Make layout_of cycles fatal errors

[Zoxc]

This makes layout_of cycles fatal errors.

This is a step towards removing query cycle recovery.

[rustbot]

Some changes occurred to the CTFE machinery

cc @RalfJung, @oli-obk, @lcnr

HIR ty lowering was modified

cc @fmease

Some changes occurred to the CTFE / Miri interpreter

cc @rust-lang/miri

[rustbot]

r? @TaKO8Ki

rustbot has assigned @TaKO8Ki.
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 69 candidates
• Random selection from 15 candidates

[Zoxc]

These notes are removed without #153194

[Zoxc]

Looks like Miri treats fatal errors as panics.

[rustbot]

The Miri subtree was changed

cc @rust-lang/miri

[RalfJung]

What is happening here, why is this needed now?

[Zoxc]

layout_of cycles will now unwind with FatalErrorMarker (like other fatal errors). Miri is currently treating any unwinding as a panic, which this changes by skipping the ICE message for fatal errors.

It might make sense to also print the backtrace for fatal errors case, as I guess it's similar to #85633 though a bit noisy.

[RalfJung]

Oh I see, that unwind is started via resume_unwind so the panic hook never gets called... feels kind of hacky given the function is called "resume" but oh well.

Suggested change

	if !panic_payload.is::<FatalErrorMarker>() {
	ecx.handle_ice();
	}
	// rustc "handles" some errors by unwinding with FatalErrorMarker
	// (after emitting suitable diagnostics), so do not treat those as ICEs.
	if !panic_payload.is::<FatalErrorMarker>() {
	ecx.handle_ice();
	}

[Zoxc]

resume_unwind doesn't have the best name, it always starts unwinding. It probably should have been just unwind.

[TaKO8Ki]

Thanks. After resolving the conflicts, r=me.

[rustbot]

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.