regression: as_place unwrap None

[INFO] [stderr] thread 'rustc' panicked at compiler/rustc_mir_build/src/build/expr/as_place.rs:253:31:
[INFO] [stderr] called `Option::unwrap()` on a `None` value

Might be related to #110453

@rustbot

I've tried the first 2 crates (bonedaddy.ellidri and edu-sync), they compile on the current stable fail to compile on beta and nightly.

Provisionally assigning P-high waiting for more investigation (Zulip discussion).

@rustbot label -I-prioritize +P-high +T-compiler

@rustbot

bisected to commit 7e4f5ecde93e629a44ad3678059c5cd364791d27, where in this commit originates the issue:
Regression in rust-lang-ci@7e4f5ec i.e. rollup PR #113120

Implement proposed API for proc_macro_span #111571 (Implement proposed API for proc_macro_span)
Simplify computation of killed borrows #112236 (Simplify computation of killed borrows)
More ImplSource nits #112867 (More ImplSource nits)
add note for non-exhaustive matches with guards #113019 (add note for non-exhaustive matches with guards)
Fix invalid HTML DIV tag used in HEAD #113094 (Fix invalid HTML DIV tag used in HEAD)
add myself to review for t-types stuff #113111 (add myself to review for t-types stuff)

@rustbot label -e-needs-bisection

@jhpratt

Bisection probably leads to #111571 cc @jhpratt

TIL to look at the perf builds results, they help getting more context. They told me tha PR 111571 is actually commit 7e4f5ecde93e629a44ad3678059c5cd364791d27

I'll take a look at this after RustConf.

@apiraino

@apiraino Is it possible the bisection contained an error? I'm not seeing any possible connection. The errors seem unrelated, .unwrap() is not used anywhere in my PR, and proc_macro is not even present in any of the three linked logs.

@oli-obk

Ok, I give up on reproducing using the crate ellidri (in the latest nightlies the error changed, I can't track the original error anymore). So I've started again using the crate edu-sync I now have a stable reproducible.

I am now a bit more confident the regression is in 34ccd04 (specifically #114022) cc @oli-obk (I see in that patch some code that is in the stacktraces).

sorry @jhpratt for the ping and thanks for pointing me into the right direction :)

No worries! Someone mentioned that it may have been incorrect at RustConf, but I wanted to have a chance to look into it myself before saying anything.

I'll be unsubscribing from this issue; please ping if something needs my attention :)

small, but not yet minimal repro. Should be good enough for the test suite. I'm looking into a fix now.

use std::future::Future;

async fn bop() {
    fold(run(), |mut foo| async move {
        &mut foo.bar;
    })
}

fn fold<Fut, F, U>(_: Foo<U>, f: F)
where
    F: FnMut(Foo<U>) -> Fut,
{
    loop {}
}

struct Foo<F> {
    bar: Vec<F>,
}

fn run() -> Foo<impl Future<Output = ()>> {
    loop {}
}

I reverted all changes from #114022 and it does not appear to change behaviour. trying a local bisect on the minimal repro.

edit: hmm weird, it also points to that PR. Trying again
edit2: ok now it works, weird

Auto merge of rust-lang#116156 - oli-obk:opaque_place_unwrap, r=compi…

Auto merge of #116156 - oli-obk:opaque_place_unwrap, r=compiler-errors

Navigation Menu

Repository navigation

regression: as_place unwrap None #115778

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

regression: as_place unwrap None #115778

Description

Activity

matthiaskrgr commented on Sep 12, 2023

apiraino commented on Sep 13, 2023

apiraino commented on Sep 14, 2023

apiraino commented on Sep 14, 2023

jhpratt commented on Sep 14, 2023

jhpratt commented on Sep 20, 2023

apiraino commented on Sep 20, 2023

jhpratt commented on Sep 20, 2023

oli-obk commented on Sep 22, 2023

oli-obk commented on Sep 22, 2023

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions