Introduce `TypingMode::Codegen` to avoid layout cycles on coroutines by cjgillot · Pull Request #145477 · rust-lang/rust

cjgillot · 2025-08-16T00:04:12Z

Computing layout of coroutines depends on their optimized_mir. At the same time, MIR opts can require using layouts to work. For instance to evaluate constants. This leads to cycles and clumsy workarounds.

This PR creates a new typing mode for layout computations:

when a coroutine's layout is requested with TypingMode::PostAnalysis or earlier, return LayourError::TooGeneric;
when a coroutine's layout is requested with TypingMode::Codegen, actually compute it.

TypingMode::Codegen is meant be be used by codegen code, and analyses that require coroutine layout, like transmute check and coroutine recursion check.

With this PR, we can remove all is_coroutine checks from rustc_mir_transform and unlock simplifying coroutine MIR.

Perf is not terrific. This PR causes recomputation of a few queries, and I had to insert workarounds.

cjgillot · 2025-08-16T00:05:55Z

@bors try @rust-timer queue

[EXPERIMENT] Introduce `TypingMode::Codegen` to avoid layout cycles

rust-bors · 2025-08-16T02:19:21Z

☀️ Try build successful (CI)
Build commit: 34dd67f (34dd67f42f0f2050c773b332e2dd333618c22743, parent: cd7cbe818e4a66d46fe2df993d1b8518eba8a5cd)

rust-timer · 2025-08-16T07:24:00Z

Finished benchmarking commit (34dd67f): comparison URL.

Overall result: ❌ regressions - please read the text below

Benchmarking this pull request means it may be perf-sensitive – we'll automatically label it not fit for rolling up. You can override this, but we strongly advise not to, due to possible changes in compiler perf.

Next Steps: If you can justify the regressions found in this try perf run, please do so in sufficient writing along with @rustbot label: +perf-regression-triaged. If not, please fix the regressions and do another perf run. If its results are neutral or positive, the label will be automatically removed.

@bors rollup=never
@rustbot label: -S-waiting-on-perf +perf-regression

Instruction count

Our most reliable metric. Used to determine the overall result above. However, even this metric can be noisy.

	mean	range	count
Regressions ❌ (primary)	0.5%	[0.2%, 3.3%]	58
Regressions ❌ (secondary)	1.8%	[0.0%, 19.4%]	58
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-0.2%	[-0.3%, -0.0%]	4
All ❌✅ (primary)	0.5%	[0.2%, 3.3%]	58

Max RSS (memory usage)

Results (primary 1.4%, secondary 4.3%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	1.7%	[0.5%, 7.3%]	17
Regressions ❌ (secondary)	5.4%	[1.2%, 12.3%]	6
Improvements ✅ (primary)	-1.3%	[-1.4%, -1.3%]	2
Improvements ✅ (secondary)	-2.3%	[-2.3%, -2.3%]	1
All ❌✅ (primary)	1.4%	[-1.4%, 7.3%]	19

Cycles

Results (primary 2.9%, secondary 4.5%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	2.9%	[2.9%, 2.9%]	1
Regressions ❌ (secondary)	5.3%	[2.1%, 12.0%]	12
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-4.0%	[-4.0%, -4.0%]	1
All ❌✅ (primary)	2.9%	[2.9%, 2.9%]	1

Binary size

Results (primary 0.1%, secondary -1.0%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	0.4%	[0.1%, 0.6%]	2
Regressions ❌ (secondary)	0.3%	[0.1%, 0.4%]	3
Improvements ✅ (primary)	-0.0%	[-0.0%, -0.0%]	3
Improvements ✅ (secondary)	-1.8%	[-7.1%, -0.1%]	5
All ❌✅ (primary)	0.1%	[-0.0%, 0.6%]	5

Bootstrap: 469.562s -> 472.464s (0.62%)
Artifact size: 377.55 MiB -> 377.67 MiB (0.03%)

cjgillot · 2025-08-16T14:48:02Z

@bors try @rust-timer queue

[EXPERIMENT] Introduce `TypingMode::Codegen` to avoid layout cycles

rust-bors · 2025-08-16T17:07:25Z

☀️ Try build successful (CI)
Build commit: 4e9488c (4e9488c147d4057600076145eccb670b130147d5, parent: 2e2642e641a941f0a1400c7827fd89aa86fef8f4)

rust-timer · 2025-08-16T18:40:34Z

Finished benchmarking commit (4e9488c): comparison URL.

