Structurally normalize types as needed in projection_ty_core
#141703
Merged
Conversation
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
rustbot
added
S-waiting-on-review
compiler-errors
commented
May 28, 2025
| ); | ||
| curr_projected_ty = projected_ty; | ||
| } | ||
| trace!(?curr_projected_ty); | ||
|
|
||
| let ty = curr_projected_ty.ty; | ||
| let ty = self.normalize(curr_projected_ty.ty, locations); |
There was a problem hiding this comment.
We have to normalize the final output type only in the old solver, since it's no longer guaranteed to be normalized.
There was a problem hiding this comment.
This normalize call actually makes it easier to hit #141708, which is why CI was failing on icu_provider.
The bug report demonstrates that it's still possible to hit today, and it's fixed by the second commit which always normalizes the type for a.
Unfortunate, but necessary unless we wanted to (e.g.) eagerly normalize all closure signatures in writeback, which presumably has its own issues.
| @@ -164,8 +168,9 @@ impl<'tcx> PlaceTy<'tcx> { | |||
| self, | |||
| tcx: TyCtxt<'tcx>, | |||
| elem: &ProjectionElem<V, T>, | |||
| mut handle_field: impl FnMut(&Self, FieldIdx, T) -> Ty<'tcx>, | |||
| mut handle_opaque_cast_and_subtype: impl FnMut(&Self, T) -> Ty<'tcx>, | |||
There was a problem hiding this comment.
It's a bit awkward for these callbacks to take a PlaceTy, since then we can't structurally normalize it without mutating it in place or something.
So I blew up the struct into its fields, which is most of the changes in this PR.
compiler-errors
commented
May 28, 2025
| pub fn field_ty(self, tcx: TyCtxt<'tcx>, f: FieldIdx) -> Ty<'tcx> { | ||
| if let Some(variant_index) = self.variant_index { | ||
| match *self.ty.kind() { | ||
| pub fn field_ty( |
There was a problem hiding this comment.
See comment below for why I split the self arg up into its fields.
| PlaceTy::from_ty(handle_opaque_cast_and_subtype(&self, ty)) | ||
| } | ||
| ProjectionElem::Field(f, fty) => PlaceTy::from_ty(handle_field( | ||
| structurally_normalize(self.ty), |
There was a problem hiding this comment.
See above, here we'd need to do something like:
{
self.ty = structurally_normalize(self.ty);
PlaceTy::from(handle_field(self, f, fty))
}
which I think is kinda awkward.
This comment has been minimized.
This comment has been minimized.
compiler-errors
force-pushed
the
deref-place
branch
from
6d0a158 to
2316797
Compare
rustbot
added
the
WG-trait-system-refactor
This comment has been minimized.
This comment has been minimized.
|
A wonderful and preexisting bug in the compiler: #141708 |
compiler-errors
force-pushed
the
deref-place
branch
from
39e741f to
26e592e
Compare
lcnr
reviewed
May 29, 2025
compiler-errors
force-pushed
the
deref-place
branch
from
26e592e to
c7821bb
Compare
compiler-errors
force-pushed
the
deref-place
branch
from
c7821bb to
9871e16
Compare
|
@bors r=lcnr |
bors
added
S-waiting-on-bors
bors
added a commit
that referenced
this pull request
May 30, 2025
Rollup of 5 pull requests Successful merges: - #141703 (Structurally normalize types as needed in `projection_ty_core`) - #141719 (Add tls_model for cygwin and enable has_thread_local) - #141736 (resolve stage0 sysroot from rustc) - #141746 (Rework `#[doc(cfg(..))]` checks as distinct pass in rustdoc) - #141749 (Remove RUSTC_RETRY_LINKER_ON_SEGFAULT hack) r? `@ghost` `@rustbot` modify labels: rollup
rust-timer
added a commit
that referenced
this pull request
May 30, 2025
Rollup merge of #141703 - compiler-errors:deref-place, r=lcnr Structurally normalize types as needed in `projection_ty_core` Introduce a `structurally_normalize` callback to `projection_ty_core`, and then use it before we match on the ty kind in `projection_ty_core`. Previously we were only structurally normalizing the return type of the `handle_field` struct, but if we were to (e.g.) apply a deref projection to that type, then the resulting type is not guaranteed to be structurally normalized and any subsequent projections applied would ICE. Fixes rust-lang/trait-system-refactor-initiative#221 I'll leave a few comments inline to explain the changes. r? lcnr --- Also fixes #141708
|
still in the queue @bors r- |
bors
added
S-waiting-on-author
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Introduce a
structurally_normalizecallback toprojection_ty_core, and then use it before we match on the ty kind inprojection_ty_core.Previously we were only structurally normalizing the return type of the
handle_fieldstruct, but if we were to (e.g.) apply a deref projection to that type, then the resulting type is not guaranteed to be structurally normalized and any subsequent projections applied would ICE.Fixes rust-lang/trait-system-refactor-initiative#221
I'll leave a few comments inline to explain the changes.
r? lcnr
Also fixes #141708