doing so requires overwriting global cache entries and
generally adds significant complexity to the solver. This is
also only ever done for root goals, so it feels easier to wrap
the `evaluate_canonical_goal` in an ordinary query if
necessary.

this allows us to only sometimes disable the global cache.

this makes it easier to maintain and modify going forward.
There may be a small performance cost as we now need to
access the provisional cache *and* walk through the stack
to detect cycles. However, the provisional cache should be
mostly empty and the stack should only have a few elements
so the performance impact is likely minimal.

Given the complexity of the search graph maintainability
trumps linear performance improvements.

To work around coherence issue. Also adds regression test.

…Amanieu

Stabilize `asm_const`

tracking issue: rust-lang#93332

reference PR: rust-lang/reference#1556

this will probably require some CI wrangling (and a rebase), so let's get that over with even though the final required PR is not merged yet.

r? `@ghost`

…rrors

`-Znext-solver` caching

This PR has two major changes while also fixing multiple issues found via fuzzing.

The main optimization is the ability to not discard provisional cache entries when popping the highest cycle head the entry depends on. This fixes the hang in Fuchsia with `-Znext-solver=coherence`.

It also bails if the result of a fixpoint iteration is ambiguous, even without reaching a fixpoint. This is necessary to avoid exponential blowup if a coinductive cycle results in ambiguity, e.g. due to unknowable candidates in coherence.

Updating stack entries pretty much exclusively happens lazily now, so `fn check_invariants` ended up being mostly useless and I've removed it. See https://gist.github.com/lcnr/8de338fdb2685581e17727bbfab0622a for the invariants we would be able to assert with it.

For a general overview, see the in-process update of the relevant rustc-dev-guide chapter: https://hackmd.io/1ALkSjKlSCyQG-dVb_PUHw

r? ```@compiler-errors```

…=scottmcm

Explicitly specify type parameter on FromResidual for Option and ControlFlow.

~~Remove type parameter default `R = <Self as Try>::Residual` from `FromResidual`~~ _Specify default type parameter on `FromResidual` impls in the stdlib_ to work around rust-lang#99940 / rust-lang#87350 ~~as mentioned in rust-lang#84277 (comment).

This does not completely fix the issue, but works around it for `Option` and `ControlFlow` specifically (`Result` does not have the issue since it already did not use the default parameter of `FromResidual`).

~~(Does this need an ACP or similar?)~~ ~~This probably needs at least an FCP since it changes the API described in [the RFC](rust-lang/rfcs#3058). Not sure if T-lang, T-libs-api, T-libs, or some combination (The tracking issue is tagged T-lang, T-libs-api).~~ This probably doesn't need T-lang input, since it is not changing the API of `FromResidual` from the RFC? Maybe needs T-libs-api FCP?

…type, r=lcnr

Record the correct target type when coercing fn items/closures to pointers

Self-explanatory. We were previously not recording the *target* type of a coercion as the output of an adjustment. This should remedy that.

We must also modify the function pointer casts in MIR typeck to use subtyping, since those broke since rust-lang#118247.

r? lcnr

Port `run-make/sysroot-crates-are-unstable` to rmake

I already have a more elaborate draft at rust-lang#126231 that tries to port the underlying Python script to rmake, but there's no need for the removal of Makefiles to be held up on complex tasks like that, so this PR simply takes the trivial Makefile and converts it into a trivial rmake recipe.

Part of rust-lang#121876.

r? ``@jieyouxu``

Make the rendered html doc for rustc better

This PR adds `|` to make the html doc of [`rustc_error::Level`](https://doc.rust-lang.org/1.80.0/nightly-rustc/rustc_errors/enum.Level.html) rendered better. Previsouly it looks good in the source code, but not rendered correctly in the html doc.

r? `@GuillaumeGomez`