Overall result: ❌ regressions - please read the text below

Benchmarking this pull request means it may be perf-sensitive – we'll automatically label it not fit for rolling up. You can override this, but we strongly advise not to, due to possible changes in compiler perf.

Next Steps: If you can justify the regressions found in this try perf run, please do so in sufficient writing along with @rustbot label: +perf-regression-triaged. If not, please fix the regressions and do another perf run. If its results are neutral or positive, the label will be automatically removed.

@bors rollup=never
@rustbot label: -S-waiting-on-perf +perf-regression

Instruction count

Our most reliable metric. Used to determine the overall result above. However, even this metric can be noisy.

	mean	range	count
Regressions ❌ (primary)	0.4%	[0.2%, 0.9%]	23
Regressions ❌ (secondary)	1.5%	[0.1%, 9.7%]	59
Improvements ✅ (primary)	-1.6%	[-1.6%, -1.6%]	1
Improvements ✅ (secondary)	-0.3%	[-0.3%, -0.3%]	2
All ❌✅ (primary)	0.4%	[-1.6%, 0.9%]	24

Max RSS (memory usage)

Results (primary 1.7%, secondary 5.0%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	1.9%	[0.5%, 7.5%]	19
Regressions ❌ (secondary)	6.6%	[0.4%, 12.9%]	5
Improvements ✅ (primary)	-1.3%	[-1.3%, -1.3%]	1
Improvements ✅ (secondary)	-2.8%	[-2.8%, -2.8%]	1
All ❌✅ (primary)	1.7%	[-1.3%, 7.5%]	20

Cycles

Results (secondary 2.7%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	4.2%	[2.1%, 10.2%]	8
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-3.0%	[-3.8%, -2.2%]	2
All ❌✅ (primary)	-	-	0

Binary size

Results (primary 0.0%, secondary -1.1%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	0.1%	[0.1%, 0.1%]	1
Regressions ❌ (secondary)	0.2%	[0.1%, 0.4%]	4
Improvements ✅ (primary)	-0.0%	[-0.0%, -0.0%]	3
Improvements ✅ (secondary)	-2.9%	[-6.8%, -0.0%]	3
All ❌✅ (primary)	0.0%	[-0.0%, 0.1%]	4

Bootstrap: 470.222s -> 472.094s (0.40%)
Artifact size: 377.51 MiB -> 377.42 MiB (-0.02%)

bors · 2025-08-23T06:22:24Z

☔ The latest upstream changes (presumably #145469) made this pull request unmergeable. Please resolve the merge conflicts.

rustbot · 2026-06-03T14:16:59Z

This PR was rebased onto a different main commit. Here's a range-diff highlighting what actually changed.

Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers.

cjgillot · 2026-06-03T16:16:21Z

@bors r=oli-obk

rust-bors · 2026-06-03T16:16:25Z

📌 Commit ac71b54 has been approved by oli-obk

It is now in the queue for this repository.

🌲 The tree is currently closed for pull requests below priority 5. This pull request will be tested once the tree is reopened.

rust-bors · 2026-06-04T02:28:39Z

☀️ Test successful - CI
Approved by: oli-obk
Duration: 3h 13m 59s
Pushing 76dfce2 to main...

github-actions · 2026-06-04T02:31:27Z

What is this?

This is an experimental post-merge analysis report that shows differences in test outcomes between the merged PR and its parent PR.

Comparing b354133 (parent) -> 76dfce2 (this PR)

Test differences

Show 12 test diffs

12 doctest diffs were found. These are ignored, as they are noisy.

Test dashboard

Run

cargo run --manifest-path src/ci/citool/Cargo.toml -- \
test-dashboard 76dfce2cb2d3f7b7f34d62e6ffe044f7e7d76948 --output-dir test-dashboard

And then open test-dashboard/index.html in your browser to see an overview of all executed tests.

