Filter PhantomData fields in CoerceShared borrowck field matching

[ikow]

Summary

The add_generic_reborrow_constraint function in compiler/rustc_borrowck/src/type_check/mod.rs matches source and dest struct fields by name when establishing borrow constraints for CoerceShared. However, it uses all_fields() which includes PhantomData fields, while the coherence check (coerce_shared_info) filters them out via collect_struct_data_fields.

This mismatch means PhantomData fields with different names between source and dest structs are silently skipped via continue (line 2524), potentially dropping lifetime constraints that should be established.

Fix

Filter out PhantomData fields before matching, consistent with the coherence check's use of collect_struct_data_fields.

Test plan

• Added tests/ui/reborrow/coerce_shared_phantom_field_names.rs — a run-pass test with CoerceShared source/dest structs that have differently-named PhantomData fields (_marker vs _lifetime)
• This test verifies the borrow constraints are correctly established even with mismatched PhantomData field names

Tracking issue: #145612 (reborrow feature)

[rustbot]

r? @JohnTitor

rustbot has assigned @JohnTitor.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

Why was this reviewer chosen?

The reviewer was selected based on:

• Owners of files modified in this PR: compiler
• compiler expanded to 73 candidates
• Random selection from 20 candidates

[workingjubilee]

@ikow Hello! Could you cite the actual code pattern in Rust source that should or should not compile but (on the current main) does or does not due to the bug in question?

It is not actually clear to me what your regression test is actually testing, because it is a run-pass test but the code also compiles and passes using a nightly compiler. I did not build main's b954122 to verify since nightly's e50aa6f worked. There is no regression that this test can prevent because a run-pass test first has to be something that doesn't compile on some previous Rust version.

[jieyouxu]

@rustbot author

[rustbot]

Reminder, once the PR becomes ready for a review, use @rustbot ready.

[oli-obk]

Please do not send us PRs that were generated that. We're a community, not just a code repository. We want contributors who want to meaningfully contribute, not just bugs that are fixed (which in your case the generated tests and code don't do). Feel free to reach out on Zulip to ask for how to best contribute and do read the rustc-dev-guide and our contribution guidelines.

This is a moderation warning. If you want to discuss your warning, reach out to moderation via email or on Zulip.