold one.
@@ -231,8 +375,13 @@ namespace MobileGL::MG_Pipe {
 
     // The half of the render state that must NOT mint a CSO: viewport, scissor, depth
     // range, blend colour, line width, polygon offset, stencil ref/write mask, clear
-    // values, sample coverage, hints and the point-size family. This is what keeps
-    // glViewport from evicting Magma's pipeline memo (D-B1).
+    // values, hints, the point-size family and the primitive-restart index. This is what
+    // keeps glViewport from evicting Magma's pipeline memo (D-B1).
+    //
+    // SAMPLE COVERAGE IS NOT IN IT, and this comment used to say it was. P2's rule is that
+    // a byte is pipeline state if and only if a public RenderState setter that calls
+    // BumpVersions() writes it, and SetSampleCoverage does - so SampleCoverageValue and
+    // SampleCoverageInvert are in pipeline chunk P2 (MGPipeRenderStateSpans.h).
     struct MGPDynamicState {
         Uint32 ChunkMask;
         Uint16 Version;
@@ -245,11 +394,29 @@ namespace MobileGL::MG_Pipe {
     // VertexAttribute[32] AND the binding points, because a pointer-call stride of 0 means
     // "element size" while a binding-model stride of 0 means "every vertex reads the same
     // element" (section 4.5.3). IsLong and Type == Float64 are carried separately.
+    //
+    // THE BLOB RULE - ONE RULE FOR EVERY RECORD IN THIS HEADER THAT CARRIES AN MGPBlobRef,
+    // and MGPSubData below is the other one. `Blob.Size` is the record's own statement of how
+    // many bytes its blob holds, and the applier holds the record to that statement WHENEVER
+    // THE RECORD MAKES IT: a non-zero Blob.Size that disagrees with the byte length the
+    // record's other fields describe is Fatal{ProtocolCorruption} and the call is refused. A
+    // ZERO Blob.Size means "this record does not declare its blob", which is what a monolith
+    // emission is - the bytes travel beside the record through the apply entry point's
+    // companion `const void*` and no MGPBlobRef is filled - and it is not a fault. Either way
+    // the bytes read are bounded by the record's OTHER fields (the two counts here, the
+    // destination range there), so the length is a cross-check and never the safety property.
+    // A transport that fills these in gets a real gate on the first record it truncates; a
+    // client that leaves them zero gets no verify-build abort for a field it never used.
     struct MGPVertexElements {
         MGPipeHandle Cso;
         Uint32 AttributeCount;
         Uint32 BindingPointCount;
-        MGPBlobRef Blob; // VertexAttribute[] followed by VertexBufferBindingPoint[]
+        // MGPVertexAttribWire[AttributeCount] followed by
+        // MGPVertexBindingPointWire[BindingPointCount] - the WIRE views of MGPipeValueTypes.h,
+        // not VertexAttribute / VertexBufferBindingPoint, which cannot travel at all (each
+        // holds a SharedPtr<BufferObject>). The layout is stated once at
+        // MGPipeValueTypes.h's P3a block and unpacked once in PipeApply.cpp.
+        MGPBlobRef Blob;
     };
     MGP_ASSERT_POD(MGPVertexElements, 40);
 
@@ -281,18 +448,49 @@ namespace MobileGL::MG_Pipe {
     MGP_ASSERT_POD(MGPSamplerView, 36);
 
     // Per texture OBJECT, independent of any view.
+    //
+    // P4a, D-E1: 32 -> 40 bytes. BuiltinSampler is the SamplerCso carrying the
+    // SamplerParameters of the SamplerObject every ITextureObject owns
+    // (TextureState/TextureObject.h's m_sampler, constructed by TextureObjectBase's
+    // constructor). GL 4.6 core table 23.18 makes filter/wrap/compare/border SAMPLER state,
+    // and Espryt pushes it with glTexParameter* onto the TEXTURE rather than with
+    // glBindSampler onto the unit - behaviour P4a preserves exactly. Naming the CSO rather
+    // than widening this payload with a filter/wrap/border block is what keeps ONE authority
+    // for one value: SyncTextureParamsToBackend reads this record, SyncBuiltinSamplerToBackend
+    // reads that CSO's SamplerParameters, and the two pushes stay two pushes.
     struct MGPTextureParams {
-        MGPipeHandle Res;
-        Uint16 BaseLevel, MaxLevel;
-        Uint8 Swizzle[4];
-        Uint8 DepthStencilMode;
+        MGPipeHandle Res;            //  0
+        // Kind SamplerCso. kMGPipeNullHandle is ILLEGAL - every texture object owns a sampler
+        // object, so a null here is Fatal{ProtocolCorruption} rather than "no sampler".
+        MGPipeHandle BuiltinSampler; //  8
+        Uint16 BaseLevel, MaxLevel;  // 16
+        Uint8 Swizzle[4];            // 20
+        Uint8 DepthStencilMode;      // 24
         // Mirrors m_forceTextureParamsResync: the widened-channel carrier needs a swizzle
         // override that the frontend params version does not move for.
-        Uint8 ForceResync;
-        Uint8 Pad0[2];
-        Float MinLod, MaxLod, LodBias;
+        Uint8 ForceResync;           // 25
+        // Mirrors m_forceSamplerResync, which had no wire spelling at all before P4a. What it
+        // guards is not mis-filtering but an INCOMPLETE texture sampling (0,0,0,1) after a
+        // driver re-mint, which is why it is a second bit and not folded into ForceResync.
+        Uint8 SamplerResync;         // 26
+        Uint8 Pad0;                  // 27
+        Float MinLod, MaxLod, LodBias; // 28
     };
-    MGP_ASSERT_POD(MGPTextureParams, 32);
+    MGP_ASSERT_POD(MGPTextureParams, 40);
+
+    // MGPTextureParams::DepthStencilMode's two legal values, and the ONLY spelling of them
+    // (P4a, ID-12 / esprytobj DV-2). The frontend keeps a GLenum - GL_DEPTH_COMPONENT 0x1902,
+    // GL_STENCIL_INDEX 0x1901 - and a Uint8 cannot hold one, so the aspect is NUMBERED here
+    // rather than truncated there. The GLenum -> byte helper belongs to the client emitter;
+    // this header owns the two numbers, so the emitter and both backends cannot disagree.
+    //
+    // 0 IS DEPTH, AND THAT IS THE WHOLE REASON FOR THIS ORDER RATHER THAN THE ENUM'S LOW BYTE.
+    // GL_DEPTH_COMPONENT is the GL initial value of GL_DEPTH_STENCIL_TEXTURE_MODE and a
+    // texture that never asks for the stencil aspect never emits the call at all, so A ZEROED
+    // RECORD MUST DECODE TO EXACTLY WHAT AN UNTOUCHED TEXTURE ALREADY HAS. Numbering by the
+    // low byte would have made depth 0x02 and stencil 0x01 and left zero meaning nothing.
+    inline constexpr Uint8 kMGPipeDepthStencilModeDepth = 0;   // GL_DEPTH_COMPONENT
+    inline constexpr Uint8 kMGPipeDepthStencilModeStencil = 1; // GL_STENCIL_INDEX
 
     // create_shader_state. The reflection blob is the whole LinkArtifacts + SpirvArtifacts
     // archive; P0.5 extracts those types out of ProgramObject.h so a server can
@@ -315,39 +513,208 @@ namespace MobileGL::MG_Pipe {
     // set_*
     // ---------------------------------------------------------------------------------
 
+    // MGPSurface::Kind's three values (P4a, ID-12 / esprytobj DV-4). MGPipeKind is REUSED
+    // rather than a second three-value enum minted beside it: it already spells Texture and
+    // Renderbuffer, its None is 0, and a zero-initialised MGPSurface is therefore ALREADY the
+    // empty attachment point this record describes - {Res = kMGPipeNullHandle, Kind = None}
+    // and every other field zero. The static_assert is what keeps that true if MGPipeKind is
+    // ever reordered.
+    inline constexpr Uint8 kMGPipeSurfaceKindNone = static_cast<Uint8>(MGPipeKind::None);
+    inline constexpr Uint8 kMGPipeSurfaceKindTexture = static_cast<Uint8>(MGPipeKind::Texture);
+    inline constexpr Uint8 kMGPipeSurfaceKindRenderbuffer =
+        static_cast<Uint8>(MGPipeKind::Renderbuffer);
+    static_assert(kMGPipeSurfaceKindNone == 0,
+                  "a zero-initialised MGPSurface must already be the empty attachment point");
+
+    // MGPSurface::TextureTarget for a point that names no texture: the renderbuffer point and
+    // the empty point both carry it. It is MobileGL::TextureTarget::Unknown, which is -1 and
+    // therefore 0xFFFF in the field's Uint16 - a value no real target has, so a reader that
+    // forgets to gate on Kind gets a nonsense target rather than a plausible wrong one.
+    inline constexpr Uint16 kMGPipeSurfaceNoTextureTarget = 0xFFFF;
+    static_assert(kMGPipeSurfaceNoTextureTarget ==
+                      static_cast<Uint16>(MobileGL::TextureTarget::Unknown),
+                  "kMGPipeSurfaceNoTextureTarget is TextureTarget::Unknown widened to the "
+                  "field, and MG_State moved Unknown off -1");
+
     // = pipe_surface. internalFormat is INLINE so the four cross-object masks fall out at
     // push time with no lookup (section 4.5.6).
     struct MGPSurface {
         MGPipeHandle Res;
         Uint32 InternalFormat;
-        Uint8 Kind; // Texture | Renderbuffer | None
+        Uint8 Kind; // kMGPipeSurfaceKind{None,Texture,Renderbuffer}, above
         Uint8 Layered;
         Uint16 Level;
         Uint32 Layer;
-        Uint16 UploadTarget;
-        Uint16 Pad0;
+        Uint16 UploadTarget; // static_cast<Uint16>(MobileGL::TextureUploadTarget)
+        // P4a, ID-12 / esprytobj DV-5: WAS Pad0, and the size did not move - the two bytes
+        // were already here. static_cast<Uint16>(MobileGL::TextureTarget), and
+        // kMGPipeSurfaceNoTextureTarget on every point that is not a texture.
+        //
+        // THE FOUR CROSS-OBJECT MASKS ARE WHY IT EXISTS. IsSnormFallbackAttachment,
+        // IsUnormFallbackAttachment and IsAlphaWidenedColorAttachment all reduce to
+        // (format, TEXTURE TARGET) - ShouldUseCaveatTextureFormat(format, target) and
+        // BackendTextureFormatAddsAlpha(format, target) - and no TextureUploadTarget ->
+        // TextureTarget inverse exists anywhere in the tree, so UploadTarget cannot answer
+        // them. Without this field D-C1's promise that the inline InternalFormat makes the
+        // masks "fall out at push time with no lookup" is unkeepable and the backend keeps
+        // reading the frontend attachment objects.
+        //
+        // CONSULTED ONLY WHEN Kind == kMGPipeSurfaceKindTexture. A zero-initialised record
+        // carries 0, which is TextureTarget::Texture1D and not the sentinel; that is not a
+        // defect, because such a record is Kind == None and names no texture at all. Gating
+        // on Kind is the reader's contract.
+        Uint16 TextureTarget;
     };
     MGP_ASSERT_POD(MGPSurface, 24);
 
+    // P4a, D-C2/D-C3 and ID-19: THE RECORD DESCRIBES A FRAMEBUFFER OBJECT, and Target says
+    // whether it ALSO moves a binding.
+    //
+    // A `Named` record describes the framebuffer object it names (Fbo) and changes NO
+    // binding. `Draw` / `Read` / `Both` records describe the same object AND set the bound
+    // handle(s) of the target(s) they name.
+    //
+    // The applier therefore keeps records PER FRAMEBUFFER OBJECT, keyed by the handle's slot
+    // (the generation is checked on lookup and a stale one refuses; a framebuffer has no wire
+    // lifetime - D-I2, the catalogue has no framebuffer delete - so a successor's record
+    // simply OVERWRITES the slot), plus the two bound handles. And every DSA entry point that
+    // hands a framebuffer to the server BY NAME - BlitNamedFramebuffer, the four
+    // ClearNamedFramebuffer*, and the DSA attachment / draw-buffer / read-buffer setters at
+    // their validate point - is PRECEDED BY A Named RECORD, so that any framebuffer the
+    // server is about to receive by name already has one.
+    //
+    // That last rule is the phase's main correction, not a nicety. With only the two
+    // bound-target records, glClearNamedFramebufferfv(fbo) on an unbound fbo made the backend
+    // mint a fresh driver framebuffer with NO ATTACHMENTS, find no record for it, decline,
+    // and then issue the clear against it anyway - GL_INVALID_FRAMEBUFFER_OPERATION and
+    // nothing cleared, where the legacy arm cleared correctly. Writing such an object into
+    // the bound-target record instead would have been worse: the applier would then claim it
+    // is bound.
+    //
+    // GL has two independent framebuffer bindings and one record carries one Fbo, so a
+    // Draw/Read pair is two records and one object bound to both targets is one record with
+    // Both. The draw-buffer array belongs to the OBJECT the record names; it reaches the
+    // driver's bound draw framebuffer only for a record whose Target is Draw or Both -
+    // Espryt's own comment records the Minecraft 26.x OIT bug where a READ-only sync landed
+    // glDrawBuffers on the wrong framebuffer - and a Named record's draw buffers are applied
+    // when that object is next configured, never to whatever happens to be bound. ReadSurface
+    // is resolved from THAT framebuffer's own read buffer in EVERY record, Named included,
+    // which is what makes the read-buffer-shared-FBO defect class unrepresentable rather than
+    // merely fixed.
+    //
+    // Target IS A ContentHash INPUT (the hash covers the whole record), and the emitter's
+    // suppressor must be keyed by the framebuffer the record names, not by one global slot:
+    // two different objects' Named records in a row must both go out, and a Named record must
+    // never be suppressed against the same object's bound record or the reverse.
+    enum class MGPipeFramebufferTarget : Uint8 {
+        Draw = 0,
+        Read = 1,
+        Both = 2,
+        // Describes the framebuffer named by Fbo and changes no binding (ID-19). Emitted
+        // ahead of every DSA entry point that hands that framebuffer over by name.
+        Named = 3,
+        Count,
+    };
+
+    // MGPFramebufferState::Color[] and DrawBuffers[] are ONE array width, and it is the wire's
+    // bound rather than the driver's: GetDynamicParameters().MaxColorAttachments is the raw ES
+    // cap and is not clamped to 8 on the GLES path, so a driver reporting more would silently
+    // truncate this record. The framebuffer subsystem bit is REFUSED at its first lookup in
+    // that case, with one ERROR naming the cap, and the legacy arm runs - the same shape the
+    // backend's existing bit-8-requires-bit-7 refusal already ships
+    // (ResolveFramebufferSubsystemArm, beside ResolveResourceSubsystemArm). Widening the
+    // payload is a wire change nobody has evidence for, and truncating silently is the bug
+    // class this phase is closing.
+    inline constexpr Uint32 kMGPipeMaxColorAttachments = 8;
+    static_assert(kMGPipeMaxColorAttachments == MobileGL::kMGMaxDrawBuffers,
+                  "MGPFramebufferState::Color[] and DrawBuffers[] are one array width");
+
+    // ONE RECORD DESCRIBES ONE FRAMEBUFFER OBJECT - the one named by Fbo - and Target says
+    // whether it also moves a binding (P4a, D-C2 as corrected by ID-19; see
+    // MGPipeFramebufferTarget above for the failure that forced it).
+    //
+    // A Named record describes that object and changes NO binding. Draw / Read / Both records
+    // describe that object AND set the bound handle(s) of the target(s) they name. The applier
+    // keeps these records PER FRAMEBUFFER OBJECT, keyed by Fbo's slot (generation checked on
+    // lookup; a framebuffer has no wire lifetime - D-I2 - so a successor's record simply
+    // overwrites the slot), plus the two bound handles; every DSA entry point that hands a
+    // framebuffer to the server by name is preceded by a Named record.
+    //
+    // WHAT Target CHANGES, FIELD BY FIELD. NO FIELD IN THIS RECORD REFERS TO "the currently
+    // bound framebuffer" - every one of them describes the object named by Fbo - and that is
+    // the invariant a reader depends on:
+    //
+    //   Fbo, Color[], Depth, Stencil, ReadSurface, Width/Height/Layers/Samples,
+    //   FixedSampleLocations, IsDefault, Complete
+    //       properties of the object named by Fbo, identical in meaning under every Target.
+    //       In particular ReadSurface is resolved from THAT framebuffer's own read buffer -
+    //       on a Named record too - never from whichever framebuffer is bound to GL_READ.
+    //   DrawBuffers[]
+    //       a property of the named object; it reaches the driver's bound draw framebuffer
+    //       only when Target is Draw or Both. Under Named it is stored with the object and
+    //       applied when that object is next configured.
+    //   Target
+    //       the only binding-specific field: Draw/Read/Both name the binding(s) this record
+    //       also sets, Named names none. It is a ContentHash input.
+    //   ContentHash
+    //       per RECORD, not per object, and the emitter's suppressor is keyed by the
+    //       framebuffer named: a Named record must never be suppressed against the same
+    //       object's bound record, nor one object's Named record against another's.
     struct MGPFramebufferState {
         MGPipeHandle Fbo; // kMGPipeDefaultFramebuffer for the default framebuffer
         MGPSurface Color[8];
         MGPSurface Depth, Stencil;
-        // The RESOLVED read surface, not an index. This is what structurally closes the
+        // The RESOLVED read surface, not an index, and it is THIS framebuffer's own read
+        // buffer under every Target - Named included. This is what structurally closes the
         // read-buffer-shared-FBO defect class.
         MGPSurface ReadSurface;
-        Int8 DrawBuffers[8]; // attachment index, -1 = NONE
+        // attachment index, -1 = NONE. The named object's array; applied to the bound draw
+        // framebuffer only when Target is Draw or Both.
+        Int8 DrawBuffers[8];
         Uint16 Width, Height, Layers, Samples;
-        Uint8 FixedSampleLocations, IsDefault, Complete, Pad0;
+        // Complete is FramebufferObject::CheckCompleteness(), the frontend-only answer - NOT
+        // glCheckFramebufferStatus's. CheckFramebufferStatus_State additionally consults
+        // ActiveBackendRejectsDistinctDepthStencil() and HasNonRenderableColorAttachment,
+        // which read the backend's probed format-capability cache; a client emitting that
+        // answer would be reading the backend from the client side, which is the exact
+        // coupling this boundary exists to remove. A later phase must not assume the stronger
+        // answer, and glCheckFramebufferStatus keeps answering from the frontend as it does
+        // today.
+        Uint8 FixedSampleLocations, IsDefault, Complete;
+        // MGPipeFramebufferTarget, above (P4a, D-C2; was Pad0). Draw/Read/Both also set the
+        // named binding(s); Named sets none (ID-19). The ONLY binding-specific field.
+        Uint8 Target;
         Uint32 Pad1;
         // Two jobs (section 4.5.6): the server's render-pass memo key, and the CLIENT's
         // emission suppressor - an unchanged hash means this record is not sent at all.
         // The same pattern is mandatory for every kVarTail set_* below, or 26.2's
         // redundant glBindSampler traffic reappears as a variable-length record per batch.
+        // Target is one of its inputs, and the suppressor is keyed per framebuffer.
         Uint64 ContentHash;
     };
     MGP_ASSERT_POD(MGPFramebufferState, 304);
 
+    // GL_MAX_VERTEX_ATTRIBS as MobileGL advertises it, on the MG_Pipe side of the boundary.
+    // It bounds the two declared counts of MGPVertexElements, the applier's two per-CSO
+    // arrays and the vertex-buffer set. It MUST equal VertexArrayObject::MAX_VERTEX_ATTRIBS;
+    // MG_Impl/Pipe/PipeFill.cpp is the one translation unit that sees both and carries the
+    // static_assert, because this header may not include a frontend one.
+    inline constexpr Uint32 kMGPipeMaxVertexAttribs = 32;
+
+    // P4a, D-G2. MobileGL's texture-unit space is ONE MERGED array of
+    // TextureState::MAX_TEXTURE_IMAGE_UNITS = 192 - there is no stage dimension on
+    // set_sampler_views / bind_sampler_states / set_shader_images, because the same unit may
+    // be sampled from two stages and per-stage 32 is an advertised number rather than a
+    // storage shape. These two bound the three var-tail sets' Start + Count, and a record
+    // that names a window outside them is Fatal{ProtocolCorruption} - the var-tail window IS
+    // the bound and entries outside it are not cleared.
+    //
+    // Pinned against the frontend constant in MG_Impl/Pipe/PipeFill.cpp, the one translation
+    // unit that sees both, exactly as kMGPipeMaxVertexAttribs is: this header may not include
+    // a frontend one.
+    inline constexpr Uint32 kMGPipeMaxTextureUnits = 192;
+    inline constexpr Uint32 kMGPipeMaxImageUnits = 192;
+
     struct MGPVertexBuffer {
         MGPipeHandle Res;
         Uint64 Offset;
@@ -361,9 +728,25 @@ namespace MobileGL::MG_Pipe {
     // Var-tail header: MGPVertexBuffer[Count] follows.
     struct MGPVertexBuffers {
         Uint32 Start, Count;
+        // The vertex-FETCH base instance these offsets are valid for (P3a, D-H1). It is DRAW
+        // state, not VAO state, and it is NOT the same thing as MGPDrawInfo::StartInstance:
+        // that one is the GL draw's baseInstance and feeds gl_BaseInstance, this one is the
+        // shift the fetch address of an instanced array needs when the device has no native
+        // base-instance support. The server decides whether to emulate it or let
+        // GL_EXT_base_instance do the work - emulation is server-owned - so the client sends
+        // the draw's raw value and never a pre-shifted offset.
+        //
+        // IT IS A ContentHash INPUT, and that is a requirement rather than a nicety:
+        // set_vertex_buffers is suppressed on an unchanged hash (MG_Impl/Pipe/
+        // SetHashSuppressor.h's SetVertexBuffers slot), so a baseInstance that moved while
+        // the buffer set did not would be suppressed and the server would keep the previous
+        // shift. It rides ONE PER EMITTED SET rather than per entry: per entry the shift is
+        // redundant and lets a malformed record disagree with itself.
+        Uint32 BaseInstance;
+        Uint32 Pad0;
         Uint64 ContentHash;
     };
-    MGP_ASSERT_POD(MGPVertexBuffers, 16);
+    MGP_ASSERT_POD(MGPVertexBuffers, 24);
 