Job duration changes

dist-apple-various: 1h 25m -> 2h 16m (+59.5%)
dist-x86_64-msvc-alt: 2h 50m -> 1h 40m (-41.1%)
x86_64-gnu-miri: 1h 11m -> 1h 38m (+37.3%)
x86_64-mingw-1: 2h 54m -> 1h 50m (-36.4%)
dist-armv7-linux: 1h 31m -> 59m 29s (-35.3%)
dist-aarch64-apple: 2h 12m -> 1h 26m (-34.3%)
x86_64-gnu-distcheck: 1h 51m -> 2h 26m (+30.8%)
x86_64-msvc-2: 2h 42m -> 1h 55m (-29.0%)
dist-loongarch64-musl: 1h 27m -> 1h 50m (+26.7%)
x86_64-msvc-1: 2h 1m -> 2h 30m (+24.6%)

How to interpret the job duration changes?

Job durations can vary a lot, based on the actual runner instance
that executed the job, system noise, invalidated caches, etc. The table above is provided
mostly for t-infra members, for simpler debugging of potential CI slow-downs.

rust-timer · 2026-06-04T03:10:37Z

Finished benchmarking commit (76dfce2): comparison URL.

Overall result: ❌✅ regressions and improvements - please read:

Our benchmarks found a performance regression caused by this PR.
This might be an actual regression, but it can also be just noise.

Next Steps:

If the regression was expected or you think it can be justified,
please write a comment with sufficient written justification, and add
@rustbot label: +perf-regression-triaged to it, to mark the regression as triaged.
If you think that you know of a way to resolve the regression, try to create
a new PR with a fix for the regression.
If you do not understand the regression or you think that it is just noise,
you can ask the @rust-lang/wg-compiler-performance working group for help (members of this group
were already notified of this PR).

@rustbot label: +perf-regression
cc @rust-lang/wg-compiler-performance

Instruction count

Our most reliable metric. Used to determine the overall result above. However, even this metric can be noisy.

	mean	range	count
Regressions ❌ (primary)	0.5%	[0.1%, 1.1%]	51
Regressions ❌ (secondary)	1.5%	[0.1%, 9.9%]	43
Improvements ✅ (primary)	-0.5%	[-0.5%, -0.5%]	1
Improvements ✅ (secondary)	-0.4%	[-0.4%, -0.4%]	1
All ❌✅ (primary)	0.4%	[-0.5%, 1.1%]	52

Max RSS (memory usage)

Results (primary 1.2%, secondary -1.8%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	1.6%	[0.7%, 2.8%]	13
Regressions ❌ (secondary)	2.8%	[2.8%, 2.8%]	1
Improvements ✅ (primary)	-1.0%	[-1.4%, -0.6%]	2
Improvements ✅ (secondary)	-2.9%	[-8.2%, -1.0%]	4
All ❌✅ (primary)	1.2%	[-1.4%, 2.8%]	15

Cycles

Results (secondary 1.9%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	3.3%	[2.1%, 6.6%]	6
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-2.4%	[-2.6%, -2.2%]	2
All ❌✅ (primary)	-	-	0

Binary size

This perf run didn't have relevant results for this metric.

Bootstrap: 511.076s -> 511.359s (0.06%)
Artifact size: 400.68 MiB -> 400.66 MiB (-0.01%)

lcnr · 2026-06-04T09:37:43Z

two thoughts, this feels related to the addition of TypingMode::ErasedNonCoherence, i.e. the trait solver mostly doesn't care which typingmode we're in between these two post-analysis modes, so we can probably extend this Erased behavior to also handle different codegen modes if perf requires it

the other one is that I am vaguely unhappy about adding new TypingMode. They have a fairly high complexity cost and I want to really stay on top of the ones we add.

