marfrit/marfrit-packages
0
0
Fork 1
Code Issues 2 Pull Requests Actions Packages Projects Releases Wiki Activity

mesa-panvk-bifrost: r4 -> r5 — advertise fragmentStoresAndAtomics on Bifrost (closes panvk-bifrost#2) #92

Merged
marfrit merged 1 commits from claude-noether/marfrit-packages:noether/mesa-panvk-bifrost-r5 into main 2026-05-23 12:19:11 +00:00
Conversation 0 Commits 1 Files Changed 2
+66 -1
claude-noether commented 2026-05-23 12:05:10 +00:00
Contributor
Copy Link
Copy Source

Summary

mesa-panvk-bifrost r4 → r5: backports Mesa main's unconditional .fragmentStoresAndAtomics = true flip for panvk. Closes the Dawn WebGPU adapter rejection at PhysicalDeviceVk.cpp:250 that was causing brave-vulkan to fall back to SwiftShader CPU adapter on Mali-G52. Tracks marfrit/panvk-bifrost#2.

The disjunction with instance->force_enable_shader_atomics is kept as a documented kill-switch (compiler folds true || X to true but the link to the DRI option pan_force_enable_shader_atomics stays visible for rebases / downstream debugging).

Phase 7 verify (on ohm, PineTab2 / RK3566 / Mali-G52 r1 MC1)

Locally-built r5 lib installed to a test path (/tmp/r5_test_lib/) and exercised vs the existing r4 production lib:

Subset r4 r5 Delta
dEQP-VK.glsl.atomic_operations.* 48 P / 0 F / 992 NS 80 P / 0 F / 960 NS +32 newly-passing, 0 new failures
dEQP-VK.image.store.* 2772 P / 0 F / 238 NS 2772 P / 0 F / 238 NS identical

vulkaninfo on the r5 test lib confirms fragmentStoresAndAtomics = true (was false on r4). vertexPipelineStoresAndAtomics remains false per upstream gating.

Process

Followed 8-step bugfix-process per feedback-bugfix-process memory:

  1. Re-analyzed: the cited Dawn warning IS load-bearing for WebGPU HW-accel.
  2. Prior art researched: Mesa main flipped this same line unconditional, our 26.0.6 base lags.
  3. Drafted patch (kept disjunction per reviewer guidance).
  4. 2nd-model review completed — recommended keeping || force_enable_shader_atomics, suggested CTS subset, advised against narrower Bifrost-only gate.
  5. Implemented (this PR).
  6. Deploy + test: built locally on ohm, installed to test path.
  7. Confirm: dEQP subsets show zero regressions, 32 newly-passing.
  8. Documentation: this PR body + memory update + closes-issue link.

Test plan

  • CI builds + signs + publishes mesa-panvk-bifrost-26.0.6.r5-1-aarch64.pkg.tar.xz
  • pacman -Sy && pacman -Si mesa-panvk-bifrost on ohm shows r5 version (three-point ship check per feedback-package-done-means-installable)
  • vulkaninfo on installed r5 reports fragmentStoresAndAtomics = true
  • Original symptom from issue #2 (the "fragmentStoresAndAtomics feature required" Dawn warning) no longer appears in brave-vulkan stderr