     // An independent call, NOT a subset of the VAO configuration version (D5).
     struct MGPIndexBuffer {
@@ -514,14 +897,17 @@ namespace MobileGL::MG_Pipe {
     // to it because both sides are the same translation unit. G3 emits the offsetof
     // assertions; under split the block is serialized field-wise rather than memcpy'd.
     struct ResidualValueBlock {
-        RenderStateParameters RenderState; // until create/bind_render_state + set_dynamic_state land
-        PixelStoreParameters Pack;         // until set_pixel_pack_state lands
+        // The 35 CapabilityInput bits, packed in enum order. P2 retired everything else:
+        // RenderStateParameters to create/bind_render_state + set_dynamic_state, Pack to
+        // set_pixel_pack_state, and the patch quintet to set_patch_state.
+        //
+        // What is left is deliberately REDUNDANT. Every one of the 35 capabilities is
+        // answerable from the assembled working block now that P2 gave FramebufferSrgb,
+        // DepthClamp and TextureCubeMapSeamless real storage - which is the point: the
+        // applier compares the two answers bit by bit, so the day a later call takes a
+        // capability over and forgets to carry it, the block says so on the next draw
+        // (Fatal{PipeResidualDiverged, "<Cap>"}, MG_Pipe/PipeApply.cpp).
         Uint64 CapabilityBits;
-        Uint32 PatchVertices;
-        Uint32 Pad0;
-        Float PatchOuter[4];
-        Float PatchInner[2];
-        Uint32 Pad1[2];
     };
     static_assert(std::is_trivially_copyable_v<ResidualValueBlock>);
 // The retirement ratchet. This number only ever goes DOWN: every stage that lands a real
@@ -530,10 +916,12 @@ namespace MobileGL::MG_Pipe {
 // gone. Shrinking the block without lowering the number, or growing it at all, is a build
 // break - which is the point.
 //
-// Stable across the ABIs MobileGL ships on: every member of RenderStateParameters and
-// PixelStoreParameters is a fixed-width scalar or an array of one, with no pointer and no
-// SizeT.
-#define MGL_RESIDUAL_BLOCK_SIZE 1248
+// Stable across the ABIs MobileGL ships on: the one member is a fixed-width scalar.
+//
+// P2: 1248 -> 8. RenderStateParameters (1168) retired to create/bind_render_state and
+// set_dynamic_state, PixelStoreParameters (28) to set_pixel_pack_state, and the patch
+// quintet (52 with its padding) to set_patch_state.
+#define MGL_RESIDUAL_BLOCK_SIZE 8
     static_assert(sizeof(ResidualValueBlock) == MGL_RESIDUAL_BLOCK_SIZE,
                   "the residual value block changed size; lower MGL_RESIDUAL_BLOCK_SIZE if a field "
                   "retired, and do not raise it");
@@ -566,16 +954,50 @@ namespace MobileGL::MG_Pipe {
     // decision belongs on the side that pays the GPU cost. Mali prices texture upload by
     // JOB COUNT: ~100 sprite rects against one union box measured +6 ms/frame.
     //
+    // `Target` IS TWO FACTS IN ONE Uint16 (P4a, D-D3 / ID-12), and MGPipePackSubDataTarget
+    // under the struct is the only spelling of the encoding - nothing may open-code a half:
+    //
+    //   low byte  = MGPipeResourceTarget          - WHICH KIND of storage the destination is.
+    //                                               The applier branches on it: a buffer
+    //                                               target dispatches into MGPipeResourceOps,
+    //                                               every other target accumulates a pending
+    //                                               upload for the texture sync to consume.
+    //   high byte = MobileGL::TextureUploadTarget - WHICH cube face / upload target the level
+    //                                               belongs to. It is NOT derivable from the
+    //                                               resource target - six faces share TexCube
+    //                                               - and 26 enumerators leave a byte ample.
+    //
+    // WHY THAT WAY ROUND, AND WHY THE ENCODING LIVES HERE RATHER THAN IN EACH EMITTER. The
+    // applier's SubDataNamesABuffer tests the WHOLE field == 0, and
+    // TextureUploadTarget::Texture1D is 0 - so a texture record carrying the bare upload
+    // enumerator is indistinguishable from a buffer record exactly when its owner is a 1D
+    // texture, and that texture's upload is dispatched into the buffer path. With the
+    // resource target in the LOW byte a buffer record's Target stays EXACTLY
+    // kMGPipeResourceTargetBuffer - P3a's buffer records are unchanged on the wire, their
+    // upload byte being zero too - while a texture record can never be zero, because no
+    // texture's MGPipeResourceTarget is. The static_assert under the struct holds that
+    // invariant, and the applier's whole-field test stays right either way.
+    //
     // THE BUFFER HALF. With Target == Buffer there is no level and no box, so the destination
     // byte range rides in the box's first coordinate and first extent: UnionBox.X is the byte
-    // offset, UnionBox.W the byte size, Y = Z = 0, H = D = 1, Level = 0, RegionCount = 0, and
-    // Blob holds exactly Size source bytes. That caps ONE record at a 2^31-1 offset and a
+    // offset, UnionBox.W the byte size, Y = Z = 0, H = D = 1, Level = 0, RegionCount = 0.
+    //
+    // THE BLOB RULE, THE SAME ONE MGPVertexElements ABOVE CARRIES: `Blob.Size` is the record's
+    // own statement of how many bytes its blob holds and the applier holds the record to that
+    // statement whenever the record makes it - a non-zero Blob.Size that is not exactly the
+    // record's own byte size (MGPipeSubDataBufferSize below) is Fatal{ProtocolCorruption} and
+    // the write is refused. A zero Blob.Size means "this record does not declare its blob",
+    // which is what a monolith emission is: the bytes travel beside the record through
+    // MGPipeApplyResourceSubData's companion `const void*`. The destination range is what
+    // bounds the write in both cases. That caps ONE record at a 2^31-1 offset and a
     // 2^32-1 size; a range beyond either is split by the emitter - the same rule, and at
     // SEG_STAGE's 32 MiB the far tighter one, that the ring's half-capacity bound already
     // imposes on it. MGPipeSetSubDataBufferRange / MGPipeSubDataBufferOffset / Size below are
     // the only spelling of this convention; nothing else reads the box for a buffer.
     struct MGPSubData {
         MGPipeHandle Res;
+        // Target is PACKED - see the block above, and read it only through
+        // MGPipeSubDataResourceTargetOf / MGPipeSubDataUploadTargetOf below.
         Uint16 Target, Level;
         // Replaces the backend's `uploadData == mipData` pointer comparison: are these
         // bytes an untransformed level shadow?
@@ -588,6 +1010,38 @@ namespace MobileGL::MG_Pipe {
     };
     MGP_ASSERT_POD(MGPSubData, 72);
 
+    // The one spelling of MGPSubData::Target's encoding, stated above the struct.
+    //
+    // Uint32 ARGUMENTS RATHER THAN THE TWO ENUM TYPES, and that is deliberate. This header is
+    // the contract: MGPipeResourceTarget is minted in it, but the upload half is MG_State's
+    // TextureUploadTarget, and nobody who reads the packed field ever needs that type - the
+    // applier and both backends read the halves BACK, as bytes, through the two accessors.
+    // Naming it in a signature would pin the contract's own API to a frontend enum for no
+    // reader's benefit, and would stop kMGPipeResourceTargetBuffer being passed as it stands.
+    // Callers pass static_cast<Uint32>(MobileGL::TextureUploadTarget) for `uploadTarget` and
+    // static_cast<Uint32>(MGPipeResourceTarget) - or kMGPipeResourceTargetBuffer - for
+    // `resourceTarget`.
+    constexpr inline Uint16 MGPipePackSubDataTarget(Uint32 resourceTarget, Uint32 uploadTarget) {
+        return static_cast<Uint16>((resourceTarget & 0xFFu) | ((uploadTarget & 0xFFu) << 8));
+    }
+    // Comparable against static_cast<Uint8>(MGPipeResourceTarget) / kMGPipeResourceTargetBuffer.
+    constexpr inline Uint8 MGPipeSubDataResourceTargetOf(Uint16 packed) {
+        return static_cast<Uint8>(packed & 0xFFu);
+    }
+    // Comparable against static_cast<Uint8>(MobileGL::TextureUploadTarget).
+    constexpr inline Uint8 MGPipeSubDataUploadTargetOf(Uint16 packed) {
+        return static_cast<Uint8>((packed >> 8) & 0xFFu);
+    }
+    // THE INVARIANT P3a's records and the applier's buffer test both rest on: a buffer
+    // record's Target is exactly kMGPipeResourceTargetBuffer, whole field, upload byte and
+    // all. TextureUploadTarget::Texture1D is 0, so the buffer case is the one place where the
+    // packed form and a bare enumerator agree - and it has to stay that place.
+    static_assert(MGPipePackSubDataTarget(kMGPipeResourceTargetBuffer, 0u) ==
+                      kMGPipeResourceTargetBuffer,
+                  "a buffer sub-data record's Target must stay exactly "
+                  "kMGPipeResourceTargetBuffer: the applier's SubDataNamesABuffer tests the "
+                  "whole field == 0");
+
     // Encodes a buffer byte range into the record's box. False, with the record untouched,
     // when the range does not fit one record: the emitter has to split it.
     inline Bool MGPipeSetSubDataBufferRange(MGPSubData& record, Uint64 offset, Uint64 size) {
@@ -606,6 +1060,76 @@ namespace MobileGL::MG_Pipe {
     }
     inline Uint64 MGPipeSubDataBufferSize(const MGPSubData& record) { return record.UnionBox.W; }
 
+    // P3a, D-A5: the per-record half of resource_respecify's kNeedsAck.
+    //
+    // Flags are a PER-CALL static property and resource_respecify serves BOTH glBufferData
+    // and glBufferStorage. A bare kNeedsAck on the call would acknowledge every glBufferData
+    // in a world upload; only glBufferStorage is a real synchronous allocation and only it is
+    // allowed a synchronous ack. So kNeedsAck on the call means "records of this call MAY
+    // require an acknowledgement" and THIS predicate decides per record. In monolith the ack
+    // is ((void)0) - the applier is one function call away - and the transport wires the
+    // doorbell to this predicate when it lands.
+    //
+    // P4a, D-A2: THE PREDICATE IS NARROWED TO NAME THE BUFFER TARGET, and that is a
+    // requirement rather than a tidy-up. glTexStorage* also sets Immutable - it is a real
+    // descriptor fact the backend reads, and the client must set it - but texture allocation
+    // is already deferred to sync time in monolith (glTexImage*/glTexStorage* only
+    // MarkStorageDirty; even glRenderbufferStorage* allocates lazily inside SyncToBackend), so
+    // splitting changes no observable behaviour and this batch must NOT ack. glBufferStorage
+    // stays the only entry point allowed a synchronous acknowledgement.
+    //
+    // PipeCatalogueTest.ResourceRespecifyAcksOnlyImmutableStorage drives glTexStorage2D and
+    // glRenderbufferStorage idioms through it, and is the negative control for a future
+    // widening.
+    inline Bool MGPipeResourceRespecifyNeedsAck(const MGPResourceDesc& desc) {
+        return desc.Immutable != 0 && desc.Target == kMGPipeResourceTargetBuffer;
+    }
+
+    // P4a, ID-18 M4: A RESPECIFY WHOSE STORAGE-DEFINING FIELDS ALL EQUAL THE STORED
+    // DESCRIPTOR IS A METADATA UPDATE, NOT A REALLOCATION.
+    //
+    // MGPResourceDesc::BindMask and ImageBindableHint are STICKY facts the client discovers
+    // AFTER allocation - a texture first bound as a shader image, first used as a render
+    // target - and they ride resource_create and every resource_respecify. An IMMUTABLE
+    // texture never has a later respecify, so without a rule those two would reach the server
+    // only by accident, or never; with one, a mask change after allocation emits a
+    // resource_respecify that REPEATS the storage the resource already has.
+    //
+    // The applier and both twins must classify such a record as a metadata update:
+    //   - NO reallocation acknowledgement. MGPipeResourceRespecifyNeedsAck above still
+    //     answers the per-record question, but a metadata update allocates nothing, so a
+    //     record it classifies as metadata is not acked even when that predicate says the
+    //     call may require one.
+    //   - NO PendingUploads clear when the call names NO level. This REFINES the whole-resource
+    //     clear: identical storage fields with a null MGPRespecifiedLevel clear NOTHING. (The
+    //     level-scoped rule exists because clearing the whole vector on a level-1 definition
+    //     silently dropped level 0's accepted texels; a metadata update must drop neither.) A
+    //     call that NAMES a level is that level's redefinition whatever the descriptor says -
+    //     a non-base level's extent is not a descriptor field - and drops exactly that level
+    //     (P4a final review C-1); the client's mask republish passes null on purpose.
+    //   - The stored descriptor's BindMask and ImageBindableHint ARE updated: the applier
+    //     replaces the descriptor WHOLE with the one the client sent (PipeApply.cpp), and the
+    //     mask in it is the CLIENT's sticky OR (TextureEmit.h's entry, never cleared), so the
+    //     replacement can never lose a bit the record once carried. The twin re-derives its
+    //     storage flags from the new mask on its next sync, recreating backend storage only
+    //     where the backend actually needs it. The record itself is not a request to recreate.
+    //
+    // THE STORAGE-DEFINING FIELD SET, named here so that neither side has to guess and a
+    // later field cannot join it by silence. It is every MGPResourceDesc member except the
+    // three metadata ones and the padding:
+    //
+    //     Target, StorageKind, InternalFormat, Width, Height, Depth, ArrayLayers, Levels,
+    //     Samples, FixedSampleLocations, Immutable, Usage, StorageFlags, HasDefinedContent,
+    //     ViewOf, BufferForTexBuffer, BufOffset, BufSize.
+    //
+    // `Resource` is the identity the stored descriptor is looked up BY, not a comparand. The
+    // three fields that may differ on a metadata update are exactly BindMask,
+    // ImageBindableHint and GlNameForDiag (diagnostics only, never an identity, never a memo
+    // key). HasDefinedContent is storage-defining ON PURPOSE: glBufferData(size, NULL) at an
+    // unchanged size is an orphaning reallocation and has to keep clearing, rather than being
+    // mistaken for a mask change. A field added to MGPResourceDesc must be placed in one of
+    // the two lists in the same commit - PipeCatalogue pins the struct's size for that.
+
     // The forward terminator for a server-initiated texture pull (section 7.1). May carry
     // zero regions - that is how a pull that needs nothing is answered.
     struct MGPSubDataComplete {

MobileGL/MG_Pipe/MGPipeValueTypes.h

@@ -289,6 +289,21 @@ namespace MobileGL {
         // Every entry is initialized to all-true in RenderState's constructor.
         Array<BoolVec4, kMGMaxDrawBuffers> ColorMasks;
 
+        // GL_FRAMEBUFFER_SRGB / GL_DEPTH_CLAMP / GL_TEXTURE_CUBE_MAP_SEAMLESS. Until P2 these
+        // three fell to SetCapability's "not supported currently" arm - glEnable was swallowed
+        // and IsCapabilityEnabled answered a compile-time false, so DirectGLES' sRGB block and
+        // the DirectVulkan read points consumed a constant while glIsEnabled lied about it.
+        // Placed HERE, in the three alignment bytes between ColorMasks (32 bytes, align 1) and
+        // ClearColor (align 4), so sizeof(RenderStateParameters) stays 1168 and no existing
+        // offset moves: the Espryt span constants and the P2 chunk table both depend on that.
+        // All three are PIPELINE state (their setters call BumpVersions): FramebufferSrgb is
+        // what ARCHITECTURE.md 5.3 asks for, DepthClamp is
+        // VkPipelineRasterizationStateCreateInfo::depthClampEnable, and TextureCubeMapSeamless
+        // changes sampler interpretation.
+        Bool FramebufferSrgbEnabled = false;
+        Bool DepthClampEnabled = false;
+        Bool TextureCubeMapSeamlessEnabled = false;
+
         // Clear State
         FloatVec4 ClearColor = FloatVec4(0.0f, 0.0f, 0.0f, 1.0f);
         Float ClearDepth = 1.0f;
@@ -529,6 +544,61 @@ namespace MobileGL {
         };
     } // namespace MG_State::GLState
 
+    // ---- P3a: the WIRE forms of the two views above (ARCHITECTURE.md section on vertex
+    // elements; brief D-G2). Neither VertexAttribute nor VertexBufferBindingPoint can travel
+    // as itself: both hold a SharedPtr<BufferObject>, and a payload never contains a pointer.
+    // They live here rather than in MGPipeTypes.h so the structs they mirror are one screen
+    // away and a member added above has its wire twin in view; MGPipeTypes.h includes this
+    // header, so MG_Pipe sees them unqualified like every other value type.
+    //
+    // Both ride the create_vertex_elements BLOB, in ascending index order, attributes first:
+    // MGPVertexAttribWire[AttributeCount] then MGPVertexBindingPointWire[BindingPointCount],
+    // each count <= VertexArrayObject::MAX_VERTEX_ATTRIBS (32). The applier refuses a record
+    // whose declared counts do not match the blob's declared size.
+
+    // The resolved flat attribute view. Buffer identity does NOT travel here - it travels in
+    // set_vertex_buffers, which is what keeps this record stable while buffers change under
+    // it. Stride is the RESOLVED distance and a surviving 0 can only have come from the
+    // binding model (see VertexAttribute::Stride above); collapsing it back into the element
+    // size is what made KHR-GL43.vertex_attrib_binding.basic-input-case7/8 read past the
+    // buffer. Divisor is deliberately ABSENT: it is resolved per binding point and travels in
+    // MGPVertexBuffer::Divisor, which is where the backend's glVertexAttribDivisor reads it.
+    // LegacyStride / LegacyPointer are likewise absent - they are the glGetVertexAttrib*
+    // query answers and stay client-side, because nothing but the query path reads them.
+    struct MGPVertexAttribWire {
+        Uint64 Offset;       //  0
+        Int32 Stride;        //  8
+        Uint32 Type;         // 12  DataType
+        Uint8 Size;          // 16  1..4; GL_BGRA keeps 4
+        Uint8 Enabled;       // 17
+        Uint8 Normalized;    // 18
+        Uint8 IsInteger;     // 19
+        // CARRIED SEPARATELY from Type == Float64, and it has to be: VertexAttribFormat(
+        // GL_DOUBLE) also reads doubles from memory but asks for them converted to float,
+        // while VertexAttribLFormat keeps all 64 bits. The backend's fp64 narrowing and its
+        // Adreno disabled-attribute workaround both key on telling the two apart.
+        Uint8 IsLong;        // 20
+        Uint8 IsBgra;        // 21
+        Uint8 BindingIndex;  // 22  which MGPVertexBuffer entry feeds it (< MAX_VERTEX_ATTRIBS)
+        Uint8 Pad0;          // 23
+    };
+
+    // The ARB_vertex_attrib_binding view. Buffer identity is again in set_vertex_buffers.
+    //
+    // WHY IT TRAVELS AT ALL, since no backend has ever read a binding point (the frontend
+    // resolves them eagerly into the flat view above, and grep finds zero backend reads of
+    // VertexBufferBindingPoint / GetAttributeBindingIndex / GetAttributeRelativeOffset): the
+    // record DECLARES BindingPointCount, PipeFields.def names it, and a record whose declared
+    // counts do not describe its own blob is a shape the applier's bounds gate would have to
+    // police forever. Carrying both views keeps the record self-describing, and the cost is
+    // paid once per configuration change rather than per draw - the blob rides only on
+    // create_vertex_elements.
+    struct MGPVertexBindingPointWire {
+        Uint64 Offset;  // 0
+        Int32 Stride;   // 8   GL 4.6 core table 23.4: the INITIAL value is 16, not 0
+        Uint32 Divisor; // 12
+    };
+
     // ---- trip wires (P0.5). Sizes are what every ABI MobileGL ships on produces: every
     // member is a fixed-width scalar, an enum of one, or an array of those - no pointer, no
     // SizeT - except the vertex types, which carry SharedPtr<BufferObject> by design and are
@@ -545,5 +615,16 @@ namespace MobileGL {
     static_assert(std::is_trivially_copyable_v<SamplerParameters> && sizeof(SamplerParameters) == 100);
     static_assert(std::is_trivially_copyable_v<MG_State::GLState::VertexAttributeVersion> &&
                   sizeof(MG_State::GLState::VertexAttributeVersion) == 6);
+    // The two P3a wire views. Unlike the structs they mirror these ARE flat PODs with
+    // explicit padding, so the trip wire is the same one every MGPipe payload carries: the
+    // blob they ride in is memcpy'd, and a field silently changing width is a protocol break
+    // no test would otherwise see. (MGP_ASSERT_POD is MGPipeTypes.h's and that header
+    // includes this one, so the assertions are spelled out here instead.)
+    static_assert(std::is_trivially_copyable_v<MGPVertexAttribWire> &&
+                  sizeof(MGPVertexAttribWire) == 24);
+    static_assert(std::is_standard_layout_v<MGPVertexAttribWire>);
+    static_assert(std::is_trivially_copyable_v<MGPVertexBindingPointWire> &&
+                  sizeof(MGPVertexBindingPointWire) == 16);
+    static_assert(std::is_standard_layout_v<MGPVertexBindingPointWire>);
 } // namespace MobileGL
 #endif // MOBILEGL_MG_PIPE_VALUE_TYPES_H

MobileGL/MG_Pipe/PipeCalls.def

@@ -16,6 +16,12 @@
 //             kScreen lands in struct MGPipeScreen, every other class in struct
 //             MGPipeContext (plan section 4.3).
 //     Flags : kNone | kNeedsAck | kHasBlob | kVarTail | kHostSpan | kReplySlot | kOptional
+//             kNeedsAck on a call means records of this call MAY require an ack; a
+//             per-record predicate decides. resource_respecify carries it for
+//             glBufferStorage - a real synchronous allocation, and the only entry point
+//             allowed a synchronous ack - and MGPipeResourceRespecifyNeedsAck(desc)
+//             (MGPipeTypes.h) is what says so, which is why the same call still carries
+//             every glBufferData without acknowledging one.
 //
 // RECORD NUMBERING NEVER CHURNS. Entries that are not implemented yet still occupy their
 // line (plan section 11, P0: "the complete call catalogue, placeholders included"). A new
@@ -73,7 +79,7 @@
     /* ---- screen: caps, resources, persistent map, fences (plan 4.4.1) ---- */                    \
     X(GetCaps,                  MGPCaps,                    kScreen,    kReplySlot)                 \
     X(ResourceCreate,           MGPResourceDesc,            kScreen,    kNone)                      \
-    X(ResourceRespecify,        MGPResourceDesc,            kScreen,    kNone)                      \
+    X(ResourceRespecify,        MGPResourceDesc,            kScreen,    kNeedsAck)                  \
     X(ResourceDestroy,          MGPHandleOnly,              kScreen,    kNone)                      \
     X(MapPersistent,            MGPHandleOnly,              kScreen,    kReplySlot|kOptional)       \
     X(UnmapPersistent,          MGPHandleOnly,              kScreen,    kOptional)                  \

MobileGL/MG_Pipe/PipeFields.def

@@ -78,26 +78,36 @@
     F(Cso) F(Texture) F(InternalFormat) F(Target) F(MinLevel) F(NumLevels) F(MinLayer) F(NumLayers) \
     F(Samples) F(FixedSampleLocations)
 
+// P4a, D-E1: BuiltinSampler and SamplerResync. Pad0 stays unlisted - gen_pipe.py's
+// PADDING_MEMBER_RE (^Pad\d*$) excludes it, and a member that stops being called Pad<n> MUST
+// gain a row here or pipe-gates goes red.
 #define MGP_FIELDS_MGPTextureParams(F) \
-    F(Res) F(BaseLevel) F(MaxLevel) F(Swizzle) F(DepthStencilMode) F(ForceResync) F(MinLod) F(MaxLod) \
-    F(LodBias)
+    F(Res) F(BuiltinSampler) F(BaseLevel) F(MaxLevel) F(Swizzle) F(DepthStencilMode) F(ForceResync) \
+    F(SamplerResync) F(MinLod) F(MaxLod) F(LodBias)
 
 #define MGP_FIELDS_MGPProgramDesc(F) \
     F(Cso) F(StageMask) F(GlobalUboSize) F(ReservedNumSamplesOffset) F(SpirvStatus) F(NativeFloat64) \
     F(PointSizeDemoted) F(EnableSpirvValidation) F(Spirv) F(Reflection)
 
+// P4a, ID-12 / esprytobj DV-5: Pad0 became Uint16 TextureTarget. Same trip wire as
+// MGPFramebufferState's Target below - PADDING_MEMBER_RE only excludes a member still NAMED
+// Pad<n>, so the rename without this row is a pipe-gates failure, and the row without the
+// rename is one too. A meaning-carrying byte cannot enter this record silently.
 #define MGP_FIELDS_MGPSurface(F) \
-    F(Res) F(InternalFormat) F(Kind) F(Layered) F(Level) F(Layer) F(UploadTarget)
+    F(Res) F(InternalFormat) F(Kind) F(Layered) F(Level) F(Layer) F(UploadTarget) F(TextureTarget)
 
+// P4a, D-C2: Pad0 became Uint8 Target, and gen_pipe.py's PADDING_MEMBER_RE only excludes a
+// member still NAMED Pad<n> - so the rename without this row is a pipe-gates failure, which
+// is exactly the trip wire that makes the byte impossible to add silently.
 #define MGP_FIELDS_MGPFramebufferState(F) \
     F(Fbo) F(Color) F(Depth) F(Stencil) F(ReadSurface) F(DrawBuffers) F(Width) F(Height) F(Layers) \
-    F(Samples) F(FixedSampleLocations) F(IsDefault) F(Complete) F(ContentHash)
+    F(Samples) F(FixedSampleLocations) F(IsDefault) F(Complete) F(Target) F(ContentHash)
 
 #define MGP_FIELDS_MGPVertexBuffer(F) \
     F(Res) F(Offset) F(Stride) F(Divisor) F(BindingIndex)
 
 #define MGP_FIELDS_MGPVertexBuffers(F) \
-    F(Start) F(Count) F(ContentHash)
+    F(Start) F(Count) F(BaseInstance) F(ContentHash)
 
 #define MGP_FIELDS_MGPIndexBuffer(F) \
     F(Res) F(Offset) F(IndexSize)
@@ -144,8 +154,11 @@
 #define MGP_FIELDS_MGPPatchState(F) \
     F(Vertices) F(Outer) F(Inner)
 
+// P2 ratcheted this block from six rows to one: RenderStateParameters retired to
+// create/bind_render_state + set_dynamic_state, Pack to set_pixel_pack_state and the
+// patch trio to set_patch_state. What is left is the redundant capability trip wire.
 #define MGP_FIELDS_ResidualValueBlock(F) \
-    F(RenderState) F(Pack) F(CapabilityBits) F(PatchVertices) F(PatchOuter) F(PatchInner)
+    F(CapabilityBits)
 
 #define MGP_FIELDS_MGPResidualValueState(F) \
     F(Version) F(Blob)
@@ -231,7 +244,8 @@
     F(Viewports) F(LineWidth) F(PointSize) F(PatchVertices) F(PatchDefaultOuterLevel) \
     F(PatchDefaultInnerLevel) F(PolygonOffsetFactor) F(PolygonOffsetUnits) F(PolygonOffsetClamp) \
     F(ClipOrigin) F(ClipDepthMode) F(BlendStates) F(LogicOp) F(DepthTestEnabled) F(DepthFunc) \
-    F(DepthMask) F(ColorMasks) F(ClearColor) F(ClearDepth) F(ClearStencil) F(BlendColor) \
+    F(DepthMask) F(ColorMasks) F(FramebufferSrgbEnabled) F(DepthClampEnabled) \
+    F(TextureCubeMapSeamlessEnabled) F(ClearColor) F(ClearDepth) F(ClearStencil) F(BlendColor) \
     F(DepthRanges) F(SampleCoverageValue) F(SampleCoverageInvert) F(SampleMaskValue) \
     F(MinSampleShadingValue) F(StencilStates) F(CullFaceEnabled) F(CullFaceModeSetting) \
     F(FrontFaceModeSetting) F(ProvokingVertexModeSetting) F(LineSmoothHint) F(PolygonSmoothHint) \
@@ -250,6 +264,18 @@
     F(SwapBytes) F(LSBFirst) F(RowLength) F(ImageHeight) F(SkipPixels) F(SkipRows) F(SkipImages) \
     F(Alignment)
 
