Get rid of `ToPolyTraitRef`
It's generally a footgun, since it throws away `PredicatePolarity`.
This PR doesn't attempt to fix any related bugs having to do with binders or polarity; it just tries to pass through `TraitPredicate`s around instead of `TraitRef`s. There should be basically no functional changes.
Fix ICE in resolving associated items as non-bindings
Fixes#135614 so that imported associated functions of traits can be shadowed by local bindings and associated constants of traits can be used in patterns.
Drop MIPS glibc 2.23 patches that reside in crosstool-ng now
These patches were added to crosstool-ng in b88d338516 and are therefore duplicate and fail to apply, breaking builds of `dist-mips*-linux`.
I have compile tested `dist-mipsel-linux`, I assume the other targets will work just as fine now.
Rollup of 7 pull requests
Successful merges:
- #133700 (const-eval: detect more pointers as definitely not-null)
- #135290 (Encode constraints that hold at all points as logical edges in location-sensitive polonius)
- #135478 (Run clippy for rustc_codegen_gcc on CI)
- #135583 (Move `std::pipe::*` into `std::io`)
- #135612 (Include x scripts in tarballs)
- #135624 (ci: mirror buildkit image to ghcr)
- #135661 (Stabilize `float_next_up_down`)
r? `@ghost`
`@rustbot` modify labels: rollup
Update cargo
13 commits in 088d496082726091024f1689c124a0c3dccbd775..045bf21b36a2e1f3ed85e38278d1c3cc4305e134
2025-01-10 20:10:21 +0000 to 2025-01-17 14:59:36 +0000
- created a function for user defined aliases (rust-lang/cargo#15076)
- took the functionality of the third party subcommand from the list_commands function (rust-lang/cargo#15075)
- fix: wrong concat and field name (rust-lang/cargo#15074)
- fix(publish): Report all unpublishable packages (rust-lang/cargo#15070)
- docs(cargo-clippy): correct typo (rust-lang/cargo#15072)
- docs(cargo-package): alwasy include the lockfile (rust-lang/cargo#15067)
- docs(ref): Deprecate 'package.authors' (rust-lang/cargo#15068)
- fix(build-std): parse as comma-separated list (rust-lang/cargo#15065)
- Fix benchsuite issue with newer versions of git (rust-lang/cargo#15069)
- Document that cargo automatically registers variables used in env! macro to trigger rebuilds (rust-lang/cargo#15062)
- perf(cargo-package): match certain path prefix with pathspec (rust-lang/cargo#14997)
- Clarify note in example (rust-lang/cargo#15054)
- chore(deps): update msrv (3 versions) to v1.82 (rust-lang/cargo#15050)
Encode constraints that hold at all points as logical edges in location-sensitive polonius
Currently, with the full setup in #134980 (but is from #134268), the polonius location-sensitive analysis converts `Locations::All` typeck constraints as edges at all points in the CFG. This was temporary.
There's a FIXME about that already, and this PR implements it: we now use the constraints that hold at all points during traversal instead of eagerly materializing them as physical edges.
Another easy one `@jackh726.`
This fixes the slowness that was happening on the big CFG from the `saturating-float-casts` test (because of its 12M materialized edges) without, AFAICT, simply moving this overhead to traversal: materializing the logical edges is done on-demand.
r? `@jackh726` (no rush either)
const-eval: detect more pointers as definitely not-null
This fixes https://github.com/rust-lang/rust/issues/133523 by making the `scalar_may_be_null` check smarter: for instance, an odd offset in any 2-aligned allocation can never be null, even if it is out-of-bounds.
More generally, if an allocation with unknown base address B is aligned to alignment N, and a pointer is at offset X inside that allocation, then we know that `(B + X) mod N = B mod N + X mod N = X mod N`. Since `0 mod N` is definitely 0, if we learn that `X mod N` is *not* 0 we can deduce that `B + X` is not 0.
This is immediately visible on stable, via `ptr.is_null()` (and, more subtly, by not raising a UB error when such a pointer is used somewhere that a non-null pointer is required). Therefore nominating for `@rust-lang/lang.`
Making these separate types from `CovTerm` and `Expression` was historically
very helpful, but now that most of the counter-creation work is handled by
`node_flow` they are no longer needed.
- Move `make_bcb_counters` out of `CoverageCounters`
- Split out `make_node_counter_priority_list`
- Flatten `Transcriber` into the function `transcribe_counters`
Bump Fuchsia integration commit
This advances Fuchsia to a checkout from 2025-01-13, which corresponds to a recent Rust roll, and hopefully avoids #135667, where a repository used by the older version of Rust was accidentally archived and broke checking out the prior version.
try-job: x86_64-fuchsia
cc `@ehuss`
new solver: prefer trivial builtin impls
As discussed [on zulip](https://rust-lang.zulipchat.com/#narrow/channel/364551-t-types.2Ftrait-system-refactor/topic/needs_help.3A.20trivial.20builtin.20impls), this PR:
- adds a new `BuiltinImplSource::Trivial` source, and marks the `Sized` builtin impls as trivial
- prefers these trivial builtin impls in `merge_trait_candidates`
The comments can likely be wordsmithed a bit better, and I ~stole~ was inspired by the old solver ones. Let me know how you want them improved.
When enabling the new solver for tests, 3 UI tests now pass:
- `regions/issue-26448-1.rs` and its sibling `regions/issue-26448-2.rs` were rejected by the new solver but accepted by the old one
- and `issues/issue-42796.rs` where the old solver emitted some overflow errors in addition to the expected error
(For some reason one of these tests is run-pass, but I can take care of that another day)
r? lcnr
Make tidy warn on unrecognized directives
This PR makes it so tidy warns on unrecognized directives, as recommended on [the discussion of #130984](https://github.com/rust-lang/rust/issues/130984#issuecomment-2589284620). This is edited from the previous version of this PR, which only warned on "tidy-ignore" and no other tidy directive typos.
Fixes#130984.
``@rustbot`` label A-tidy C-enhancement
This advances Fuchsia to a checkout from 2025-01-13, which corresponds
to a recent Rust roll, and hopefully avoids #135667, where a repository
used by the older version of Rust was accidentally archived and broke
checking out the prior version.
try-job: x86_64-fuchsia
Stable Hash: Ignore all HirIds that just identify the node itself
This should provide better incremental caching, but it seems there is more to it.
These IDs also serve no purpose being in the stable hash of the item they refer to, only when referring to *another* item is it important that we hash the `HirId`. So we can at least avoid the cost during stable hashing, even if we don't benefit from it by avoiding some queries' caches from being invalidated
Unsure how to make sure we do this right by construction. Would be nice to do something type based
Instead of materializing `Locations::All` constraints as physical edges
at all the points in the CFG, we record them as logical edges and only
materialize them during traversal as successors for a given node.
This fixes the slowness/hang in the `saturating-float-casts.rs` test.
