T-compiler Meeting Agenda 2026-02-26

Announcements

Heads-up: release of 1.94.0 is next week

New! Backports can now be accepted/declined directly from Zulip, example:

@triagebot backport stable accept 123456
@triagebot backport beta decline 123456

Reminder: if you see a PR/issue that seems like there might be legal implications due to copyright/IP/etc, please let us know (or at least message @davidtwco or @Wesley Wiser so we can pass it along).

MCPs/FCPs

(sorry the list could not be fully generated)

New MCPs (take a look, see if you like them!)
- Remove soft_unstable
- Remove -Csoft-float

Backport nominations

T-compiler beta / T-compiler stable

None

T-types beta / T-types stable

None

PRs S-waiting-on-t-compiler

T-compiler

Issues in progress or waiting on other teams

Issues of Note

Short Summary

Stable regression

“#[no_link] attribute does not prevent crate linking” rust#152854
- doesn’t seem super urgent, @_bjorn3 opened #152856 to investigate removing the attribute. Possibly it should go through a T-lang FCP first (comment)
- (FIY in case anyone has opinions)

P-critical

T-compiler

None

T-types

None

P-high regressions

P-high beta regressions

“Regression: “rust-lld: error: undefined symbol: Rlog1p” · rust#150663
- Failed a crater run a while ago. Seems a flaky build or something. We tried to reproduce to no avail. Unsure what we can do here.
“Chromium hitting SIGILL crashes on aarch64 after #144938 enabled outlined atomics by default” rust#151486
- Should be fixed by #151896
“regression: undefined symbol: __gxx_personality_v0” rust#152353
- Another crate that failed a crate run a while ago, still unclear but possibly a flaky build

Unassigned P-high nightly regressions

None

Performance logs

Triage logs for 2026-02-17

Several pull requests introduced (usually very small) regressions across the board this week. On the other hand, #151380 provided a nice performance win in the inference engine. I would also like to bring attention to #152375, which improved the parallel frontend. It is not shown in this report, because we don’t yet have many benchmarks for the parallel frontend, but this PR seemingly improved the check (wall-time) performance with multiple frontend threads on several real-world crates by 5-10%!

Triage done by @kobzol. Revision range: 39219ceb..3c9faa0d

Summary:

(instructions:u)	mean	range	count
Regressions (primary)	0.7%	[0.2%, 3.1%]	96
Regressions (secondary)	1.1%	[0.0%, 5.7%]	62
Improvements (primary)	-0.4%	[-0.9%, -0.2%]	8
Improvements (secondary)	-2.6%	[-7.0%, -0.0%]	45
All (primary)	0.6%	[-0.9%, 3.1%]	104

2 Regressions, 0 Improvements, 9 Mixed; 4 of them in rollups 36 artifact comparisons made in total

Regressions

Rollup of 17 pull requests #152562 (Comparison Link)

(instructions:u)	mean	range	count
Regressions (primary)	-	-	0
Regressions (secondary)	0.3%	[0.2%, 0.5%]	11
Improvements (primary)	-	-	0
Improvements (secondary)	-	-	0
All (primary)	-	-	0

Noise that got reverted in a follow-up PR.
Already marked as triaged.

replace box_new with lower-level intrinsics #148190 (Comparison Link)

(instructions:u)	mean	range	count
Regressions (primary)	0.8%	[0.5%, 1.9%]	8
Regressions (secondary)	1.8%	[0.4%, 5.5%]	24
Improvements (primary)	-0.8%	[-0.8%, -0.8%]	1
Improvements (secondary)	-0.3%	[-0.3%, -0.3%]	1
All (primary)	0.6%	[-0.8%, 1.9%]	9

Small performance hit across the board caused by refactoring how Box works internally.
Was deemed acceptable pre-merge.
Marked as triaged.

Mixed

Rollup of 12 pull requests #152399 (Comparison Link)

(instructions:u)	mean	range	count
Regressions (primary)	0.2%	[0.2%, 0.2%]	3
Regressions (secondary)	0.8%	[0.1%, 1.4%]	26
Improvements (primary)	-	-	0
Improvements (secondary)	-0.2%	[-0.3%, -0.1%]	8
All (primary)	0.2%	[0.2%, 0.2%]	3

The largest regressions seem to be caused by #152327.
That PR fixed some compiler bugs, so the regression on a secondary benchmarks seems justifiable.
Marked as triaged.