+// P4a, D-F1: THE PADDING TRAP. SamplerParameters is sizeof == 100 with THREE BYTES OF
+// TRAILING PADDING (96 bytes of members plus the 1-byte borderColorForm) and had no field
+// table and no MGP_VERIFY_PAYLOAD_LIST row at all, so MGPSamplerDesc's blob was compared as
+// BYTES and MOBILEGL_PIPE_VERIFY could false-differ on uninitialised padding - a coin flip
+// rather than a gate. With this list the comparator sees the sixteen members and the three
+// bytes can never enter the answer. The client-side CSO cache hashes and memcmp-confirms over
+// a ZERO-INITIALISED canonical copy for the same reason, which is the other half of D-F1.
+#define MGP_FIELDS_SamplerParameters(F) \
+    F(wrapS) F(wrapT) F(wrapR) F(minFilter) F(magFilter) F(mipmapMode) F(minLod) F(maxLod) \
+    F(lodBias) F(maxAnisotropy) F(compareFunc) F(compareMode) F(borderColor) F(borderColorI) \
+    F(borderColorUI) F(borderColorForm)
+
 #define MGP_FIELDS_PerBufferBlendState(F) \
     F(Enabled) F(SrcFactorRGB) F(DstFactorRGB) F(SrcFactorAlpha) F(DstFactorAlpha) F(ColorEquation) \
     F(AlphaEquation)
@@ -291,6 +317,21 @@
 #define MGP_FIELDS_MGHostSpan(F) \
     F(Ptr) F(Seg) F(Size) F(Offset)
 
+// P3a's two vertex wire views (MGPipeValueTypes.h). They are not call payloads either: they
+// are the ELEMENTS of create_vertex_elements' blob, and the comparator has to see into them
+// for the same reason it sees into the value structs - a blob compared with memcmp would
+// false-differ on MGPVertexAttribWire::Pad0. Divisor is deliberately not in the attribute
+// list (it travels in MGPVertexBuffer) and the two Legacy* query answers are deliberately not
+// on the wire at all; both absences are argued in MGPipeValueTypes.h and both are enforced
+// here by gen_pipe.py's "every list names exactly its struct's direct members" rule.
+
+#define MGP_FIELDS_MGPVertexAttribWire(F) \
+    F(Offset) F(Stride) F(Type) F(Size) F(Enabled) F(Normalized) F(IsInteger) F(IsLong) F(IsBgra) \
+    F(BindingIndex)
+
+#define MGP_FIELDS_MGPVertexBindingPointWire(F) \
+    F(Offset) F(Stride) F(Divisor)
+
 // Every payload above, in the order the comparator is generated. Keep in sync with the
 // macros; gen_pipe.py reads THIS list to know what to emit.
 #define MGP_VERIFY_PAYLOAD_LIST(P) \
@@ -307,7 +348,9 @@
     P(MGPDrawRange) P(MGPDrawIndirect) P(MGPGridInfo) P(MGPMemoryBarrier) P(MGPStreamOutputBegin) \
     P(MGPXfbAccounting) P(MGPStreamOutputControl) P(MGPFlush) P(MGPPresent) P(MGPSwapInterval) \
     P(MGPSurfaceInfo) \
-    P(RenderStateParameters) P(PixelStoreParameters) P(PerBufferBlendState) P(StencilFaceState) \
-    P(DynamicBackendParameters) P(MGHostSpan)
+    P(RenderStateParameters) P(PixelStoreParameters) P(SamplerParameters) P(PerBufferBlendState) \
+    P(StencilFaceState) \
+    P(DynamicBackendParameters) P(MGHostSpan) \
+    P(MGPVertexAttribWire) P(MGPVertexBindingPointWire)
 
 // clang-format on

MobileGL/MG_Pipe/PipeMutation.h

@@ -33,10 +33,395 @@ namespace MobileGL::MG_Pipe {
     // A no-op unless a context is live, a verb has been filled, and `field` is in that verb
     // class's may-read mask; a forwarded (sticky) field has no storage and is never copied.
     void MGPipeNoteFrontendMutation(MGPipeInputField field);
+
+    // ---- the aggregate generations (P2 brief D4, ARCHITECTURE.md 5.2) ----
+    //
+    // MGP_NOTE_MUTATION answers "a backend moved a frontend value INSIDE its own verb".
+    // MGP_NOTE_AGGREGATE answers a different question, which is why it is a second macro
+    // and not an overload: "did ANY object of this class move since the last time the
+    // tracker looked", collapsed onto one monotonic Uint64 per class so a per-verb dirty
+    // walk is a handful of compares rather than a scan over 32 attributes, 16 attachments,
+    // 32 texture units and 84 binding points.
+    //
+    // The counters are members of the owning MG_State container, all guarded by
+    // MOBILEGL_PIPE_PUSH so the pull build's state objects do not change size (G1). The
+    // bump points sit on OBJECTS, which have no back-pointer to their state, so the macro
+    // goes through a free function that finds the live GLContext - the same shape, and for
+    // the same reason, as MGP_NOTE_MUTATION (MG_Impl/Pipe/PipeFill.cpp). It costs a global
+    // load on a path that has just written object state.
+    //
+    // Monotonic and never reset: the tracker widens and compares, it never subtracts.
+    // Over-firing is free (one extra re-push); under-firing renders stale, which is why
+    // every counter here is deliberately COARSER than the state it guards.
+    enum class MGPipeAggregate : Uint32 {
+        // VertexArrayState: any VAO attribute format / buffer / enable moved.
+        VaoAttribute = 0,
+        // FramebufferState: any FBO attachment or default-geometry write, or a bind - and, since
+        // P4a (fable seam F-3), any STORAGE DEFINITION of a texture or a renderbuffer, because
+        // set_framebuffer_state inlines an attachment's format, extent and samples and those
+        // setters are the only writers of what it inlines (TextureObject.cpp /
+        // RenderbufferObject.cpp, PipePublishDescriptor).
+        FramebufferAttachment,
+        // TextureState: any texture object CONTENT moved (an upload, a dirty region).
+        TextureContent,
+        // TextureState: any texture object or sampler object PARAMETER moved.
+        TextureParams,
+        // BufferState: any buffer object contents moved.
+        BufferChange,
+        // GLContext: a glVertexAttrib* default value moved. Not one of D4 five: the bit it
+        // shutters (NEW_VERTEX_ATTRIB_DEFAULTS) is specified there as a ContentHash over
+        // all 32 CurrentVertexAttributeValues, and hashing 768 bytes on EVERY draw does not
+        // fit inside the T1 ceiling. The hash still decides whether to EMIT (D11 set-hash
+        // suppressor); this decides whether to hash at all.
+        VertexAttribDefault,
+        Count,
+    };
+
+    // MG_Impl/Pipe/PipeFill.cpp. A no-op unless a context is live.
+    void MGPipeNoteAggregate(MGPipeAggregate aggregate);
+
+    // ---- P3a: the resource family's emission points (brief D-A1) ----
+    //
+    // The seven BufferBackendOps hooks already dispatch at the GL call that causes them
+    // (ARCHITECTURE.md 5.1 names them as the ONE exception to push-at-validate), so their
+    // pipe calls are emitted from the same BufferObject dispatchers rather than from the
+    // validate point. That puts the emission inside MG_State, which is why these are
+    // DECLARED here beside the two notices and DEFINED in MG_Impl/Pipe/PipeFill.cpp: this
+    // header is the one MG_State already includes for exactly this, and the closure gate
+    // (check_include_closure.py's mutation-header probe) keeps it a declaration - reaching
+    // MG_Impl/Pipe/ResourceTracker.h from BufferObject.cpp would pull the client's tracker
+    // into the state machine that calls it.
+    //
+    // The forward declaration is the whole coupling: none of these needs the definition of
+    // BufferObject, and this header must not gain it.
+} // namespace MobileGL::MG_Pipe
+
+namespace MobileGL::MG_State::GLState {
+    class BufferObject;
+    // P4a's five, for the BIRTH half at the tail of this header. Declarations only, exactly as
+    // BufferObject is: none of the hooks below needs a definition, and this header must not
+    // gain one - reaching a frontend class header from here would put the state machine's own
+    // types in front of every mutator that spells MGP_NOTE_MUTATION.
+    class ITextureObject;
+    class RenderbufferObject;
+    class FramebufferObject;
+    class SamplerObject;
+    class ProgramObject;
+}
+
+namespace MobileGL::MG_Pipe {
+    // (Features.PipePush & kMGPipeSubsystemResources) != 0 && MGPipeGetResourceOps() != nullptr.
+    //
+    // BOTH HALVES MATTER. The bit is the operator's per-subsystem A/B; the table is "has a
+    // backend taken this family over at all". Until one has, every dispatch below falls
+    // through to the BufferBackendOps table it replaces and the tree behaves exactly as it
+    // did - which is what lets the client half land on its own.
+    Bool MGPipeResourceSubsystemEnabled();
+    // The nullable member, asked the way the frontend asks g_bufferBackendOps->ResidentSubData
+    // today: one backend deliberately does not implement it and the caller has a different
+    // path when it is absent (BufferObject::FillSubData).
+    Bool MGPipeResourceOpsHaveSubDataResident();
+
+    // Minted from the constructor and released from the destructor, both unconditionally in
+    // a push build: a handle is CLIENT state and set_vertex_buffers names it whether or not
+    // the resource family is switched on. The CALLS are what the predicate above gates.
+    void MGPipeMintResourceHandle(MG_State::GLState::BufferObject& buffer);
+    // In this order, and it is not negotiable (D-L): the destroy resolves the handle, and
+    // MGPipeSlotAllocator::Free erases the lifetimeId -> slot mapping it resolves through.
+    //
+    // RETURNS whether resource_destroy was emitted, which is the LATCH taken at this buffer's
+    // create and not a second reading of MGPipeResourceSubsystemEnabled(). The destructor
+    // needs that answer to decide whether the legacy OnDestroy still owes a call: asking the
+    // predicate twice pairs a create emitted under one registration with a destroy gated on
+    // another, and either direction leaks - a live applier record on a slot about to be
+    // re-handed-out, or a backend object nobody releases.
+    Bool MGPipeEmitResourceDestroyAndFree(MG_State::GLState::BufferObject& buffer);
+
+    // THE VERTEX-ELEMENTS CSO's DEATH, and it is BACKEND-NEUTRAL - which is the whole point.
+    // Before this, the only thing that ever returned a VertexElementsCso slot was DirectGLES'
+    // StateObjectDeathOps table; under any backend that installs none - DirectVulkan/Magma,
+    // which keeps its own age-reclaimed identity table on purpose - every VAO ever created
+    // held its slot and its ~1.3 KB applier record for the life of the process, on the shipped
+    // 0x1ff mask, and past 65536 slots every create_vertex_elements became a permanent
+    // Fatal{ProtocolCorruption}. The client mints the slot, so the client is where the death
+    // has to be spoken from.
+    //
+    // Takes the lifetime id and not the object for StateObjectDeathNotice.h's reason: the last
+    // SharedPtr has already dropped by the time this runs, and the lifetime id is what the
+    // slot allocator resolves the handle from. Returns whether delete_vertex_elements went
+    // out, i.e. whether the applier actually held a record - see the definition for why that
+    // is asked rather than assumed.
+    Bool MGPipeEmitVertexElementsDestroyAndFree(Uint64 lifetimeId);
+
+    // ---- P4a: ONE CLIENT-SIDE DEATH HELPER PER KIND P4a MINTS (brief D-I1) ----
+    //
+    // BACKEND-NEUTRAL FROM DAY ONE, and this is the P3a final-review lesson taken forward
+    // rather than repeated. Before it, the only thing that ever returned a VertexElementsCso
+    // slot was DirectGLES' StateObjectDeathOps table; under a backend that installs none -
+    // DirectVulkan/Magma, which keeps its own age-reclaimed identity table on purpose - every
+    // VAO ever created held its slot and its applier record for the life of the process, and
+    // past 65536 slots every create became a permanent Fatal{ProtocolCorruption}. P4a mints
+    // SIX kinds, so the rule is stated once and obeyed six times: whatever mints a handle owns
+    // the death of that handle, the client mints all six, and a backend death notice is a
+    // redundant SECOND path that must be idempotent - which it is, because it resolves through
+    // the same lifetimeId -> slot map these free, and MGPipeSlotAllocator::Free refuses a slot
+    // that is not live at that generation.
+    //
+    // THE ORDER INSIDE EACH IS FIXED AND IS NOT A PACKAGE'S CHOICE:
+    //   1. emit the wire delete FIRST - it drops the applier's record while the record still
+    //      exists, so a recycled slot cannot inherit a field;
+    //   2. raise NotifyStateObjectDestroyed SECOND - it resolves the handle through the
+    //      allocator, and a backend told after the Free could no longer find its twin, which
+    //      moves the leak from the client to the driver object;
+    //   3. free the slot LAST, and a double free on a stale generation is a proven no-op
+    //      because Free bumps no generation (the bump rides the next handout).
+    //
+    // ALL SIX TAKE THE LIFETIME ID rather than the object, for MGPipeEmitVertexElementsDestroy
+    // AndFree's reason: they run from a destructor, where the last SharedPtr has already
+    // dropped, and the lifetime id is what the slot allocator resolves the handle from. It is
+    // also what keeps this header a declaration-only coupling - no frontend class needs
+    // forward-declaring for any of them.
+    //
+    // Each returns whether its wire delete actually went out, which is the LATCH taken at the
+    // object's create and not a second reading of the subsystem predicate: an object born
+    // while a subsystem bit was clear and destroyed after it was set would otherwise free its
+    // slot with the applier's record still Live, on a slot about to be handed out again. The
+    // legacy path runs only when the answer is false.
+
+    // ResourceDestroy, and then the SamplerViewCso minted off this same lifetime id (P4a
+    // D-F2: one sampler view per ITextureObject). Called from TextureObjectBase's VIRTUAL
+    // destructor, so 2D / 3D / cube / buffer / view all announce exactly once.
+    Bool MGPipeEmitTextureDestroyAndFree(Uint64 lifetimeId);
+    // ResourceDestroy.
+    Bool MGPipeEmitRenderbufferDestroyAndFree(Uint64 lifetimeId);
+    // NO WIRE CALL AT ALL (D-I2). PipeCalls.def has no framebuffer delete, because a
+    // framebuffer is not a resource and is not a CSO - it is STATE, and set_framebuffer_state
+    // is the only call that names one - and the catalogue is closed, so P4a does not invent a
+    // row. The handle is minted and freed entirely client-side and this helper does steps 2
+    // and 3 only. A recycled framebuffer handle is distinguished by Gen, which is inside the
+    // record's ContentHash, so it can never be suppressed against its predecessor's record.
+    Bool MGPipeEmitFramebufferDestroyAndFree(Uint64 lifetimeId);
+    // DeleteSamplerState. Also the path the content-addressed CSO cache's LRU eviction takes,
+    // which is why it is addressed by lifetime id and not by "the object that owns it".
+    Bool MGPipeEmitSamplerCsoDestroyAndFree(Uint64 lifetimeId);
+    // DeleteSamplerView. Called by the texture helper above; a sampler view has no frontend
+    // object of its own, so this is the only path there is.
+    Bool MGPipeEmitSamplerViewCsoDestroyAndFree(Uint64 lifetimeId);
+    // DeleteShaderState, for an ordinary program AND for a program-pipeline COMPOSITE, whose
+    // slot has two independent release paths - the pipeline cache's LRU eviction and the
+    // composite ProgramObject's own destructor. One helper for both, and the second call is a
+    // proven no-op.
+    Bool MGPipeEmitShaderCsoDestroyAndFree(Uint64 lifetimeId);
+
+    void MGPipeEmitResourceCreate(MG_State::GLState::BufferObject& buffer);
+    void MGPipeEmitResourceRespecify(MG_State::GLState::BufferObject& buffer);
+    void MGPipeEmitResourceSubData(MG_State::GLState::BufferObject& buffer, SizeT offset, SizeT size);
+    void MGPipeEmitBufferSubDataResident(MG_State::GLState::BufferObject& buffer, SizeT offset,
+                                         const void* bytes, SizeT size);
+    void MGPipeEmitResourceFlushRange(MG_State::GLState::BufferObject& buffer, SizeT offset, SizeT size,
+                                      Uint32 accessFlags);
+    void MGPipeEmitResourceReadback(MG_State::GLState::BufferObject& buffer);
+    // Returns the coherent host pointer the resource owner donated, or null for a DECLINE -
+    // which is a real answer. Every call, mint or decline, is one map-persistent roundtrip.
+    void* MGPipeEmitMapPersistent(MG_State::GLState::BufferObject& buffer);
+
+    // ================================================================================
+    // P4a: THE BIRTH HALF, one hook per client path MG_State owns (D-C .. D-I)
+    // ================================================================================
+    //
+    // The death helpers above are half a lifetime. The other half is emitted from MG_State
+    // too - a texture's create from its constructor, a renderbuffer's respecify from its
+    // storage mutators, a texture's params from glTexParameter*, a sampler CSO from the
+    // sampler object, a shader CSO from the program - because that is where the event
+    // happens, exactly as P3a's buffer family emits from BufferObject's own dispatchers
+    // (ARCHITECTURE.md 5.1 names those as the ONE exception to push-at-validate). Only the
+    // texture sub-data DRAIN runs at the validate point, and even it is fed from here: the
+    // drain list is appended on a level's first dirty mark.
+    //
+    // WHY THEY ARE DECLARED HERE. This header is the one door MG_State has into the client
+    // (check_include_closure.py's mutation-header probe pins it: reaching
+    // MG_Impl/Pipe/*Emit.h from a frontend mutator would pull the client's emitters into the
+    // state machine that calls them). So a hook a frontend mutator calls is DECLARED here and
+    // DEFINED in MG_Impl/Pipe/PipeFill.cpp, which is package A's for the whole phase - the
+    // same "declaration here, definition there" split MGPipeMintResourceHandle and
+    // MGPipeEmitResourceCreate use, and the reason no file is touched twice.
+    //
+    // WHAT EACH BODY DOES, and the division is fixed:
+    //   * PipeFill.cpp owns the GATE - the subsystem bit in MOBILEGL_PIPE_PUSH *and* the
+    //     family's own kMGPipeWired*Subsystem constant, the same pair the validate point's
+    //     `wants()` applies to every emission - and the four MINTS, which are pure allocator
+    //     work and need no family knowledge;
+    //   * the FAMILY EMITTER (MG_Impl/Pipe/<Family>Emit.h, owned by package B or C) owns the
+    //     payload build, the handle rule for its own kind and the PUBLICATION LATCH below.
+    //     PipeFill.cpp forwards to it through an entry point that is compiled only while that
+    //     family's wired constant is non-zero, so this tree links against the STUB emitters
+    //     and against the finished ones with no edit to PipeFill.cpp - and a family that sets
+    //     its constant without providing the entry point is a COMPILE ERROR in its own commit
+    //     rather than a surprise at the merge. The entry point each hook forwards to is named
+    //     beside it and spelled out in PipeFill.cpp's contract block.
+    //
+    // NOTHING CALLS ANY OF THEM AT THE CONTRACT COMMIT. B and C add the call sites in the
+    // five MG_State directories C.7 gives them, in the SAME commit that gives the emitter its
+    // body - by EDITING an existing constructor/mutator body, never by adding one (G1).
+
+    // ---- the publication latch (D-I1), and it is the ONE answer both halves read ----
+    //
+    // The create is gated at its call site and the destroy inside the death helper, so the
+    // two ask the same question at two different moments. An object born while its subsystem
+    // bit was clear and destroyed after it was set would otherwise free its slot with the
+    // applier's record still Live - on a slot the allocator is about to hand out again. A
+    // slot is NOT evidence of a record either: a backend twin table mints one through
+    // MGPipeSlots().Acquire whether or not the subsystem ever asked this client to emit a
+    // create, and a delete_* on such a handle is a refused call the applier asserts on.
+    //
+    // So the emitter latches the answer when its create actually goes out, the death helper
+    // reads the latch, and the latch is keyed by {kind, slot, gen} so a recycled slot cannot
+    // inherit its predecessor's answer. Defined in PipeFill.cpp beside the six death helpers,
+    // declared here because both the helpers and the five emit headers read it.
+    void MGPipeNoteHandlePublished(MGPipeKind kind, MGPipeHandle handle);
+    Bool MGPipeHandleIsPublished(MGPipeKind kind, MGPipeHandle handle);
+    void MGPipeNoteHandleUnpublished(MGPipeKind kind, MGPipeHandle handle);
+
+    // ---- the four mints (pure allocator work, no family knowledge) ----
+    //
+    // UNCONDITIONAL in a push build, for MGPipeMintResourceHandle's reason: a handle is CLIENT
+    // state and other subsystems name these objects by handle whether or not their own family
+    // is switched on - MGPSurface::Res names a Texture or a Renderbuffer out of the framebuffer
+    // subsystem, MGPBoundView::Texture and MGPImageView::Res name a Texture out of the sampler
+    // one. Gating the mint on the family bit would make those emit null handles in exactly the
+    // A/B arm that exists to isolate the families. Each costs one free-list pop and one map
+    // insert per object and emits nothing.
+    void MGPipeMintTextureHandle(MG_State::GLState::ITextureObject& texture);
+    void MGPipeMintRenderbufferHandle(MG_State::GLState::RenderbufferObject& renderbuffer);
+    // A framebuffer has a handle and NO wire lifetime (D-I2): set_framebuffer_state is the only
+    // call that names one, and there is no create or destroy for the kind. The mint is still
+    // the object's, so the identity exists before the first validate point that pushes it.
+    void MGPipeMintFramebufferHandle(MG_State::GLState::FramebufferObject& framebuffer);
+    // Ordinary programs only. A program-pipeline COMPOSITE is minted by the composite resolver
+    // out of the reserved band through MGPipeSlotAllocator::AllocateComposite, which is the one
+    // door into it, and it is not a frontend construction event.
+    void MGPipeMintShaderCsoHandle(MG_State::GLState::ProgramObject& program);
+
+    // ---- textures and renderbuffers: MG_Impl/Pipe/TextureEmit.h, package B ----
+    //
+    // resource_create from ITextureObject's constructor and RenderbufferObject's;
+    // resource_respecify from every storage-defining entry point, including
+    // RenderbufferObject::{SetInternalFormat, AllocateStorage, SetSamples}, which publish
+    // nothing at all today (D-D2); set_texture_params from the parameter mutators, which is
+    // where the READ-attachment-only gap D-E3 closes.
+    //
+    // Entry points MGPipeTextureEmitter must provide, all taking the frontend object by
+    // reference and returning void:
+    //   EmitResourceCreate(ITextureObject&)
+    //   EmitResourceRespecify(ITextureObject&, MGPipeTextureRespecifyScope, Uint32 uploadTarget,
+    //                         Uint32 level)
+    //   EmitTextureParams(ITextureObject&)
+    //   NoteLevelDirty(ITextureObject& storageOwner, Uint32 uploadTarget, Uint32 level)
+    //   EmitRenderbufferCreate(RenderbufferObject&) / EmitRenderbufferRespecify(RenderbufferObject&)
+    void MGPipeEmitTextureResourceCreate(MG_State::GLState::ITextureObject& texture);
+
+    // WHICH STORAGE A TEXTURE RESPECIFY REPLACES (P4a final review C-1). The applier scopes
+    // its pending-upload clear on this answer and not on the descriptor, because the
+    // descriptor cannot give it: AllocateStorage is per (uploadTarget, level) and
+    // TruncateMipmapLevels removes every level at or above a cut, while MGPResourceDesc
+    // carries only the base extent and the level count. A level the applier had ACCEPTED at
+    // one verb (the client's dirty flag already clear, D-D5 step 1) and that a later per-level
+    // definition redefined AROUND was dropped by the whole-resource arm with nobody owing its
+    // texels - so every respecify states its scope, and "whole resource" is said, never
+    // defaulted. The emitter builds wire's MGPRespecifiedLevel from the pair, packed exactly
+    // as the drain packs a sub-data record's Target (MGPipePackSubDataTarget), so the key it
+    // drops is the key that level's emission made.
+    enum class MGPipeTextureRespecifyScope : Uint32 {
+        // The whole store is redefined or restated: a format, sample-count or
+        // fixed-sample-locations change, an immutable allocation completing
+        // (SetImmutableLevels), a texture view's creation. Every pending upload goes.
+        WholeResource = 0,
+        // ONE (uploadTarget, level) was (re)allocated: glTexImage*D, glCompressedTexImage*D,
+        // glCopyTexImage*D, one level of a glTexStorage* loop, one level of a generated-mipmap
+        // grow. That level's pending upload goes; every other level's stays. `uploadTarget` and
+        // `level` name it.
+        OneLevel = 1,
+        // The chain was cut: every level of `uploadTarget` at or above `level` is gone and the
+        // levels below it are untouched (glGenerateMipmap fitting the chain, a base-level
+        // redefinition discarding its tail, glTexStorage* fitting the chain to its level
+        // count). `level` is the first level removed; a cut at 0 is the whole resource.
+        LevelsFrom = 2,
+    };
+    void MGPipeEmitTextureResourceRespecify(MG_State::GLState::ITextureObject& texture,
+                                            MGPipeTextureRespecifyScope scope, Uint32 uploadTarget,
+                                            Uint32 level);
+    void MGPipeEmitTextureParams(MG_State::GLState::ITextureObject& texture);
+    // The DRAIN LIST's append, on a level's FIRST dirty mark, keyed on the STORAGE OWNER from
+    // day one (D-D4: a view and its owner already share one dirty state, so an upload through
+    // either lands on the same key). The record itself is emitted at the validate point by
+    // MGPipeTextureEmitter::DrainTextureSubData; this is only what puts the level on the list,
+    // and walking every live texture per verb is the cost it exists to avoid.
+    void MGPipeNoteTextureLevelDirty(MG_State::GLState::ITextureObject& storageOwner, Uint32 uploadTarget,
+                                     Uint32 level);
+    void MGPipeEmitRenderbufferResourceCreate(MG_State::GLState::RenderbufferObject& renderbuffer);
+    void MGPipeEmitRenderbufferResourceRespecify(MG_State::GLState::RenderbufferObject& renderbuffer);
+
+    // ---- D-A4's two sticky bind-mask producers (P4a final review M-A) ----
+    //
+    // kMGPipeBindSampler is "any texture the sampler-view resolution names in an emitted
+    // MGPBoundView" and kMGPipeBindShaderImage "any texture named in an emitted MGPImageView"
+    // - both the SAMPLER package's emitters (SamplerEmit.h, ImageEmit.h), which the texture
+    // emitter's header includes and which therefore cannot include it back - and, earliest of
+    // all, glBindImageTexture's state setter (TextureState.h, MG_State), which may include no
+    // emit header at all. So the note goes through this door, exactly as the birth hooks do.
+    // Nothing produced either bit before the fix round: ImageBindableHint was always 0, the
+    // metadata respecify (ID-18 M4) had no live trigger, and the remint pull the hint exists to
+    // prevent was neither prevented nor counted.
+    //
+    // UNCONDITIONAL IN A PUSH BUILD, like the mints: the mask is CLIENT state the framebuffer
+    // emitter ORs into whether or not the texture family is on, and the emission a mask move
+    // causes (the metadata respecify) is gated inside the emitter on the family's own pair.
+    void MGPipeNoteTextureBoundAs(MGPipeHandle texture, Uint32 bindBit);
+    // glBindImageTexture. The hint is the PREVENTION half of the texture-remint stall class -
+    // a texture the server knows may be image-bound is allocated image-bindable up front - so it
+    // has to reach the applier before the texture's first sync, i.e. at the bind itself, not at
+    // the validate point's image walk (which notes it as well, D-A4's letter).
+    void MGPipeNoteTextureImageBound(MG_State::GLState::ITextureObject& texture);
+
+
+    // ---- sampler CSOs and sampler views: MG_Impl/Pipe/SamplerEmit.h, package C ----
+    //
+    // Entry points MGPipeSamplerEmitter must provide, returning void:
+    //   EmitSamplerCso(SamplerObject&)     - D-F1's content-addressed mint-or-share at
+    //                                        capacity 256, hashed field-wise over a canonical
+    //                                        zero-initialised copy, behind the version-first
+    //                                        skip. The HANDLE RULE FOR THIS KIND IS THE
+    //                                        EMITTER'S, not this file's: two identical
+    //                                        samplers share one CSO, so there is deliberately
+    //                                        no per-object mint above, and it is the emitter
+    //                                        that decides which lifetime id (if any) owns the
+    //                                        slot the death helper will resolve.
+    //   EmitSamplerView(ITextureObject&)   - D-F2's ONE view per texture object, minted off
+    //                                        the texture's own lifetime id and re-issued on
+    //                                        the SAME handle when the restrictions move.
+    void MGPipeEmitSamplerCsoCreate(MG_State::GLState::SamplerObject& sampler);
+    void MGPipeEmitSamplerViewCreate(MG_State::GLState::ITextureObject& texture);
+
+    // ---- programs: MG_Impl/Pipe/ProgramEmit.h, package C ----
+    //
+    // Entry point MGPipeProgramEmitter must provide, returning void:
+    //   EmitShaderCso(ProgramObject&)
+    //
+    // Re-issued on the SAME handle whenever the link version moves, exactly as
+    // create_vertex_elements is (Gen moves only on slot reuse). D-H4 keeps the TRACKER out of
+    // it - bit 6's shutter reads GetCurrentProgram() and deliberately not GetProgramForDraw(),
+    // because the tracker must not force a compile to answer "did the shader move" - so the
+    // ordinary emission is the validate point's, from the join the verb was going to make
+    // anyway. This hook exists for the paths that are NOT a draw: a link that completes off
+    // the draw path still owns its own publication.
+    void MGPipeEmitShaderCsoCreate(MG_State::GLState::ProgramObject& program);
 } // namespace MobileGL::MG_Pipe
 #define MGP_NOTE_MUTATION(Field)                                                                                       \
     ::MobileGL::MG_Pipe::MGPipeNoteFrontendMutation(::MobileGL::MG_Pipe::MGPipeInputField::Field)
+#define MGP_NOTE_AGGREGATE(Aggregate)                                                                                  \
+    ::MobileGL::MG_Pipe::MGPipeNoteAggregate(::MobileGL::MG_Pipe::MGPipeAggregate::Aggregate)
 #else
 #define MGP_NOTE_MUTATION(Field) ((void)0)
+#define MGP_NOTE_AGGREGATE(Aggregate) ((void)0)
 #endif
 #endif

