[MG_IntegrationTest, CI, Docs] (Disaggregated): P5e - Magma's named-blit entries get a lane of their own

- Both DirectVulkan.Split.NamedBlit entries carried LABELS "integration-split" and abort under
  strict on GetFramebufferBindingSlot@Clear (VulkanRenderer.cpp:7648). P5e does not unbarrier that
  field on Magma - it retires there in P7 - so the Espryt lane could not go hard green with them
  in it.
- ARGUED RATHER THAN ASSUMED, because the brief allowed either reading: admitting the row instead
  is the WORSE option. FieldOwnership.def carries ONE retiring-phase string for both backends
  ("P5e (Espryt unbarriered), P7 (Magma)"), so a rule that admitted the pair for Magma would admit
  it for Espryt too - and an Espryt Clear reading the frontend's binding slot again is exactly the
  regression fb's handle arm exists to prevent. The lane split costs two entries; the admission
  would cost the gate.
- THE LABEL IS `integration-magma-split`, NOT `integration-split-magma`, and the spelling is
  load-bearing: `ctest -L` takes a REGEX, so `-L integration-split` matches a label spelled
  `integration-split-magma` as a substring. Measured with the obvious name first: both DirectVulkan
  entries were still listed by `ctest -N -L integration-split`, i.e. the split looked done in every
  listing and had not happened.
- Its CI step asserts the run is RED **and** that a private log carries the named marker, so
  "Magma is expected red" cannot decay into "Magma did not run". Measured: 0/2 under strict with
  two logs carrying Fatal{UnmigratedPipeInput, "GetFramebufferBindingSlot@Clear"}, and 2/2 green
  without the knob. integration-split is 117 and integration-magma-split is 2.
- CONTRACT-P5E section 7 states the split and the reason it is not an allowlist entry.

.github/workflows/test.yml

@@ -1341,6 +1341,45 @@ jobs:
           fi
           echo "integration-split under strict is GREEN, two-sided, with the arming control present"
 
+      # ---- P5e (gl), CONTRACT-P5E §7: MAGMA'S OWN LANE, EXPECTED RED -----------------------
+      #
+      # The two DirectVulkan.Split.NamedBlit entries abort on GetFramebufferBindingSlot@Clear
+      # (VulkanRenderer.cpp:7648). That field retires in P5e ON ESPRYT and in P7 ON MAGMA, and
+      # FieldOwnership.def carries ONE retiring-phase string for both - so admitting the pair to
+      # keep them in the Espryt lane would forgive an Espryt Clear reading the frontend's binding
+      # slot again, which is exactly the regression fb's handle arm exists to prevent. Splitting
+      # the lane costs two entries; admitting the row would cost the gate.
+      #
+      # THE LABEL IS `integration-magma-split` AND THE SPELLING MATTERS: `ctest -L` is a REGEX, so
+      # a label named `integration-split-magma` is still matched by `-L integration-split` and the
+      # entries would never have left. Measured before the rename.
+      #
+      # IT IS ASSERTED RED, WITH ITS MARKER. "Magma is red" must not be allowed to decay into
+      # "Magma did not run": the step fails both when the run goes green (P7 landed early, or the
+      # entries stopped drawing) and when it goes red without the named abort in a private log.
+      - name: Integration-magma-split under strict - expected red until P7 (CONTRACT-P5E section 7)
+        working-directory: build-split
+        env:
+          MOBILEGL_ITEST_REQUIRE_GPU: "1"
+          MOBILEGL_IPC_STRICT_ERRORS: "1"
+          MOBILEGL_TRANSPORT: "inproc"
+        run: |
+          ulimit -c unlimited
+          rc=0
+          ctest -L integration-magma-split --no-tests=error -j 2 > "${RUNNER_TEMP}/strict-magma.txt" 2>&1 || rc=$?
+          tail -5 "${RUNNER_TEMP}/strict-magma.txt"
+          if [ "${rc}" -eq 0 ]; then
+            echo "::error::integration-magma-split went GREEN under strict. Magma is in lockstep for the whole of P5e (ID-90) and GetFramebufferBindingSlot retires there in P7, so these entries are expected to abort. If P7 has landed, delete this step and fold the entries back into integration-split."
+            exit 1
+          fi
+          marker='Fatal{UnmigratedPipeInput, "GetFramebufferBindingSlot@Clear"}'
+          if ! grep -rqF "${marker}" MobileGL/MG_IntegrationTest/p5b-DirectVulkan-*.log 2>/dev/null; then
+            echo "::error::integration-magma-split is red but no private log carries ${marker} - the lane failed for some OTHER reason, which is a defect rather than the expected-red control"
+            grep -rhoE '(Fatal|Admitted)\{[A-Za-z]*, "[^"]*"\}' MobileGL/MG_IntegrationTest/p5b-DirectVulkan-*.log 2>/dev/null | sort | uniq -c | sort -rn | head
+            exit 1
+          fi
+          echo "integration-magma-split is red by design on ${marker}"
+
       - name: Upload strict lane logs
         if: always()
         uses: actions/upload-artifact@v7

