Fix ICE in pointer layout for DSTs with bogus Sized impls#155260
Open
GokhanKabar wants to merge 1 commit intorust-lang:mainfrom
Open
Fix ICE in pointer layout for DSTs with bogus Sized impls#155260GokhanKabar wants to merge 1 commit intorust-lang:mainfrom
GokhanKabar wants to merge 1 commit intorust-lang:mainfrom
Conversation
A bogus `impl<T: ?Sized> Sized for T {}` (rejected as E0322) was
processed by the new trait solver (`-Znext-solver=globally`), making
DSTs like `[T]` appear sized. This caused `*const [T]` to receive a
thin-pointer (Scalar) layout instead of a fat-pointer (ScalarPair)
layout. When const evaluation then performed an array-to-slice unsizing
coercion and tried to write a ScalarPair value to a Scalar-layout place,
it triggered an ICE:
"invalid immediate for given destination place: value ScalarPair(...)
does not match ABI Scalar(...)"
The fix adds a structural cross-check in the pointer layout computation:
after `is_sized` returns true, we verify via `struct_tail_for_codegen`
(which is purely structural and unaffected by bogus trait impls) that
the pointee tail is not a known DST kind (Slice/Str/Dynamic). If it is,
we skip the thin-pointer early return and fall through to compute the
correct fat-pointer layout.
Collaborator
|
r? @nnethercote rustbot has assigned @nnethercote. Use Why was this reviewer chosen?The reviewer was selected based on:
|
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Fixes #155252.
Summary
A bogus
impl<T: ?Sized> Sized for T {}(rejected as E0322) was processed by the new trait solver (-Znext-solver=globally), making DSTs like[T]appear sized. This caused*const [T]to receive a thin-pointer (Scalar) layout instead of a fat-pointer (ScalarPair) layout inrustc_ty_utils/src/layout.rs.When const evaluation then performed an array-to-slice unsizing coercion (
unsize_into_ptr) and tried to write the resultingScalarPairvalue to aScalar-layout place, it triggered an ICE:Fix
In
compiler/rustc_ty_utils/src/layout.rs, afteris_sizedreturnstruefor the pointee, we cross-check usingstruct_tail_for_codegen— a purely structural traversal that is unaffected by bogus trait impls. If the tail is a known DST kind (Slice/Str/Dynamic), we skip the thin-pointer early return and fall through to compute the correct fat-pointer (ScalarPair) layout.This approach is safe because
struct_tail_for_codegendoes not invoke the trait solver, so it cannot be misled by the bogusSizedimpl.Test
Added
tests/ui/traits/next-solver/dont-ice-on-bogus-sized-impl-in-const-block.rswhich previously ICE'd and now produces the expected E0322 + E0210 errors.