Repository navigation
Conversation
This was referenced Oct 31, 2026
MarcZX10
suggested changes
Oct 31, 2026
MarcZX10
left a comment
MarcZX10
left a comment
Contributor
There was a problem hiding this comment.
Reviewed — changes requested
The linked final-head GPU report records the new agc_driver_native_depth_initial_owner test exiting with Windows status 0xc0000409, so the added regression test has not demonstrated a passing result on the submitted head.
What I ran
Reviewed exact head b5829490036c0760bc5eeca0c3090e2e0c85c121 against base ca073ec3ad0d766ea9f4658e418a5c977001a353. Individual no-build merge previews against current main were clean; the stack report records declared merge-order dependency #12862, whose head is not included. Environment: Linux x86-64, Python 3.11.2, Git 2.39.5. I did not build or run the C++ test locally.
python3 /root/.codex/skills/anyps5-pr-review/scripts/pr_snapshot.py 14084 --repo boykopovar/AnyPS5
python3 tools/pr_precheck.py --repo boykopovar/AnyPS5 --pr 14084 --output-dir /tmp/anyps5-14084-review
python3 tools/pr_stack_check.py 14084 --repo boykopovar/AnyPS5 --output-dir /tmp/anyps5-14084-stack-extra
curl -L --fail --silent --show-error https://gist.githubusercontent.com/Amoenus/87d6c1bdbfac4dca23171afd7ce59602/raw/ -o /tmp/anyps5-14084-report-gist.txt
grep -n -E 'initial_owner|0xc0000409|tests FAILED' /tmp/anyps5-14084-report-gist.txtResults
| Check | Result |
|---|---|
| Diff and whitespace | Passed. |
| Conventions and PR template | Passed locally and hosted. |
| Current-main merge previews | Both PR #12862 and #14084 individually merge cleanly; #12862's head is not included in #14084, consistent with the declared merge-order dependency. |
| ABI audit | Not applicable to these paths; local helper unavailable in isolated PR snapshot. |
| Test-quality heuristic | Could not run because its helper is absent from the isolated PR snapshot. |
Linked final-head --initial-owner CTest report |
Failed: Exit code 0xc0000409 on Windows, 0/1 tests passed. |
| Hosted CI | Conventions, eligibility, intake, labels, portability and report passed; Windows job was in progress at snapshot. Product Linux and specialized jobs were skipped. |
Unedited linked test output
1/1 Test #1734: agc_driver_native_depth_initial_owner ...Exit code 0xc0000409***Exception: 0.85 sec
0% tests passed, 1 tests failed out of 1
The following tests FAILED:
1734 - agc_driver_native_depth_initial_owner (Exit code 0xc0000409)
Errors while running CTest
Negative control
| Assertion | Change to code under test | Outcome |
|---|---|---|
| The regression detects the old behavior | Use main production code with the new fixture | Author reports the control fails at the expected refusal on two main configurations; the report includes a final-head run that crashes before passing. |
Findings
Blocking
The PR description says the new case passes on both GPU vendors, but the linked final-head raw log ends with agc_driver_native_depth_initial_owner failing with Windows status 0xc0000409. That is a process failure, not a passing regression run. Please diagnose the crash and provide a passing run of the exact current head on the affected setup (or explain with reproducible evidence why this log is unrelated to the claimed final-head runs).
Non-blocking
None.
Sources
- Exact-head production change and regression fixture
- Linked test report and raw output
- Current hosted checks
Not verified
I did not build or run the C++ fixture locally, independently reproduce the Windows crash, or verify the author's GPU runs. Product Linux and specialized hosted jobs were skipped; hosted Windows was still running at the last snapshot. The linked report includes vendor-specific GPU and Vulkan validation evidence, but I did not rerun those measurements.
To get this to a good state
Resolve the 0xc0000409 failure and publish a passing exact-head run, including the command and raw output. I will rerun the no-build precheck and review the new head.
