Jan 2023

5 commits
Use dlists instead of SHM_QUEUE for predicate locking
20 Jan 2023Andres Freundmaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1Also in: Improve heavyweight locks instead of building new lock managers? (not this thread)
LLM analysis, 99% suremodified before commitlight changesfrom patchset #10 of 1 reviewer saw the final version

Matches the predicate-locking conversion in #1, with light cleanup in src/backend/storage/lmgr/predicate.c: explanatory unconstify comments, additional const qualifiers and removal of a redundant cast. The list conversion and control flow are retained.

  • Thomas Munrono review foundtrailerThe commit credits review of an older version, but Thomas has no review activity on the submitted patchset in this thread.
Remove SHM_QUEUE
20 Jan 2023Andres Freundmaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1Also in: Improve heavyweight locks instead of building new lock managers? (not this thread)
LLM analysis, 100% surelanded as postedfrom patchset #10 of 1 reviewer saw the final version

Matches the SHM_QUEUE removal patch in #1, deleting shmqueue.c, declarations in storage/shmem.h, build entries and the typedef entry. The changed copyright year in the deleted file is a rebase difference.

  • Thomas Munrono review foundtrailerThe commit credits review of an older version, but Thomas has no review activity on the submitted patchset in this thread.
Use dlists instead of SHM_QUEUE for syncrep queue
18 Jan 2023Andres Freundmaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1Also in: Improve heavyweight locks instead of building new lock managers? (not this thread)
LLM analysis, 99% surelanded as postedfrom patchset #10 of 1 reviewer saw the final version

Matches the synchronous-replication queue patch in #1 across all five files, including reverse insertion traversal and detached-node deletion. Apparent differences in the SyncRepWakeQueue diff are hunk alignment; the resulting code matches modulo rebase and whitespace.

  • Thomas Munrono review foundtrailerThe commit credits review of an older version, but Thomas has no review activity on the submitted patchset in this thread.
Add detached node functions to ilist
18 Jan 2023Andres Freundmaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1Also in: Improve heavyweight locks instead of building new lock managers? (not this thread)
LLM analysis, 100% surelanded as postedfrom patchset #10 of 1 reviewer saw the final version

All 63 added lines implementing the five detached-node helpers in src/include/lib/ilist.h match the first patch in #1. Differences in surrounding const-qualified signatures come from the newer base.

  • Thomas Munrono review foundtrailerAndres recounts Thomas's earlier naming suggestions in #1, but no review of this thread's submitted patchset by Thomas is present.
Use dlist/dclist instead of PROC_QUEUE / SHM_QUEUE for heavyweight locks
18 Jan 2023Andres Freundmaster16beta1REL_19_STABLE16beta1REL_18_STABLE16beta1REL_17_STABLE16beta1REL_16_STABLE16beta1Also in: Improve heavyweight locks instead of building new lock managers? (not this thread)
LLM analysis, 99% suremodified before commitlight changesfrom patchset #10 of 1 reviewer saw the final version

Matches the heavyweight-lock conversion in #1, retaining dclist wait queues, dlist process lists and PROC_QUEUE removal. Additional cleanup reuses proc_iter in deadlock.c and corrects obsolete ProcWakeup commentary in proc.c.

  • Thomas Munrono review foundtrailerThe commit credits review of an older version, but Thomas has no review activity on the submitted patchset in this thread.

Credits across 5 commits

Author
Andres FreundLLM5
Reviewer
Thomas Munro5
Committer
Andres Freund5

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