Implement coercions between `&pin (mut|const) T` and `&(mut) T` when `T: Unpin` by frank-king · Pull Request #149130 · rust-lang/rust

frank-king · 2025-11-20T09:23:55Z

This allows the following (mutual) coercions when T: Unpin:

&T <-> Pin<&T>
&mut T <-> Pin<&mut T>
&mut T -> Pin<&T>
Pin<&mut T> -> &T

Part of Pin Ergonomics.

rustbot · 2025-11-20T11:57:14Z

Some changes occurred in need_type_info.rs

cc @lcnr

Some changes occurred in src/tools/clippy

cc @rust-lang/clippy

Some changes occurred in exhaustiveness checking

cc @Nadrieril

rustbot · 2025-11-20T11:57:17Z

r? @nnethercote

rustbot has assigned @nnethercote.
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

nnethercote · 2025-11-21T10:52:54Z

I am not a good person to review this. How about...

r? @lcnr

lcnr · 2025-11-24T18:57:32Z

r? types

spastorino · 2025-11-24T21:03:36Z

r? types

bors · 2025-12-05T16:16:58Z

☔ The latest upstream changes (presumably #148602) made this pull request unmergeable. Please resolve the merge conflicts.

bors · 2025-12-18T04:09:04Z

☔ The latest upstream changes (presumably #150115) made this pull request unmergeable. Please resolve the merge conflicts.

jackh726

Incredibly sorry about the review latency here.

Overall, the changes look not bad. That being said, there are two very mechanical changes that would be useful to split into a separate PR (or really, a separate commit is fine, but I prefer a separate PR since they stand alone well).

View changes since this review

compiler/rustc_middle/src/ty/adjustment.rs

compiler/rustc_middle/src/ty/sty.rs

rustbot · 2026-01-23T07:21:47Z

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

frank-king · 2026-01-30T03:52:32Z

I've split the two refactors to separated PRs:

refactor: remove Ty::pinned_ref in favor of Ty::maybe_pinned_ref #151849
refactor: add an enum DerefAdjustKind in favor of Option<OverloadedDeref> #151850

and a post-refactor PR #151932 (remove Adjust::ReborrowPin)

@rustbot ready

…r=jackh726 refactor: remove `Ty::pinned_ref` in favor of `Ty::maybe_pinned_ref` Also returns the `Region` of the reference type in `Ty::maybe_pinned_Ref`. Part of rust-lang#149130. r? jackh726

…r=jackh726 refactor: add an `enum DerefAdjustKind` in favor of `Option<OverloadedDeref>` Part of rust-lang#149130. r? jackh726

Rollup merge of #151849 - frank-king:refactor/pin-coerce-1, r=jackh726 refactor: remove `Ty::pinned_ref` in favor of `Ty::maybe_pinned_ref` Also returns the `Region` of the reference type in `Ty::maybe_pinned_Ref`. Part of #149130. r? jackh726

Rollup merge of #151850 - frank-king:refactor/pin-coerce-2, r=jackh726 refactor: add an `enum DerefAdjustKind` in favor of `Option<OverloadedDeref>` Part of #149130. r? jackh726

rustbot · 2026-02-01T01:37:23Z

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.

jackh726 · 2026-02-01T03:47:19Z

compiler/rustc_hir_typeck/src/coercion.rs

+                if self.tcx.features().pin_ergonomics()
+                    && a.pinned_ty().is_some_and(|ty| ty.is_ref())
+                    && let Ok(coerce) = self.commit_if_ok(|_| self.coerce_maybe_pinned_ref(a, b))
+                {
+                    return Ok(coerce);
+                }


I am sort of thinking that this code (and the lines added under the ty::Adt arm, should go in coerce_to_ref and coerce_to_pin_ref respectively - keeping this function relatively clean.

This also would help to eliminate some span_bugs from coerce_maybe_pinned_ref

compiler/rustc_hir_typeck/src/coercion.rs

jackh726 · 2026-02-01T03:58:10Z

compiler/rustc_hir_typeck/src/coercion.rs

+        let Some((a_ty, a_pinnedness, a_mutbl, a_region)) = a.maybe_pinned_ref() else {
+            span_bug!(span, "expect pinned reference or reference, found {:?}", a);
+        };
+        let Some((_b_ty, b_pinnedness, b_mutbl, _b_region)) = b.maybe_pinned_ref() else {
+            span_bug!(span, "expect pinned reference or reference, found {:?}", b);
+        };


Something feels a bit off here to me...I like have one function that covers "all the pinned ref" coercions - but simultaneously all the span_bugs here scream this function being ripe for getting misused or logic changing elsewhere but not updated here.