Contributor
Author
|
@MarcZX10 I made the gist navigation ambiguous: its bare The crashing log belongs to recorded head 74689a2, SHA256 Exact submitted head b582949 passing records:
The bare gist raw URL now returns an evidence index; use the named raw files above for test output. Every linked file was downloaded/hash-verified. This corrects presentation, not test outcomes or source: head remains b582949, and these are the already-completed exact-head executions. The hosted Windows job has since passed. Skipped Linux/specialized jobs remain untested, and no independent Linux execution is claimed. Please re-review this unchanged head using the explicit final-head evidence. AI-assisted:yes(Codex);author responsible. |
MarcZX10
approved these changes
Oct 31, 2026
MarcZX10
left a comment
MarcZX10
left a comment
Contributor
There was a problem hiding this comment.
Re-checked — new head, looks good
The evidence-index correction resolves the reported failure: the immutable logs for the exact reviewed head show the targeted test passing on both listed GPU vendors.
What I ran
Reviewed head b5829490036c0760bc5eeca0c3090e2e0c85c121, PR base ca073ec3ad0d766ea9f4658e418a5c977001a353. The PR remains at that head. I fetched both linked raw logs and confirmed their SHA-256 values match the manifest; I did not run the GPU tests locally.
python3 - <<'PY'
import hashlib, urllib.request
urls = [
'https://gist.githubusercontent.com/Amoenus/87d6c1bdbfac4dca23171afd7ce59602/raw/8d1a942e729c7efd28e060e62beb7456c6bd3ee1/native-depth-initial-owner-mainca073-20261031-candidate-nvidia-agc_driver_native_depth_initial_owner-qualified-v1.log',
'https://gist.githubusercontent.com/Amoenus/87d6c1bdbfac4dca23171afd7ce59602/raw/ee0093cbce2aeb3dad501acb4ecba221763ca093/native-depth-initial-owner-mainca073-20261031-candidate-amd-agc_driver_native_depth_initial_owner-qualified-v1.log',
'https://gist.githubusercontent.com/Amoenus/87d6c1bdbfac4dca23171afd7ce59602/raw/7463a0c2b2be1b4e0d2d58b1ecf42f988777516c/01-final-head-manifest.json',
]
for url in urls:
data = urllib.request.urlopen(url, timeout=20).read()
print(len(data), hashlib.sha256(data).hexdigest())
PY
gh api repos/boykopovar/AnyPS5/commits/b5829490036c0760bc5eeca0c3090e2e0c85c121/check-runsResults
| Check | Result |
|---|---|
| Exact-head NVIDIA GPU log | Author-reported test passes, 24 cases; fetched bytes match manifest SHA-256 |
| Exact-head AMD GPU log | Author-reported test passes, 24 cases; fetched bytes match manifest SHA-256 |
| Hosted Windows job | Passed |
| Hosted conventions, eligibility, labels, library-portability, and report checks | Passed |
| Hosted Linux, relinker, ASan, register-context, and write-watch jobs | Skipped |
Fetched log output
nvidia bytes= 48160 sha256= a6fc39c26d331a3cb801555e780a4a5200b8359034040972608b63912a6f084a
Start 1734: agc_driver_native_depth_initial_owner
1/1 Test #1734: agc_driver_native_depth_initial_owner ... Passed 0.78 sec
100% tests passed, 0 tests failed out of 1
amd bytes= 49050 sha256= 8bb7c6be36b02fbd124c2bccf856465363b7533da3c3856f205f8b7391a83f83
Start 1734: agc_driver_native_depth_initial_owner
1/1 Test #1734: agc_driver_native_depth_initial_owner ... Passed 0.61 sec
100% tests passed, 0 tests failed out of 1
Findings
Blocking
None found in this re-check. The prior blocker was the ambiguous raw gist URL; the corrected manifest and named immutable logs identify the crashing run as an earlier fixture setup and provide passing records for the submitted head.
Non-blocking
The GPU runs are author-provided evidence, not independently reproduced. The PR also notes that a Sonic Unleashed replay remains pending; this approval does not assert title-specific validation.
