Previously, we were checking the LLVM stamp before fetching the submodule
which leads to not being able to compile llvm on submodule updates.

Signed-off-by: onur-ozkan <work@onurozkan.dev>

Signed-off-by: onur-ozkan <work@onurozkan.dev>

…ts passed to the driver

It's been 7 months since we deprecated this. It should be fine to remove it now.

Signed-off-by: onur-ozkan <work@onurozkan.dev>

Signed-off-by: onur-ozkan <work@onurozkan.dev>

Windows: add basic support for FormatMessageW

fix error display for './miri run --dep'

Fixes rust-lang/miri#3463

Handle Miri sysroot entirely outside the Miri driver

(Extracted from rust-lang/miri#3409)

This entirely moves the responsibility of setting miri-sysroot to whatever *invokes* the Miri driver.  cargo-miri knows whether it is inside rustdoc or not and can adjust accordingly. I previously avoided doing that because there are a bunch of places that are invoking the driver (cargo-miri, the ui test suite, `./miri run`, `./x.py run miri`) and they all need to be adjusted now. But it is also somewhat less fragile as we usually have more information there -- and we can just decide that `./miri run file.rs --sysroot path` is not supported. The advantage of this is that the driver is reasonably clean and doesn't need magic environment variables like MIRI_SYSROOT, and we don't have to fight rustc_driver to use a different default sysroot. Everything is done in cargo-miri (and the other much simpler driver wrappers) where it can hopefully be debugged much better.

add some basic support for GetFullPathNameW

This is the last missing piece to make std `path::` tests work on Windows.

It will allow more flexible uses in the future. This makes `split_simd_to_128bit_chunks` simpler, moving some of the complexity to its callers.

We assert that we are in the new solver in the line above

Make `split_simd_to_128bit_chunks` take only one operand

It will allow more flexible uses in the future. This makes `split_simd_to_128bit_chunks` simpler, moving some of the complexity to its callers.

Bump rustc-build-sysroot to 0.4.6

Fixes rust-lang#3313

Rustfix remains TODO

The lint is unstable, and the lint group `rust_2024_compatibility` must keep working on stable

Co-authored-by: Guillaume Boisseau <Nadrieril@users.noreply.github.com>

…=albertlarsan68

fetch submodule before checking llvm stamp

Previously, we were checking the LLVM stamp before fetching the submodule which leads to not being able to compile llvm on submodule updates.

Fixes rust-lang#122612
Fixes rust-lang#122787

…meGomez

Support type '/' to search

Related topic on IRLO: https://internals.rust-lang.org/t/rustdoc-use-key-to-search-instead-of-s/20559

…chenkov

Stabilize checking of cfgs at compile-time: `--check-cfg` option

This PR stabilize the `--check-cfg` CLI option of `rustc` (and `rustdoc`) 🎉.

In particular this PR does two things:
  1. it makes the `--check-cfg` option stable
  2. and it moves the documentation to the stable books

FCP: rust-lang#82450 (comment)

Resolves rust-lang#82450

``@rustbot`` labels +S-blocked +F-check-cfg
r? ``@petrochenkov``

…ding_mode_2024, r=Nadrieril

Match ergonomics 2024: `mut` doesn't reset binding mode

r? ``@Nadrieril``

cc rust-lang#123076

``@rustbot`` label A-edition-2024 A-patterns

…albertlarsan68

drop `changelog-seen`

It's been 7 months since we deprecated this. It should be fine to remove it now.

…mpiler-errors

The new solver ignores `DefineOpaqueTypes`, so switch it to `Yes`

We assert that we are in the new solver in the line above

r? ``@compiler-errors``

Miri subtree update

r? ``@ghost``

…mpiler-errors

Change a diagnostics-path-only `DefineOpaqueTypes` to `Yes`.

This can't possibly affect compilation, so it's safe to flip, even if I couldn't come up with an affected test

r? ``@compiler-errors``

…i-obk

interpret: pass MemoryKind to before_memory_deallocation

This will be needed for rust-lang/miri#3475.

r? ``@oli-obk``

interpret: remove outdated comment

In rust-lang#107756, allocation became generally fallible, so the "only panic if there is provenance" no longer applies.

r? ``@oli-obk``