Feb 2026

1 commit
Convert SpinLock* macros to static inline functions.
23 Feb 2026Nathan Bossartmaster19beta1REL_19_STABLE19beta1Also in: convert SpinLock* macros to static inline functions (part of a posted patchset)

Jun 2020

5 commits
Clean up includes of s_lock.h.
19 Jun 2020Andres Freundmaster14beta1REL_19_STABLE14beta1REL_18_STABLE14beta1REL_17_STABLE14beta1REL_16_STABLE14beta1REL_15_STABLE14beta1REL_14_STABLE14beta1
LLM analysis, 100% surewritten from the discussion, no matching patch

The three include changes implement Andres's explicit proposal in #42: remove storage/s_lock.h from main.c and lwlock.h, and replace it with storage/spin.h in condition_variable.h. Neither attached patchset contains these changes.

Fix deadlock danger when atomic ops are done under spinlock.
18 Jun 2020Andres Freundmaster14beta1REL_19_STABLE14beta1REL_18_STABLE14beta1REL_17_STABLE14beta1REL_16_STABLE14beta1REL_15_STABLE14beta1REL_14_STABLE14beta1REL_13_STABLE13beta2+5 more
LLM analysis, 100% suremodified before commitlight changesfrom patchset #281 of 1 reviewer saw the final version

The fourth patch in #28 supplies the separate semaphore pools and nested-atomic regression test. The commit renames NUM_SIMULATION_SEMAPHORES to NUM_EMULATION_SEMAPHORES and extends and relocates the regression test to exercise both 32-bit and 64-bit atomics.

  • Robert Haassaw the final versionLLM analysis, 97% sureRobert explicitly reviewed the fourth patch's functionality and questioned its tests in #29, before the naming and test-only changes.
Add basic spinlock tests to regression tests.
18 Jun 2020Andres Freundmaster14beta1REL_19_STABLE14beta1REL_18_STABLE14beta1REL_17_STABLE14beta1REL_16_STABLE14beta1REL_15_STABLE14beta1REL_14_STABLE14beta1REL_13_STABLE13beta2+5 more
LLM analysis, 100% suremodified before commitlight changesfrom patchset #281 of 1 reviewer saw the final version

The third patch in #28 supplies test_spinlock() and its overflow stress test in src/test/regress/regress.c. The commit adds storage/spin.h, clarifies test intent, fixes the reported typo, and replaces integer sentinels with byte arrays checked using memcmp.

  • Robert Haassaw the final versionLLM analysis, 100% sureRobert examined s_lock portability, sentinel checks, failure behavior, and a comment typo in #29, prompting clearer explanations before commitment.
spinlock emulation: Fix bug when more than INT_MAX spinlocks are initialized.
17 Jun 2020Andres Freundmaster14beta1REL_19_STABLE14beta1REL_18_STABLE14beta1REL_17_STABLE14beta1REL_16_STABLE14beta1REL_15_STABLE14beta1REL_14_STABLE14beta1REL_13_STABLE13beta2+5 more
LLM analysis, 100% surelanded as postedfrom patchset #281 of 1 reviewer saw the final version

The commit exactly matches the first patch in #28, changing the static counter in s_init_lock_sema() from int to uint32 in src/backend/storage/lmgr/spin.c; their stable patch IDs match.

  • Robert Haassaw the final versiontrailerRobert explicitly reported reading the series and having no objection to the overflow fix in #29, which landed unchanged.
Avoid potential spinlock in a signal handler as part of global barriers.
17 Jun 2020Andres Freundmaster14beta1REL_19_STABLE14beta1REL_18_STABLE14beta1REL_17_STABLE14beta1REL_16_STABLE14beta1REL_15_STABLE14beta1REL_14_STABLE14beta1REL_13_STABLE13beta2
LLM analysis, 100% surelanded as postedfrom patchset #281 of 1 reviewer saw the final version

The commit exactly matches the second patch in #28, introducing PROCSIG_BARRIER and moving generation reads outside the signal handler in procsignal.c and procsignal.h; their stable patch IDs match.

  • Robert Haassaw the final versiontrailerRobert reviewed the replacement barrier patch and reported no objection in #29, after supporting deferred processing in #20.

Credits across 6 commits

Author
Andres Freund4
Nathan BossartLLM1
Reviewer
Robert Haas2
Andres Freund1
Fabrízio de Royes Mello1
Committer
Andres Freund5
Nathan Bossart1

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