Sources
- Reviewed source and test changes at exact head
- Exact-head evidence manifest
- NVIDIA test log
- AMD test log
Not verified
I did not independently run the GPU tests, a Linux build/test, console behavior, or the pending Sonic Unleashed replay. The PR's #12862 registration relocation remains an explicitly declared merge-order coordination item; its individual preview against the stated base is clean, but I did not perform a combined integration build.
Collaborator
|
@Amoenus, reviewed at Please return |
|
Other open pull requests touch the same code. Merging one will break the other, or both implement the same thing. If this one builds on them, list them in Depends on.
Some comparisons could not be completed because a pull request conflicts with the base branch. Review these shared files manually before concluding that there is no overlap.
|
Contributor
Author
|
@oneandonlydean addressed in 03dae7c619. Verified: NVIDIA and AMD each24/0 pass; temporary test-only capability controls produce0/24 → direct exit77 and CTest Skipped, and partial4/20 → pass. Controls removed; submitted source restored/rebuilt and passes again. Required Vulkan and loaded Khronos1.4.363 verified, zero validation warnings/errors/VUIDs. Production code/registration unchanged; conventions pass. Pinned report, controls and raw logs. After #12862 merged,I refreshed to e269d111c4 on main121b using its resolver. Registration/properties preserved;production/fixture code unchanged. Both GPUs freshly pass24/0;pinned evidence. Depends on:none now. This also resolves the live merge conflict;Exact-head hosted Windows job passed; Linux/specialized jobs were skipped. @MarcZX10 I also reconciled the direct-control log hash you noted. The local capture uses CRLF ( |
MarcZX10
approved these changes
Oct 31, 2026
MarcZX10
left a comment
MarcZX10
left a comment
Contributor
There was a problem hiding this comment.
Re-checked — new head, looks good
The test-only update now skips the initial-owner fixture only when the device supports none of its depth/sample combinations, and retains coverage for supported combinations.
What I ran
Reviewed exact head 03dae7c6195042f2087ced2b6ecfec84d52e2c36 against base ca073ec3ad0d766ea9f4658e418a5c977001a353; the merge preview is clean. Environment: Debian Linux x86_64, Python 3.11.2, GCC 12.2.0 (c++). I did not build or run the GPU test locally. I downloaded and hash-checked the author-provided pinned report and CTest logs.
git diff --check ca073ec3ad0d766ea9f4658e418a5c977001a353...03dae7c6195042f2087ced2b6ecfec84d52e2c36
python3 tools/pr_precheck.py --repo boykopovar/AnyPS5 --pr 14084
python3 - <<'PY'
import hashlib, json, urllib.request
files = json.load(urllib.request.urlopen("https://api.github.com/gists/87d6c1bdbfac4dca23171afd7ce59602"))["files"]
for name in ["pr14084-skip-review-20261031-final-nvidia.log", "pr14084-skip-review-20261031-final-amd.log", "pr14084-skip-review-20261031-control-zero-v3.log", "pr14084-skip-review-20261031-control-partial-v3.log", "pr14084-skip-review-20261031-control-zero-direct-v3.log"]:
data = urllib.request.urlopen(files[name]["raw_url"]).read()
print(name, hashlib.sha256(data).hexdigest())
PYResults
| Check | Result |
|---|---|
| Diff check | Passed. The new head changes only NativeMultisampleDepth.cpp; the production fix and CTest registration are unchanged from the previously approved head. |
| Conventions and PR template | Passed. |
| Merge preview | Clean against current main. |
| Author-provided NVIDIA and AMD CTest runs | Both logs match the published hashes; each reports checked=24 unavailable=0, with zero VUID- or Validation Error strings. These were run by the author on Windows 11 with NVIDIA RTX 5080 and AMD Radeon Graphics 1002:13c0; I did not run them. |
| Author-provided capability controls | The pinned CTest zero-capability log matches its published hash and reports 0 checked / 24 unavailable with CTest Skipped. The partial-capability run matches its hash and reports 4 checked / 20 unavailable, passing. |
| Precheck helper results | Diff, conventions/template and dependency check passed. ABI audit was not applicable to this test-only diff; the test-quality helper was unavailable in the isolated PR worktree. I manually confirmed the test registration and SKIP_RETURN_CODE 77. |
| Hosted CI for this head | Report passed; conventions, eligibility, labels and library portability were pending at the final refresh. |
Unedited excerpts from the downloaded CTest logs
1734: Native depth initial owner cases checked=24 unavailable=0
1/1 Test #1734: agc_driver_native_depth_initial_owner ... Passed 0.38 sec
100% tests passed, 0 tests failed out of 1
1734: Native depth initial owner cases checked=0 unavailable=24
1/1 Test #1734: agc_driver_native_depth_initial_owner ...***Skipped 0.25 sec
The following tests did not run:
1734 - agc_driver_native_depth_initial_owner (Skipped)
Negative control
| Assertion | Change made for the control | Result |
|---|---|---|
| No supported depth/sample combination must skip rather than fail | Author temporarily forced all capability checks to report unsupported | Author reports direct exit 77 and CTest skipped; the CTest log was hash-verified. |
| Partial device support still exercises available combinations | Author temporarily limited support to D32/4 samples | Author reports 4 checks passed and 20 unavailable; the CTest log was hash-verified. |
Findings
Blocking
None found in the new head.
