Skip to content

CI: add st-a5 and st-sim-a5 DFX per-feature smoke steps. - #1278

Merged
ChaoZheng109 merged 1 commit into
hw-native-sys:mainfrom
yanghaoran29:fix/issue-1249-a5-dfx-ci-smoke
Jul 8, 2026
Merged

ChaoZheng109 merged 1 commit into
hw-native-sys:mainfrom
yanghaoran29:fix/issue-1249-a5-dfx-ci-smoke

Conversation

@yanghaoran29

@yanghaoran29 yanghaoran29 commented Jul 6, 2026 •

Copy link
Copy Markdown
Contributor

Fixes #1249

The a2a3/a2a3sim job already runs four dedicated DFX smoke steps (dep_gen, l2_swimlane, PMU, args_dump) with the matching --enable-* flags. The a5/a5sim job only ran the default pytest sweep, so a5 DFX scene tests were collected but their capture pipelines were never validated in sim CI.

Mirror the st-a2a3/st-sim-a2a3 smoke set in st-a5/st-sim-a5 with a5/a5sim paths and flags. graphviz is already installed in the a5/a5sim job for dep_gen.

Items A.1-A.3 (PMU artifact validation hardening in test_pmu.py) were already landed on main; this commit closes the remaining sim CI gap.

Also update a5's DeviceRunner::finalize_collectors() to use finalize() instead of stop() for diagnostics collectors, matching a2a3's implementation. This ensures shared memory is properly released on any exit from run() — success or early error — preventing stale state issues when workers are reused across runs (e.g., pytest session-scoped worker pool). Add scope_stats_collector_ cleanup with args.scope_stats_data_base reset.

@gemini-code-assist

Copy link
Copy Markdown

Note

Gemini is unable to generate a review for this pull request due to the file types involved not being currently supported.

@coderabbitai

coderabbitai Bot commented Jul 6, 2026 •

Copy link
Copy Markdown

Review Change Stack

Important

Review skipped

Auto incremental reviews are disabled on this repository.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: ae03a145-0ea9-4e3f-96e8-428a25382c29

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

  • 🔍 Trigger review
📝 Walkthrough

Walkthrough

Four new pytest smoke-test steps (dep_gen, l2_swimlane, PMU, args_dump) were added to the st-sim-a5 job in the CI workflow, mirroring the existing st-sim-a2a3 coverage with corresponding --enable-* flags and platform/device settings.

Changes

CI Smoke Coverage

Layer / File(s) Summary
Add DFX per-feature smoke steps
.github/workflows/ci.yml
Four pytest steps for dep_gen, l2_swimlane, PMU, and args_dump added to st-sim-a5 job with feature-specific --enable-* flags and a5sim platform/device settings.

Estimated code review effort: 1 (Trivial) | ~5 minutes

Poem

A rabbit hops through CI's green fields,
Four new smoke steps, now sim yields.
dep_gen, swimlane, PMU, args galore,
a5's coverage matches what a2a3 wore. 🐇✅

🚥 Pre-merge checks | ✅ 5
✅ Passed checks (5 passed)
Check name Status Explanation
Linked Issues check ✅ Passed The diff adds the requested st-sim-a5 DFX smoke steps, and the PMU hardening items are noted as already on main.
Out of Scope Changes check ✅ Passed The only change is the CI workflow update needed for the requested smoke steps; nothing unrelated was introduced.
Docstring Coverage ✅ Passed No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Title check ✅ Passed The title clearly reflects the main change: adding DFX per-feature smoke steps to the CI workflow.
Description check ✅ Passed The description is related to the PR and explains the CI coverage gap and intended smoke-step additions.

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

Comment @coderabbitai help to get the list of available commands.

