Fix hang in where-clause suggestion with predicate_can_apply

[compiler-errors]

Using predicate_may_hold during error reporting causes an evaluation overflow, which (because we use evaluate_obligation_no_overflow) then causes the predicate to need to be re-evaluated locally, which results in a hang.

... but since the "add a where clause" suggestion is best-effort, just throw any overflow errors. No need for 100% accuracy.

r? @lcnr who has been thinking about overflows... Let me know if you want more context about this issue, and as always, feel free to reassign.

Fixes #104225

[lcnr]

generally this seems fine, but why does this not result in an infinite hang if we manually try to check Vec<[[[?0; 1]; 1]; 1]>: PartialEq<?0> 🤔 something like

#![recursion_limit = "128"]

fn f<B>(x: (B,)) -> Vec<[[[B; 1]; 1]; 1]> {
    vec![]
}

fn is_eq<T: PartialEq<B>, B>(_: (B,), _: T) {}

fn main() {
    let b = loop {};
    is_eq(b, f(b));
}

[compiler-errors]

I actually tested something similar while investigating this issue -- I actually don't know why this doesn't hang but instead quickly reports an error when registered as an obligation during typeck'ing. I'll have to investigate.

[compiler-errors]

generally this seems fine, but why does this not result in an infinite hang if we manually try to check Vec<[[[?0; 1]; 1]; 1]>: PartialEq<?0>

Oh, @lcnr, this is because of this check that the inference context is tainted.

rust/compiler/rustc_trait_selection/src/traits/select/mod.rs

Lines 1098 to 1103 in 6d651a2

	if self.infcx.is_tainted_by_errors() {
	return Err(OverflowError::Error(
	ErrorGuaranteed::unchecked_claim_error_was_emitted(),
	));
	}
	self.infcx.err_ctxt().report_overflow_error(error_obligation, true);

If it's not tainted, we fatally abort on an overflow, but if it is tainted, then we'll keep evaluating a bunch of nonsense predicates leading to the hang.

Notice how this code hangs when the commented line is uncommented (which causes infcx to become tainted):

fn f() {
    // 1i32.cos();
    foo::<_>();
}

fn foo<B>() where Vec<[[[B; 1]; 1]; 1]>: PartialEq<B> {}

fn main() {}

[lcnr]

can you still add the example where the infcx is not tainted as a test? It looks like this hang is something we will have to deal with for non-fatal overflow.

r=me after that

[compiler-errors]

@bors r=lcnr

[bors]

📌 Commit 65b3a30 has been approved by lcnr

It is now in the queue for this repository.

[matthiaskrgr]

@bors r- this needs to be reblessed it seems
#104671

ailures:

---- [ui] src/test/ui/typeck/hang-in-overflow.rs stdout ----
diff of stderr:

8	   = note: required for `...` to implement `...`
9	   = note: 127 redundant requirements hidden
10	   = note: required for `...` to implement `...`
+	   = note: the full type name has been written to '$TEST_BUILD_DIR/typeck/hang-in-overflow/hang-in-overflow.long-type-13273308321832961787.txt'
11	note: required by a bound in `foo`
12	  --> $DIR/hang-in-overflow.rs:14:28
13	   |


The actual stderr differed from the expected stderr.
Actual stderr saved to /checkout/obj/build/x86_64-unknown-linux-gnu/test/ui/typeck/hang-in-overflow/hang-in-overflow.stderr
To update references, rerun the tests and pass the `--bless` flag
To only update this specific test, also pass `--test-args typeck/hang-in-overflow.rs`

error: 1 errors occurred comparing output.
status: exit status: 1
command: "/checkout/obj/build/x86_64-unknown-linux-gnu/stage2/bin/rustc" "/checkout/src/test/ui/typeck/hang-in-overflow.rs" "-Zthreads=1" "--target=x86_64-unknown-linux-gnu" "--error-format" "json" "--json" "future-incompat" "-Ccodegen-units=1" "-Zui-testing" "-Zdeduplicate-diagnostics=no" "-Cstrip=debuginfo" "--emit" "metadata" "-C" "prefer-dynamic" "--out-dir" "/checkout/obj/build/x86_64-unknown-linux-gnu/test/ui/typeck/hang-in-overflow" "-A" "unused" "-Crpath" "-Cdebuginfo=0" "-Lnative=/checkout/obj/build/x86_64-unknown-linux-gnu/native/rust-test-helpers" "-L" "/checkout/obj/build/x86_64-unknown-linux-gnu/test/ui/typeck/hang-in-overflow/auxiliary"
stdout: none
--- stderr -------------------------------
error[E0275]: overflow evaluating the requirement `&&[&&mut [&&[&[&_]]; _]]: PartialEq<&_>`
  --> /checkout/src/test/ui/typeck/hang-in-overflow.rs:8:5
   |
LL |     foo::<_>();
Some tests failed in compiletest suite=ui mode=ui host=x86_64-unknown-linux-gnu target=x86_64-unknown-linux-gnu
   |     ^^^^^^^^
   |
   = help: consider increasing the recursion limit by adding a `#![recursion_limit = "256"]` attribute to your crate (`hang_in_overflow`)
   = note: required for `[&&[&&mut [&&[&[&_]]; _]]]` to implement `PartialEq<[&_]>`
   = note: 127 redundant requirements hidden
   = note: required for `Vec<[[[&[...; _]; 1]; 1]; 1]>` to implement `PartialEq<&[&mut [&mut [&mut [&&mut [&[&[&[&&[&[&[&[&[&[&[&[&&mut [&[&[[&&[&[&[&mut [&&[&[&[&mut [&&[&[&[&&mut [&&[&[&&mut [&&[&&[&[&&[&&[&&[&[&[&&[&&[&[&&mut [&[&&[&[&&[&&mut [&&[&[&&[&&mut [&&[&[&_]]; _]]]]]]]]; _]]]]]; _]]]]]]]]]]]]; _]]]; _]]]]]]]]; _]]]]]]]; _]]]]]]]]]]]]; _]>`
   = note: the full type name has been written to '/checkout/obj/build/x86_64-unknown-linux-gnu/test/ui/typeck/hang-in-overflow/hang-in-overflow.long-type-13273308321832961787.txt'
note: required by a bound in `foo`
  --> /checkout/src/test/ui/typeck/hang-in-overflow.rs:14:28
   |
LL | fn foo<B>()
   |    --- required by a bound in this
LL | where
LL |     Vec<[[[B; 1]; 1]; 1]>: PartialEq<B>,
   |                            ^^^^^^^^^^^^ required by this bound in `foo`

[compiler-errors]

@bors r=lcnr

[bors]

📌 Commit 9decfff has been approved by lcnr

It is now in the queue for this repository.

[bors]

☔ The latest upstream changes (presumably #104776) made this pull request unmergeable. Please resolve the merge conflicts.