Update mgca to use type const syntax instead of the #[type_const] attribute. #152324 (Comparison Link)

(instructions:u)	mean	range	count
Regressions (primary)	0.4%	[0.1%, 0.7%]	18
Regressions (secondary)	0.5%	[0.0%, 1.7%]	13
Improvements (primary)	-	-	0
Improvements (secondary)	-0.2%	[-0.3%, -0.0%]	4
All (primary)	0.4%	[0.1%, 0.7%]	18

It is unclear whether the regressions were expected or not.
Asked the reviewer to follow up.
Not marking as triaged yet.

Update to LLVM 22.1.0-rc3 #152428 (Comparison Link)

(instructions:u)	mean	range	count
Regressions (primary)	0.2%	[0.2%, 0.2%]	1
Regressions (secondary)	0.1%	[0.0%, 0.2%]	4
Improvements (primary)	-	-	0
Improvements (secondary)	-0.3%	[-0.5%, -0.1%]	10
All (primary)	0.2%	[0.2%, 0.2%]	1

A few tiny changes that got over the noise threshold, but overall a wash.
Marked as triaged.

Rollup of 7 pull requests #152506 (Comparison Link)

(instructions:u)	mean	range	count
Regressions (primary)	0.2%	[0.2%, 0.2%]	1
Regressions (secondary)	0.2%	[0.1%, 0.2%]	3
Improvements (primary)	-	-	0
Improvements (secondary)	-0.1%	[-0.2%, -0.1%]	2
All (primary)	0.2%	[0.2%, 0.2%]	1

Very small regressions across the board, but I don’t think it’s worth it to dig deeper.
Marked as triaged.

Start using pattern types in libcore (NonZero and friends) #148537 (Comparison Link)

(instructions:u)	mean	range	count
Regressions (primary)	1.9%	[0.1%, 3.8%]	12
Regressions (secondary)	2.2%	[0.1%, 3.6%]	4
Improvements (primary)	-	-	0
Improvements (secondary)	-0.1%	[-0.1%, -0.1%]	2
All (primary)	1.9%	[0.1%, 3.8%]	12

The regression was deemed acceptable pre-merge, and the post-merge results correspond to the previous perf. runs.
Marked as triaged.

Make operational semantics of pattern matching independent of crate and module #150681 (Comparison Link)

(instructions:u)	mean	range	count
Regressions (primary)	0.6%	[0.1%, 1.1%]	7
Regressions (secondary)	0.2%	[0.1%, 0.2%]	12
Improvements (primary)	-2.6%	[-2.6%, -2.6%]	1
Improvements (secondary)	-2.3%	[-2.3%, -2.3%]	1
All (primary)	0.2%	[-2.6%, 1.1%]	8

A few small regressions across the board, mainly in doc builds, the rest of the profiles are a wash. Given the change made here, I suppose the perf. regression is expected.
Marked as triaged.

Rollup of 7 pull requests #152632 (Comparison Link)

(instructions:u)	mean	range	count
Regressions (primary)	0.4%	[0.1%, 0.8%]	6
Regressions (secondary)	0.4%	[0.1%, 0.8%]	3
Improvements (primary)	-0.5%	[-1.0%, -0.2%]	3
Improvements (secondary)	-1.0%	[-2.0%, -0.5%]	5
All (primary)	0.1%	[-1.0%, 0.8%]	9

Performance-wise it’s a wash.
Marked as triaged.

Shallow resolve ty and const vars to their root vars #151380 (Comparison Link)

(instructions:u)	mean	range	count
Regressions (primary)	-	-	0
Regressions (secondary)	0.3%	[0.2%, 0.4%]	4
Improvements (primary)	-0.3%	[-0.4%, -0.1%]	14
Improvements (secondary)	-2.7%	[-7.1%, -0.1%]	50
All (primary)	-0.3%	[-0.4%, -0.1%]	14

Many more wins than losses.
Marked as triaged.

Big query system cleanups #152636 (Comparison Link)

(instructions:u)	mean	range	count
Regressions (primary)	0.3%	[0.2%, 0.4%]	10
Regressions (secondary)	0.2%	[0.1%, 0.3%]	9
Improvements (primary)	-	-	0
Improvements (secondary)	-0.4%	[-0.4%, -0.3%]	4
All (primary)	0.3%	[0.2%, 0.4%]	10

The small performance regressions were investigated and deemed to be caused mostly by inlining perturbations.
Already marked as triaged.

Nominated Issues

T-compiler

None

RFC

None

Oldest PRs waiting for review

None this week

Next meeting’s agenda draft: hackmd link