Oct 2019

2 commits
For all ppc compilers, implement compare_exchange and fetch_add with asm.
19 Oct 2019Noah Mischmaster13beta1REL_19_STABLE13beta1REL_18_STABLE13beta1REL_17_STABLE13beta1REL_16_STABLE13beta1REL_15_STABLE13beta1REL_14_STABLE13beta1REL_13_STABLE13beta1
LLM analysis, 100% surelanded as postedfrom patchset #501 of 1 reviewer saw the final version

The complete unification attachment in #50, applied to the commit's parent, produces byte-identical contents for all eight affected paths, including configure, arch-ppc.h and both header-check scripts, and the same deletion of generic-xlc.h.

  • Tom Lanesaw the final versionLLM analysis, 100% sureTom examined successive implementations, identified the register-constraint failure, requested readability and header-check fixes, and confirmed final compilation and regression tests in #51; the commit also explicitly credits his review.
For PowerPC instruction "addi", use constraint "b".
19 Oct 2019Noah Mischmaster13beta1REL_19_STABLE13beta1REL_18_STABLE13beta1REL_17_STABLE13beta1REL_16_STABLE13beta1REL_15_STABLE13beta1REL_14_STABLE13beta1REL_13_STABLE13beta1+6 more
LLM analysis, 100% surelanded as postedfrom patchset #501 of 1 reviewer saw the final version

Exactly matches the separate s_lock.h attachment in #50, including the explanatory comment and replacement of "=&r" with "=&b" in tas(); applying it to the parent reproduces the committed file.

  • Tom Lanesaw the final versionLLM analysis, 100% sureTom confirmed that the final patch set compiled and passed regression tests on his older Apple machines in #51, and the commit explicitly credits his review.

Sep 2019

3 commits
For all ppc compilers, implement pg_atomic_fetch_add_ with inline asm.
14 Sep 2019Noah Mischmaster13beta1REL_19_STABLE13beta1REL_18_STABLE13beta1REL_17_STABLE13beta1REL_16_STABLE13beta1REL_15_STABLE13beta1REL_14_STABLE13beta1REL_13_STABLE13beta1
LLM analysis, 100% surelanded as postedfrom patchset #431 of 1 reviewer saw the final version

Matches the complete PPC fetch-add unification attachment in #43. Applying that attachment to the commit's parent reproduces all five committed files exactly; the apparent configure difference is only hunk placement.

  • Tom Lanesaw the final versionLLM analysis, 100% sureTom questioned the complexity devoted to the addi optimization in #44, prompting Noah's code-generation justification in #45, and the commit explicitly credits his review.
Replace xlc __fetch_and_add() with inline asm.
14 Sep 2019Noah Mischmaster13beta1REL_19_STABLE13beta1REL_18_STABLE13beta1REL_17_STABLE13beta1REL_16_STABLE13beta1REL_15_STABLE13beta1REL_14_STABLE13beta1REL_13_STABLE13beta1+5 more
LLM analysis, 100% surelanded as postedfrom patchset #431 of 1 reviewer saw the final version

Exactly matches the complete XLC assembly-replacement attachment in #43, replacing both fetch-add intrinsics in src/include/port/atomics/generic-xlc.h with explicit synchronized assembly loops.

  • Tom Lanesaw the final versionLLM analysis, 100% sureThe commit explicitly credits Tom Lane as reviewer; his acknowledgment in #44 does not independently document substantive review of this particular attachment.
Test pg_atomic_fetch_add_ with variable addend and 16-bit edge cases.
14 Sep 2019Noah Mischmaster13beta1REL_19_STABLE13beta1REL_18_STABLE13beta1REL_17_STABLE13beta1REL_16_STABLE13beta1REL_15_STABLE13beta1REL_14_STABLE13beta1REL_13_STABLE13beta1+5 more
LLM analysis, 100% surelanded as postedfrom patchset #431 of 1 reviewer saw the final version

Exactly matches the complete regression-test attachment in #43, adding variable fetch-add operands and 16-bit boundary cases in src/test/regress/regress.c.

  • Tom Lanesaw the final versionLLM analysis, 100% sureThe commit explicitly credits Tom Lane as reviewer; his acknowledgment in #44 does not independently document substantive review of the test attachment.

Credits across 5 commits

Author
Noah MischLLM5
Committer
Noah Misch5

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