MobileGL/MG_Pipe/generated/PipeFilled.inc

@@ -299,6 +299,107 @@ inline constexpr const char* kMGPipeInputFieldFilledBy[kMGPipeInputFieldCount] =
     "SetStreamOutputTargets",
 };
 
+// P2 brief D5: the call that now SUPPLIES a field, so the residual fill loop no
+// longer pulls it out of GLContext. kNone means the field is still pulled - which
+// is what makes MOBILEGL_PIPE_PUSH a true per-subsystem A/B instead of a single
+// switch. Rows come from Coverage.def's MGP_COVERAGE_EMITTED_LIST.
+enum class MGPipeFieldEmitter : Uint8 {
+    kNone = 0,
+    BindRenderState,
+    BindVertexElements,
+    CreateRenderState,
+    SetDispatchProgram,
+    SetDrawProgram,
+    SetDynamicState,
+    SetFramebufferState,
+    SetPatchState,
+    SetSamplerViews,
+    SetShaderImages,
+    SetVertexAttribDefaults,
+};
+
+inline constexpr const char* kMGPipeFieldEmitterNames[] = {
+    "kNone",
+    "BindRenderState",
+    "BindVertexElements",
+    "CreateRenderState",
+    "SetDispatchProgram",
+    "SetDrawProgram",
+    "SetDynamicState",
+    "SetFramebufferState",
+    "SetPatchState",
+    "SetSamplerViews",
+    "SetShaderImages",
+    "SetVertexAttribDefaults",
+};
+
+inline constexpr MGPipeFieldEmitter kMGPipeFieldEmittedBy[kMGPipeInputFieldCount] = {
+    MGPipeFieldEmitter::kNone, // GetActiveTextureUnit
+    MGPipeFieldEmitter::SetDynamicState, // GetBlendColor
+    MGPipeFieldEmitter::CreateRenderState, // GetBlendEquationIndexed
+    MGPipeFieldEmitter::CreateRenderState, // GetBlendFuncIndexed
+    MGPipeFieldEmitter::kNone, // GetBoundTransformFeedbackName
+    MGPipeFieldEmitter::BindVertexElements, // GetBoundVertexArray
+    MGPipeFieldEmitter::kNone, // GetBufferBindingSlot
+    MGPipeFieldEmitter::kNone, // GetBufferBindingPoint
+    MGPipeFieldEmitter::kNone, // GetBufferBindingPointCount
+    MGPipeFieldEmitter::kNone, // GetTouchedBufferBindingPointCount
+    MGPipeFieldEmitter::SetDynamicState, // GetClampReadColor
+    MGPipeFieldEmitter::SetDynamicState, // GetClearColor
+    MGPipeFieldEmitter::SetDynamicState, // GetClearDepth
+    MGPipeFieldEmitter::SetDynamicState, // GetClearStencil
+    MGPipeFieldEmitter::CreateRenderState, // GetColorMaskIndexed
+    MGPipeFieldEmitter::CreateRenderState, // GetCullFaceMode
+    MGPipeFieldEmitter::SetVertexAttribDefaults, // GetCurrentVertexAttribute
+    MGPipeFieldEmitter::CreateRenderState, // GetDepthFunc
+    MGPipeFieldEmitter::CreateRenderState, // GetDepthMask
+    MGPipeFieldEmitter::SetDynamicState, // GetDepthRangeIndexed
+    MGPipeFieldEmitter::SetFramebufferState, // GetFramebufferBindingSlot
+    MGPipeFieldEmitter::SetShaderImages, // GetImageTextureBinding
+    MGPipeFieldEmitter::SetDynamicState, // GetLineWidth
+    MGPipeFieldEmitter::CreateRenderState, // GetLogicOp
+    MGPipeFieldEmitter::SetSamplerViews, // GetMaxTouchedTextureUnit
+    MGPipeFieldEmitter::CreateRenderState, // GetMinSampleShadingValue
+    MGPipeFieldEmitter::SetPatchState, // GetPatchDefaultInnerLevel
+    MGPipeFieldEmitter::SetPatchState, // GetPatchDefaultOuterLevel
+    MGPipeFieldEmitter::SetPatchState, // GetPatchVertices
+    MGPipeFieldEmitter::BindRenderState, // GetPipelineStateVersion
+    MGPipeFieldEmitter::kNone, // GetPixelStoreParameters
+    MGPipeFieldEmitter::CreateRenderState, // GetPolygonModeFront
+    MGPipeFieldEmitter::SetDynamicState, // GetPolygonOffsetFactor
+    MGPipeFieldEmitter::SetDynamicState, // GetPolygonOffsetUnits
+    MGPipeFieldEmitter::SetDynamicState, // GetPrimitiveRestartIndex
+    MGPipeFieldEmitter::SetDispatchProgram, // GetProgramForDispatch
+    MGPipeFieldEmitter::SetDrawProgram, // GetProgramForDraw
+    MGPipeFieldEmitter::kNone, // GetProgramObject
+    MGPipeFieldEmitter::CreateRenderState, // GetProvokingVertexMode
+    MGPipeFieldEmitter::CreateRenderState, // GetRenderStateParameters
+    MGPipeFieldEmitter::BindRenderState, // GetRenderStateParametersVersion
+    MGPipeFieldEmitter::kNone, // GetSamplingResolutionGeneration
+    MGPipeFieldEmitter::SetDynamicState, // GetScissorBox
+    MGPipeFieldEmitter::CreateRenderState, // GetStencilState
+    MGPipeFieldEmitter::kNone, // GetTextureBindGeneration
+    MGPipeFieldEmitter::kNone, // GetTextureContextId
+    MGPipeFieldEmitter::kNone, // GetTextureObject
+    MGPipeFieldEmitter::SetSamplerViews, // GetTextureUnitObject
+    MGPipeFieldEmitter::kNone, // GetTransformFeedbackCapturedVertices
+    MGPipeFieldEmitter::kNone, // GetTransformFeedbackGeneration
+    MGPipeFieldEmitter::kNone, // GetTransformFeedbackPausedPrimitiveCounter
+    MGPipeFieldEmitter::kNone, // GetTransformFeedbackProgram
+    MGPipeFieldEmitter::SetDynamicState, // GetViewport
+    MGPipeFieldEmitter::SetDynamicState, // GetViewportIndexed
+    MGPipeFieldEmitter::CreateRenderState, // IsCapabilityEnabled
+    MGPipeFieldEmitter::CreateRenderState, // IsCapabilityEnabledIndexed
+    MGPipeFieldEmitter::kNone, // IsTransformFeedbackActive
+    MGPipeFieldEmitter::kNone, // IsTransformFeedbackPaused
+    MGPipeFieldEmitter::kNone, // InvalidateCompileEnv
+    MGPipeFieldEmitter::kNone, // ValidateProgramName
+    MGPipeFieldEmitter::kNone, // RecordError
+    MGPipeFieldEmitter::kNone, // GetBoundTransformFeedbackLifetimeId
+    MGPipeFieldEmitter::kNone, // HasOpenTransformFeedbackSpan
+};
+inline constexpr SizeT kMGPipeEmittedFieldCount = 40;
+
 struct MGPipeFilledState {
     Uint64 CurrentVerbSerial;
     Uint64 FilledGen[kMGPipeInputFieldCount];

MobileGL/MG_Pipe/generated/PipeSpanTable.inc

@@ -14,54 +14,80 @@
 
 // D-B1 rejected three CSOs and demanded this table instead, so the table needs its own
 // completeness trip wire: MG_Test walks every public RenderState setter and asserts that
-// the pipeline-subset hash moves IF AND ONLY IF m_pipelineStateVersion moves. That test
-// and MGPipeRenderStateSpans.cpp land with P2; what P0 pins is the MEMBER LIST, taken from
-// what VulkanRenderer::ComputePipelineStateHash hashes today, so the later offsets are
-// derived from a list that was reviewed rather than invented.
+// the pipeline-subset hash moves IF AND ONLY IF m_pipelineStateVersion moves
+// (MG_Test/Pipe/RenderStateSpansTest.cpp).
 //
-// Deliberately absent, and each absence is a question P2 has to answer before the chunk
-// table freezes:
-//   - FramebufferSrgb and DepthClamp have NO STORAGE at all (RenderState.cpp's SetCapability
-//     falls to "not supported currently" and IsCapabilityEnabled returns false), so six
-//     backend read points are constant false today. Pipeline state or dead capability?
-//   - ProvokingVertexModeSetting is Vulkan pipeline state but is not hashed today.
-//   - FrontFaceModeSetting, ClipOrigin and ClipDepthMode are pipeline state on Vulkan and
-//     are handled elsewhere in the payload path rather than in the memo word.
+// P2 replaced P0's provenance with a RULE, and the rule is the only thing that decides
+// membership: a member is pipeline state IF AND ONLY IF some public RenderState setter that
+// calls BumpVersions() writes it. That is what makes the G7 invariant true by construction
+// rather than by inspection, and it turns the subset into a strict SUPERSET of the 24
+// members VulkanRenderer::ComputePipelineStateHash used to hash.
+//
+// The three questions P0 left open are ANSWERED here, and the answers are in this list:
+//   - FramebufferSrgb, DepthClamp and TextureCubeMapSeamless had NO STORAGE at all -
+//     SetCapability fell to "not supported currently" and IsCapabilityEnabled answered a
+//     compile-time false. P2 gave all three real storage in the three padding bytes between
+//     ColorMasks and ClearColor, and their setters call BumpVersions(), so: pipeline state.
+//   - ProvokingVertexModeSetting: SetProvokingVertexMode calls BumpVersions(), so pipeline.
+//   - FrontFaceModeSetting likewise. ClipOrigin and ClipDepthMode do NOT (SetClipControl is
+//     ++m_version only), so they are dynamic, in chunk D1.
 //
 // The complement of this list is the DYNAMIC subset - the half whose whole purpose is that
 // glViewport must not mint a new CSO.
 
 inline constexpr const char* const kMGPipePipelineStateMembers[] = {
-    "CullFaceEnabled",
-    "DepthTestEnabled",
-    "PolygonOffsetFillEnabled",
-    "RasterizerDiscardEnabled",
-    "ColorLogicOpEnabled",
-    "StencilTestEnabled",
-    "PrimitiveRestartEnabled",
-    "PrimitiveRestartFixedIndexEnabled",
-    "DepthMask",
-    "SampleShadingEnabled",
-    "MultisampleEnabled",
-    "SampleMaskEnabled",
-    "SampleMaskValue",
-    "MinSampleShadingValue",
     "PatchVertices",
     "PatchDefaultOuterLevel",
     "PatchDefaultInnerLevel",
-    "PolygonModeFront",
-    "CullFaceModeSetting",
-    "DepthFunc",
-    "LogicOp",
-    "StencilStates",
     "BlendStates",
+    "LogicOp",
+    "DepthTestEnabled",
+    "DepthFunc",
+    "DepthMask",
     "ColorMasks",
+    "FramebufferSrgbEnabled",
+    "DepthClampEnabled",
+    "TextureCubeMapSeamlessEnabled",
+    "SampleCoverageValue",
+    "SampleCoverageInvert",
+    "SampleMaskValue",
+    "MinSampleShadingValue",
+    "StencilStates",
+    "CullFaceEnabled",
+    "CullFaceModeSetting",
+    "FrontFaceModeSetting",
+    "ProvokingVertexModeSetting",
+    "PolygonModeFront",
+    "PolygonModeBack",
+    "ColorLogicOpEnabled",
+    "DebugOutputEnabled",
+    "DebugOutputSynchronousEnabled",
+    "DitherEnabled",
+    "LineSmoothEnabled",
+    "MultisampleEnabled",
+    "PolygonOffsetFillEnabled",
+    "PolygonOffsetLineEnabled",
+    "PolygonOffsetPointEnabled",
+    "PolygonSmoothEnabled",
+    "PrimitiveRestartEnabled",
+    "PrimitiveRestartFixedIndexEnabled",
+    "RasterizerDiscardEnabled",
+    "SampleAlphaToCoverageEnabled",
+    "SampleAlphaToOneEnabled",
+    "SampleCoverageEnabled",
+    "SampleMaskEnabled",
+    "SampleShadingEnabled",
+    "StencilTestEnabled",
+    "ProgramPointSizeEnabled",
+    "ScissorTestEnabledMask",
 };
-inline constexpr SizeT kMGPipePipelineStateMemberCount = 24;
+inline constexpr SizeT kMGPipePipelineStateMemberCount = 44;
 static_assert(kMGPipePipelineStateMemberCount ==
                   sizeof(kMGPipePipelineStateMembers) / sizeof(kMGPipePipelineStateMembers[0]));
 
-// Filled in by MG_Pipe/MGPipeRenderStateSpans.cpp (P2), which computes the offsets
-// in C++ with offsetof rather than guessing them in python.
+// Defined by MG_Pipe/MGPipeRenderStateSpans.cpp (P2), which computes every
+// boundary in C++ with offsetof rather than guessing it in python. 7 pipeline
+// chunks / 396 bytes and 8 dynamic chunks / 772 bytes, and the two halves
+// partition [0, sizeof(RenderStateParameters)) exactly - asserted there.
 extern const MGPStateChunk kMGPipePipelineChunks[];
 extern const MGPStateChunk kMGPipeDynamicChunks[];

MobileGL/MG_Pipe/generated/PipeVerify.inc

@@ -109,10 +109,13 @@ inline Bool MGPipeVerify(const MGPSwapInterval& a, const MGPSwapInterval& b, con
 inline Bool MGPipeVerify(const MGPSurfaceInfo& a, const MGPSurfaceInfo& b, const char** outField);
 inline Bool MGPipeVerify(const RenderStateParameters& a, const RenderStateParameters& b, const char** outField);
 inline Bool MGPipeVerify(const PixelStoreParameters& a, const PixelStoreParameters& b, const char** outField);
+inline Bool MGPipeVerify(const SamplerParameters& a, const SamplerParameters& b, const char** outField);
 inline Bool MGPipeVerify(const PerBufferBlendState& a, const PerBufferBlendState& b, const char** outField);
 inline Bool MGPipeVerify(const StencilFaceState& a, const StencilFaceState& b, const char** outField);
 inline Bool MGPipeVerify(const DynamicBackendParameters& a, const DynamicBackendParameters& b, const char** outField);
 inline Bool MGPipeVerify(const MGHostSpan& a, const MGHostSpan& b, const char** outField);
+inline Bool MGPipeVerify(const MGPVertexAttribWire& a, const MGPVertexAttribWire& b, const char** outField);
+inline Bool MGPipeVerify(const MGPVertexBindingPointWire& a, const MGPVertexBindingPointWire& b, const char** outField);
 
 template <>
 struct MGPipeHasFieldVerifier<MGPBlobRef> : std::true_type {};
@@ -245,6 +248,8 @@ struct MGPipeHasFieldVerifier<RenderStateParameters> : std::true_type {};
 template <>
 struct MGPipeHasFieldVerifier<PixelStoreParameters> : std::true_type {};
 template <>
+struct MGPipeHasFieldVerifier<SamplerParameters> : std::true_type {};
+template <>
 struct MGPipeHasFieldVerifier<PerBufferBlendState> : std::true_type {};
 template <>
 struct MGPipeHasFieldVerifier<StencilFaceState> : std::true_type {};
@@ -252,6 +257,10 @@ template <>
 struct MGPipeHasFieldVerifier<DynamicBackendParameters> : std::true_type {};
 template <>
 struct MGPipeHasFieldVerifier<MGHostSpan> : std::true_type {};
+template <>
+struct MGPipeHasFieldVerifier<MGPVertexAttribWire> : std::true_type {};
+template <>
+struct MGPipeHasFieldVerifier<MGPVertexBindingPointWire> : std::true_type {};
 
 template <class T>
 inline Bool MGPipeFieldEqual(const T& a, const T& b) {
@@ -623,6 +632,11 @@ inline Bool MGPipeVerify(const PixelStoreParameters& a, const PixelStoreParamete
     return true;
 }
 
+inline Bool MGPipeVerify(const SamplerParameters& a, const SamplerParameters& b, const char** outField) {
+    MGP_FIELDS_SamplerParameters(MGP_VERIFY_FIELD)
+    return true;
+}
+
 inline Bool MGPipeVerify(const PerBufferBlendState& a, const PerBufferBlendState& b, const char** outField) {
     MGP_FIELDS_PerBufferBlendState(MGP_VERIFY_FIELD)
     return true;
@@ -643,6 +657,16 @@ inline Bool MGPipeVerify(const MGHostSpan& a, const MGHostSpan& b, const char**
     return true;
 }
 
+inline Bool MGPipeVerify(const MGPVertexAttribWire& a, const MGPVertexAttribWire& b, const char** outField) {
+    MGP_FIELDS_MGPVertexAttribWire(MGP_VERIFY_FIELD)
+    return true;
+}
+
+inline Bool MGPipeVerify(const MGPVertexBindingPointWire& a, const MGPVertexBindingPointWire& b, const char** outField) {
+    MGP_FIELDS_MGPVertexBindingPointWire(MGP_VERIFY_FIELD)
+    return true;
+}
+
 #undef MGP_VERIFY_FIELD
 
-inline constexpr SizeT kMGPipeVerifiedPayloadCount = 69;
+inline constexpr SizeT kMGPipeVerifiedPayloadCount = 72;

MobileGL/MG_Pipe/generated/PipeWire.inc

@@ -929,3 +929,11 @@ inline Bool MGPipeApplyWireRecord(MGPWireOp op, const void* record, Uint64 size,
 }
 
 #undef MGP_WIRE_CHECK_BOUNDS
+
+// The ResidualValueBlock layout, from PipeFields.def's
+// MGP_FIELDS_ResidualValueBlock. Retiring a field without lowering
+// MGL_RESIDUAL_BLOCK_SIZE is a build break, which is the point.
+static_assert(offsetof(ResidualValueBlock, CapabilityBits) == 0,
+              "the residual block's first member must sit at offset 0");
+static_assert(sizeof(ResidualValueBlock) == MGL_RESIDUAL_BLOCK_SIZE,
+              "the residual ratchet only ever goes down");

MobileGL/MG_State/GLState/BufferState/BufferObject.cpp

@@ -11,6 +11,7 @@
 #include <Config.h>
 
 #include <atomic>
+#include <MG_Pipe/PipeMutation.h>
 
 namespace MobileGL::MG_State::GLState {
     namespace {
@@ -33,9 +34,31 @@ namespace MobileGL::MG_State::GLState {
 
     BufferObject::BufferObject(Uint externalIndex)
         : m_externalIndex(externalIndex), m_size(0), m_usage(BufferUsage::StaticDraw), m_isMapped(false),
-          m_mappingAccess(BufferMappingAccessBit::Null), m_mappedRange({0, 0}), m_ownsStagingData{} {}
+          m_mappingAccess(BufferMappingAccessBit::Null), m_mappedRange({0, 0}), m_ownsStagingData{} {
+#if MOBILEGL_PIPE_PUSH
+        // P3a D-A2: a resource EXISTS before anything can name it, so resource_create is
+        // emitted from the constructor and carries no storage - the store is defined lazily
+        // by the first respecify and every backend already tolerates a resource with none.
+        // The handle itself is minted whatever the subsystem bitmask says, because
+        // set_vertex_buffers names this buffer by handle out of a different subsystem.
+        MG_Pipe::MGPipeMintResourceHandle(*this);
+        if (MG_Pipe::MGPipeResourceSubsystemEnabled()) MG_Pipe::MGPipeEmitResourceCreate(*this);
+#endif
+    }
 
