Mar 2025

1 commit
Fix crash if LockErrorCleanup() is called twice
28 Mar 2025Heikki Linnakangasmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1Also in: SIGSEGV in GrantLockLocal() (not this thread)Also in: BUG #18832: Segfault in GrantLockLocal (not this thread)
LLM analysis, 100% surewritten from the discussion, no matching patch1 of 1 reviewer saw the final version

Alexander's reproducer exposed repeated LockErrorCleanup() calls accessing an already-released LOCALLOCK #11; Heikki diagnosed the missing reset and announced the fix #12. The commit adds ResetAwaitedLock() in lock.c and calls it from proc.c, with no matching attachment.

  • Alexander Lakhinsaw the final versionLLM analysis, 98% sureSupplied a reproducible lock-timeout crash, backtrace, and bisection that enabled the fix, without an attached fix patch to review #11 #12.

Nov 2024

4 commits
Split ProcSleep function into JoinWaitQueue and ProcSleep
4 Nov 2024Heikki Linnakangasmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 99% surepart of a posted patchsetmoderate changesfrom patchset #90 of 1 reviewer saw the final version

Combines the LOCK_PRINT removal, redundant-global removal, local-lock update relocation, earlier partition-lock release, and function split from #9, as confirmed in #10. Commit-time edits flatten LockAcquireExtended()'s wait-result handling and remove the now-local lockAwaited reset in LockErrorCleanup().

  • Maxim Orlovsaw an earlier versiontrailerReviewed Heikki's series positively and supplied the assertion-only variable fix #9, before the subsequent control-flow and cleanup edits.
Move TRACE calls into WaitOnLock()
4 Nov 2024Heikki Linnakangasmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% surelanded as postedfrom patchset #91 of 1 reviewer saw the final version

The TRACE-call relocation in src/backend/storage/lmgr/lock.c has the same stable patch-id as the corresponding patch in #9. Its patch header credits Heikki, whose series Maxim rebased from #8.

  • Maxim Orlovsaw the final versiontrailerExplicitly reviewed the series and endorsed its preparatory refactorings #9, including this unchanged TRACE-call relocation.
Set MyProc->heldLocks in ProcSleep
4 Nov 2024Heikki Linnakangasmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% suremodified before commitmoderate changesfrom patchset #90 of 1 reviewer saw the final version

Implements Heikki's heldLocks relocation from #9, but replaces the immediate chained assignment with a separate myProcHeldLocks value. In src/backend/storage/lmgr/proc.c, MyProc->heldLocks is assigned later alongside the other PGPROC wait fields.

  • Maxim Orlovsaw an earlier versiontrailerEndorsed the preparatory refactorings in #9, before the committed version changed when MyProc->heldLocks is assigned.
Fix comment in LockReleaseAll() on when locallock->nLock can be zero
4 Nov 2024Heikki Linnakangasmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% surelanded as postedfrom patchset #91 of 1 reviewer saw the final version

The LockReleaseAll() comment correction in src/backend/storage/lmgr/lock.c has the same stable patch-id as the corresponding patch in #9. Maxim forwarded Heikki's patch from #8 without changing its content.

  • Maxim Orlovsaw the final versiontrailerReviewed and endorsed the series' small preparatory changes #9, including this unchanged comment correction.

Credits across 5 commits

Author
Heikki LinnakangasLLM4
Reviewer
Maxim Orlov4
Reported by
Alexander Lakhin1
Richard Guo1
Robins Tharakan1
Committer
Heikki Linnakangas5

N LLM means N of the credits come from LLM analysis, not from commit trailers.