I sort of wonder if everything before let mut coerce = self.unify_and needs to be done in the calling functions, and necessarily bits get passed in.

I'm not sure if I fully get your ideas. I removed the coerce_maybe_pinned_ref and split it into coerce_to_pin_ref and coerce_pin_ref_to_ref.

Theoretically, coerce_pin_ref_to_ref can be merged into coerce_to_ref, but I have no idea how to add an autoref before the autoderef (I mean, the Autoderef iterator handles &mut Pin<&mut T> to &mut T and &Pin<&T> to &T well, but for Pin<&mut T> and Pin<&T>, there need to be an autoref at the beginning). I'm also not sure if doing that would mess up the existing code.

Skimming over, I think your changes look better, but I'm noticing there are test changes, so there are semantic changes i need to think about. Will do a proper review this week.

compiler/rustc_hir_typeck/src/expr_use_visitor.rs

Use `coerce_to_pin_ref` and `coerce_pin_ref_to_ref` instead.

jackh726 · 2026-02-20T17:54:02Z

compiler/rustc_hir_typeck/src/coercion.rs

-                if self.tcx.features().pin_ergonomics()
-                    && self.tcx.is_lang_item(pin.did(), hir::LangItem::Pin) =>
+            _ if self.tcx.features().pin_ergonomics()
+                && let Some((_, ty::Pinnedness::Pinned, mut_b, _)) = b.maybe_pinned_ref() =>


Should be able to replace this with something like let Ok(pin_coerce) = self.commit_if_ok(|_| self.coerce_to_pin_ref(a, b)) like above?

jackh726 · 2026-02-20T18:22:34Z

compiler/rustc_hir_typeck/src/coercion.rs

+                && let Some((_, ty::Pinnedness::Pinned, mut_b, _)) = b.maybe_pinned_ref() =>
            {
-                if a.is_ref() && b.pinned_ty().is_some_and(|ty| ty.is_ref()) {
-                    return self.coerce_maybe_pinned_ref(a, b);


Okay, so I think this is (one of the causes) of the test changes (but similar analysis for the above arm too). Basically, before we "committed" to this coercion, and returned whatever result that gave - and that could be mutability errors.

Now, we aren't committing to this coercion, but are actually falling back to other coercions if this ever encounters a type error (including mutability) - even though we'll never actually be able to fulfill that coercion.

I think keeping the types differ in mutability error is useful. So, we sort of need to separate the notion of "this was the right coercion to try, but failed" from "this wasn't the right coercion to try". Unfortunately, because we do want this to be wrapped in a snapshot, it's kind of not super clean. I think there are two options:

Split the first part of coerce_to_pin_ref and coerce_pin_ref_to_ref into separate functions that return an Option<XYZ> (where XYZ is going to be the data that you pass to coerce_to_pin_ref and coerce_pin_ref_to_ref (either a tuple, or a new struct - probably a new struct would be better).

You make coerce_to_pin_ref and coerce_pin_ref_to_ref return not CoerceResult<'tcx>, but instead a InferResult<'tcx, Option<(Vec<Adjustment<'tcx>>, Ty<'tcx>)>>, then coerce_to_pin_ref and coerce_pin_ref_to_ref return Ok(None) if the coercion path didn't make sense. Technically, this isn't quite perfect, because we actually commit a snapshot for a path that we didn't want - but there should be no vars or anything created, so nothing should actually be committed. But, it means that it's easier for somebody to mess up and reorder things in those functions that do.

I think option 1 is better.

rustbot added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. T-compiler Relevant to the compiler team, which will review and decide on the PR/issue. labels Nov 20, 2025

This comment has been minimized.