     BufferObject::~BufferObject() {
+#if MOBILEGL_PIPE_PUSH
+        // P3a D-L: the buffer's death crosses as resource_destroy, which is the catalogue
+        // call for it - no seventh NotifyStateObjectDestroyed raiser is added, because that
+        // header exists for kinds that have no such call. The emit-then-free ORDER is fixed
+        // inside the helper and is not negotiable.
+        // The answer is the helper's LATCH - "was resource_create emitted for this buffer" -
+        // not a second reading of MGPipeResourceSubsystemEnabled(): a buffer constructed
+        // while a backend's table was registered and destroyed after it was unregistered has
+        // a pipe record to drop and no legacy backend object, and one constructed the other
+        // way round has the opposite, so the create's answer is the only one that pairs.
+        if (MG_Pipe::MGPipeEmitResourceDestroyAndFree(*this)) return;
+#endif
         if (m_resource.Backend() && g_bufferBackendOps && g_bufferBackendOps->OnDestroy) {
             g_bufferBackendOps->OnDestroy(m_resource.ReleaseBackend());
         }
@@ -43,6 +66,13 @@ namespace MobileGL::MG_State::GLState {
 
     void BufferObject::NotifyRespecify() {
         ++m_changeSerial;
+        MGP_NOTE_AGGREGATE(BufferChange);
+#if MOBILEGL_PIPE_PUSH
+        if (MG_Pipe::MGPipeResourceSubsystemEnabled()) {
+            MG_Pipe::MGPipeEmitResourceRespecify(*this);
+            return;
+        }
+#endif
         if (g_bufferBackendOps && g_bufferBackendOps->Respecify) {
             g_bufferBackendOps->Respecify(*this);
         }
@@ -50,8 +80,15 @@ namespace MobileGL::MG_State::GLState {
 
     void BufferObject::NotifySubData(SizeT offset, SizeT size) {
         ++m_changeSerial;
+        MGP_NOTE_AGGREGATE(BufferChange);
         if (size == 0) return;
         m_hasDefinedContent = true;
+#if MOBILEGL_PIPE_PUSH
+        if (MG_Pipe::MGPipeResourceSubsystemEnabled()) {
+            MG_Pipe::MGPipeEmitResourceSubData(*this, offset, size);
+            return;
+        }
+#endif
         if (g_bufferBackendOps && g_bufferBackendOps->SubData) {
             g_bufferBackendOps->SubData(*this, offset, size);
         }
@@ -59,8 +96,19 @@ namespace MobileGL::MG_State::GLState {
 
     void BufferObject::NotifyFlushMappedRange(Range1D range, Flags<BufferMappingAccessBit> appAccess) {
         ++m_changeSerial;
+        MGP_NOTE_AGGREGATE(BufferChange);
         if (range.start >= range.end) return;
         m_hasDefinedContent = true;
+#if MOBILEGL_PIPE_PUSH
+        if (MG_Pipe::MGPipeResourceSubsystemEnabled()) {
+            static_assert(sizeof(appAccess.GetRaw()) <= sizeof(Uint32),
+                          "MGPFlushRange::AccessFlags is a Uint32 and carries the application's "
+                          "real Flags<BufferMappingAccessBit>, unnormalised");
+            MG_Pipe::MGPipeEmitResourceFlushRange(*this, range.start, range.end - range.start,
+                                                  static_cast<Uint32>(appAccess.GetRaw()));
+            return;
+        }
+#endif
         if (g_bufferBackendOps && g_bufferBackendOps->FlushMappedRange) {
             g_bufferBackendOps->FlushMappedRange(*this, range, appAccess);
         }
@@ -73,6 +121,7 @@ namespace MobileGL::MG_State::GLState {
             // undefined store to "has content" - that would cost the next orphaning
             // respecification a full-size upload of bytes the application never wrote.
             ++m_changeSerial;
+            MGP_NOTE_AGGREGATE(BufferChange);
             return;
         }
         m_hasDefinedContent = true;
@@ -80,6 +129,7 @@ namespace MobileGL::MG_State::GLState {
             // The write already landed in coherent GPU memory; the backend has no separate
             // copy to sync. Only bump the serial so cached transient slices invalidate.
             ++m_changeSerial;
+            MGP_NOTE_AGGREGATE(BufferChange);
             return;
         }
         NotifySubData(offset, size);
@@ -183,6 +233,12 @@ namespace MobileGL::MG_State::GLState {
         if (m_size < kLargeBufferAdoptBytes) return;
         if (m_resource.IsGpuResident()) return;
         if (m_isMapped) return;
+#if MOBILEGL_PIPE_PUSH
+        if (MG_Pipe::MGPipeResourceSubsystemEnabled()) {
+            if (void* base = MG_Pipe::MGPipeEmitMapPersistent(*this)) m_resource.AdoptPersistentMap(base);
+            return;
+        }
+#endif
         if (g_bufferBackendOps == nullptr || g_bufferBackendOps->AcquirePersistentMap == nullptr) return;
         if (void* base = g_bufferBackendOps->AcquirePersistentMap(*this)) {
             m_resource.AdoptPersistentMap(base);
@@ -302,6 +358,7 @@ namespace MobileGL::MG_State::GLState {
                         data.size, m_size);
         Memcpy(m_resource.Bytes() + atOffset, data.data, data.size);
         ++m_changeSerial;
+        MGP_NOTE_AGGREGATE(BufferChange);
     }
 
     void BufferObject::MarkGpuWritten() {
@@ -314,6 +371,17 @@ namespace MobileGL::MG_State::GLState {
         // Cleared unconditionally: without a readback op the shadow can never catch up,
         // and retrying on every subsequent read would only repeat the same no-op.
         m_gpuWritePending = false;
+#if MOBILEGL_PIPE_PUSH
+        if (m_size != 0 && MG_Pipe::MGPipeResourceSubsystemEnabled()) {
+            // The answer comes back through the reverse channel's OnBufferWriteback, which
+            // resolves this handle to this object and writes the shadow before the server
+            // bumps its mutation epoch. In monolith the whole sequence is synchronous inside
+            // the applier, so the caller sees the reconciled shadow on return exactly as it
+            // does today.
+            MG_Pipe::MGPipeEmitResourceReadback(*this);
+            return;
+        }
+#endif
         if (m_size == 0 || g_bufferBackendOps == nullptr || g_bufferBackendOps->ReadbackFromGpu == nullptr) {
             return;
         }
@@ -362,10 +430,35 @@ namespace MobileGL::MG_State::GLState {
     // NotifyContentWrite on a resident store only bumps the serial: the backend has no
     // separate copy to sync, so no transfer op runs.
     void BufferObject::LandBytesIntoResidentStore(SizeT offset, DataPtr bytes) {
+#if MOBILEGL_PIPE_PUSH
+        // buffer_subdata_resident stays NULLABLE and stays asymmetric: one backend
+        // deliberately does not implement it, and the frontend checks the pipe table exactly
+        // as it checks the op table it replaces, so a backend without it keeps the legacy
+        // ordered in-place host write below.
+        if (bytes.size > 0 && MG_Pipe::MGPipeResourceSubsystemEnabled() &&
+            MG_Pipe::MGPipeResourceOpsHaveSubDataResident()) {
+            MG_Pipe::MGPipeEmitBufferSubDataResident(*this, offset, bytes.data, bytes.size);
+            m_hasDefinedContent = true;
+            ++m_changeSerial;
+            MGP_NOTE_AGGREGATE(BufferChange);
+            m_gpuWritePending = true;
+            return;
+        }
+        // No resident op: the write lands in place below, after retiring the GPU writes this
+        // store is known to be waiting on - which is the same answer, and the same code, a
+        // backend with a null ResidentSubData gets today.
+        if (MG_Pipe::MGPipeResourceSubsystemEnabled()) {
+            SyncGpuWrites();
+            Memcpy(m_resource.Bytes() + offset, bytes.data, bytes.size);
+            NotifyContentWrite(offset, bytes.size);
+            return;
+        }
+#endif
         if (bytes.size > 0 && g_bufferBackendOps && g_bufferBackendOps->ResidentSubData) {
             g_bufferBackendOps->ResidentSubData(*this, offset, bytes);
             m_hasDefinedContent = true;
             ++m_changeSerial;
+            MGP_NOTE_AGGREGATE(BufferChange);
             m_gpuWritePending = true;
             return;
         }
@@ -394,7 +487,16 @@ namespace MobileGL::MG_State::GLState {
         // op the landing would memcpy the expansion into the mapping the loop below
         // fills in place anyway, so a whole-arena clear would allocate a whole arena
         // for nothing.
-        if (m_resource.IsGpuResident() && g_bufferBackendOps && g_bufferBackendOps->ResidentSubData) {
+        if (m_resource.IsGpuResident() &&
+#if MOBILEGL_PIPE_PUSH
+            // The same question, asked of whichever table owns the family in this build.
+            (MG_Pipe::MGPipeResourceSubsystemEnabled()
+                 ? MG_Pipe::MGPipeResourceOpsHaveSubDataResident()
+                 : (g_bufferBackendOps && g_bufferBackendOps->ResidentSubData))
+#else
+            g_bufferBackendOps && g_bufferBackendOps->ResidentSubData
+#endif
+        ) {
             Vector<Uint8> expanded(size);
             if (pattern.size == 1) {
                 Memset(expanded.data(), *static_cast<const Uint8*>(pattern.data), size);
@@ -496,6 +598,14 @@ namespace MobileGL::MG_State::GLState {
         if (m_isMapped) {
             return false;
         }
+#if MOBILEGL_PIPE_PUSH
+        if (m_size != 0 && MG_Pipe::MGPipeResourceSubsystemEnabled()) {
+            void* pushedBase = MG_Pipe::MGPipeEmitMapPersistent(*this);
+            if (pushedBase == nullptr) return false;
+            m_resource.AdoptPersistentMap(pushedBase);
+            return true;
+        }
+#endif
         if (m_size == 0 || g_bufferBackendOps == nullptr || g_bufferBackendOps->AcquirePersistentMap == nullptr) {
             return false;
         }
@@ -540,6 +650,16 @@ namespace MobileGL::MG_State::GLState {
             // returning; AdoptPersistentMap then releases the shadow. Falls back to the
             // shadow when the backend declines (returns null). Only attempted once - the
             // storage is immutable and outlives unmap/remap.
+#if MOBILEGL_PIPE_PUSH
+            if (!m_resource.IsGpuResident() && (access & BufferMappingAccessBit::Write) &&
+                !(access & BufferMappingAccessBit::FlushExplicit) &&
+                MG_Pipe::MGPipeResourceSubsystemEnabled()) {
+                if (void* pushedBase = MG_Pipe::MGPipeEmitMapPersistent(*this)) {
+                    m_resource.AdoptPersistentMap(pushedBase);
+                }
+                return m_resource.Bytes() + range.start;
+            }
+#endif
             if (!m_resource.IsGpuResident() && (access & BufferMappingAccessBit::Write) &&
                 !(access & BufferMappingAccessBit::FlushExplicit) && g_bufferBackendOps &&
                 g_bufferBackendOps->AcquirePersistentMap) {

MobileGL/MG_State/GLState/BufferState/BufferState.h

@@ -64,7 +64,19 @@ namespace MobileGL::MG_State::GLState {
         Bool ValidateName(Uint index) const;
         Bool ValidateBufferObject(Uint index) const;
 
+#if MOBILEGL_PIPE_PUSH
+    // P2 brief D4: "did the contents of ANY buffer object move". One counter for every
+    // BufferObject ++m_changeSerial site, which is what NEW_VERTEX_BUFFERS /
+    // NEW_INDEX_BUFFER / NEW_CONST_BUFFERS / NEW_SHADER_BUFFERS / NEW_SO_TARGETS all
+    // shutter on in P2 - five bits over one aggregate until P3b splits them.
+    void NoteBufferChanged() { ++m_anyBufferChangeGeneration; }
+    Uint64 GetAnyBufferChangeGeneration() const { return m_anyBufferChangeGeneration; }
+#endif
+
     private:
+#if MOBILEGL_PIPE_PUSH
+    Uint64 m_anyBufferChangeGeneration = 0;
+#endif
         UnorderedMap<Uint, SharedPtr<BufferObject>> m_bufferObjects;
         IndexGenerator<Uint> m_indexGenerator;
         Array<BindingSlot<BufferObject>, GlobalBufferTargets.size()> m_bindingSlots;

MobileGL/MG_State/GLState/Core.cpp

@@ -15,6 +15,7 @@
 #include <Config.h>
 
 #include <atomic>
+#include <MG_Pipe/PipeMutation.h>
 
 namespace MobileGL::MG_State {
     void Init() {
@@ -213,6 +214,12 @@ namespace MobileGL::MG_State {
                 current.intValue[component] = static_cast<Int32>(value[component]);
                 current.uintValue[component] = static_cast<Uint32>(value[component]);
             }
+#if MOBILEGL_PIPE_PUSH
+            // The two views above are CONVERSIONS, not bit copies, so which one was written
+            // is part of the value; set_vertex_attrib_defaults carries it.
+            m_currentVertexAttributeClasses[index] = kVertexAttribValueClassFloat;
+#endif
+        MGP_NOTE_AGGREGATE(VertexAttribDefault);
         }
 
         void GLContext::SetCurrentVertexAttributeInt(Uint index, const Array<Int32, 4>& value) {
@@ -227,6 +234,10 @@ namespace MobileGL::MG_State {
                 current.floatValue[component] = static_cast<Float>(value[component]);
                 current.uintValue[component] = static_cast<Uint32>(value[component]);
             }
+#if MOBILEGL_PIPE_PUSH
+            m_currentVertexAttributeClasses[index] = kVertexAttribValueClassInt;
+#endif
+        MGP_NOTE_AGGREGATE(VertexAttribDefault);
         }
 
         void GLContext::SetCurrentVertexAttributeUint(Uint index, const Array<Uint32, 4>& value) {
@@ -241,6 +252,10 @@ namespace MobileGL::MG_State {
                 current.floatValue[component] = static_cast<Float>(value[component]);
                 current.intValue[component] = static_cast<Int32>(value[component]);
             }
+#if MOBILEGL_PIPE_PUSH
+            m_currentVertexAttributeClasses[index] = kVertexAttribValueClassUint;
+#endif
+        MGP_NOTE_AGGREGATE(VertexAttribDefault);
         }
 
         const CurrentVertexAttributeValue& GLContext::GetCurrentVertexAttribute(Uint index) const {

MobileGL/MG_State/GLState/Core.h

@@ -30,10 +30,26 @@ namespace MobileGL {
         void Init();
 
         namespace GLState {
+#if MOBILEGL_PIPE_PUSH
+            // MGPAttribValue::ValueClass' encoding (MG_Pipe/MGPipeTypes.h documents the order
+            // "Float | Int | Uint | Double"). It lives here rather than in MG_Pipe because the
+            // FRONTEND is the only thing that knows which of the three views below a value was
+            // written through - the other two are numeric conversions of it - and MG_Pipe has
+            // no enum for the field yet. If package A introduces one, this becomes its alias.
+            inline constexpr Uint32 kVertexAttribValueClassFloat = 0;
+            inline constexpr Uint32 kVertexAttribValueClassInt = 1;
+            inline constexpr Uint32 kVertexAttribValueClassUint = 2;
+#endif
+
             struct CurrentVertexAttributeValue {
                 Array<Float, 4> floatValue{0.f, 0.f, 0.f, 1.f};
                 Array<Int32, 4> intValue{0, 0, 0, 1};
                 Array<Uint32, 4> uintValue{0u, 0u, 0u, 1u};
+                // Three scalar arrays and NOTHING ELSE. MG_Backend/MGPipe/PipeInputs.cpp
+                // compares this storage with one memcmp and asserts that size, so a fourth
+                // member here is a build break in a file P2 package B does not own. The
+                // written-class discriminator set_vertex_attrib_defaults needs therefore
+                // lives beside the array on GLContext, not inside the value.
             };
 
             // Which of the three views above a shader input of a given GLSL type consumes.
@@ -198,6 +214,64 @@ namespace MobileGL {
                 Uint GetBoundProgramPipelineName() const { return m_boundProgramPipeline; }
                 const SharedPtr<ProgramPipelineObject>& GetBoundProgramPipeline() const;
 
+#if MOBILEGL_PIPE_PUSH
+                // ---- the aggregate generations (P2 brief D4) ----
+                //
+                // The bump points sit on OBJECTS - a VertexArrayObject, a TextureObject, a
+                // BufferObject - which have no back-pointer to the state container that owns
+                // them, so MGP_NOTE_AGGREGATE goes through MGPipeNoteAggregate, which finds
+                // the live context and lands here. This facade is the whole reason the
+                // objects need no back-pointer, and it is push-only so the pull build's
+                // GLContext is byte-identical (G1).
+                void NoteVaoAttributeChanged() { m_vertexArrayState.NoteAttributeChanged(); }
+                Uint64 GetAnyVaoAttributeGeneration() const {
+                    return m_vertexArrayState.GetAnyAttributeGeneration();
+                }
+                void NoteFramebufferAttachmentChanged() { m_framebufferState.NoteAttachmentChanged(); }
+                Uint64 GetAnyFramebufferAttachmentGeneration() const {
+                    return m_framebufferState.GetAnyAttachmentGeneration();
+                }
+                void NoteTextureContentChanged() { m_textureState.NoteTextureContentChanged(); }
+                Uint64 GetAnyTextureContentGeneration() const {
+                    return m_textureState.GetAnyTextureContentGeneration();
+                }
+                void NoteTextureParamsChanged() { m_textureState.NoteTextureParamsChanged(); }
+                Uint64 GetAnyTextureParamsGeneration() const {
+                    return m_textureState.GetAnyTextureParamsGeneration();
+                }
+                void NoteBufferChanged() { m_bufferState.NoteBufferChanged(); }
+                Uint64 GetAnyBufferChangeGeneration() const {
+                    return m_bufferState.GetAnyBufferChangeGeneration();
+                }
+                // The sixth aggregate lives here rather than on a state container because
+                // the values it guards do too (m_currentVertexAttributes).
+                void NoteVertexAttribDefaultChanged() { ++m_anyVertexAttribDefaultGeneration; }
+                Uint64 GetAnyVertexAttribDefaultGeneration() const {
+                    return m_anyVertexAttribDefaultGeneration;
+                }
+
+                // Which of the three views of m_currentVertexAttributes[index] the last
+                // glVertexAttrib* write filled DIRECTLY. The other two are NUMERIC
+                // conversions of it (SetCurrentVertexAttribute* below), not bit copies, so
+                // four words on a wire are not the value unless the class travels with them:
+                // glVertexAttrib4f(loc, 1.5f, ...) leaves 1 in intValue and 0x3FC00000 in
+                // floatValue. set_vertex_attrib_defaults carries this as MGPAttribValue's
+                // ValueClass so the applier can redo the conversion instead of memcpying one
+                // view into all three.
+                //
+                // It is kept BESIDE the array rather than inside CurrentVertexAttributeValue
+                // because that struct is mirrored into PipeInputs and compared there by a
+                // memcmp whose size assertion (MG_Backend/MGPipe/PipeInputs.cpp) is a file
+                // this package does not own - and because it need not be mirrored: the class
+                // only decides how to REBUILD the three views, so two writes that leave the
+                // three views identical rebuild identically whichever class they carried.
+                Uint32 GetCurrentVertexAttributeClass(Uint index) const {
+                    return index < m_currentVertexAttributeClasses.size()
+                               ? m_currentVertexAttributeClasses[index]
+                               : kVertexAttribValueClassFloat;
+                }
+#endif
+
                 // RenderState
                 Uint GetRenderStateParametersVersion() const;
                 // Only the pipeline-relevant subset - see RenderState::m_pipelineStateVersion.
@@ -515,6 +589,11 @@ namespace MobileGL {
                 Bool m_transformFeedbackPaused = false;
                 GLenum m_transformFeedbackPrimitiveMode = GL_POINTS;
                 SharedPtr<ProgramObject> m_transformFeedbackProgram;
+#if MOBILEGL_PIPE_PUSH
+                Uint64 m_anyVertexAttribDefaultGeneration = 0;
+                // Parallel to m_currentVertexAttributes; see GetCurrentVertexAttributeClass.
+                Array<Uint32, VertexArrayObject::MAX_VERTEX_ATTRIBS> m_currentVertexAttributeClasses{};
+#endif
                 Uint64 m_transformFeedbackGeneration = 0;
                 // Source of the per-span ids above; never rolls back with an object switch.
                 Uint64 m_transformFeedbackNextGeneration = 0;

MobileGL/MG_State/GLState/FramebufferState/FramebufferObject.cpp

@@ -7,9 +7,11 @@
 // End of Source File Header
 
 #include "FramebufferObject.h"
+#include "MG_State/GLState/StateObjectDeathNotice.h"
 #include "MG_Util/Types.h"
 
 #include <atomic>
+#include <MG_Pipe/PipeMutation.h>
 
 namespace MobileGL::MG_State::GLState {
     // Starts at 1 so a zero-initialized memo slot can never carry a live object's id.
@@ -21,6 +23,27 @@ namespace MobileGL::MG_State::GLState {
         return s_nextFramebufferLifetimeId.fetch_add(1, std::memory_order_relaxed);
     }
 
+#if MOBILEGL_PIPE_PUSH
+    FramebufferObject::~FramebufferObject() {
+        // P4a D-I2: A FRAMEBUFFER HAS A HANDLE AND NO WIRE LIFETIME. PipeCalls.def carries
+        // resource_destroy and five delete_* rows and NO framebuffer delete, because a
+        // framebuffer is not a resource and is not a CSO - it is STATE, and
+        // set_framebuffer_state is the only call that names one - and the catalogue is closed,
+        // so P4a invents no row. The helper is therefore steps 2 and 3 only: the death notice,
+        // raised while the handle still resolves (this is the P2 step-e2 announcement that used
+        // to stand here alone - the last SharedPtr to this object dropping, not the glDelete*
+        // that only marks the name and leaves a still-bound object very much alive), and then
+        // the slot.
+        //
+        // What makes a dangling Fbo unreachable is the frontend's own
+        // MarkFramebufferObjectForDeletion path, which already rebinds any slot holding the
+        // victim to framebuffer 0; and a RECYCLED framebuffer handle can never be suppressed
+        // against its predecessor's record, because Fbo carries Gen and Gen is inside the
+        // record's ContentHash.
+        MG_Pipe::MGPipeEmitFramebufferDestroyAndFree(m_lifetimeId);
+    }
+#endif
+
     // FramebufferAttachmentObject
     FramebufferAttachmentObject::FramebufferAttachmentObject(
         const SharedPtr<MG_State::GLState::ITextureObject>& texture, TextureUploadTarget textureUploadTarget, Int level,
@@ -190,6 +213,7 @@ namespace MobileGL::MG_State::GLState {
         if (m_readBuffer == buf) return;
         m_readBuffer = buf;
         ++m_objectVersion;
+        MGP_NOTE_AGGREGATE(FramebufferAttachment);
     }
 
     Uint FramebufferObject::GetExternalIndex() const {
@@ -201,6 +225,7 @@ namespace MobileGL::MG_State::GLState {
         if (member == value) return;                                                                                    \
         member = value;                                                                                                 \
         ++m_objectVersion;                                                                                              \
+        MGP_NOTE_AGGREGATE(FramebufferAttachment); \
     }
 
     MOBILEGL_DEFINE_FRAMEBUFFER_DEFAULT_SETTER(DefaultWidth, m_defaultWidth, Int)
@@ -213,5 +238,6 @@ namespace MobileGL::MG_State::GLState {
     void FramebufferObject::BumpAttachmentVersion(FramebufferAttachmentType type) {
         ++m_attachmentVersions[static_cast<SizeT>(type)];
         ++m_objectVersion;
+        MGP_NOTE_AGGREGATE(FramebufferAttachment);
     }
 } // namespace MobileGL::MG_State::GLState

MobileGL/MG_State/GLState/FramebufferState/FramebufferObject.h

@@ -115,6 +115,12 @@ namespace MobileGL {
                 Array<Uint16, static_cast<SizeT>(FramebufferAttachmentType::FramebufferAttachmentTypeCount)>;
 
             FramebufferObject(Uint externalIndex);
+#if MOBILEGL_PIPE_PUSH
+            // P2 step e2. Out of line, and declared only where there is a notice to raise:
+            // in a pull build this class keeps its implicit destructor, which is what keeps
+            // the pull build's symbol set byte-for-byte the pre-P2 one (G1).
+            ~FramebufferObject();
+#endif
 
             void AttachTexture(FramebufferAttachmentType type, const SharedPtr<ITextureObject>& texture,
                                TextureUploadTarget textureUploadTarget = TextureUploadTarget::Unknown, int level = 0,

MobileGL/MG_State/GLState/FramebufferState/FramebufferState.h

@@ -25,7 +25,19 @@ namespace MobileGL::MG_State::GLState {
         Bool ValidateName(Uint index) const;
         Bool ValidateFramebufferObject(Uint index) const;
 
+#if MOBILEGL_PIPE_PUSH
+        // P2 brief D4: "did the attachment set or the default geometry of ANY framebuffer
+        // move". It does NOT cover a BIND - a bind writes a BindingSlot, not the object - so
+        // MGPipeTracker pairs this counter with the bound draw framebuffer identity, which
+        // is one extra load and keeps the bump points on the object where they belong.
+        void NoteAttachmentChanged() { ++m_anyAttachmentGeneration; }
+        Uint64 GetAnyAttachmentGeneration() const { return m_anyAttachmentGeneration; }
+#endif
+
     private:
+#if MOBILEGL_PIPE_PUSH
+        Uint64 m_anyAttachmentGeneration = 0;
+#endif
         UnorderedMap<Uint, SharedPtr<FramebufferObject>> m_framebufferObjects;
         IndexGenerator<Uint> m_indexGenerator;
         Array<BindingSlot<FramebufferObject>, static_cast<SizeT>(FramebufferTarget::FramebufferTargetCount)>

