branch-4.1: [fix](cloud) Use bthread-aware shared mutex for tablet header lock

[liaoxin01]

Pick #64574

[hello-stephen]

Thank you for your contribution to Apache Doris.
Don't know what should be done next? See How to process your PR.

Please clearly describe your PR:

1. What problem was fixed (it's best to include specific error reporting information). How it was fixed.
2. Which behaviors were modified. What was the previous behavior, what is it now, why was it modified, and what possible impacts might there be.
3. What features were added. Why was this function added?
4. Which code was refactored and why was this part of the code refactored?
5. Which functions were optimized and what is the difference before and after the optimization?

[Copilot]

Pull request overview

Note

Copilot couldn't run its full agentic review because it didn't start before the timeout. Make sure your repository has a runner available, or add a copilot-code-review.yml file specifying one with the runs-on attribute. See the docs for more details.

Ports the tablet header lock from std::shared_mutex to a bthread-aware BthreadSharedMutex to avoid cross-thread unlock UB / wedges and writer-preferring recursive shared-lock deadlocks, and updates call sites accordingly.

Changes:

• Added BthreadSharedMutex (bthread mutex/CV-based shared mutex) plus a dedicated unit test suite.
• Replaced tablet/header lock types and updated lock guards (std::lock_guard / std::unique_lock / std::shared_lock) to use the new mutex.
• Introduced “*_unlocked” helper variants to avoid recursive shared acquisition under a writer-preferring lock.

Reviewed changes

Copilot reviewed 24 out of 24 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
be/src/util/bthread_shared_mutex.h	Adds the bthread-compatible shared mutex implementation.
be/test/util/bthread_shared_mutex_test.cpp	Adds behavioral + regression tests for the new mutex (writer pending, migration, cross-thread unlock).
be/src/storage/tablet/base_tablet.h	Switches _meta_lock to BthreadSharedMutex and adds an unlocked compaction score accessor.
be/src/storage/tablet/base_tablet.cpp	Implements get_real_compaction_score_unlocked() and routes the locked version through it.
be/src/storage/tablet/tablet.h	Updates header lock accessor type and adds unlocked compaction rowset-picking declarations.
be/src/storage/tablet/tablet.cpp	Replaces lock types and avoids recursive shared locking via unlocked helpers.
be/src/storage/task/index_builder.cpp	Updates header lock guards to use the new lock type (CTAD).
be/src/storage/task/engine_clone_task.cpp	Updates header lock guard to use the new lock type (CTAD).
be/src/storage/tablet/tablet_manager.cpp	Updates header lock guard to use the new lock type (CTAD).
be/src/storage/schema_change/schema_change.cpp	Updates header lock guards to use the new lock type (CTAD).
be/src/storage/rowset_builder.cpp	Updates header lock guard to use the new lock type (CTAD).
be/src/storage/compaction/compaction.cpp	Updates header lock guards to use the new lock type (CTAD).
be/src/storage/compaction/full_compaction.cpp	Updates header lock guard to use the new lock type (CTAD).
be/src/storage/compaction/cumulative_compaction_time_series_policy.cpp	Switches to unlocked rowset-picking to avoid recursive shared acquisition.
be/src/cloud/cloud_tablet.h	Updates lock-typed parameters and adds unlocked rowset-picking APIs.
be/src/cloud/cloud_tablet.cpp	Removes inner shared locking in some paths and introduces unlocked rowset-picking implementations.
be/src/cloud/cloud_meta_mgr.h	Updates fill_version_holes to accept unique_lock<BthreadSharedMutex>&.
be/src/cloud/cloud_meta_mgr.cpp	Updates header-lock guards and fill_version_holes signature/usage.
be/src/cloud/cloud_base_compaction.cpp	Uses unlocked candidate selection under an existing shared header lock.
be/src/cloud/cloud_full_compaction.cpp	Uses unlocked candidate selection under an existing shared header lock.
be/src/cloud/cloud_cumulative_compaction.cpp	Switches traversal to an unlocked variant under an existing lock.
be/test/cloud/cloud_tablet_test.cpp	Updates test lock guards to use the new lock type (CTAD).
be/test/cloud/cloud_meta_mgr_test.cpp	Updates test lock guards to use the new lock type (CTAD).
be/test/cloud/cloud_empty_rowset_compaction_test.cpp	Updates test lock guards to use the new lock type (CTAD).
be/src/cloud/cloud_tablet.h	Updates stats comment to reflect BthreadSharedMutex.

Comments suppressed due to low confidence (1)

be/src/cloud/cloud_tablet.cpp:1

• Same issue as capture_versions_prefer_cache: the internal _meta_lock acquisition has been removed, turning this into an implicitly *_unlocked method without changing its name/signature. Recommendation (mandatory): provide an explicit locked entrypoint and a separate *_unlocked variant (or rename this one to *_unlocked) so callers can’t accidentally invoke it without holding _meta_lock.

// Licensed to the Apache Software Foundation (ASF) under one

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[liaoxin01]

run buildall

[hello-stephen]

BE UT Coverage Report

Increment line coverage 83.90% (99/118) 🎉

Increment coverage report
Complete coverage report

Category	Coverage
Function Coverage	57.11% (23325/40841)
Line Coverage	41.11% (231162/562353)
Region Coverage	37.39% (183021/489483)
Branch Coverage	38.47% (82226/213744)

[hello-stephen]

BE Regression && UT Coverage Report

Increment line coverage 94.00% (94/100) 🎉

Increment coverage report
Complete coverage report

Category	Coverage
Function Coverage	73.08% (29143/39878)
Line Coverage	56.82% (317864/559435)
Region Coverage	53.93% (265577/492475)
Branch Coverage	54.78% (117208/213975)