I don't think this PR is necessarily wrong, I would like this to have gone through a types MCP or at least nomination.

nbdd0121 · 2026-06-05T15:25:43Z

+ // We are in codegen. It's very likely this constant has been evaluated in PostAnalysis
+ // before. Try to reuse this evaluation, and only re-run if we hit a `TooGeneric` error.
+ ty::TypingMode::Codegen => {
+ let with_postanalysis =
+ ty::TypingEnv::new(typing_env.param_env, ty::TypingMode::PostAnalysis);
+ let with_postanalysis = f(with_postanalysis.as_query_input(value));
+ match with_postanalysis {
+ Ok(_) | Err(ErrorHandled::Reported(..)) => return Some(with_postanalysis),
+ Err(ErrorHandled::TooGeneric(_)) => {}
+ }


I think the fact that this exists make me feel that this is not the right approach.
Perhaps we should just not use optimized_mir for coroutine layout computation? Coud we just run a fixed list of passes that does not require knowing layouts for coroutine layout computation?
View changes since the review

Upstream rustc changes [1] `fully_monomophized` to `TypingMode::Codegen` instead of `TypingMode::post_analysis`. This causes many queries to run multiple times and cause issues because these modes do not compare equal. Switch everything to use `TypingEnv::codegen` to still have a single typing env shared by all analysis. Link: rust-lang/rust#145477 [1] Signed-off-by: Gary Guo <gary@garyguo.net>

Introduce `TypingMode::Codegen` to avoid layout cycles on coroutines Computing layout of coroutines depends on their `optimized_mir`. At the same time, MIR opts can require using layouts to work. For instance to evaluate constants. This leads to cycles and clumsy workarounds. This PR creates a new typing mode for layout computations: - when a coroutine's layout is requested with `TypingMode::PostAnalysis` or earlier, return `LayourError::TooGeneric`; - when a coroutine's layout is requested with `TypingMode::Codegen`, actually compute it. `TypingMode::Codegen` is meant be be used by codegen code, and analyses that require coroutine layout, like transmute check and coroutine recursion check. With this PR, we can remove all `is_coroutine` checks from `rustc_mir_transform` and unlock simplifying coroutine MIR. Perf is not terrific. This PR causes recomputation of a few queries, and I had to insert workarounds.

frewsxcv · 2026-06-23T21:13:08Z

This fixed an ICE. See: #158328

danderson · 2026-06-23T21:32:02Z

Just to be a pessimist: it's possible my reproducer on that bug is just fragile, and this change incidentally makes that specific repro avoid the panic codepath. I defer to rustc experts though, I don't have anywhere near the knowledge of rustc internals to evaluate that.

With the new trait solver enabled by default in nightly-2026-08-22 [1], large_futures can ICE while inspecting an awaited future. The lint switches to Codegen mode to obtain coroutine layouts but passes a type whose opaque aliases are still marked rigid from type checking. Because layout_of receives an already-Codegen environment, its environment-change path does not clear those markers. Clear alias rigidity and normalize the awaited type in that environment before querying its layout. Codegen mode must remain: post-analysis layout queries cannot compute coroutine sizes [2]. The regression covers direct and pinned futures under both solvers, without other lint errors that could hide a delayed ICE. Fixesrust-lang/rust#161495. [1]: rust-lang/rust#160619 [2]: rust-lang/rust#145477

With the new trait solver enabled by default in nightly-2026-08-22 [1], large_futures can ICE while inspecting an awaited future. The lint switches to Codegen mode to obtain coroutine layouts but passes a type whose opaque aliases are still marked rigid from type checking. Because layout_of receives an already-Codegen environment, its environment-change path does not clear those markers. Clear alias rigidity and normalize the awaited type in that environment before querying its layout. Codegen mode must remain: post-analysis layout queries cannot compute coroutine sizes [2]. The regression covers direct and pinned futures under both solvers, without other lint errors that could hide a delayed ICE. Fixesrust-lang/rust#161495. changelog: [`large_futures`]: Fix an ICE on awaited futures with the next trait solver. [1]: rust-lang/rust#160619 [2]: rust-lang/rust#145477

This comment has been minimized.
Sign in to view

rust-borsBot added a commit that referenced this pull request Aug 16, 2025
Auto merge of#145477- cjgillot:codegen-mode, r=<try>
34dd67f
[EXPERIMENT] Introduce `TypingMode::Codegen` to avoid layout cycles

This comment has been minimized.
Sign in to view

rustbot added the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Aug 16, 2025

