fix(layout): pre-compute coroutine field layouts iteratively #153620
Open
weihanglo wants to merge 2 commits intorust-lang:mainfrom
Open
fix(layout): pre-compute coroutine field layouts iteratively #153620weihanglo wants to merge 2 commits intorust-lang:mainfrom
weihanglo wants to merge 2 commits intorust-lang:mainfrom
Conversation
When computing coroutine layout, each saved local is wrapped in `MaybeUninit<T>`, which chains through `ManuallyDrop<T>` and `MaybeDangling<T>` before reaching `T`. This adds 3 extra layout_of queries per field. For coroutines with deeply nested field types, these extra queries push the query depth past the recursion limit. Issue: rust-lang/rust 152942
This commit fixes the regression described in oroutine-query-depth.rs by walking all coroutine field types iteratively, collecting transitive dependencies, then computing their layouts bottom-up. Each `layout_of` call then finds its dependencies already cached, and then we can bound max query depth to `current_depth + 1` regardless of type nesting depth. Issue: rust-lang/rust 152942
rustbot
added
S-waiting-on-author
Collaborator
|
|
rustbot
added
S-waiting-on-review
weihanglo
commented
Mar 9, 2026
Comment on lines
571
to
577
| .iter() | ||
| .map(|local| { | ||
| let field_ty = EarlyBinder::bind(local.ty); | ||
| let uninit_ty = Ty::new_maybe_uninit(tcx, field_ty.instantiate(tcx, args)); | ||
| cx.spanned_layout_of(uninit_ty, local.source_info.span) | ||
| }) | ||
| .try_collect::<IndexVec<_, _>>()?; |
Member
Author
There was a problem hiding this comment.
Or the fix could be that we don't wrap it in new_maybe_uninit but then we need to set uninhabited = false in all variants. I don't know what that implies though especially for some niche optimization
Member
Author
|
😭 it failed. |
Collaborator
|
The job Click to see the possible cause of the failure (guessed by this bot) |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
When computing coroutine layout, each saved local is wrapped in
MaybeUninit<T>, which chains throughManuallyDrop<T>andMaybeDangling<T>before reachingT. This adds 3 extra layout_of queries per field.For coroutines with deeply nested field types, these extra queries push the query depth past the recursion limit.
This commit fixes the regression described in
coroutine-query-depth.rsby walking all coroutine field types iteratively, collecting transitive dependencies, then computing their layouts bottom-up.Each
layout_ofcall then finds its dependencies already cached, and then we can bound max query depth tocurrent_depth + 1regardless of type nesting depth.Fixes #152942
r? WaffleLapkin