Docs: keep the log-level split where both files say it is

[ChaoWao]

Summary

docs/logging.md opens with

For the user-facing model — one Python knob and the CLI flags — see testing.md § Log levels. This file documents the implementation.

and docs/testing.md § Log levels points back for implementation details. The split is declared in both directions. Neither file held it:

• testing.md carried the simpler_log_init / HostLogger / InitArgs chain, the cann_level() → CANN mapping, and the fork-time threshold propagation — all mechanism, in the file that says it is not about mechanism.
• logging.md restated the one-shot snapshot semantics that testing.md also spelled out.

That is the shape that produces a doc which silently lies: #1475 retired the verbose V-tiers and had two places to land.

After

testing.md answers only what a test author needs — the integer layout, the macro→level table, how to set the threshold, and the two consequences:

• the threshold is snapshotted once at Worker.init(), so logger.setLevel(...) afterwards does not reach a live worker;
• onboard AICPU severity comes from CANN, and a pre-set ASCEND_GLOBAL_LOG_LEVEL wins over the Python logger.

logging.md gains what moved, under ## Configuration flow:

• the CANN mapping as a table (timing and warn both → CANN WARN, which is why the default threshold does not open CANN's INFO stream);
• the ASCEND_GLOBAL_LOG_LEVEL-wins rule;
• forked chip subprocesses, which were documented nowhere else: L3/L4 parents snapshot the threshold before fork() and pass it explicitly to _chip_process_loop, rather than letting the child re-read a logger it never configured.

Not a size reduction

testing.md loses 8 lines (788 → 780); logging.md gains 30 (282 → 312). The gain is that one topic now has one owner per aspect. testing.md is still 780 lines / 16 H2 and well past the soft target in .claude/rules/doc-consistency.md §6 — splitting it is a separate decision, not folded in here.

Testing

• mkdocs build --strict passes
• markdownlint-cli2 --config tests/lint/.markdownlint.yaml — 0 errors on both files
• Both cross-links still resolve: logging.md#configuration-flow and testing.md#log-levels (neither heading renamed)
• Every moved claim checked against source — dlog_setlevel / cann_level() in src/common/platform/*/host/c_api_shared.cpp, the --log-level choices in conftest.py

Docs only.

[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: f8c29960-ada8-474f-aee9-8d6ffc270593

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

Logging documentation now describes worker threshold propagation and CANN-owned onboard severity. Testing documentation reorganizes log-level configuration guidance, adds CLI examples, and links detailed behavior to the logging configuration documentation.

Changes

Logging documentation

Layer / File(s)	Summary
Configuration flow documentation docs/logging.md	Documents threshold propagation to forked chip subprocesses, CANN severity ownership, environment precedence, and Simple-to-CANN level mapping.
Testing log-level guidance docs/testing.md	Adds log-level CLI examples, clarifies one-time threshold snapshots and output routing, and links to the logging configuration flow.

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

Possibly related PRs

• hw-native-sys/simpler#1409: Updates the default ASCEND_GLOBAL_LOG_LEVEL for multi-round timing runs, matching this PR’s CANN severity and TIMING documentation.

Poem

I’m a rabbit with logs in a row,
Watching thresholds fork and flow.
CANN keeps onboard levels bright,
TIMING rests at WARN tonight.
Docs now guide each debugging hop!

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly reflects the documentation split between logging.md and testing.md.
Description check	✅ Passed	The description matches the docs reorganization and cross-linking described in the changeset.
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.}

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@docs/logging.md`:
- Around line 278-281: Update the dlog_setlevel guidance in the CANN logging
documentation to state that it must be called before the device context is
opened; calls made after device initialization do not affect device-side
logging. Keep the existing ASCEND_GLOBAL_LOG_LEVEL and dlog_setlevel
configuration details unchanged.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro Plus

Run ID: cbf21621-9dcd-4819-be36-1bc047976a53

📥 Commits

Reviewing files that changed from the base of the PR and between ca22a8f and 55d6a54.

📒 Files selected for processing (2)

• docs/logging.md
• docs/testing.md