MobileGL/MG_State/GLState/ProgramState/ProgramArtifacts.h

@@ -408,6 +408,17 @@ namespace MobileGL::MG_State::GLState {
     // Free constrained templates rather than members so the struct bodies above stay a verbatim
     // move. The sizeof trip wires below are what keep these tables honest: a member added to a
     // struct changes its size, trips the assertion, and the message sends the author here.
+    //
+    // THE SERIALIZER NOW EXISTS (P4a): MG_State/GLState/ProgramState/ProgramArtifactsCodec.
+    // {h,cpp}, beside this header rather than inside it so the check_include_closure.py
+    // "artifacts-header" probe stays untouched. It is two visitors over the tables below - a
+    // writer that appends to a Vector<Uint8> and a reader that consumes one - length-prefixed,
+    // little-endian, with a format-version word first and a MGL_LINKARTIFACTS_SIZE echo
+    // second, so a struct that gained a field and a codec that did not is a mismatch at READ
+    // time rather than a silent truncation. Adding a member to any struct above therefore
+    // means: add its VisitFields row here, update the sizeof number below, and bump
+    // kProgramArtifactsCodecVersion. `LinkArtifacts::program` stays the one deliberate
+    // omission, and the codec has no arm for it.
     template <class Self, class V>
         requires std::same_as<std::remove_const_t<Self>, TypeFacts>
     void VisitFields(Self& a, V&& v) {
@@ -549,7 +560,7 @@ namespace MobileGL::MG_State::GLState {
     // ---- trip wires ----
     // TypeFacts is a POD on every ABI: 13 Bool + 3 bytes of padding + 7 x 4-byte scalars.
     static_assert(std::is_trivially_copyable_v<TypeFacts> && sizeof(TypeFacts) == 44,
-                  "TypeFacts changed: add the field to VisitFields(TypeFacts) (and its serializer when one exists), then update this number");
+                  "TypeFacts changed: add the field to VisitFields(TypeFacts) (and ProgramArtifactsCodec.cpp's serializer), then update this number");
     // The container-bearing structs have one size per standard library (std::string and
     // std::set differ between libstdc++ and libc++), so their numbers are pinned PER STL:
     // libstdc++ (the Linux CI toolchain) here, libc++ (the NDK) by the integrator, MSVC
@@ -565,12 +576,12 @@ namespace MobileGL::MG_State::GLState {
 #endif
 #ifdef MGL_LINKARTIFACTS_SIZE
     static_assert(sizeof(ResourceReflection) == MGL_RESOURCEREFLECTION_SIZE,
-                  "ResourceReflection changed size: add the field to VisitFields(ResourceReflection) (and its serializer when one exists), then update this number");
+                  "ResourceReflection changed size: add the field to VisitFields(ResourceReflection) (and ProgramArtifactsCodec.cpp's serializer), then update this number");
     static_assert(sizeof(XfbVarying) == MGL_XFBVARYING_SIZE,
-                  "XfbVarying changed size: add the field to VisitFields(XfbVarying) (and its serializer when one exists), then update this number");
+                  "XfbVarying changed size: add the field to VisitFields(XfbVarying) (and ProgramArtifactsCodec.cpp's serializer), then update this number");
     static_assert(sizeof(LinkArtifacts) == MGL_LINKARTIFACTS_SIZE,
-                  "LinkArtifacts changed size: add the field to VisitFields(LinkArtifacts) (and its serializer when one exists), then update this number");
+                  "LinkArtifacts changed size: add the field to VisitFields(LinkArtifacts) (and ProgramArtifactsCodec.cpp's serializer), then update this number");
     static_assert(sizeof(SpirvArtifacts) == MGL_SPIRVARTIFACTS_SIZE,
-                  "SpirvArtifacts changed size: add the field to VisitFields(SpirvArtifacts) (and its serializer when one exists), then update this number");
+                  "SpirvArtifacts changed size: add the field to VisitFields(SpirvArtifacts) (and ProgramArtifactsCodec.cpp's serializer), then update this number");
 #endif
 } // namespace MobileGL::MG_State::GLState

MobileGL/MG_State/GLState/ProgramState/ProgramArtifactsCodec.cpp

@@ -0,0 +1,303 @@
+// MobileGL - MobileGL/MG_State/GLState/ProgramState/ProgramArtifactsCodec.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
+
+// ProgramArtifactsCodec.h. Compiled only under MOBILEGL_PIPE_PUSH (the root CMakeLists.txt
+// appends it inside `if (MOBILEGL_PIPE_PUSH)`), so the pull build gains no symbol from it.
+#include "ProgramArtifactsCodec.h"
+
+#include <bit>
+#include <cstring>
+#include <set>
+#include <string>
+#include <type_traits>
+#include <vector>
+
+namespace MobileGL::MG_State::GLState {
+    namespace {
+        // Little-endian, stated rather than assumed. Every ABI MobileGL ships on is
+        // little-endian; the day one is not, this is a compile error and not a silently
+        // byte-swapped reflection table.
+        static_assert(std::endian::native == std::endian::little,
+                      "the program-archive codec writes scalars in native order and MobileGL's "
+                      "ABIs are little-endian; a big-endian target needs explicit byte order");
+
+        // ---- the four container shapes the archive is built out of ----
+        //
+        // Detected by SHAPE rather than by naming std::vector / ska::flat_hash_map, because
+        // MobileGL's aliases are not all std:: types (UnorderedMap is ska::flat_hash_map) and
+        // a codec that named them would stop compiling the day one is swapped. The order the
+        // arms are tested in is what makes them unambiguous: String before every container,
+        // maps before sets (a map has both key_type and mapped_type), fixed arrays before
+        // resizable ones.
+        template <class T>
+        concept ArchiveString = std::same_as<T, String>;
+
+        template <class T>
+        concept ArchiveMap = requires {
+            typename T::key_type;
+            typename T::mapped_type;
+        };
+
+        template <class T>
+        concept ArchiveSet = requires { typename T::key_type; } && !ArchiveMap<T> && !ArchiveString<T>;
+
+        template <class T>
+        concept ArchiveFixedArray = requires { std::tuple_size<T>::value; };
+
+        template <class T>
+        concept ArchiveVector = !ArchiveString<T> && !ArchiveFixedArray<T> && requires(T& t) {
+            t.resize(SizeT{0});
+            t.size();
+            t.begin();
+        };
+
+        template <class T>
+        concept ArchiveScalar = std::is_arithmetic_v<T> || std::is_enum_v<T>;
+
+        // The ONE hand-written arm, and it is hand-written because ProgramArtifacts.h gives it
+        // no VisitFields table: glslang::TIntermediate::TUniformInitializer is a plain
+        // aggregate that merely LOOKS like a glslang type (std::string + scalars + two
+        // std::vectors), which is exactly what ProgramTranslationCache.h audited it as when it
+        // decided the archive holds no glslang-owned memory. If a field is added there, this
+        // arm and the format version below both have to move.
+        using UniformInitializer = glslang::TIntermediate::TUniformInitializer;
+
+        template <class T>
+        concept ArchiveUniformInitializer = std::same_as<T, UniformInitializer>;
+
+        // ---- the writer ----
+
+        template <class T>
+        void PutRaw(Vector<Uint8>& out, const T& value) {
+            static_assert(std::is_trivially_copyable_v<T>);
+            const SizeT at = out.size();
+            out.resize(at + sizeof(T));
+            std::memcpy(out.data() + at, &value, sizeof(T));
+        }
+
+        void PutCount(Vector<Uint8>& out, SizeT count) {
+            PutRaw(out, static_cast<Uint64>(count));
+        }
+
+        template <class T>
+        void WriteValue(Vector<Uint8>& out, const T& value);
+
+        template <class T>
+        void WriteSequence(Vector<Uint8>& out, const T& value) {
+            PutCount(out, value.size());
+            for (const auto& element : value) WriteValue(out, element);
+        }
+
+        template <class T>
+        void WriteValue(Vector<Uint8>& out, const T& value) {
+            if constexpr (ArchiveScalar<T>) {
+                PutRaw(out, value);
+            } else if constexpr (ArchiveString<T>) {
+                PutCount(out, value.size());
+                const SizeT at = out.size();
+                out.resize(at + value.size());
+                if (!value.empty()) std::memcpy(out.data() + at, value.data(), value.size());
+            } else if constexpr (ArchiveMap<T>) {
+                PutCount(out, value.size());
+                for (const auto& entry : value) {
+                    WriteValue(out, entry.first);
+                    WriteValue(out, entry.second);
+                }
+            } else if constexpr (ArchiveSet<T>) {
+                WriteSequence(out, value);
+            } else if constexpr (ArchiveFixedArray<T>) {
+                // No count: the width is part of the type, and writing one would let a reader
+                // believe a stream that disagrees with the struct.
+                for (const auto& element : value) WriteValue(out, element);
+            } else if constexpr (ArchiveVector<T>) {
+                WriteSequence(out, value);
+            } else if constexpr (ArchiveUniformInitializer<T>) {
+                WriteValue(out, value.name);
+                WriteValue(out, value.basicType);
+                WriteValue(out, value.vectorSize);
+                WriteValue(out, value.matrixCols);
+                WriteValue(out, value.matrixRows);
+                WriteValue(out, value.arraySize);
+                WriteValue(out, value.intValues);
+                WriteValue(out, value.floatValues);
+            } else {
+                // The archive's own structs: TypeFacts, ResourceReflection, XfbVarying. ONE
+                // table serves both directions, so a member added to any of them is carried by
+                // both halves of this codec the moment its VisitFields row is added - and a
+                // type with no table at all is a compile error here rather than a silently
+                // skipped field.
+                VisitFields(value, [&out](const char*, const auto& field) { WriteValue(out, field); });
+            }
+        }
+
+        // ---- the reader ----
+
+        struct ReadCursor {
+            const Uint8* Bytes = nullptr;
+            SizeT Size = 0;
+            SizeT Pos = 0;
+            Bool Ok = true;
+
+            SizeT Remaining() const { return Size - Pos; }
+        };
+
+        template <class T>
+        Bool TakeRaw(ReadCursor& in, T& value) {
+            static_assert(std::is_trivially_copyable_v<T>);
+            if (!in.Ok || in.Remaining() < sizeof(T)) {
+                in.Ok = false;
+                return false;
+            }
+            std::memcpy(&value, in.Bytes + in.Pos, sizeof(T));
+            in.Pos += sizeof(T);
+            return true;
+        }
+
+        // A COUNT IS CHECKED AGAINST THE BYTES THAT REMAIN BEFORE ANYTHING IS ALLOCATED. Every
+        // element this codec writes costs at least one byte, so a count larger than the
+        // remaining bytes cannot describe this stream - and refusing it here is what stops a
+        // corrupt or truncated archive from turning into a multi-gigabyte resize before the
+        // element loop notices it has run out.
+        Bool TakeCount(ReadCursor& in, SizeT& count) {
+            Uint64 raw = 0;
+            if (!TakeRaw(in, raw)) return false;
+            if (raw > static_cast<Uint64>(in.Remaining())) {
+                in.Ok = false;
+                return false;
+            }
+            count = static_cast<SizeT>(raw);
+            return true;
+        }
+
+        template <class T>
+        void ReadValue(ReadCursor& in, T& value);
+
+        template <class T>
+        void ReadValue(ReadCursor& in, T& value) {
+            if constexpr (ArchiveScalar<T>) {
+                TakeRaw(in, value);
+            } else if constexpr (ArchiveString<T>) {
+                SizeT count = 0;
+                if (!TakeCount(in, count)) return;
+                value.assign(reinterpret_cast<const char*>(in.Bytes + in.Pos), count);
+                in.Pos += count;
+            } else if constexpr (ArchiveMap<T>) {
+                SizeT count = 0;
+                if (!TakeCount(in, count)) return;
+                value.clear();
+                for (SizeT i = 0; i < count && in.Ok; ++i) {
+                    typename T::key_type key{};
+                    typename T::mapped_type mapped{};
+                    ReadValue(in, key);
+                    ReadValue(in, mapped);
+                    if (!in.Ok) return;
+                    value.emplace(Move(key), Move(mapped));
+                }
+            } else if constexpr (ArchiveSet<T>) {
+                SizeT count = 0;
+                if (!TakeCount(in, count)) return;
+                value.clear();
+                for (SizeT i = 0; i < count && in.Ok; ++i) {
+                    typename T::key_type key{};
+                    ReadValue(in, key);
+                    if (!in.Ok) return;
+                    value.insert(Move(key));
+                }
+            } else if constexpr (ArchiveFixedArray<T>) {
+                for (auto& element : value) {
+                    ReadValue(in, element);
+                    if (!in.Ok) return;
+                }
+            } else if constexpr (ArchiveVector<T>) {
+                SizeT count = 0;
+                if (!TakeCount(in, count)) return;
+                value.clear();
+                value.resize(count);
+                for (auto& element : value) {
+                    ReadValue(in, element);
+                    if (!in.Ok) return;
+                }
+            } else if constexpr (ArchiveUniformInitializer<T>) {
+                ReadValue(in, value.name);
+                ReadValue(in, value.basicType);
+                ReadValue(in, value.vectorSize);
+                ReadValue(in, value.matrixCols);
+                ReadValue(in, value.matrixRows);
+                ReadValue(in, value.arraySize);
+                ReadValue(in, value.intValues);
+                ReadValue(in, value.floatValues);
+            } else {
+                VisitFields(value, [&in](const char*, auto& field) {
+                    if (in.Ok) ReadValue(in, field);
+                });
+            }
+        }
+
+        // The struct-size echo. Under a toolchain whose sizes are not pinned yet
+        // (ProgramArtifacts.h's libc++ branch until the integrator fills it in) this is 0,
+        // which still round-trips within one build - the echo compares what THIS build wrote
+        // against what THIS build expects - and stops mattering the moment the pin lands.
+#ifdef MGL_LINKARTIFACTS_SIZE
+        inline constexpr Uint64 kLinkArtifactsSizeEcho = MGL_LINKARTIFACTS_SIZE;
+#else
+        inline constexpr Uint64 kLinkArtifactsSizeEcho = 0;
+#endif
+    } // namespace
+
+    void EncodeProgramArtifacts(const LinkArtifacts& link, const SpirvArtifacts& spirv,
+                                Vector<Uint8>& out) {
+        PutRaw(out, kProgramArtifactsCodecVersion);
+        PutRaw(out, kLinkArtifactsSizeEcho);
+        // `link` is walked through its own VisitFields table, which omits the live
+        // SharedPtr<glslang::TProgram>: 57 of the 58 members. There is no arm here for it and
+        // there must not be one - it points into a glslang arena that no archived instance
+        // owns, and ProgramTranslationCache asserts it is null at insert.
+        WriteValue(out, link);
+        WriteValue(out, spirv);
+    }
+
+    Bool DecodeProgramArtifacts(const Uint8* bytes, SizeT size, LinkArtifacts& link,
+                                SpirvArtifacts& spirv) {
+        // Both outputs are left in a DEFINED state on every exit, including every failure:
+        // a caller that ignores the return value gets an empty archive rather than half of a
+        // truncated one.
+        link = LinkArtifacts{};
+        spirv = SpirvArtifacts{};
+        if (bytes == nullptr) return false;
+
+        ReadCursor in{bytes, size, 0, true};
+        Uint32 version = 0;
+        Uint64 sizeEcho = 0;
+        if (!TakeRaw(in, version) || !TakeRaw(in, sizeEcho)) return false;
+        // REFUSED, NOT GUESSED. A different version word or a struct that changed width means
+        // the bytes describe a layout this build does not have; deserialising them anyway
+        // writes garbage into the tail of a reflection table, which is exactly the failure the
+        // two words exist to turn into a clean false.
+        if (version != kProgramArtifactsCodecVersion) return false;
+        if (sizeEcho != kLinkArtifactsSizeEcho) return false;
+
+        ReadValue(in, link);
+        ReadValue(in, spirv);
+        if (!in.Ok) {
+            link = LinkArtifacts{};
+            spirv = SpirvArtifacts{};
+            return false;
+        }
+        // Trailing bytes are a mismatch too: the format accounts for every byte it writes, so
+        // anything left over means the reader and the writer disagree about the shape and the
+        // agreement so far was luck.
+        if (in.Pos != in.Size) {
+            link = LinkArtifacts{};
+            spirv = SpirvArtifacts{};
+            return false;
+        }
+        // Never written, never read, and stated here so it cannot be added by reflex.
+        link.program = nullptr;
+        return true;
+    }
+} // namespace MobileGL::MG_State::GLState

MobileGL/MG_State/GLState/ProgramState/ProgramArtifactsCodec.h

@@ -0,0 +1,76 @@
+// MobileGL - MobileGL/MG_State/GLState/ProgramState/ProgramArtifactsCodec.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 "ProgramArtifacts.h"
+
+// The reflection ARCHIVE's serializer (P4a, D-H2): create_shader_state's payload is per-stage
+// SPIR-V plus LinkArtifacts + SpirvArtifacts, whole structs, and until now nothing could turn
+// those into bytes. Every VisitFields comment in ProgramArtifacts.h said "(and its serializer
+// when one exists)"; this is it.
+//
+// IT LIVES BESIDE THE HEADER RATHER THAN INSIDE IT, deliberately: ProgramArtifacts.h carries
+// the check_include_closure.py "artifacts-header" probe, which pins that the header is
+// glslang-free by symbol and reaches no ShaderObject, no SpvcSession, no Config and no
+// MG_Backend. A codec inside it would have to be inspected against that probe on every edit;
+// a codec beside it leaves the probe untouched, and this file is compiled only in a push
+// build (the root CMakeLists.txt appends it inside `if (MOBILEGL_PIPE_PUSH)`).
+//
+// WHEN IT ACTUALLY RUNS, and the answer is "not on the monolith hot path at all". In monolith
+// the archive does not travel: MGPProgramDesc's seven blob refs are declared with Size 0 -
+// "this record does not declare its blob" - and MGPipeApplyCreateShaderState takes the two
+// structs by pointer beside the record, so the applier reads the frontend's own archive and
+// this codec is never called. The VERIFY build is where it is exercised, and it is exercised
+// as LIVE CODE WITH A GATE rather than as dead code with a unit test: the applier serialises,
+// deserialises and field-compares before storing, and a mismatch is
+// Fatal{PipeVerifyDiffer, "program-archive"}. Under split, P5 is what makes it the transport's
+// path.
+//
+// THE FORMAT, and every part of it is a refusal rather than a guess:
+//   * a VERSION word first, and a MGL_LINKARTIFACTS_SIZE echo second, so a struct that gained
+//     a field and a codec that did not is a MISMATCH AT READ TIME rather than a silent
+//     truncation that deserialises garbage into the tail of a reflection table;
+//   * length-prefixed everything - strings, vectors, maps, sets - with the count checked
+//     against the bytes that remain before a single element is allocated, so a corrupt count
+//     cannot turn into a four-billion-element resize;
+//   * little-endian, which is asserted rather than assumed;
+//   * and `LinkArtifacts::program` is NEVER visited. It is the live glslang TProgram, it is
+//     null for every archived instance by construction, and VisitFields deliberately omits it
+//     (57 of the 58 members). Decode leaves it null.
+namespace MobileGL::MG_State::GLState {
+
+    // Bumped whenever the byte format changes in a way a previous reader would misread. A
+    // reader that sees a different word REFUSES; it never tries to guess a layout.
+    inline constexpr Uint32 kProgramArtifactsCodecVersion = 1;
+
+    // Appends the archive to `out` (which is not cleared, so a caller may frame it). Never
+    // fails: everything it walks is owned plain data.
+    void EncodeProgramArtifacts(const LinkArtifacts& link, const SpirvArtifacts& spirv,
+                                Vector<Uint8>& out);
+
+    // Replaces `link` and `spirv` with what `bytes` describes. Returns false - with both
+    // outputs left in a defined, default state - for a truncated stream, a version mismatch, a
+    // struct-size mismatch, or trailing bytes the format does not account for. `link.program`
+    // is always null on return.
+    Bool DecodeProgramArtifacts(const Uint8* bytes, SizeT size, LinkArtifacts& link,
+                                SpirvArtifacts& spirv);
+
+    // How many fields a type's VisitFields table actually visits. The codec walks exactly that
+    // table, so this is what pins "the codec did not quietly grow an arm of its own" - most of
+    // all for LinkArtifacts, whose 58th member is the live TProgram the table omits. It is a
+    // runtime count rather than a static_assert because VisitFields needs an INSTANCE and
+    // these structs carry strings, vectors and maps: none of them is a constant expression.
+    // ProgramArtifactsCodecTest is where it is asserted.
+    template <class T>
+    inline SizeT ProgramArtifactsVisitedFieldCount() {
+        T probe{};
+        SizeT count = 0;
+        VisitFields(probe, [&count](const char*, auto&) { ++count; });
+        return count;
+    }
+} // namespace MobileGL::MG_State::GLState

MobileGL/MG_State/GLState/ProgramState/ProgramObject.cpp

@@ -14,6 +14,8 @@
 #include <MG_Util/Async/ShaderCompilePool.h>
 #include <MG_Util/Converters/GLToStr/GLEnumConverter.h>
 #include <MG_Util/ShaderTranspiler/CompileEnv.h>
+#include <MG_Pipe/PipeMutation.h>
+#include <MG_State/GLState/StateObjectDeathNotice.h>
 
 const char* kDefaultFragmentShaderSource = R"(#version 460 core
 layout(location = 0) out vec4 FragColor;
@@ -28,7 +30,34 @@ namespace MobileGL::MG_State::GLState {
         return s_nextProgramLifetimeId.fetch_add(1, std::memory_order_relaxed);
     }
 
-    ProgramObject::~ProgramObject() { CancelLink(); }
+    ProgramObject::~ProgramObject() {
+        CancelLink();
+#if MOBILEGL_PIPE_PUSH
+        // P2 step e2: ANNOUNCE the death instead of leaving the backend to discover it in a
+        // garbage sweep. This is the last SharedPtr to this object dropping - not
+        // glDeleteProgram, which only marks the name and leaves a still-bound object very much
+        // alive - so it is the exact moment the backend's twin, and the driver storage that
+        // twin owns, stop being reachable. The notice carries the lifetime id because the
+        // object no longer exists to be passed, and because the lifetime id is what the client
+        // slot allocator resolves the handle from. No-op unless a backend registered the ops
+        // (a pull build declares none at all).
+        //
+        // P4a D-I1: the notice is no longer raised directly - it is step 2 of the ONE
+        // client-side death helper for this kind, which emits delete_shader_state first,
+        // raises the notice second and frees the slot last. The client mints the ShaderCso, so
+        // the client is where its death has to be spoken from: a backend death-ops table is a
+        // redundant, idempotent SECOND path, and under a backend that installs none it was
+        // previously the ONLY one, which is how a slot leaks for the life of the process.
+        //
+        // AN ORDINARY PROGRAM AND A PIPELINE COMPOSITE TAKE THIS SAME LINE. A composite is an
+        // ordinary ProgramObject with its own lifetime id, its slot merely comes out of the
+        // reserved band, and its OTHER release path - the pipeline cache dropping it when the
+        // draw-program signature moves - goes through the same helper. Whichever runs second is
+        // a proven no-op, because the slot allocator refuses a slot that is not live at that
+        // generation.
+        MG_Pipe::MGPipeEmitShaderCsoDestroyAndFree(m_lifetimeId);
+#endif
+    }
 
     // EnsureLinkJoined() is defined inline in ProgramObject.h (see the comment there for
     // why: ~1200 call sites, no LTO). Only its blocking half lives here.