MobileGL/MG_IntegrationTest/CMakeLists.txt

@@ -2257,19 +2257,45 @@ if (MOBILEGL_BUILD_DISAGGREGATED)
 endif()
 
 # P5b measured named-blit blockers, exercising both backend bound-form lowerings.
+#
+# P5e (gl), CONTRACT-P5E §7: THE TWO BACKENDS GET DIFFERENT LABELS, because under strict they are
+# answering different questions.
+#
+# Both DirectVulkan entries abort on GetFramebufferBindingSlot@Clear (VulkanRenderer.cpp:7648).
+# That field's row retires in P5e ON ESPRYT and in P7 ON MAGMA, and FieldOwnership.def carries ONE
+# retiring-phase string for both - so the derived allowlist cannot admit the pair without admitting
+# it for Espryt too, where it would forgive exactly the regression fb's handle arm exists to
+# prevent (an Espryt Clear reading the frontend's binding slot again would become a warning line).
+# The alternative to a label is therefore not "admit one row"; it is "stop the Espryt lane from
+# being able to see that row at all", which is a real loss of gate strength.
+#
+# So the lane is SPLIT rather than the allowlist WIDENED: `integration-split` stays a statement
+# about Espryt and goes hard green, and the Magma entries get the expected-red step the contract
+# already promises them for the backend it keeps in lockstep all phase (ID-90).
 if (MOBILEGL_BUILD_DISAGGREGATED)
     foreach(namedBlitBackend DirectGLES DirectVulkan)
         mgl_itest_join_environment(MGL_ITEST_NAMED_BLIT_ENVIRONMENT
             "MOBILEGL_BACKEND_TYPE=${namedBlitBackend}" "MOBILEGL_TRANSPORT=inproc"
             "MOBILEGL_ITEST_REQUIRE_GPU=1" "MGITEST_SPLIT_LANE=1"
             ${MGL_ITEST_CAPABILITY_ENV} ${MGL_ITEST_COMMON_ENV})
+        # THE LABEL IS `integration-magma-split` AND NOT `integration-split-magma`, and that is not
+        # a preference: `ctest -L` takes a REGEX, not an exact name, so `-L integration-split`
+        # MATCHES a label spelled `integration-split-magma` as a substring. The obvious name would
+        # have left both entries in the Espryt lane while every listing showed them relabelled - a
+        # split that looks done and is not. Measured: with that spelling
+        # `ctest -N -L integration-split` still listed both DirectVulkan entries.
+        if (namedBlitBackend STREQUAL "DirectVulkan")
+            set(MGL_ITEST_NAMED_BLIT_LABELS "integration-magma-split")
+        else()
+            set(MGL_ITEST_NAMED_BLIT_LABELS "integration-split")
+        endif()
         foreach(namedBlitCase NamedBlitPreservesBindingsAndRestoresNextVerbsPixels NamedBlitDefaultEndpointPixels)
             gtest_discover_tests(MobileGLIntegrationTest
                 TEST_PREFIX "${namedBlitBackend}.Split.NamedBlit."
                 TEST_FILTER "F1WireScenario.${namedBlitCase}"
                 DISCOVERY_TIMEOUT 30
                 PROPERTIES
-                    LABELS "integration-split"
+                    LABELS "${MGL_ITEST_NAMED_BLIT_LABELS}"
                     TIMEOUT ${MGL_ITEST_TIMEOUT}
                     ENVIRONMENT "${MGL_ITEST_NAMED_BLIT_ENVIRONMENT}\;MOBILEGL_LOG_FILE_PATH=${CMAKE_CURRENT_BINARY_DIR}/p5b-${namedBlitBackend}-${namedBlitCase}.log")
         endforeach()

MobileGL/MG_Remote/CONTRACT-P5E.md

@@ -518,6 +518,12 @@ and asserts `vbs > 0` — the server stamped a verb boundary. Everything strict
 of that stamp and the monolith arm never stamps, so without it "the lane is green" and "strict was
 never armed" are the same observation.
 
+**Magma keeps the expected-red step, in a lane of its own.** The two `DirectVulkan.Split.NamedBlit`
+entries carry `integration-magma-split` (so spelled: `ctest -L` is a regex and
+`-L integration-split` matches `integration-split-magma`) and are asserted RED on
+`Fatal{UnmigratedPipeInput, "GetFramebufferBindingSlot@Clear"}`. Admitting that pair instead would
+forgive the same read on Espryt, where it is this phase's debt — one retiring-phase string serves
+both backends.
 
 ---