Non-blocking
The results.json manifest lists SHA256 4d9dda362ef223423ce9b3073f3c219e4534219c8e5a3dd23aab5c19c82182e8 for control-zero-direct-v3.log, while the currently linked raw file hashes to 37fd58c29f2bc3f4d393f2c19c0a7fdee0295ae300a2a47c2d3e87b5849c0e7a. The CTest skip log itself matches its manifest hash and confirms the required skip behavior; please reconcile the direct-log hash in the report artifacts when convenient.
Sources
- Changed fixture at the reviewed head
- CTest skip configuration at the reviewed head
- Pinned author report, NVIDIA log, AMD log, and zero-capability CTest log.
Not verified
I did not build or run the fixture locally, use a GPU or console, or independently reproduce the author's Windows test runs. The hosted checks still pending above were not treated as passed.
MarcZX10
approved these changes
Oct 31, 2026
MarcZX10
left a comment
MarcZX10
left a comment
Contributor
There was a problem hiding this comment.
Re-checked — new head, looks good
The refreshed branch preserves the reviewed depth-clear fix, and the new-head regression evidence passes on both reported GPUs.
What I ran
Reviewed exact head e269d111c4d72747e8864ad603fef730e2ce1247 against base 121b2e37af75278e3d84a2f43869730030b112ed; GitHub reports it mergeable. Local environment: Debian 12, Python 3.11.2, GCC 12.2.0. I did not build or run GPU tests locally.
python3 tools/pr_precheck.py --repo boykopovar/AnyPS5 --pr 14084
sha256sum /tmp/anyps5-14084-main121b-nvidia.log /tmp/anyps5-14084-main121b-amd.log
gh pr checks 14084 --repo boykopovar/AnyPS5Results
| Check | Result |
|---|---|
| Diff check, conventions, dependency/mergeability | Passed; merge preview is clean. |
| Hosted Windows job | Passed on this head. |
| Author-supplied exact-head GPU logs | SHA-256 matched the supplied manifest; NVIDIA RTX 5080 and AMD Radeon Graphics each report 24 checked, 0 unavailable, and CTest passed. |
| Hosted Linux and specialized jobs | Skipped. |
| Local ABI and test-quality helpers | Not executable in the isolated PR checkout because those local helper scripts are absent there; this is a tooling limitation, not a PR failure. |
Unedited relevant output
All conventions checks passed.
9cde0a96c9593bb26878d09cbc050a40cb531d2ae870627d096a9ab40c5ac448 /tmp/anyps5-14084-main121b-nvidia.log
04b18ad591cd1b63841d8d40714d18091d41d584fa8b2856a67b6c44b3a9dba1 /tmp/anyps5-14084-main121b-amd.log
Native depth initial owner cases checked=24 unavailable=0
1/1 Test #1556: agc_driver_native_depth_initial_owner ... Passed 0.36 sec
Native depth initial owner cases checked=24 unavailable=0
1/1 Test #1556: agc_driver_native_depth_initial_owner ... Passed 0.29 sec
windows pass 4m51s https://github.com/boykopovar/AnyPS5/actions/runs/41235257138/job/124638538959
linux skipping 0 https://github.com/boykopovar/AnyPS5/actions/runs/41235257138/job/124638540222
Findings
Blocking
None.