🤖 Generated with Claude Code

## Summary `mesa-panvk-bifrost` r4 → r5: backports Mesa main's unconditional `.fragmentStoresAndAtomics = true` flip for panvk. Closes the Dawn WebGPU adapter rejection at `PhysicalDeviceVk.cpp:250` that was causing brave-vulkan to fall back to SwiftShader CPU adapter on Mali-G52. Tracks [marfrit/panvk-bifrost#2](https://git.reauktion.de/marfrit/panvk-bifrost/issues/2). The disjunction with `instance->force_enable_shader_atomics` is kept as a documented kill-switch (compiler folds `true || X` to `true` but the link to the DRI option `pan_force_enable_shader_atomics` stays visible for rebases / downstream debugging). ## Phase 7 verify (on ohm, PineTab2 / RK3566 / Mali-G52 r1 MC1) Locally-built r5 lib installed to a test path (`/tmp/r5_test_lib/`) and exercised vs the existing r4 production lib: | Subset | r4 | r5 | Delta | |---|---|---|---| | `dEQP-VK.glsl.atomic_operations.*` | 48 P / 0 F / 992 NS | 80 P / 0 F / 960 NS | **+32 newly-passing, 0 new failures** | | `dEQP-VK.image.store.*` | 2772 P / 0 F / 238 NS | 2772 P / 0 F / 238 NS | identical | `vulkaninfo` on the r5 test lib confirms `fragmentStoresAndAtomics = true` (was `false` on r4). `vertexPipelineStoresAndAtomics` remains `false` per upstream gating. ## Process Followed 8-step bugfix-process per `feedback-bugfix-process` memory: 1. Re-analyzed: the cited Dawn warning IS load-bearing for WebGPU HW-accel. 2. Prior art researched: Mesa main flipped this same line unconditional, our 26.0.6 base lags. 3. Drafted patch (kept disjunction per reviewer guidance). 4. **2nd-model review** completed — recommended keeping `|| force_enable_shader_atomics`, suggested CTS subset, advised against narrower Bifrost-only gate. 5. Implemented (this PR). 6. Deploy + test: built locally on ohm, installed to test path. 7. Confirm: dEQP subsets show zero regressions, 32 newly-passing. 8. Documentation: this PR body + memory update + closes-issue link. ## Test plan - [ ] CI builds + signs + publishes `mesa-panvk-bifrost-26.0.6.r5-1-aarch64.pkg.tar.xz` - [ ] `pacman -Sy && pacman -Si mesa-panvk-bifrost` on ohm shows r5 version (three-point ship check per `feedback-package-done-means-installable`) - [ ] `vulkaninfo` on installed r5 reports `fragmentStoresAndAtomics = true` - [ ] Original symptom from issue #2 (the "fragmentStoresAndAtomics feature required" Dawn warning) no longer appears in brave-vulkan stderr 🤖 Generated with [Claude Code](https://claude.com/claude-code)
claude-noether added 1 commit 2026-05-23 12:05:11 +00:00
mesa-panvk-bifrost: r4 -> r5 — advertise fragmentStoresAndAtomics on Bifrost 6896853544 
Backports Mesa main's unconditional flip of .fragmentStoresAndAtomics
to true in src/panfrost/vulkan/panvk_vX_physical_device.c. Closes
the Dawn-WebGPU adapter rejection at PhysicalDeviceVk.cpp:250 that
caused brave-vulkan to fall back to the SwiftShader CPU adapter on
PineTab2/Mali-G52, per marfrit/panvk-bifrost#2.

Phase 7 verify on ohm (PineTab2, RK3566, Mali-G52 r1 MC1) with a
locally-built r5 lib installed to /tmp/r5_test_lib/:

  dEQP-VK.glsl.atomic_operations.*:
    r4:  48 pass /   0 fail / 992 NotSupported  (1040 total)
    r5:  80 pass /   0 fail / 960 NotSupported  (1040 total)
    delta: +32 newly-passing, zero new failures

  dEQP-VK.image.store.*:
    r4: 2772 pass /  0 fail / 238 NotSupported  (3010 total)
    r5: 2772 pass /  0 fail / 238 NotSupported  (3010 total)
    delta: identical (image.store is independent of the flag)

The disjunction with instance->force_enable_shader_atomics is kept as
a documented kill-switch even though the compiler folds it away —
it leaves the DRI option pan_force_enable_shader_atomics semantically
wired for future rebases or downstream debugging.

Patch reviewed via 2nd-model pass (per bugfix-process step 4):
recommended keeping the disjunction (applied), Bifrost-only-vs-unconditional
left unconditional to match upstream (applied), pre-ship CTS subset
(applied with results above).

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
marfrit merged commit 685f85c22e into main 2026-05-23 12:19:11 +00:00
marfrit deleted branch noether/mesa-panvk-bifrost-r5 2026-05-23 12:19:11 +00:00
Sign in to join this conversation.
Reviewers
No Reviewers
Labels
No items
No Label
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
marfrit (Markus Fritsche)
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: marfrit/marfrit-packages#92
Delete Branch "claude-noether/marfrit-packages:noether/mesa-panvk-bifrost-r5"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?