Sign in to view

frank-king force-pushed the feature/pin-coerce branch from 29a89ae to 92596db Compare November 20, 2025 09:56

rustbot added the T-clippy Relevant to the Clippy team. label Nov 20, 2025

frank-king marked this pull request as ready for review November 20, 2025 11:57

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. labels Nov 20, 2025

rustbot assigned nnethercote Nov 20, 2025

rustbot assigned lcnr and unassigned nnethercote Nov 21, 2025

rustbot added the T-types Relevant to the types team, which will review and decide on the PR/issue. label Nov 24, 2025

rustbot assigned spastorino and unassigned lcnr Nov 24, 2025

rustbot assigned jackh726 and unassigned spastorino Nov 24, 2025

frank-king force-pushed the feature/pin-coerce branch from 92596db to 500d786 Compare December 15, 2025 02:53

This comment has been minimized.

Sign in to view

frank-king mentioned this pull request Dec 18, 2025

Continue Experimentation with Pin Ergonomics rust-lang/rust-project-goals#389

Open

4 tasks

jackh726 requested changes Jan 23, 2026

View reviewed changes

compiler/rustc_middle/src/ty/adjustment.rs Show resolved Hide resolved

compiler/rustc_middle/src/ty/sty.rs Outdated Show resolved Hide resolved

rustbot added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Jan 23, 2026

This was referenced Jan 30, 2026

refactor: remove Ty::pinned_ref in favor of Ty::maybe_pinned_ref #151849

Merged

refactor: add an enum DerefAdjustKind in favor of Option<OverloadedDeref> #151850

Merged

frank-king force-pushed the feature/pin-coerce branch from 500d786 to 2ef4a8a Compare January 30, 2026 03:48

This comment has been minimized.

Sign in to view

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. labels Jan 30, 2026

This comment has been minimized.

Sign in to view

rust-timer added a commit that referenced this pull request Jan 31, 2026

Unrolled build for #151850

2828499

Rollup merge of #151850 - frank-king:refactor/pin-coerce-2, r=jackh726 refactor: add an `enum DerefAdjustKind` in favor of `Option<OverloadedDeref>` Part of #149130. r? jackh726

frank-king force-pushed the feature/pin-coerce branch from 2ef4a8a to 8dd903c Compare February 1, 2026 01:37

Support coercion between references and pinned references

0422097

frank-king force-pushed the feature/pin-coerce branch from 8dd903c to 0422097 Compare February 1, 2026 01:52

frank-king mentioned this pull request Feb 1, 2026

refactor: remove Adjust::ReborrowPin #151932

Open

jackh726 requested changes Feb 1, 2026

View reviewed changes

rustbot added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Feb 1, 2026

Remove coerce_maybe_pinned_ref.

ae0c73c

Use `coerce_to_pin_ref` and `coerce_pin_ref_to_ref` instead.

frank-king force-pushed the feature/pin-coerce branch from bc088dd to ae0c73c Compare February 4, 2026 09:02

jackh726 requested changes Feb 20, 2026

View reviewed changes

Uh oh!

Comments

Conversation

frank-king commented Nov 20, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

This comment has been minimized.

rustbot commented Nov 20, 2025

Uh oh!

rustbot commented Nov 20, 2025

Uh oh!

nnethercote commented Nov 21, 2025

Uh oh!

lcnr commented Nov 24, 2025

Uh oh!

spastorino commented Nov 24, 2025

Uh oh!

bors commented Dec 5, 2025

Uh oh!

This comment has been minimized.

bors commented Dec 18, 2025

Uh oh!

jackh726 left a comment • 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!

Uh oh!

Uh oh!

rustbot commented Jan 23, 2026

Uh oh!

This comment has been minimized.

frank-king commented Jan 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

This comment has been minimized.

rustbot commented Feb 1, 2026

Uh oh!

jackh726 Feb 1, 2026

Choose a reason for hiding this comment

Uh oh!

jackh726 Feb 1, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

jackh726 Feb 1, 2026

Choose a reason for hiding this comment

Uh oh!

frank-king Feb 4, 2026

Choose a reason for hiding this comment

Uh oh!

jackh726 Feb 8, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

jackh726 Feb 20, 2026

Choose a reason for hiding this comment

Uh oh!

jackh726 Feb 20, 2026

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

8 participants

frank-king commented Nov 20, 2025 •

edited

Loading

jackh726 left a comment •

edited by rustbot

Loading

frank-king commented Jan 30, 2026 •

edited

Loading