Non-blocking
No new issues found in this re-check.
Sources
- Depth-surface guard at the reviewed commit
- Initial-owner test and unsupported-device skip at the reviewed commit
- NVIDIA raw log, pinned gist revision
- AMD raw log, pinned gist revision
Not verified
I did not build or run GPU tests locally. Linux product tests, specialized jobs, title execution, and physical-console behavior were not verified.
Contributor
✅ NVIDIA GPU test: passedPR head
Failing on main too (3, not counted against this PR)Validated but not compared (1, not in the main baseline's validation pass)Validation IDs already present on main (5 IDs in 480 tests)Skipped (44)ReproduceAutomated build and test on a contributor's NVIDIA machine; it does not replace upstream CI. A new comment is posted for each new push. Full logs. |
Contributor
Author
|
@JAB-dev I checked the updated overlap warning against #13021 I propose merging this small independent #14084 first, then rebasing #13021 and its #13023 child while preserving both early dispatch paths and this fresh-owner initialization guard. Would that order work for you? This is a proposal; I have not assumed agreement or added your work to this branch. The new independent Linux/NVIDIA result is linked in Tested, with the remaining suite/timing caveats retained. |
Supplemental Linux/RADV validation: initial depth clear with a previous ownerTested the current head on a discrete Radeon with RADV, supplementing the existing Windows NVIDIA/AMD and Linux NVIDIA evidence. This includes a real test-only negative control, not a main run where the new test is absent. Pins: main
The new test actually executes all 24 cases per fixed run, with no capability exclusions: D16/D32 x 2/4/8 samples x clear 0/1 x the two HTILE encodings. Its readback/reuse/subsequent-write assertions all complete. This is 72 completed cases across the three independent fixed runs, not a zero-case success. The negative and reversed controls fail specifically with: The fixed runs print: The five existing tests are Vulkan validation: serial runs, no CTest skips/timeouts, VUIDs, Khronos validation errors/warnings, or sync hazards in any selected phase. Every selected test has confirmed real RX 9060 XT device selection and Khronos layer insertion in its own verbose output. The loader reports 16 Khronos insertions per six-test suite pass because some tests create multiple instances; this is not a claim of 16 separate CTest processes. Neither side logs Environment: CachyOS Linux Environment caveat: both trees equally emit loader errors about a pre-existing implicit Reproduction and test-source identityFor the negative controls use the pinned main with only these two test files from this head, under
cmake -S . -B build-main -G Ninja -DCMAKE_BUILD_TYPE=Release -DBUILD_TESTING=ON -DANYPS5_ENABLE_SPIRV_TOOLS=OFF
cmake --build build-main --parallel 4 --target agc_driver_native_multisample_depth_tests
export ANYPS5_REQUIRE_VULKAN=1 ANYPS5_NO_SHADER_CACHE=1
export VK_INSTANCE_LAYERS=VK_LAYER_KHRONOS_validation VK_LOADER_DEBUG=layer
for repeat in 1 2 3; do
ctest --test-dir build-main -R '^(agc_driver_native_multisample_depth|agc_driver_native_depth_(initial_owner|retirement|clears|clear_refusals|buffer_clear))$' --parallel 1 --no-tests=error -V > "depth-${repeat}.log" 2>&1
doneSet Scope: no title, console comparison, performance, or broader guest-import compatibility claim. This does not identify the old owner's type in the motivating title or establish a title unblock. The temporary test overlays were removed; the checkout and selected binaries were restored to main. AI-assisted hardware test run and report: yes (Codex). |
Contributor
✅ NVIDIA GPU test: passedPR head
Intermittent validation messages (1, seen once in the parallel run, not reproduced when the test was re-run alone 2 times; not counted against this PR)Unstable on main (2, these also failed in recent main runs on this machine; not counted against this PR)Failing on main too (3, not counted against this PR)Passed on retry (1, failed in the parallel run, passed when re-run with a 360 s timeout; not counted against this PR)Validated but not compared (1, not in the main baseline's validation pass)Validation IDs already present on main (4 IDs in 495 tests)Skipped (45)ReproduceAutomated build and test on a contributor's NVIDIA machine; it does not replace upstream CI. A new comment is posted for each new push. Full logs. |
Contributor
✅ NVIDIA GPU test: passedPR head
