Replacing an empty span (which an empty line is) with an empty string triggers a
debug assertion in rustc. This fixes the debug assertion by using contiguous
spans, with the same resulting suggestion.
Add rustfmt 2024 reformatting to git blame ignore
https://github.com/rust-lang/rust/pull/130724 essentially reformatted the world, so we should add it to the git blame ignore list.
Tested with `git blame compiler/rustc_abi/src/lib.rs -L1137,1146`. I first thought that I have to ignore the merge commit, but it seems like the actual commit that did the reformatting should be ignored instead.
r? ``@compiler-errors``
Fix `break_last_token`.
It currently doesn't handle the three-char tokens `>>=` and `<<=` correctly. These can be broken twice, resulting in three individual tokens. This is a latent bug that currently doesn't cause any problems, but does cause problems for #124141, because that PR increases the usage of lazy token streams.
r? `@petrochenkov`
bootstrap: Set the dylib path when building books with rustdoc
The library path is needed when the toolchain has been configured with
`[rust] rpath = false`. Otherwise, building the reference book will get
an error when it tries to run rustdoc, like:
rustdoc: error while loading shared libraries: librustc_driver-2ec457c3b8826b72.so
std: implement the `random` feature (alternative version)
Implements the ACP rust-lang/libs-team#393.
This PR is an alternative version of #129120 that replaces `getentropy` with `CCRandomGenerateBytes` (on macOS) and `arc4random_buf` (other BSDs), since that function is not suited for generating large amounts of data and should only be used to seed other CPRNGs. `CCRandomGenerateBytes`/`arc4random_buf` on the other hand is (on modern platforms) just as secure and uses its own, very strong CPRNG (ChaCha20 on the BSDs, AES on macOS) periodically seeded with `getentropy`.
Subtree sync for rustc_codegen_cranelift
The main highlight this time is a fix for a miscompilation of RwLock on macOS. Also a Cranelift update and a couple of extra vendor intrinsics on arm64.
r? `@ghost`
`@rustbot` label +A-codegen +A-cranelift +T-compiler
Build quine-mc_cluskey with `opt-level=3` in dev builds
While doing some profiling I noticed that debug clippy running on the `clippy_lints` crate spends 35s out of 160s in one specific code path of `nonminimal_bool`, which seemed a bit excessive.
I've found that just enabling optimizations for quine-mc_cluskey (used by nonminimal_bool) cuts down the part that took 35s to 3s
While this doesn't really change anything for users, this helps dogfood a bit as it cuts off about half a minute of runtime (in some of my tests, at least).
Something similar was attempted in #10576, however that involved compiling everything in release mode including clippy itself, whereas this only affects a single dependency that's compiled in parallel with something that takes longer so this should hopefully not have a negative impact in any case (and changing clippy doesn't require recompiling that dependency)
changelog: none
Update to LLVM 19.1.0
This is a branch rebase of the submodule, now that LLVM 19.1.0 is final.
Our *only* extra patch right now is the one we're carrying for SGX unwind.
Apply `EarlyOtherwiseBranch` to scalar value
In the future, I'm thinking of hoisting discriminant via GVN so that we only need to write very little code here.
r? `@cjgillot`
internal: Disable GitHub releases for now
These are currently throwing `Error: HttpError: Resource not accessible by integration` because of the organization change, let's disable them for today's release.
Rollup of 10 pull requests
Successful merges:
- #129550 (Add str.as_str() for easy Deref to string slices)
- #130344 (Handle unsized consts with type `str` in v0 symbol mangling)
- #130659 (Support `char::encode_utf16` in const scenarios.)
- #130705 (No longer mark RTN as incomplete)
- #130712 (Don't call `ty::Const::normalize` in error reporting)
- #130713 (Mark `u8::make_ascii_uppercase` and `u8::make_ascii_lowercase` as const.)
- #130714 (Introduce `structurally_normalize_const`, use it in `rustc_hir_typeck`)
- #130715 (Replace calls to `ty::Const::{try_}eval` in mir build/pattern analysis)
- #130723 (Add test for `available_parallelism()`)
- #130726 (tests: Remove spuriously failing vec-tryinto-array codegen test)
r? `@ghost`
`@rustbot` modify labels: rollup
tests: Remove spuriously failing vec-tryinto-array codegen test
This has failed more than a couple of times now. It costs real time, money, and energy to deal with this, far more than this test is saving us.
Add test for `available_parallelism()`
This is a redo of [this PR](https://github.com/rust-lang/rust/pull/104095).
I changed the location of the test as per comments in the original thread. Otherwise the test is practically the same.
try-job: test-various
Replace calls to `ty::Const::{try_}eval` in mir build/pattern analysis
We normalize consts in writeback: #130645. This means that consts are gonna be as normalized as they're ever gonna get in MIR building and pattern analysis. Therefore we can just use `try_to_target_usize` rather than calling `eval_target_usize`.
Regarding the `.expect` calls, I'm not totally certain whether they're correct given rigid unevaluated consts. But this PR shouldn't make *more* ICEs occur; we may have to squash these ICEs when mGCE comes around, tho 😺
Introduce `structurally_normalize_const`, use it in `rustc_hir_typeck`
Introduces `structurally_normalize_const` to typecking to separate the "eval a const" step from the "try to turn a valtree into a target usize" in HIR typeck, where we may still have infer vars and stuff around.
I also changed `check_expr_repeat` to move a double evaluation of a const into a single one. I'll leave inline comments.
r? ```@BoxyUwU```
I hesitated to really test this on the new solver where it probably matters for unevaluated consts. If you're worried about the side-effects, I'd be happy to craft some more tests 😄