MobileGL/MG_State/GLState/ProgramState/ProgramObject.h

@@ -1111,6 +1111,23 @@ namespace MobileGL::MG_State::GLState {
             return Artifacts();
         }
 
+#if MOBILEGL_PIPE_PUSH
+        // The phase-B twin of GetLinkReflection, and it exists for exactly one caller:
+        // create_shader_state publishes the WHOLE SpirvArtifacts beside the whole
+        // LinkArtifacts, because the archive the far side deserialises into is those two
+        // structs and nothing else. The granular getters above (GetGeneratedSpirv,
+        // GetSpirvStatus, UsesNativeFloat64, PointSizeDemoted, GetSpirvValidationEnabled,
+        // GetUBOSize) cover what the GL layer needs and deliberately do not reach
+        // reservedNumSamplesOffset or uniformOffsets, which the descriptor does need.
+        //
+        // PUSH-ONLY, so a pull build's symbol set and this class's layout are byte-for-byte
+        // what they were: an inline const-ref accessor adds nothing to a pull build that is
+        // never compiled into one. It joins both phases, like every other Spirv() reader.
+        const SpirvArtifacts& GetSpirvReflection() const {
+            return Spirv();
+        }
+#endif
+
         static Bool IsValidUniformLocation(const LinkArtifacts& artifacts, Int location) {
             if (location < 0 || location > static_cast<Int>(artifacts.maxUniformLocation)) return false;
             if (static_cast<SizeT>(location) >= artifacts.uniformIndexInTProgram.size()) return false;

MobileGL/MG_State/GLState/RenderState/RenderState.cpp

@@ -317,9 +317,11 @@ namespace MobileGL {
                     SET_CAPABILITY(ColorLogicOp, enabled);
                     SET_CAPABILITY(DebugOutput, enabled);
                     SET_CAPABILITY(DebugOutputSynchronous, enabled);
+                    SET_CAPABILITY(DepthClamp, enabled);
                     SET_CAPABILITY(DepthTest, enabled);
                     SET_CAPABILITY(CullFace, enabled);
                     SET_CAPABILITY(Dither, enabled);
+                    SET_CAPABILITY(FramebufferSrgb, enabled);
                     SET_CAPABILITY(LineSmooth, enabled);
                     SET_CAPABILITY(Multisample, enabled);
                     SET_CAPABILITY(PolygonOffsetFill, enabled);
@@ -335,6 +337,7 @@ namespace MobileGL {
                     SET_CAPABILITY(SampleMask, enabled);
                     SET_CAPABILITY(SampleShading, enabled);
                     SET_CAPABILITY(StencilTest, enabled);
+                    SET_CAPABILITY(TextureCubeMapSeamless, enabled);
                     SET_CAPABILITY(ProgramPointSize, enabled);
                 case CapabilityInput::Blend: {
                     Bool stateChanged = false;
@@ -378,7 +381,9 @@ namespace MobileGL {
                     ++m_version;
                     break;
                 }
-                default: // not supported currently
+                // Every CapabilityInput now has storage; the arm is a backstop for a value
+                // outside the enum, not a silent swallow of a real glEnable.
+                default:
                     break;
                 }
 #undef SET_CAPABILITY
@@ -392,9 +397,11 @@ namespace MobileGL {
                     RETURN_CAPABILITY(ColorLogicOp);
                     RETURN_CAPABILITY(DebugOutput);
                     RETURN_CAPABILITY(DebugOutputSynchronous);
+                    RETURN_CAPABILITY(DepthClamp);
                     RETURN_CAPABILITY(DepthTest);
                     RETURN_CAPABILITY(CullFace);
                     RETURN_CAPABILITY(Dither);
+                    RETURN_CAPABILITY(FramebufferSrgb);
                     RETURN_CAPABILITY(LineSmooth);
                     RETURN_CAPABILITY(Multisample);
                     RETURN_CAPABILITY(PolygonOffsetFill);
@@ -410,6 +417,7 @@ namespace MobileGL {
                     RETURN_CAPABILITY(SampleMask);
                     RETURN_CAPABILITY(SampleShading);
                     RETURN_CAPABILITY(StencilTest);
+                    RETURN_CAPABILITY(TextureCubeMapSeamless);
                     RETURN_CAPABILITY(ProgramPointSize);
                 case CapabilityInput::Blend:
                     return m_parameters.BlendStates[0].Enabled;

MobileGL/MG_State/GLState/RenderState/RenderState.h

@@ -169,11 +169,16 @@ namespace MobileGL {
                 // not evict a cached pipeline. Keeping one counter for both made a glViewport call
                 // knock the next draw off the pipeline memo AND the draw fast path.
                 Uint16 m_pipelineStateVersion = 0;
-                RenderStateParameters m_parameters;
+                // Value-initialised, PADDING INCLUDED. The MGPipe CSO key is a byte-range
+                // hash over RenderStateParameters and the residual block's trip wire is a
+                // byte-level compare, so indeterminate padding would make a CSO handle
+                // reproducible only within one context and would make the trip wire
+                // meaningless. Costs one .text resize of this constructor.
+                RenderStateParameters m_parameters{};
 
                 // Pixel Store
-                PixelStoreParameters m_pixelStorePackParameters;
-                PixelStoreParameters m_pixelStoreUnpackParameters;
+                PixelStoreParameters m_pixelStorePackParameters{};
+                PixelStoreParameters m_pixelStoreUnpackParameters{};
             };
         } // namespace GLState
     } // namespace MG_State

MobileGL/MG_State/GLState/RenderbufferState/RenderbufferObject.cpp

@@ -8,6 +8,11 @@
 
 #include "RenderbufferObject.h"
 #include <MG_Util/Metrics/TextureMetrics.h>
+#include <MG_State/GLState/StateObjectDeathNotice.h>
+// The contract's own door, exactly as the texture half takes it (c0b): the four renderbuffer
+// hooks this file calls are declared in MG_Pipe/PipeMutation.h and defined in
+// MG_Impl/Pipe/PipeFill.cpp, so no MG_State translation unit includes the client's emitter.
+#include <MG_Pipe/PipeMutation.h>
 
 #include <atomic>
 
@@ -24,7 +29,35 @@ namespace MobileGL {
                 return g_nextRenderbufferLifetimeId.fetch_add(1, std::memory_order_relaxed);
             }
 
-            RenderbufferObject::RenderbufferObject(Uint externalIndex) : m_externalIndex(externalIndex) {}
+            RenderbufferObject::RenderbufferObject(Uint externalIndex) : m_externalIndex(externalIndex) {
+#if MOBILEGL_PIPE_PUSH
+                // P4a D-D1: resource_create from the constructor, carrying no storage. A
+                // renderbuffer is an INDEPENDENT class on the wire - it shares MGPResourceDesc's
+                // shape with textures and buffers and nothing else - and its handle is minted
+                // whatever the subsystem bitmask says, because MGPSurface::Res names it out of the
+                // framebuffer subsystem.
+                // TWO CALLS AND NOT ONE (c0b): the mint is unconditional in a push build
+                // because a renderbuffer is named by handle out of the framebuffer subsystem
+                // whether or not its own family is switched on; the create is what the gate in
+                // PipeFill.cpp decides.
+                MG_Pipe::MGPipeMintRenderbufferHandle(*this);
+                MG_Pipe::MGPipeEmitRenderbufferResourceCreate(*this);
+#endif
+            }
+
+#if MOBILEGL_PIPE_PUSH
+            RenderbufferObject::~RenderbufferObject() {
+                // P4a D-I1, the fixed three-step order: the wire delete first (published-gated,
+                // because a slot is not evidence of a record), then the death notice - the P2
+                // step-e2 announcement that used to stand here alone, raised while the handle
+                // still resolves - and the slot last. Steps 2 and 3 are the contract's helper;
+                // step 1 is this package's, one statement earlier, because the helper's file
+                // belongs to the contract package for the whole phase.
+                // All three steps are the contract's helper (c0b); v1's separate step-1 call
+                // is deleted, not kept, for the reason ~TextureObjectBase states in full.
+                MG_Pipe::MGPipeEmitRenderbufferDestroyAndFree(m_lifetimeId);
+            }
+#endif
 
             Uint RenderbufferObject::GetExternalIndex() const {
                 return m_externalIndex;
@@ -81,17 +114,53 @@ namespace MobileGL {
             void RenderbufferObject::SetInternalFormat(TextureInternalFormat format) {
                 m_internalFormat = format;
                 m_componentSizes = MG_Util::GetComponentSizesForInternalFormat(format);
+#if MOBILEGL_PIPE_PUSH
+                PipePublishDescriptor();
+#endif
             }
 
             void RenderbufferObject::AllocateStorage(IntVec2 size) {
                 m_width = size.x();
                 m_height = size.y();
                 m_allocated = true;
+#if MOBILEGL_PIPE_PUSH
+                PipePublishDescriptor();
+#endif
             }
 
             void RenderbufferObject::SetSamples(Int samples) {
                 m_samples = samples;
+#if MOBILEGL_PIPE_PUSH
+                PipePublishDescriptor();
+#endif
             }
+
+#if MOBILEGL_PIPE_PUSH
+            // D-D2: THE RENDERBUFFER PUBLICATION HOLE, CLOSED BY EMISSION AND NOT BY A NEW
+            // VERSION. These three setters bump no version and raise no notice, so
+            // `glBindRenderbuffer; glRenderbufferStorage(newSize)` on an attached renderbuffer
+            // was invisible to everything downstream. Emitting from the storage entry point
+            // closes the RESOURCE half; a version counter here would resize the pull build's
+            // object and break G1.
+            //
+            // THE FRAMEBUFFER HALF IS THE AGGREGATE BUMP BELOW (P4a fable seam F-3), and the
+            // sentence that used to end the paragraph above - "widening the shutter would fire
+            // the framebuffer emission on an unrelated renderbuffer write" - was the seam:
+            // set_framebuffer_state inlines an attachment's InternalFormat, extent and Samples at
+            // emission (D-C1), so re-storaging an ATTACHED renderbuffer left the framebuffer
+            // record - and the handle arm's four cross-object masks - describing the previous
+            // storage while the resource record described the new one. The bump costs one
+            // framebuffer re-emission per storage definition, whether or not the object is
+            // attached, which the emitter's content hash suppresses when nothing it inlines
+            // moved; it is not a counter on this object.
+            //
+            // The emitter dedupes on the built descriptor, so glRenderbufferStorage's three-setter
+            // sequence publishes once rather than three times.
+            void RenderbufferObject::PipePublishDescriptor() {
+                MG_Pipe::MGPipeEmitRenderbufferResourceRespecify(*this);
+                MGP_NOTE_AGGREGATE(FramebufferAttachment);
+            }
+#endif
         } // namespace GLState
     } // namespace MG_State
 } // namespace MobileGL

MobileGL/MG_State/GLState/RenderbufferState/RenderbufferObject.h

@@ -25,6 +25,12 @@ namespace MobileGL {
                 using TargetEnum = RenderbufferTarget;
 
                 RenderbufferObject(Uint externalIndex);
+#if MOBILEGL_PIPE_PUSH
+                // P2 step e2. Out of line, and declared only where there is a notice to raise:
+                // in a pull build this class stays trivially destructible, which is what keeps
+                // the pull build's symbol set byte-for-byte the pre-P2 one (G1).
+                ~RenderbufferObject();
+#endif
 
                 Uint GetExternalIndex() const;
                 void SetInternalFormat(TextureInternalFormat format);
@@ -53,6 +59,13 @@ namespace MobileGL {
 
             private:
                 static Uint64 AllocateLifetimeId();
+#if MOBILEGL_PIPE_PUSH
+                // P4a D-D2: resource_respecify, from every storage-defining setter. Non-virtual
+                // and push-only, so the pull build's object layout is untouched (P4a's
+                // admitted-resize set is EMPTY); defined in RenderbufferObject.cpp, which is the
+                // one translation unit that includes the client emitter.
+                void PipePublishDescriptor();
+#endif
 
                 Uint m_externalIndex = 0;
                 const Uint64 m_lifetimeId = AllocateLifetimeId();

MobileGL/MG_State/GLState/SamplerState/SamplerObject.cpp

@@ -9,8 +9,10 @@
 #include "SamplerObject.h"
 
 #include <MG_State/GLState/Core.h>
+#include <MG_State/GLState/StateObjectDeathNotice.h>
 
 #include <atomic>
+#include <MG_Pipe/PipeMutation.h>
 
 namespace MobileGL {
     namespace MG_State {
@@ -24,6 +26,35 @@ namespace MobileGL {
             SamplerObject::SamplerObject(Uint externalIndex)
                 : m_externalIndex(externalIndex), m_lifetimeId(AllocateLifetimeId()) {}
 
+#if MOBILEGL_PIPE_PUSH
+            SamplerObject::~SamplerObject() {
+                // P2 step e2: ANNOUNCE the death instead of leaving the backend to discover it in a
+                // garbage sweep. This is the last SharedPtr to this object dropping - not the
+                // glDelete* that only marks the name and leaves a still-bound object very much
+                // alive - so it is the exact moment the backend's twin, and the driver storage
+                // that twin owns, stop being reachable. The notice carries the lifetime id
+                // because the object no longer exists to be passed, and because the lifetime id
+                // is what the client slot allocator resolves the handle from. No-op unless a
+                // backend registered the ops (a pull build declares none at all).
+                //
+                // P4a D-I1: the notice is no longer raised directly - it is step 2 of the ONE
+                // client-side death helper for this kind, which emits delete_sampler_state
+                // first, raises the notice second and frees the slot last. Making the client
+                // the only death path is what stops a slot leaking under a backend that
+                // installs no death-ops table at all, and the backend's own notice becomes a
+                // redundant, idempotent second path rather than the only one.
+                //
+                // FOR A CONTENT-ADDRESSED SAMPLER CSO THIS HELPER CORRECTLY FREES NOTHING, and
+                // that is the design rather than a gap: the CSO belongs to a VALUE, not to this
+                // object (two identical SamplerObjects share one), so it is allocated with no
+                // lifetime id, the helper resolves nothing for this one, and the only death
+                // path for that slot is the CSO cache's LRU eviction - which is client-side and
+                // therefore backend-neutral on day one. What still goes out, unconditionally
+                // and exactly as before, is the notice.
+                MG_Pipe::MGPipeEmitSamplerCsoDestroyAndFree(m_lifetimeId);
+            }
+#endif
+
             void SamplerObject::BumpVersion() {
                 ++m_version;
                 // Every setter early-outs on an unchanged value, so this only runs on a real
@@ -32,6 +63,9 @@ namespace MobileGL {
                 // bindings must never miss an invalidation, and over-invalidating on a wrap-mode
                 // write costs one re-resolve.
                 if (pGLContext) pGLContext->BumpSamplingResolutionGeneration();
+                // Every sampler parameter is a texture PARAMETER as far as the dirty walk is
+                // concerned, and BumpVersion is the one choke point every setter reaches.
+                MGP_NOTE_AGGREGATE(TextureParams);
             }
 
             void SamplerObject::SetWrapS(SamplerWrapMode mode) {

MobileGL/MG_State/GLState/SamplerState/SamplerObject.h

@@ -16,6 +16,12 @@ namespace MobileGL {
             class SamplerObject {
             public:
                 SamplerObject(Uint externalIndex);
+#if MOBILEGL_PIPE_PUSH
+                // P2 step e2. Out of line, and declared only where there is a notice to raise:
+                // in a pull build this class keeps its implicit destructor, which is what keeps
+                // the pull build's symbol set byte-for-byte the pre-P2 one (G1).
+                ~SamplerObject();
+#endif
 
                 void SetWrapS(SamplerWrapMode mode);
                 void SetWrapT(SamplerWrapMode mode);

MobileGL/MG_State/GLState/StateObjectDeathNotice.h

@@ -0,0 +1,62 @@
+// MobileGL - MobileGL/MG_State/GLState/StateObjectDeathNotice.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>
+
+#if MOBILEGL_PIPE_PUSH
+#include <MG_Pipe/MGPipeHandles.h>
+
+// P2 step e2, the frontend half: TELL the backend that a state object died, instead of
+// leaving it to discover the death in a garbage sweep.
+//
+// Today the only kind that announces its own death is Buffer, through BufferBackendOps
+// (BufferState/BufferObject.h) - a frontend-declared ops table that the backend fills in at
+// context bring-up. This is the same shape for the other six kinds, with two differences that
+// follow from what the notice is for:
+//
+//   * it carries {kind, lifetimeId} and NOT the object, because by the time the last
+//     SharedPtr has dropped there is no object left to pass, and the lifetime id is exactly
+//     the key the client slot allocator resolves a handle from (ARCHITECTURE.md 4.2);
+//   * it is one entry point for every kind rather than one ops table per kind, because the
+//     backend's answer is the same for all six: free the slot, drop the twin.
+//
+// It exists only under MOBILEGL_PIPE_PUSH. A pull build has no slot allocator, no handle and
+// nothing that could consume the notice, and G1 requires its symbol set to be byte-for-byte
+// the pre-P2 one - so in that build this header declares nothing at all and the call sites
+// compile to nothing.
+//
+// The pointer is written once, at backend bring-up, and read from state-object destructors.
+// It is deliberately a plain pointer and not an atomic: the destructors and the bring-up run
+// on the context thread, exactly as BufferBackendOps' g_bufferBackendOps does.
+namespace MobileGL::MG_State::GLState {
+
+    struct StateObjectDeathOps {
+        // The last SharedPtr to the frontend object with this lifetime id has dropped.
+        // Called from the object's destructor, so the object must NOT be touched.
+        void (*OnDestroyed)(MG_Pipe::MGPipeKind kind, Uint64 lifetimeId) = nullptr;
+    };
+
+    inline const StateObjectDeathOps* g_stateObjectDeathOps = nullptr;
+
+    inline void SetStateObjectDeathOps(const StateObjectDeathOps* ops) {
+        g_stateObjectDeathOps = ops;
+    }
+
+    inline const StateObjectDeathOps* GetStateObjectDeathOps() {
+        return g_stateObjectDeathOps;
+    }
+
+    inline void NotifyStateObjectDestroyed(MG_Pipe::MGPipeKind kind, Uint64 lifetimeId) {
+        const StateObjectDeathOps* ops = g_stateObjectDeathOps;
+        if (ops == nullptr || ops->OnDestroyed == nullptr) return;
+        ops->OnDestroyed(kind, lifetimeId);
+    }
+
+} // namespace MobileGL::MG_State::GLState
+#endif // MOBILEGL_PIPE_PUSH

MobileGL/MG_State/GLState/TextureState/TextureObject.cpp

@@ -8,8 +8,17 @@
 
 #include "TextureObject.h"
 #include "MG_State/GLState/Core.h"
+#include "MG_State/GLState/StateObjectDeathNotice.h"
 #include "MG_Util/Types.h"
 #include <MG_Util/Metrics/TextureMetrics.h>
+#include <MG_Pipe/PipeMutation.h>
+// NO MG_State TRANSLATION UNIT SEES THE CLIENT'S EMITTER ANY MORE (c0b, ID-13). v1 included
+// MG_Impl/Pipe/TextureEmit.h here and in RenderbufferObject.cpp because at the contract tag
+// MG_Pipe/PipeMutation.h carried no texture row; it now declares the four mints, the nine
+// emissions and the publication latch, so this file sees a DECLARATION exactly as
+// BufferObject.cpp does and the closure gate's mutation-header probe has nothing to find.
+// The three PipePublish* helpers stay on TextureObjectBase so the cube's, the view's and the
+// buffer texture's translation units keep calling an inherited member.
 
 namespace MobileGL {
     namespace MG_State {
@@ -25,6 +34,84 @@ namespace MobileGL {
                 return s_nextTextureLifetimeId.fetch_add(1, std::memory_order_relaxed);
             }
 
+#if MOBILEGL_PIPE_PUSH
+            TextureObjectBase::~TextureObjectBase() {
+                // P4a D-I1: BACKEND-NEUTRAL FROM DAY ONE, and the three-step order is fixed.
+                //
+                //   1. the wire delete FIRST - it drops the applier's record while the record
+                //      still exists, so a recycled slot cannot inherit a field. Published-gated
+                //      rather than slot-gated: a backend twin table mints a slot through
+                //      MGPipeSlots().Acquire whether or not the subsystem ever asked this client
+                //      to emit a create, and a resource_destroy on such a handle is a refused
+                //      call the applier counts and asserts on;
+                //   2. the death notice SECOND - it resolves the handle through the allocator,
+                //      and a backend told after the free could no longer find its twin. This is
+                //      the P2 step-e2 announcement that used to stand here alone: the last
+                //      SharedPtr to this object dropping, not the glDelete* that only marks the
+                //      name and leaves a still-bound object very much alive;
+                //   3. the slot LAST, and a double free on a stale generation is a proven no-op.
+                //
+                // ALL THREE STEPS ARE THE CONTRACT'S HELPER (c0b): it reads the publication
+                // latch, emits the resource_destroy itself, raises the notice while the handle
+                // still resolves and frees the slot last. v1 emitted step 1 from a second
+                // statement here because at the tag the helper hard-coded `published = false`;
+                // that statement is deleted rather than kept, since a second delete for a
+                // record the helper has already dropped is a refused call the applier asserts
+                // on. The SamplerViewCso minted off this same lifetime id goes with it. The
+                // BUILT-IN SAMPLER does not: it is a real SamplerObject with its own lifetime
+                // id and its own destructor, which takes the same helper shape.
+                MG_Pipe::MGPipeEmitTextureDestroyAndFree(m_lifetimeId);
+            }
+
+            // ---- P4a's three client emission points (see TextureObject.h) ----
+
+            void TextureObjectBase::PipePublishDescriptor() {
+                MG_Pipe::MGPipeEmitTextureResourceRespecify(*this, MG_Pipe::MGPipeTextureRespecifyScope::WholeResource,
+                                                            0, 0);
+                // AND THE FRAMEBUFFER AGGREGATE MOVES (P4a fable seam F-3). The resource record
+                // above is only half of what a storage definition changes: set_framebuffer_state
+                // INLINES an attachment's InternalFormat, TextureTarget, extent, Samples and
+                // Complete at emission (D-C1), so redefining the storage of a texture that is
+                // ATTACHED changed those fields with nothing bit 11 reads moving - the format
+                // and shape setters bump the two TEXTURE aggregates and never the attachment
+                // one, and no path from a texture reaches its framebuffers. The handle arm then
+                // answered its four cross-object masks from the stale copy while the legacy arm
+                // re-read the frontend at the same re-sync. This is the one funnel every
+                // storage-defining entry point takes (see AllocateStorage), so the bump lives
+                // here and not per setter, it is push-only like the rest of this block, and it
+                // over-fires the framebuffer bit once per storage definition of an unattached
+                // texture - at load time, where a 304-byte hash is nothing.
+                MGP_NOTE_AGGREGATE(FramebufferAttachment);
+            }
+
+            void TextureObjectBase::PipePublishLevelDescriptor(TextureUploadTarget uploadTarget, Uint mipmapLevel) {
+                // ONE level was (re)allocated: only that level's pending upload is against
+                // storage that is gone (P4a final review C-1). Every other level's stays.
+                MG_Pipe::MGPipeEmitTextureResourceRespecify(*this, MG_Pipe::MGPipeTextureRespecifyScope::OneLevel,
+                                                            static_cast<Uint32>(uploadTarget),
+                                                            static_cast<Uint32>(mipmapLevel));
+                MGP_NOTE_AGGREGATE(FramebufferAttachment); // an attached level's extent is inlined (F-3)
+            }
+
+            void TextureObjectBase::PipePublishTruncatedDescriptor(TextureUploadTarget uploadTarget, Uint levelCount) {
+                // The chain was cut at `levelCount`: the levels above the cut are gone with their
+                // pending uploads, the levels below it are untouched and keep theirs.
+                MG_Pipe::MGPipeEmitTextureResourceRespecify(*this, MG_Pipe::MGPipeTextureRespecifyScope::LevelsFrom,
+                                                            static_cast<Uint32>(uploadTarget),
+                                                            static_cast<Uint32>(levelCount));
+                MGP_NOTE_AGGREGATE(FramebufferAttachment);
+            }
+
+            void TextureObjectBase::PipePublishParams() {
+                MG_Pipe::MGPipeEmitTextureParams(*this);
+            }
+
+            void TextureObjectBase::PipeNoteLevelDirty(TextureUploadTarget uploadTarget, Uint mipmapLevel) {
+                MG_Pipe::MGPipeNoteTextureLevelDirty(*this, static_cast<Uint32>(uploadTarget),
+                                                     static_cast<Uint32>(mipmapLevel));
+            }
+#endif
+
             void TextureObjectBase::BumpShapeVersion() {
                 ++m_shapeVersion;
                 // Shape is what mipmap-completeness is computed from, and completeness decides
@@ -53,6 +140,34 @@ namespace MobileGL {
                     m_sampler->SetWrapT(SamplerWrapMode::ClampToEdge);
                     m_sampler->SetWrapR(SamplerWrapMode::ClampToEdge);
                 }
+#if MOBILEGL_PIPE_PUSH
+                // P4a D-D1: A RESOURCE EXISTS BEFORE ANYTHING CAN NAME IT, so resource_create is
+                // emitted from the constructor and carries no storage - the store is defined
+                // lazily by the first respecify and every backend already tolerates a resource
+                // with none. The handle itself is minted whatever the subsystem bitmask says,
+                // because set_framebuffer_state and set_sampler_views name this texture by handle
+                // out of two different subsystems.
+                //
+                // NOTHING THE DERIVED CLASS IMPLEMENTS IS TOUCHED HERE and that is a
+                // correctness requirement rather than a style: the derived object does not exist
+                // yet, so ITextureObject::GetStorageType and ::GetUploadTargets - PURE, with no
+                // body on this base - would be undefined behaviour. The emitter reads GetTarget()
+                // and GetExternalIndex(), which TextureObjectBase itself overrides and which
+                // therefore dispatch to this class's own bodies over members the mem-init list
+                // has already written; the storage kind is derived from the target, which is
+                // exact (TextureObjectBuffer is the only class that reports Buffer and
+                // TextureBuffer is the only target it is constructed with).
+                //
+                // TWO CALLS AND NOT ONE (c0b): the MINT is unconditional in a push build -
+                // set_framebuffer_state and set_sampler_views name this texture by handle out of
+                // two other subsystems, so gating it would make them emit null handles in exactly
+                // the A/B arm that exists to isolate the families - and the CREATE is what the
+                // subsystem gate in PipeFill.cpp decides.
+                (void)target;
+                (void)externalIndex;
+                MG_Pipe::MGPipeMintTextureHandle(*this);
+                MG_Pipe::MGPipeEmitTextureResourceCreate(*this);
+#endif
             }
 
             TextureInternalFormat TextureObjectBase::GetFormat() const {
@@ -99,6 +214,15 @@ namespace MobileGL {
                 m_internalFormat = format;
                 BumpShapeVersion();
                 ++m_textureParamsVersion;
+                MGP_NOTE_AGGREGATE(TextureParams);
+#if MOBILEGL_PIPE_PUSH
+                // The format is BOTH halves: a descriptor field the backend allocates from and a
+                // parameter the swizzle / depth-stencil push reads, so both records move. It is
+                // also the last statement of both glTexBuffer entry points, which is what
+                // publishes a buffer texture's window without a call site in MG_Impl/GLImpl.
+                PipePublishDescriptor();
+                PipePublishParams();
+#endif
             }
 
             Uint TextureObjectBase::GetExternalIndex() const {
@@ -124,6 +248,10 @@ namespace MobileGL {
 
                 m_sampler->SetBorderColor(color);
                 ++m_textureParamsVersion;
+                MGP_NOTE_AGGREGATE(TextureParams);
+#if MOBILEGL_PIPE_PUSH
+                PipePublishParams();
+#endif
             }
 
             const IntVec4& TextureObjectBase::GetBorderColorI() const {
@@ -138,6 +266,10 @@ namespace MobileGL {
 
                 m_sampler->SetBorderColorI(color);
                 ++m_textureParamsVersion;
+                MGP_NOTE_AGGREGATE(TextureParams);
+#if MOBILEGL_PIPE_PUSH
+                PipePublishParams();
+#endif
             }
 
             const UintVec4& TextureObjectBase::GetBorderColorUI() const {
@@ -152,6 +284,10 @@ namespace MobileGL {
 
                 m_sampler->SetBorderColorUI(color);
                 ++m_textureParamsVersion;
+                MGP_NOTE_AGGREGATE(TextureParams);
+#if MOBILEGL_PIPE_PUSH
+                PipePublishParams();
+#endif
             }
 