Unstable on main (3, these also failed in recent main runs on this machine; not counted against this PR)Failing on main too (3, not counted against this PR)Validated but not compared (1, not in the main baseline's validation pass)Validation IDs already present on main (4 IDs in 517 tests)Skipped (45)ReproduceAutomated build and test on a contributor's NVIDIA machine; it does not replace upstream CI. A new comment is posted for each new push. Full logs. |
Contributor
✅ AMD GPU test: passedPR head
Unstable on main (1, these also failed in recent main runs on this machine; not counted against this PR)Failing on main too (8, not counted against this PR)Validation IDs already present on main (4 IDs in 526 tests)Skipped (58)ReproduceAutomated build and test on a contributor's AMD machine; it does not replace upstream CI. A new comment is posted for each new push. Full logs. |
Contributor
✅ NVIDIA GPU test: passedPR head
Unstable on main (3, these also failed in recent main runs on this machine; not counted against this PR)Failing on main too (3, not counted against this PR)Validation IDs already present on main (4 IDs in 518 tests)Skipped (48)ReproduceAutomated build and test on a contributor's NVIDIA machine; it does not replace upstream CI. A new comment is posted for each new push. Full logs. |
Contributor
✅ AMD GPU test: passedPR head
Unstable on main (1, these also failed in recent main runs on this machine; not counted against this PR)Failing on main too (6, not counted against this PR)Validated but not compared (53, not in the main baseline's validation pass)Validation IDs already present on main (4 IDs in 500 tests)Skipped (57)ReproduceAutomated build and test on a contributor's AMD machine; it does not replace upstream CI. A new comment is posted for each new push. Full logs. |
Contributor
✅ NVIDIA GPU test: passedPR head
Unstable on main (4, these also failed in recent main runs on this machine; not counted against this PR)Failing on main too (3, not counted against this PR)Validation IDs already present on main (13 IDs in 548 tests)Skipped (49)ReproduceAutomated build and test on a contributor's NVIDIA machine; it does not replace upstream CI. A new comment is posted for each new push. Full logs. |
Contributor
Author
|
6 November full-CI refresh request: current The newest independent NVIDIA report shows no new attributed failure at old728b6d9; it is not execution of this final head. The narrow green hosted Windows job skipped full builds/AGC CTest. Please run full hosted Linux/Windows CI on the refreshed head when possible; upstream permissions remain read-only. AI-assisted: yes (Codex). |
12 tasks done
Contributor
✅ NVIDIA GPU test: passedPR head
Unstable on main (2, these also failed in recent main runs on this machine; not counted against this PR)Failing on main too (3, not counted against this PR)Validation IDs already present on main (13 IDs in 548 tests)Skipped (49)ReproduceAutomated build and test on a contributor's NVIDIA machine; it does not replace upstream CI. A new comment is posted for each new push. Full logs. |
Contributor
✅ NVIDIA GPU test: passedPR head
Unstable on main (4, these also failed in recent main runs on this machine; not counted against this PR)Failing on main too (3, not counted against this PR)Validation IDs already present on main (9 IDs in 553 tests)Skipped (56)ReproduceAutomated build and test on a contributor's NVIDIA machine; it does not replace upstream CI. A new comment is posted for each new push. Full logs. |
Contributor
✅ AMD GPU test: passedPR head
Unstable on main (1, these also failed in recent main runs on this machine; not counted against this PR)Failing on main too (8, not counted against this PR)Validated but not compared (1, not in the main baseline's validation pass)Validation IDs already present on main (4 IDs in 547 tests)Skipped (65)ReproduceAutomated build and test on a contributor's AMD machine; it does not replace upstream CI. A new comment is posted for each new push. Full logs. |
Contributor
✅ NVIDIA GPU test: passedPR head
Unstable on main (3, these also failed in recent main runs on this machine; not counted against this PR)Failing on main too (3, not counted against this PR)Passed on retry (1, failed in the parallel run, passed when re-run with a 360 s timeout; not counted against this PR)Validation IDs already present on main (9 IDs in 589 tests)Skipped (54)ReproduceAutomated build and test on a contributor's NVIDIA machine; it does not replace upstream CI. A new comment is posted for each new push. Full logs. |
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
A fresh D16/D32 multisample depth attachment can pass every initial HTILE/import check, yet fail at its first currentness check because an older attachment owns the same depth address.
