Error on recursive opaque ty in HIR typeck

[compiler-errors]

"Non-trivially recursive" opaques are opaques whose hidden types are inferred to be equal to something other than themselves. For example, if we have a TAIT like type TAIT = impl Sized, if we infer the hidden type to be TAIT := (TAIT,), that would be a non-trivial recursive definition. We don't want to support opaques that are non-trivially recursive, since they will (almost!! -- see caveat below) always result in borrowck errors, and are generally a pain to deal with.

On the contrary, trivially recursive opaques may occur today because the old solver overagerly uses replace_opaque_types_with_inference_vars. This infer var can then later be constrained to be equal to the opaque itself. These cases will not necessarily result in borrow-checker errors, since other uses of the opaque may properly constrain the opaque. If there are no other uses we may instead fall back to () today.

The only weird case that we have to unfortunately deal with was discovered in #139406:

#![allow(unconditional_recursion)]

fn what1<T>(x: T) -> impl Sized {
    what1(x)
}

fn what2<T>(x: T) -> impl Sized {
    what2(what2(x))
}

fn print_return_type<T, U>(_: impl Fn(T) -> U) {
    println!("{}", std::any::type_name::<U>())
}

fn main() {
    print_return_type(what1::<i32>); // ()
    print_return_type(what2::<i32>); // i32
}

HIR typeck eagerly replaces the return type with an infer var, ending up with RPIT<T> = RPIT<RPIT<T>> in the storage. While we return this in the TypeckResults, it's never actually used anywhere.

MIR building then results in the following statement

let _0: impl RPIT<T> /* the return place */ = build<RPIT<T>>(_some_local);

Unlike HIR typeck MIR typeck now directly equates RPIT<T> with RPIT<RPIT<T>>. This does not try to define RPIT but instead relates its generic arguments

rust/compiler/rustc_infer/src/infer/relate/type_relating.rs

Lines 185 to 190 in b9856b6

	(
	&ty::Alias(ty::Opaque, ty::AliasTy { def_id: a_def_id, .. }),
	&ty::Alias(ty::Opaque, ty::AliasTy { def_id: b_def_id, .. }),
	) if a_def_id == b_def_id => {
	super_combine_tys(infcx, self, a, b)?;
	}

This means we relate T with RPIT<T>, which results in a defining use RPIT<T> = T

I think it's pretty obvious that this is not desirable behavior, and according to the crater run there were no regressions, so let's break this so that we don't have any inference hazards in the new solver.

In the future what2 may end up compiling again by also falling back to (). However, that is not yet guaranteed and the transition to this state is made significantly harder by not temporarily breaking it on the way. It is also concerning to change the inferred hidden type like this without any notification to the user, even if likely not an issue in this concrete case.

[compiler-errors]

@bors try

[bors]

⌛ Trying commit 99e7ce9 with merge f45c1ab...

[bors]

☀️ Try build successful - checks-actions
Build commit: f45c1ab (f45c1ab62d4de6b63f69b4daa9d1f0b90c38a208)

[rust-timer]

Finished benchmarking commit (f45c1ab): comparison URL.

Overall result: no relevant changes - no action needed

Benchmarking this pull request likely means that it is perf-sensitive, so we're automatically marking it as not fit for rolling up. While you can manually mark this PR as fit for rollup, we strongly recommend not doing so since this PR may lead to changes in compiler perf.

@bors rollup=never
@rustbot label: -S-waiting-on-perf -perf-regression

Instruction count

This benchmark run did not return any relevant results for this metric.

Max RSS (memory usage)

Results (secondary -2.2%)

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-2.2%	[-2.2%, -2.2%]	1
All ❌✅ (primary)	-	-	0

Cycles

This benchmark run did not return any relevant results for this metric.

Binary size

This benchmark run did not return any relevant results for this metric.

Bootstrap: 775.534s -> 775.663s (0.02%)
Artifact size: 365.93 MiB -> 365.92 MiB (-0.00%)

[compiler-errors]

@craterbot check

[craterbot]

👌 Experiment pr-139419 created and queued.
🤖 Automatically detected try build f45c1ab
🔍 You can check out the queue and this experiment's details.

ℹ️ Crater is a tool to run experiments across parts of the Rust ecosystem. Learn more

[craterbot]

🚧 Experiment pr-139419 is now running

ℹ️ Crater is a tool to run experiments across parts of the Rust ecosystem. Learn more

[craterbot]

🎉 Experiment pr-139419 is completed!
📊 3 regressed and 2 fixed (610052 total)
📰 Open the full report.

⚠️ If you notice any spurious failure please add them to the denylist!
ℹ️ Crater is a tool to run experiments across parts of the Rust ecosystem. Learn more

[lcnr]

all spurious, can you think of a test case which actually compiled before this PR?

I would expect us to error for them already, just after rustc_borrowck

[lcnr]

This is a breaking change, see #139406. We should very much break this, explaining what's going on in that issue

[rustbot]

Some changes occurred in exhaustiveness checking

cc @Nadrieril

[rfcbot]

🔔 This is now entering its final comment period, as per the review above. 🔔

[bors]

☔ The latest upstream changes (presumably #140106) made this pull request unmergeable. Please resolve the merge conflicts.

[rfcbot]

The final comment period, with a disposition to merge, as per the review above, is now complete.

As the automated representative of the governance process, I would like to thank the author for their work and everyone else who contributed.

This will be merged soon.

[lcnr]

r=me after rebase

[lcnr]

@bors r+ rollup

[bors]

📌 Commit 47b9e37 has been approved by lcnr

It is now in the queue for this repository.