Jan 2026

1 commit
jit: Add missing inline pass for LLVM >= 17.
22 Jan 2026Thomas Munromaster19beta1REL_19_STABLE19beta1REL_18_STABLE18.2REL_17_STABLE17.8REL_16_STABLE16.12REL_15_STABLE15.16REL_14_STABLE14.21
LLM analysis, 100% surelanded as postedfrom patchset #75 of 6 reviewers saw the final version

e5d99b4d9e exactly matches the single four-insertion hunk in src/backend/jit/llvm/llvmjit.c from #7, including the always-inline comment. The commit message was polished, but the code patch is unchanged.

  • Thomas Munrosaw the final versiontrailerThomas identified the redundant condition and raised always-inline behavior in #2, announced acceptance in #13, and reviewed the final version as committer.
  • Andreas Karlssonsaw the final versiontrailerAndreas examined pass consistency and possible tuple-deforming use of always-inline in #3; the committed fix differs from that initial version only by a redundant-condition cleanup and a comment.
  • Andres Freundsaw the final versiontrailerAndres opposed removing always-inline because generated deform functions need it in #6, then approved the revised patch in #11.
  • Álvaro Herrerano review foundtrailerÁlvaro discussed JIT adoption and welcomed the discovery in #5, but showed no patch review or testing activity.
  • Pierre Ducroquetsaw the final versiontrailerPierre explicitly reported reviewing the patch and gave it a +1 in #10.
  • Matheus Alcantarasaw the final versiontrailerMatheus tested the final patch with LLVM 22 and showed that enabling inlining now inlined the deform function in #12.

Credits across 1 commit

Author
Anthonin Bonnefoy1
Reviewer
Alvaro Herrera1
Andreas Karlsson1
Andres Freund1
Matheus Alcantara1
Pierre Ducroquet1
Thomas Munro1
Committer
Thomas Munro1