DepthSurfaceViewrecordsnative multisample depth fast-clear transition is not implementedeven though the constructor has already initialized the new image and the full-clear ownership path does not copy the old depth plane. #14083Keep that proven initialization for created && samplingActive surfaces. The one-line production change retains refusal for existing surfaces and ineligible imports, first-refusal preservation, complete HTILE proof, generation/watch/race checks and unsupported later transitions. No title-specific condition, fallback, retry, environment switch or new export.
Builds on the existing initial-import contract (#5496/#5554) and JJenkx's merged fresh-plane correction #6438. Vulkan host-clear contract; no new console HTILE semantics or hardware-equivalence claim. Related #6300/#6808 retain the old bind refusal. #7256 moves that same condition into a layered helper; retain this correction during that rebase. No code from those open PRs is included.
The regression is isolated from shared helper signatures/property lists. #12862 has merged;the registration is now in its required executable fragment,relocated with the documented resolver. The earlier overlap scan/reviews remain source-pinned history,not an assertion that all peer combinations were tested. #7256 layered-rebase order proposal remains separate;no author agreement is presumed.
Review snapshot — 6 November:
de8f61508aa9a9f327c7e719b71d2352083d722dincludes tested main07fae3d172f4ffcdd8fb0eed76a475a8630e716d. The proven initial-depth-owner correction is unchanged; published history is retained. Current qualification.Tested
Final head
de8f61508aa9a9f327c7e719b71d2352083d722dis based on main07fae3d172f4ffcdd8fb0eed76a475a8630e716d. The first pinned refresh at main6185bdcfc4e05bbc5a7b61a74ba6345f941a2b6epassed 16 physical-GPU processes; the focused follow-up on this final head passes 8, with the exact group/source breakdown in the sweep report, commands, raw logs and hashes. Both use RTX5080 and AMD iGPU on the same machine. Main controls pass32 processes at the first pin and26 at the final pin; absent new capability registrations are not counted as main passes.The final selection covers the new initial-owner case, native-depth publication, depth-only aliasing and depth clears; the initial refresh also runs all four #15964 regressions and stencil/native-depth controls. The source correction is unchanged. The latest independent NVIDIA result at old728b6d9 reports no new attributed failures, not execution of this final head.
Full configured native libraries/relinker and the selected real execution targets build. Required Vulkan, app shader cache/journal disabled, serial verbose CTest with30-second bound, loaded Khronos1.4.363 and actual device identity verified per GPU process. No skips, VUIDs, SYNC diagnostics or validation warnings/errors. Selected CPU checks, conventions and whitespace pass. Added/removed topic lines are unchanged by these mechanical updates; no new implementation or title/FPS claim. The newer main changes shared f16 probing/target/request serialization, which is why the focused follow-up was run.
Earlier qualification/fault controls retain their original pins: evidence 1. They are not rerun-control claims at this new head.
The green hosted Windows job is a narrow registration/link/changed-guest-test check; full build, libs and AGC CTest were skipped. Full hosted Linux/Windows validation remains a maintainer action, and older independent reports retain their actual heads.
Checklist
main; no other open PR implements the same functionsNotImplemented_nid_no_patch); silent stubs are listed under Silent stubs in their library's technical debt file.mdfiles or images added to the repository (attach them to this PR)