             BorderColorForm TextureObjectBase::GetBorderColorForm() const {
@@ -201,6 +337,10 @@ namespace MobileGL {
                     break;
                 }
                 ++m_textureParamsVersion;
+                MGP_NOTE_AGGREGATE(TextureParams);
+#if MOBILEGL_PIPE_PUSH
+                PipePublishParams();
+#endif
             }
 
             void TextureObjectBase::SetSwizzleParamRGBA(const Vec4<TextureSwizzleParam>& values) {
@@ -208,6 +348,10 @@ namespace MobileGL {
 
                 m_swizzleParams = values;
                 ++m_textureParamsVersion;
+                MGP_NOTE_AGGREGATE(TextureParams);
+#if MOBILEGL_PIPE_PUSH
+                PipePublishParams();
+#endif
             }
 
             const UintVec2& TextureObjectBase::GetLevelRange() const {
@@ -225,7 +369,14 @@ namespace MobileGL {
                     m_levelRange.y() = m_levelRange.x();
                 }
                 ++m_textureParamsVersion;
+                MGP_NOTE_AGGREGATE(TextureParams);
                 BumpShapeVersion();
+#if MOBILEGL_PIPE_PUSH
+                // The level range is PARAMS, not a descriptor field: the record carries
+                // BaseLevel / MaxLevel and the storage is untouched. The descriptor is deduped on
+                // its own bytes, so the shape bump above costs nothing here.
+                PipePublishParams();
+#endif
             }
 
             void TextureObjectBase::SetMaxLevel(Uint maxLevel) {
@@ -236,7 +387,11 @@ namespace MobileGL {
 
                 m_levelRange.y() = maxLevel;
                 ++m_textureParamsVersion;
+                MGP_NOTE_AGGREGATE(TextureParams);
                 BumpShapeVersion();
+#if MOBILEGL_PIPE_PUSH
+                PipePublishParams();
+#endif
             }
 
             Bool TextureObjectBase::IsImmutable() const {
@@ -256,6 +411,15 @@ namespace MobileGL {
                     m_levelRange.y() = std::min(std::max(m_levelRange.y(), m_levelRange.x()), m_immutableLevels - 1);
                 }
                 ++m_textureParamsVersion;
+                MGP_NOTE_AGGREGATE(TextureParams);
+#if MOBILEGL_PIPE_PUSH
+                // Immutable is a descriptor fact the backend reads - and NOT a request for an
+                // acknowledgement: MGPipeResourceRespecifyNeedsAck names the buffer target
+                // explicitly, because texture allocation is lazy in monolith and stays lazy in
+                // split. The level clamp above is params.
+                PipePublishDescriptor();
+                PipePublishParams();
+#endif
             }
 
             Uint16 TextureObjectBase::GetTextureParamsVersion() const {
@@ -283,6 +447,7 @@ namespace MobileGL {
 
             void TextureObjectBase::BumpContentVersion() {
                 ++m_contentVersion;
+                MGP_NOTE_AGGREGATE(TextureContent);
             }
 
             Int TextureObjectBase::GetSamples() const {
@@ -292,6 +457,10 @@ namespace MobileGL {
             void TextureObjectBase::SetSamples(Int samples) {
                 m_samples = samples;
                 ++m_textureParamsVersion;
+                MGP_NOTE_AGGREGATE(TextureParams);
+#if MOBILEGL_PIPE_PUSH
+                PipePublishDescriptor();
+#endif
             }
 
             Bool TextureObjectBase::HasFixedSampleLocations() const {
@@ -301,6 +470,10 @@ namespace MobileGL {
             void TextureObjectBase::SetFixedSampleLocations(Bool fixedSampleLocations) {
                 m_fixedSampleLocations = fixedSampleLocations;
                 ++m_textureParamsVersion;
+                MGP_NOTE_AGGREGATE(TextureParams);
+#if MOBILEGL_PIPE_PUSH
+                PipePublishDescriptor();
+#endif
             }
 
             Uint64 TextureObjectBase::GetLifetimeId() const {
@@ -332,11 +505,27 @@ namespace MobileGL {
                                                              MipmapInput input) {
                 BumpShapeVersion();
                 m_textureStorage.AllocateLevel(GetIndexOfTextureUploadTarget(uploadTarget), mipmapLevel, input);
+#if MOBILEGL_PIPE_PUSH
+                // AFTER the allocation, never before: BumpShapeVersion runs first and a descriptor
+                // built there would describe the level set this call is about to change. Every
+                // storage-defining GL entry point - glTexImage*, glCompressedTexImage*,
+                // glTexStorage*, glTextureView and the generated-mip storage grow - reaches
+                // storage through here, which is what makes the emission complete without one call
+                // site per entry point in MG_Impl/GLImpl. AND IT NAMES THE LEVEL (final review
+                // C-1): this call replaced ONE level's storage, and only that level's pending
+                // upload may go with it.
+                PipePublishLevelDescriptor(uploadTarget, mipmapLevel);
+#endif
             }
 
             void TextureObjectWithOneMipmap::TruncateMipmapLevels(TextureUploadTarget uploadTarget, Uint levelCount) {
                 BumpShapeVersion();
                 m_textureStorage.TruncateToLevelCount(GetIndexOfTextureUploadTarget(uploadTarget), levelCount);
+#if MOBILEGL_PIPE_PUSH
+                // The levels at and above the cut are gone; the ones below keep their pending
+                // uploads (final review C-1).
+                PipePublishTruncatedDescriptor(uploadTarget, levelCount);
+#endif
             }
 
             void TextureObjectWithOneMipmap::UpdateMipmapSubData(TextureUploadTarget uploadTarget, Uint mipmapLevel,
@@ -352,8 +541,14 @@ namespace MobileGL {
                                                               Bool dirty) {
                 if (dirty) {
                     ++m_contentVersion;
+                    MGP_NOTE_AGGREGATE(TextureContent);
                 }
                 m_textureStorage.MarkDirty(GetIndexOfTextureUploadTarget(uploadTarget), mipmapLevel, dirty);
+#if MOBILEGL_PIPE_PUSH
+                // THE DRAIN LIST HAS NO CLEAN ARM (see TextureEmit.h): a level that goes clean
+                // is collected at the next drain, whose first test is IsStorageDirty.
+                if (dirty) PipeNoteLevelDirty(uploadTarget, mipmapLevel);
+#endif
             }
 
             Bool TextureObjectWithOneMipmap::IsStorageDirty(TextureUploadTarget uploadTarget, Uint mipmapLevel) const {
@@ -363,8 +558,16 @@ namespace MobileGL {
             void TextureObjectWithOneMipmap::MarkStorageDirtyRegion(TextureUploadTarget uploadTarget, Uint mipmapLevel,
                                                                     IntVec3 offset, IntVec3 size) {
                 ++m_contentVersion;
+                MGP_NOTE_AGGREGATE(TextureContent);
                 m_textureStorage.MarkDirtyRegion(GetIndexOfTextureUploadTarget(uploadTarget), mipmapLevel, offset,
                                                  size);
+#if MOBILEGL_PIPE_PUSH
+                // THE DRAIN LIST IS KEYED ON THE STORAGE OWNER FOR FREE: TextureObjectView
+                // forwards this call to the OWNER's method after remapping the level and the
+                // region, so an upload through a view and an upload through the owner arrive here
+                // on the same object with the same owner-side coordinates.
+                PipeNoteLevelDirty(uploadTarget, mipmapLevel);
+#endif
             }
 
             MipmapDirtyRegion TextureObjectWithOneMipmap::GetStorageDirtyRegion(TextureUploadTarget uploadTarget,

MobileGL/MG_State/GLState/TextureState/TextureObject.h

@@ -13,6 +13,7 @@
 #include "../SamplerState/SamplerObject.h"
 #include <Includes.h>
 #include <MG_Util/Math/VectorTypes.h>
+#include <MG_Pipe/PipeMutation.h>
 
 namespace MobileGL::MG_State::GLState {
     // Texture objects are always SharedPtr-owned (TextureState creates every instance via
@@ -116,7 +117,15 @@ namespace MobileGL::MG_State::GLState {
     class TextureObjectBase : public ITextureObject {
     public:
         TextureObjectBase(TextureTarget target, Uint externalIndex);
+#if MOBILEGL_PIPE_PUSH
+        // P2 step e2. Out of line, and declared only where there is a notice to raise: in a
+        // pull build this stays the implicit `= default` the pre-P2 tree had, which is what
+        // keeps the pull build's symbol set byte-for-byte the pre-P2 one (G1). Declared on the
+        // BASE, so every concrete texture class - 2D, 3D, cube, buffer, view - announces once.
+        virtual ~TextureObjectBase();
+#else
         virtual ~TextureObjectBase() = default;
+#endif
 
         TextureInternalFormat GetFormat() const override;
         TextureTarget GetTarget() const override;
@@ -180,6 +189,13 @@ namespace MobileGL::MG_State::GLState {
             if (m_depthStencilTextureMode == mode) return;
             m_depthStencilTextureMode = mode;
             ++m_textureParamsVersion;
+            MGP_NOTE_AGGREGATE(TextureParams);
+#if MOBILEGL_PIPE_PUSH
+            // D-E3's whole point, at the one site that proves it: the depth-stencil mode of a
+            // texture that is ONLY the READ framebuffer's attachment reaches the driver, because
+            // set_texture_params is addressed by resource and is independent of every binding.
+            PipePublishParams();
+#endif
         }
 
     protected:
@@ -191,6 +207,46 @@ namespace MobileGL::MG_State::GLState {
         // otherwise invisible to such a memo (no bind moved).
         void BumpShapeVersion();
 
+#if MOBILEGL_PIPE_PUSH
+        // ---- P4a's client emission points (brief D-D1, D-D3, D-E1, D-I1) ----
+        //
+        // NON-VIRTUAL AND PUSH-ONLY, both deliberately: a virtual would grow the vtable and a
+        // member would grow the object, and P4a's admitted-resize set is EMPTY - every edit
+        // that reaches the pull build is inside this guard, so the pull build's symbol set is
+        // byte-for-byte the one it had before the phase.
+        //
+        // DECLARED HERE AND DEFINED IN TextureObject.cpp, which calls the contract's own hooks
+        // in MG_Pipe/PipeMutation.h - the same door BufferObject.cpp uses, and no MG_State
+        // translation unit sees MG_Impl/Pipe/TextureEmit.h at all (c0b, ID-13). They stay
+        // members rather than free calls so the cube's, the view's and the buffer texture's
+        // translation units keep calling an inherited helper.
+        //
+        // resource_respecify, WHOLE-RESOURCE scope: the format setter and the three parameter
+        // setters that move a DESCRIPTOR field without moving the shape (immutable levels,
+        // sample count, fixed sample locations), and a view's creation. The emitter dedupes
+        // this form on the built descriptor, so an over-call costs one 88-byte compare and
+        // never an extra record.
+        void PipePublishDescriptor();
+        // The PER-LEVEL and the CHAIN-CUT forms of the same call (P4a final review C-1). The
+        // applier keeps a pending-upload set per (uploadTarget, level) and drops the entries
+        // against the storage a respecify REPLACES - and the descriptor cannot tell it which:
+        // AllocateStorage is per level and TruncateMipmapLevels removes a tail, while the
+        // descriptor carries the base extent and the level count only. So the storage entry
+        // points state the scope themselves; the whole-resource form above is for the calls
+        // that really redefine the whole store. A per-level form is NOT deduped on the
+        // descriptor: a non-base level redefined at a new size moves no descriptor field, and
+        // the applier's box against the old level has to go regardless.
+        void PipePublishLevelDescriptor(TextureUploadTarget uploadTarget, Uint mipmapLevel);
+        void PipePublishTruncatedDescriptor(TextureUploadTarget uploadTarget, Uint levelCount);
+        // set_texture_params, from every mutator that bumps m_textureParamsVersion.
+        void PipePublishParams();
+        // The sub-data DRAIN LIST's append, on a level's first dirty mark. There is no clean
+        // arm: the contract's hook (MG_Pipe/PipeMutation.h) carries no `dirty` flag, and a
+        // level that goes clean is collected at the next drain, where !IsStorageDirty is the
+        // first test EmitOneLevel makes.
+        void PipeNoteLevelDirty(TextureUploadTarget uploadTarget, Uint mipmapLevel);
+#endif
+
         const Uint m_externalIndex;
         const Uint64 m_lifetimeId;
         const TextureTarget m_target = TextureTarget::Unknown;

MobileGL/MG_State/GLState/TextureState/TextureObject2DCube.cpp

@@ -7,6 +7,7 @@
 // End of Source File Header
 
 #include "TextureObject2DCube.h"
+#include <MG_Pipe/PipeMutation.h>
 
 namespace MobileGL {
     namespace MG_State {
@@ -30,11 +31,21 @@ namespace MobileGL {
                                                       MipmapInput input) {
                 BumpShapeVersion();
                 m_textureStorage.AllocateLevel(GetIndexOfTextureUploadTarget(uploadTarget), mipmapLevel, input);
+#if MOBILEGL_PIPE_PUSH
+                // AFTER the allocation, for TextureObjectWithOneMipmap's reason: BumpShapeVersion
+                // runs first and a descriptor built there would describe the level set this call
+                // is about to change. The FACE rides in `uploadTarget`, so the key the emitter
+                // drops is that face's level and no other face's (final review C-1).
+                PipePublishLevelDescriptor(uploadTarget, mipmapLevel);
+#endif
             }
 
             void TextureObject2DCube::TruncateMipmapLevels(TextureUploadTarget uploadTarget, Uint levelCount) {
                 BumpShapeVersion();
                 m_textureStorage.TruncateToLevelCount(GetIndexOfTextureUploadTarget(uploadTarget), levelCount);
+#if MOBILEGL_PIPE_PUSH
+                PipePublishTruncatedDescriptor(uploadTarget, levelCount);
+#endif
             }
 
             void TextureObject2DCube::UpdateMipmapSubData(TextureUploadTarget uploadTarget, Uint mipmapLevel,
@@ -49,8 +60,14 @@ namespace MobileGL {
             void TextureObject2DCube::MarkStorageDirty(TextureUploadTarget uploadTarget, Uint mipmapLevel, bool dirty) {
                 if (dirty) {
                     ++m_contentVersion;
+                    MGP_NOTE_AGGREGATE(TextureContent);
                 }
                 m_textureStorage.MarkDirty(GetIndexOfTextureUploadTarget(uploadTarget), mipmapLevel, dirty);
+#if MOBILEGL_PIPE_PUSH
+                // SIX FACES, SIX BLOBS, SIX DRAIN KEYS: the upload target is the face, and it is
+                // what the sub-data record's Target byte carries beside the resource target.
+                if (dirty) PipeNoteLevelDirty(uploadTarget, mipmapLevel);
+#endif
             }
 
             bool TextureObject2DCube::IsStorageDirty(TextureUploadTarget uploadTarget, Uint mipmapLevel) const {
@@ -60,8 +77,12 @@ namespace MobileGL {
             void TextureObject2DCube::MarkStorageDirtyRegion(TextureUploadTarget uploadTarget, Uint mipmapLevel,
                                                              IntVec3 offset, IntVec3 size) {
                 ++m_contentVersion;
+                MGP_NOTE_AGGREGATE(TextureContent);
                 m_textureStorage.MarkDirtyRegion(GetIndexOfTextureUploadTarget(uploadTarget), mipmapLevel, offset,
                                                  size);
+#if MOBILEGL_PIPE_PUSH
+                PipeNoteLevelDirty(uploadTarget, mipmapLevel);
+#endif
             }
 
             MipmapDirtyRegion TextureObject2DCube::GetStorageDirtyRegion(TextureUploadTarget uploadTarget,

MobileGL/MG_State/GLState/TextureState/TextureObjectBuffer.h

@@ -29,6 +29,15 @@ namespace MobileGL {
                 void SetBufferRange(SizeT offset, SizeT size) {
                     m_bufferRangeOffset = offset;
                     m_bufferRangeSize = size;
+#if MOBILEGL_PIPE_PUSH
+                    // Both glTexBuffer entry points bind the backing buffer and then set the
+                    // window, so this is the first statement at which the descriptor's
+                    // BufferForTexBuffer / BufOffset / BufSize trio is complete. SetInternalFormat
+                    // publishes again one statement later and is deduped away when the format did
+                    // not move - which is exactly the case a re-attach of a DIFFERENT buffer at
+                    // the same format would otherwise fall through.
+                    PipePublishDescriptor();
+#endif
                 }
                 SizeT GetBufferRangeOffset() const { return m_bufferRangeOffset; }
                 // Resolved against the buffer's current size, so kWholeBuffer tracks it.

MobileGL/MG_State/GLState/TextureState/TextureObjectView.cpp

@@ -79,6 +79,18 @@ namespace MobileGL::MG_State::GLState {
         // to the view. GetImmutableLevels() forwards to the owner for the actual GL query, which
         // GL 4.6 core 8.18 defines as the ORIGINAL texture's value.
         SetImmutableLevels(numLevels);
+#if MOBILEGL_PIPE_PUSH
+        // ViewOf, and it is published from HERE rather than left to SetImmutableLevels above,
+        // which early-returns when the composed level count happens to be the base class's
+        // current value - a degenerate view (the spec's min() composition narrowed to zero
+        // levels) would otherwise never publish the one field that makes it a view. The
+        // descriptor is deduped on its own bytes, so the ordinary case pays one compare.
+        //
+        // ONE HOP ALWAYS REACHES STORAGE: glTextureView composes a view-of-a-view onto the ROOT
+        // at creation, which is what the spec's additive min-level rule describes and what the
+        // assertion above pins, so the owner named here is never itself a view.
+        PipePublishDescriptor();
+#endif
     }
 
     Uint TextureObjectView::GetImmutableLevels() const {

MobileGL/MG_State/GLState/TextureState/TextureState.h

@@ -33,6 +33,15 @@ namespace MobileGL::MG_State::GLState {
             Access = access;
             Format = format;
             ++Version;
+#if MOBILEGL_PIPE_PUSH
+            // P4a D-A4 / final review M-A: the EARLIEST producer of kMGPipeBindShaderImage. The
+            // ImageBindableHint the bit feeds is the prevention half of the texture-remint stall
+            // class (a texture the server knows may be image-bound is allocated image-bindable
+            // up front), so it has to reach the applier before the texture's first sync - at
+            // the bind, not at the next validate point's image walk. Push-only through the
+            // contract's door, like every other hook in this directory (G1).
+            if (Texture) MG_Pipe::MGPipeNoteTextureImageBound(*Texture);
+#endif
         }
     };
 
@@ -140,7 +149,22 @@ namespace MobileGL::MG_State::GLState {
         // again (the unit tests do exactly that between cases).
         Uint64 GetContextId() const { return m_contextId; }
 
+#if MOBILEGL_PIPE_PUSH
+        // P2 brief D4, the two texture aggregates. CONTENT is an upload or a dirty region;
+        // PARAMS is a glTexParameter or a glSamplerParameter. They are separate because
+        // NEW_SAMPLER_VIEWS and NEW_SAMPLERS are separate dirty bits and a Minecraft frame
+        // moves them at wildly different rates.
+        void NoteTextureContentChanged() { ++m_anyTextureContentGeneration; }
+        Uint64 GetAnyTextureContentGeneration() const { return m_anyTextureContentGeneration; }
+        void NoteTextureParamsChanged() { ++m_anyTextureParamsGeneration; }
+        Uint64 GetAnyTextureParamsGeneration() const { return m_anyTextureParamsGeneration; }
+#endif
+
     private:
+#if MOBILEGL_PIPE_PUSH
+        Uint64 m_anyTextureContentGeneration = 0;
+        Uint64 m_anyTextureParamsGeneration = 0;
+#endif
         static Uint64 AllocateContextId();
 
         const Uint64 m_contextId;

MobileGL/MG_State/GLState/VertexArrayState/VertexArrayObject.cpp

@@ -9,6 +9,7 @@
 #include "VertexArrayObject.h"
 
 #include <atomic>
+#include <MG_Pipe/PipeMutation.h>
 
 namespace MobileGL::MG_State::GLState {
     // Starts at 1 so a zero-initialized memo slot can never carry a live object's id.
@@ -37,6 +38,29 @@ namespace MobileGL::MG_State::GLState {
         }
     }
 
+#if MOBILEGL_PIPE_PUSH
+    VertexArrayObject::~VertexArrayObject() {
+        // P2 step e2 / P3a C-1: ANNOUNCE the death instead of leaving the backend to discover
+        // it in a garbage sweep, and RETURN THE CLIENT'S OWN SLOT while doing it. This is the
+        // last SharedPtr to this object dropping - not the glDelete* that only marks the name
+        // and leaves a still-bound object very much alive - so it is the exact moment the
+        // backend's twin, the driver storage that twin owns, the applier's vertex-elements
+        // record and the {slot, gen} that names all three stop being reachable.
+        //
+        // ALL FOUR OF THOSE GO THROUGH ONE HELPER, and it is the client's rather than a
+        // backend's: the handle is minted by MGPipeVertexInputEmitter on every backend, so a
+        // death path that only exists inside a backend's death-ops table is no path at all
+        // under a backend that installs none - which is what DirectVulkan/Magma does on
+        // purpose, and what made every VAO leak a slot and a ~1.3 KB applier record for the
+        // life of the process on the shipped mask. The helper emits delete_vertex_elements,
+        // raises the notice (a no-op unless a backend registered the ops) and frees the slot,
+        // in that fixed order; MG_Pipe/PipeMutation.h and its definition say why each position
+        // is where it is. The buffer's death has exactly this shape one file over
+        // (BufferObject.cpp -> MGPipeEmitResourceDestroyAndFree).
+        MG_Pipe::MGPipeEmitVertexElementsDestroyAndFree(m_lifetimeId);
+    }
+#endif
+
     void VertexArrayObject::EnableAttribute(Uint index) {
         if (index >= MAX_VERTEX_ATTRIBS) return;
 
@@ -297,18 +321,21 @@ namespace MobileGL::MG_State::GLState {
         if (index >= MAX_VERTEX_ATTRIBS) return;
         ++m_attributeVersions[index].FormatVersion;
         ++m_configVersion;
+        MGP_NOTE_AGGREGATE(VaoAttribute);
     }
 
     void VertexArrayObject::BumpAttributeBufferVersion(Uint index) {
         if (index >= MAX_VERTEX_ATTRIBS) return;
         ++m_attributeVersions[index].BufferVersion;
         ++m_configVersion;
+        MGP_NOTE_AGGREGATE(VaoAttribute);
     }
 
     void VertexArrayObject::BumpAttributeSwitchVersion(Uint index) {
         if (index >= MAX_VERTEX_ATTRIBS) return;
         ++m_attributeVersions[index].SwitchVersion;
         ++m_configVersion;
+        MGP_NOTE_AGGREGATE(VaoAttribute);
     }
 
     const VertexAttributeVersion& VertexArrayObject::GetAttributeVersion(Uint index) const {

MobileGL/MG_State/GLState/VertexArrayState/VertexArrayObject.h

@@ -25,6 +25,12 @@ namespace MobileGL {
                 static constexpr int MAX_VERTEX_ATTRIB_BINDINGS = 32;
 
                 VertexArrayObject(Uint externIndex);
+#if MOBILEGL_PIPE_PUSH
+                // P2 step e2. Out of line, and declared only where there is a notice to raise:
+                // in a pull build this class keeps its implicit destructor, which is what keeps
+                // the pull build's symbol set byte-for-byte the pre-P2 one (G1).
+                ~VertexArrayObject();
+#endif
 
                 void EnableAttribute(Uint index);
                 void DisableAttribute(Uint index);
@@ -100,6 +106,22 @@ namespace MobileGL {
                 // "any vertex-input state changed" with one compare.
                 Uint32 GetConfigVersion() const { return m_configVersion; }
 
+#if MOBILEGL_PIPE_LEGACY_MEMOS
+                // ---- THE BACKEND'S THREE MEMOS ON THE FRONTEND OBJECT ----
+                //
+                // P2 D12.5 (ARCHITECTURE.md 9.5) retires all three: a frontend state object
+                // must not hold the backend's raw pointers, and under split it cannot - the
+                // backend is in another process and its cache entry has no address the client
+                // could store. Magma's handle arm keeps the same three facts in a slot-indexed
+                // table it owns itself (VertexInputStateFactory::VaoBackendMemos), keyed on the
+                // VAO's {slot, gen} and validated by the same config version, so nothing is
+                // recomputed more often than it was.
+                //
+                // They stay compiled under MOBILEGL_PIPE_LEGACY_MEMOS - which a PULL build
+                // forces ON - because that is the arm the pre-handle A/B runs, and because G1
+                // admits no change to the pull build. They are deleted outright with the pull
+                // path at P13.
+                //
                 // Backend-owned content-hash memo, valid while the config version matches
                 // (same idea as ProgramObject's hash memo — avoids re-hashing all
                 // attributes on every draw).
@@ -148,6 +170,7 @@ namespace MobileGL {
                     m_backendAuxMemo1 = aux1;
                     m_backendAuxMemoVersion = m_configVersion;
                 }
+#endif // MOBILEGL_PIPE_LEGACY_MEMOS
 
             private:
                 void BumpAttributeFormatVersion(Uint index);
@@ -187,6 +210,10 @@ namespace MobileGL {
                 Array<Bool, MAX_VERTEX_ATTRIBS> m_attributeUsesBindingModel = {};
 
                 Uint32 m_configVersion = 0;
+#if MOBILEGL_PIPE_LEGACY_MEMOS
+                // The storage behind the three accessors above; retired with them (D12.5).
+                // A pull build forces MOBILEGL_PIPE_LEGACY_MEMOS ON, so sizeof(this) does not
+                // move there and G1 sees no change.
                 mutable Uint64 m_backendHashMemo = 0;
                 mutable Uint32 m_backendHashMemoVersion = ~0u;
                 mutable const void* m_backendStateMemo = nullptr;
@@ -195,6 +222,7 @@ namespace MobileGL {
                 mutable Uint64 m_backendAuxMemo0 = 0;
                 mutable Uint64 m_backendAuxMemo1 = 0;
                 mutable Uint32 m_backendAuxMemoVersion = ~0u;
+#endif // MOBILEGL_PIPE_LEGACY_MEMOS
             };
         } // namespace GLState
     } // namespace MG_State