@yanghaoran29
yanghaoran29 force-pushed the fix/issue-1249-a5-dfx-ci-smoke branch 3 times, most recently from c2a26fd to 0a28d23 Compare July 7, 2026 03:59
@yanghaoran29 yanghaoran29 changed the title CI: add st-sim-a5 DFX per-feature smoke steps CI: add st-a5 and st-sim-a5 DFX per-feature smoke steps. Jul 7, 2026
@yanghaoran29
yanghaoran29 force-pushed the fix/issue-1249-a5-dfx-ci-smoke branch 5 times, most recently from d9861a9 to e00c1b7 Compare July 7, 2026 11:14
Comment thread src/a5/platform/onboard/host/device_runner.cpp
@yanghaoran29
yanghaoran29 force-pushed the fix/issue-1249-a5-dfx-ci-smoke branch from e00c1b7 to ff475e7 Compare July 8, 2026 06:50
…m L2 swimlane rotation table and finalize collectors

Fixes hw-native-sys#1249

The a2a3/a2a3sim job already runs four dedicated DFX smoke steps (dep_gen, l2_swimlane, PMU, args_dump) with the matching --enable-* flags. The a5/a5sim job only ran the default pytest sweep, so a5 DFX scene tests were collected but their capture pipelines were never validated in sim CI.

Mirror the st-a2a3/st-sim-a2a3 smoke set in st-a5/st-sim-a5 with a5/a5sim paths and flags. graphviz is already installed in the a5/a5sim job for dep_gen.

Items A.1-A.3 (PMU artifact validation hardening in test_pmu.py) were already landed on main; this commit closes the remaining sim CI gap.

Also update a5's DeviceRunner::finalize_collectors() to use finalize() instead of stop() for diagnostics collectors, matching a2a3's implementation. This ensures shared memory is properly released on any exit from run() — success or early error — preventing stale state issues when workers are reused across runs (e.g., pytest session-scoped worker pool). Add scope_stats_collector_ cleanup with args.scope_stats_data_base reset.

a5 sim DeviceRunner never published l2_swimlane_aicore_rotation_table to the AICPU .so, so AICPU init could not fill per-core head slots and get_l2_swimlane_aicore_head() returned null → segfault in record_task.

Also replace stop-only perf_cleanup with finalize_collectors(), matching a2a3 sim and the onboard fix, so session-scoped Workers can re-init DFX collectors across runs.
@ChaoZheng109
ChaoZheng109 merged commit 32f2080 into hw-native-sys:main Jul 8, 2026
16 checks passed
doraemonmj pushed a commit to doraemonmj/simpler_wc that referenced this pull request Jul 14, 2026
…m L2 swimlane rotation table and finalize collectors (hw-native-sys#1278)

Fixes hw-native-sys#1249

The a2a3/a2a3sim job already runs four dedicated DFX smoke steps (dep_gen, l2_swimlane, PMU, args_dump) with the matching --enable-* flags. The a5/a5sim job only ran the default pytest sweep, so a5 DFX scene tests were collected but their capture pipelines were never validated in sim CI.

Mirror the st-a2a3/st-sim-a2a3 smoke set in st-a5/st-sim-a5 with a5/a5sim paths and flags. graphviz is already installed in the a5/a5sim job for dep_gen.

Items A.1-A.3 (PMU artifact validation hardening in test_pmu.py) were already landed on main; this commit closes the remaining sim CI gap.

Also update a5's DeviceRunner::finalize_collectors() to use finalize() instead of stop() for diagnostics collectors, matching a2a3's implementation. This ensures shared memory is properly released on any exit from run() — success or early error — preventing stale state issues when workers are reused across runs (e.g., pytest session-scoped worker pool). Add scope_stats_collector_ cleanup with args.scope_stats_data_base reset.

a5 sim DeviceRunner never published l2_swimlane_aicore_rotation_table to the AICPU .so, so AICPU init could not fill per-core head slots and get_l2_swimlane_aicore_head() returned null → segfault in record_task.

Also replace stop-only perf_cleanup with finalize_collectors(), matching a2a3 sim and the onboard fix, so session-scoped Workers can re-init DFX collectors across runs.
@yanghaoran29
yanghaoran29 deleted the fix/issue-1249-a5-dfx-ci-smoke branch July 25, 2026 08:04
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[Code Health] Harden a5 DFX scene tests: PMU artifact validation gaps + missing st-sim-a5 smoke coverage

2 participants