This comment has been minimized.
Sign in to view

rustbot added perf-regression Performance regression. and removed S-waiting-on-perf Status: Waiting on a perf run to be completed. labels Aug 16, 2025

cjgillot force-pushed the codegen-mode branch from a35a84a to 3f6e4dbCompare August 16, 2025 14:30

This comment has been minimized.
Sign in to view

rust-borsBot added a commit that referenced this pull request Aug 16, 2025
Auto merge of#145477- cjgillot:codegen-mode, r=<try>
4e9488c
[EXPERIMENT] Introduce `TypingMode::Codegen` to avoid layout cycles

rustbot added the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Aug 16, 2025

This comment has been minimized.
Sign in to view

rustbot removed the S-waiting-on-perf Status: Waiting on a perf run to be completed. label Aug 16, 2025

This comment has been minimized.
Sign in to view

cjgillot force-pushed the codegen-mode branch from 7bde1fe to bdb8bf2Compare August 23, 2025 18:31

This comment has been minimized.
Sign in to view

cjgillot force-pushed the codegen-mode branch from 57ce693 to ac71b54Compare June 3, 2026 14:16

rust-borsBot added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. labels Jun 3, 2026

This comment has been minimized.
Sign in to view

rust-borsBot added merged-by-bors This PR was explicitly merged by bors. and removed S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. labels Jun 4, 2026

rust-borsBot merged commit 76dfce2 into rust-lang:mainJun 4, 2026
13 checks passed

rustbot added this to the 1.98.0 milestone Jun 4, 2026

This was referenced Jun 4, 2026
Remove 'static requirement on try_as_dyn #150161
Merged
Reduce verbosity of cycle errors when possible #157342
Merged

cjgillot deleted the codegen-mode branch June 4, 2026 08:07

nbdd0121 reviewed Jun 5, 2026
View reviewed changes

danderson mentioned this pull request Jun 23, 2026
[ICE]: Found unstable fingerprints for evaluate_obligation(...): Ok(EvaluatedToOk) #158328
Open

matthiaskrgr mentioned this pull request Jul 5, 2026
[ICE]: enum variant discr was too generic to eval#158797
Open

chenyukang mentioned this pull request Jul 18, 2026
1.98 beta regression: "queries overflow the depth limit!" #159427
Closed

tamird mentioned this pull request Aug 22, 2026
Fix large_futures ICE with the next solver rust-lang/rust-clippy#17601
Merged

Uh oh!

Uh oh!

Conversation

cjgillot commented Aug 16, 2025• edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

cjgillot commented Aug 16, 2025

Uh oh!

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

rust-borsBot commented Aug 16, 2025

Uh oh!

This comment has been minimized.

rust-timer commented Aug 16, 2025

Overall result: ❌ regressions - please read the text below

Uh oh!

cjgillot commented Aug 16, 2025

Uh oh!

This comment has been minimized.

This comment has been minimized.

This comment has been minimized.

rust-borsBot commented Aug 16, 2025

Uh oh!

This comment has been minimized.

This comment has been minimized.

rust-timer commented Aug 16, 2025

Overall result: ❌ regressions - please read the text below

Uh oh!

This comment has been minimized.

bors commented Aug 23, 2025

Uh oh!

This comment has been minimized.

rustbot commented Jun 3, 2026

Uh oh!

cjgillot commented Jun 3, 2026

Uh oh!

rust-borsBot commented Jun 3, 2026

Uh oh!

This comment has been minimized.

rust-borsBot commented Jun 4, 2026

Uh oh!

Uh oh!

github-actionsBot commented Jun 4, 2026

Test differences

Job duration changes

Uh oh!

rust-timer commented Jun 4, 2026

Overall result: ❌✅ regressions and improvements - please read:

Uh oh!

lcnr commented Jun 4, 2026

Uh oh!

nbdd0121Jun 5, 2026• edited by rustbot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

frewsxcv commented Jun 23, 2026

Uh oh!

danderson commented Jun 23, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

11 participants

cjgillot commented Aug 16, 2025•
edited
Loading

nbdd0121Jun 5, 2026•
edited by rustbot
Loading