Nov 2026

1 commit
Centralize resetting SIGCHLD handler
10 Nov 2026Heikki Linnakangasmasterunreleased
LLM analysis, 100% suremodified before commitlight changesfrom patchset #571 of 1 reviewer saw the final version

Matches the SIGCHLD-centralization prerequisite in #57, removing per-process resets across 13 files and adding the reset to InitPostmasterChild() in src/backend/utils/init/miscinit.c. Beyond rebasing, differences are rewritten and relocated comments in miscinit.c and an expanded commit message.

  • Andres Freundsaw the final versiontrailerIn #52, Andres examined the SIGCHLD change's single-user implications and concluded it was safe; this prerequisite subsequently received only rebase, naming, and comment changes.

Jul 2026

1 commit
Refactor how some aux processes advertise their ProcNumber
8 Jul 2026Heikki Linnakangasmasterunreleased
LLM analysis, 100% suremodified before commitlight changesfrom patchset #571 of 1 reviewer saw the final version

Matches the auxiliary-process ProcNumber prerequisite in #57, including atomic publication and cleanup in src/backend/storage/lmgr/proc.c. Differences are comment wording and commit-message polish.

  • Andres Freundsaw the final versiontrailerIn #52, Andres examined the publication code's volatile accesses and proposed a generic auxiliary-process array; this prerequisite subsequently changed only lightly.

Feb 2026

1 commit
Ignore SIGINT in walwriter and walsummarizer
17 Feb 2026Heikki Linnakangasmaster19beta1REL_19_STABLE19beta1
LLM analysis, 100% suremodified before commitlight changesfrom patchset #511 of 1 reviewer saw the final version

The walwriter.c and walsummarizer.c changes exactly match the SIGINT prerequisite in #51. The commit message adds bgwriter to the behavior summary, as suggested in #52.

  • Andres Freundsaw the final versiontrailerAndres explicitly endorsed merging the SIGINT change and supplied the missing bgwriter behavior in #52.

Mar 2025

3 commits
Split WaitEventSet functions to separate source file
5 Mar 2025Heikki Linnakangasmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 99% suremodified before commitlight changesfrom patchset #411 of 1 reviewer saw the final version

Matches the WaitEventSet extraction in #41, with unused includes removed, comments corrected, and Windows preprocessor guards fixed. Differences in the moved ModifyWaitEvent body also reflect the preceding prerequisite's correction described in #43.

  • Andres Freundsaw the final versiontrailerAndres reviewed the code movement, includes, comments, and header dependencies in #42, followed by the small fixes reported in #43.
Use ModifyWaitEvent to update exit_on_postmaster_death
5 Mar 2025Heikki Linnakangasmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% suremodified before commitmoderate changesfrom patchset #410 of 1 reviewer saw the final version

Derives from the ModifyWaitEvent prerequisite in #41, but moves postmaster-death mode handling ahead of the unchanged-events fast path. This fixes switching back from WL_EXIT_ON_PM_DEATH, as explained in #43.

  • Andres Freundsaw an earlier versiontrailerAndres identified the reverse-switching problem in #42, after which Heikki changed ModifyWaitEvent control flow without a further posted review.
Remove unused ShutdownLatchSupport() function
5 Mar 2025Heikki Linnakangasmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% surelanded as postedfrom patchset #411 of 1 reviewer saw the final version

The removal of ShutdownLatchSupport from src/backend/storage/ipc/latch.c and its header has the same stable patch ID as the corresponding patch in #41.

  • Andres Freundsaw the final versiontrailerAndres confirmed why ShutdownLatchSupport had become unused and approved its removal in #42.

Nov 2024

2 commits
Rename two functions that wake up other processes
1 Nov 2024Heikki Linnakangasmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 99% surepart of a posted patchsetlight changesfrom patchset #20

Combines the WalSummarizer rename with the WaitLSN rename extracted from #20, leaving its ProcNumber conversion for a separate commit. The WaitLSN code had moved from commands/waitlsn.c to access/transam/xlogwait.c.

Use ProcNumbers instead of direct Latch pointers to address other procs
1 Nov 2024Heikki Linnakangasmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 99% surepart of a posted patchsetlight changesfrom patchset #20

Combines the walwriter/checkpointer, walreceiver, and WaitLSN ProcNumber conversions from #20, excluding its function renames and interrupt replacement. Remaining edits are comments, declaration formatting, and include adjustments.

Oct 2024

3 commits
Clean up WaitLatch calls that passed latch without WL_LATCH_SET
5 Oct 2024Heikki Linnakangasmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% surelanded as postedfrom patchset #20

Stable patch IDs match the WaitLatch cleanup in #20: four calls in be-secure-gssapi.c, be-secure-openssl.c, and walsummarizer.c pass NULL when WL_LATCH_SET is absent.

Remove unneeded #include
5 Oct 2024Heikki Linnakangasmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% surelanded as postedfrom patchset #20

Exactly matches the removal of storage/latch.h from src/include/storage/buf_internals.h in #20, which repeats the earlier cleanup from #18.

Remove unused latch
5 Oct 2024Heikki Linnakangasmaster18beta1REL_19_STABLE18beta1REL_18_STABLE18beta1
LLM analysis, 100% surelanded as postedfrom patchset #20

Stable patch IDs match the unused WalSnd latch removal in #20, covering walsender.c and walsender_private.h.

Credits across 11 commits

Author
Heikki LinnakangasLLM11
Reviewer
Andres Freund6
Committer
Heikki Linnakangas11

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