CI: run A5 SDMA tests on x86

[doraemonmj]

Summary

• Run the full A5 onboard scene-test corpus, including SDMA, on x86_64 and ARM64 runners.
• Remove the A5 x86_64 not sdma marker filter while preserving the A2/A3 SDMA isolation policy.
• Update CI and testing documentation to match the runner capability.

Testing

• Pre-commit hooks passed.
• Workflow YAML and Markdown lint passed.
• Hardware tests not run locally; the updated A5 CI job exercises the full corpus.

[coderabbitai]

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 Plus

Run ID: a6893cda-44f4-477d-8fd0-e67f0c73382e

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

A5 scene-test execution now runs the complete pytest corpus, including SDMA tests, on x86_64 and ARM64. The workflow, CI documentation, testing guide, and testing skill now use one consistent command.

Changes

A5 SDMA execution

Layer / File(s)	Summary
Run the full A5 test corpus .github/workflows/_st-npu-a5.yml	The workflow submits pytest directly through task-submit without the x86_64 sdma exclusion.
Align CI and testing guidance .claude/skills/testing/SKILL.md, docs/ci.md, docs/testing.md	The guidance states that A5 runs the complete corpus on both architectures. It retains the separate SDMA step for a2a3.
Estimated code review effort: 2 (Simple)	~10 minutes

Possibly related PRs

• hw-native-sys/simpler#1709: Changes the A5 x86_64 SDMA exclusion that this PR removes.
• hw-native-sys/simpler#1373: Modifies A5 scene-test execution to use task-submit.
• hw-native-sys/simpler#1609: Changes SDMA test handling and quarantine guidance for related CI flows.

Poem

A5 runs every test in line,
SDMA joins the pytest shine.
x86 and ARM agree,
One command hops so happily.
— A reviewing rabbit 🐇

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately identifies the A5 SDMA CI change, but it omits the ARM64 scope and full-corpus change.
Description check	✅ Passed	The description clearly covers the A5 SDMA CI change, preserved A2/A3 policy, documentation updates, and testing status.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

---

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

• X
• Mastodon
• Reddit
• LinkedIn

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

[ChaoZheng109]

The A5 marker-selection wording is still inconsistent in two testing skills:

• .claude/skills/test-all-device/SKILL.md says st-onboard-a5 has no marker filter.
• .claude/skills/test-runtime-device/SKILL.md says the same.

After this PR, the A5 workflow still uses -m "not pod"; it only stops excluding sdma. Saying there is no marker filter can make local/agent reproductions run the pod tests in the ordinary single-machine A5 sweep.

Could we update both skills to say something like: st-onboard-a5 uses -m "not pod" and does not exclude sdma?

[doraemonmj]

@ChaoZheng109 Addressed — both testing skills now state that st-onboard-a5 uses -m "not pod" without excluding sdma, and include matching A5 command examples.