[perf] Delay the construction of early lint diag structs
Attacks some of the perf regressions from https://github.com/rust-lang/rust/pull/124417#issuecomment-2123700666.
See individual commits for details. The first three commits are not strictly necessary.
However, the 2nd one (06bc4fc671, *Remove `LintDiagnostic::msg`*) makes the main change way nicer to implement.
It's also pretty sweet on its own if I may say so myself.
Remove `DefId` from `EarlyParamRegion`
Currently we represent usages of `Region` parameters via the `ReEarlyParam` or `ReLateParam` variants. The `ReEarlyParam` is effectively equivalent to `TyKind::Param` and `ConstKind::Param` (i.e. it stores a `Symbol` and a `u32` index) however it also stores a `DefId` for the definition of the lifetime parameter.
This was used in roughly two places:
- Borrowck diagnostics instead of threading the appropriate `body_id` down to relevant locations. Interestingly there were already some places that had to pass down a `DefId` manually.
- Some opaque type checking logic was using the `DefId` field to track captured lifetimes
I've split this PR up into a commit for generate rote changes to diagnostics code to pass around a `DefId` manually everywhere, and another commit for the opaque type related changes which likely require more careful review as they might change the semantics of lints/errors.
Instead of manually passing the `DefId` around everywhere I previously tried to bundle it in with `TypeErrCtxt` but ran into issues with some call sites of `infcx.err_ctxt` being unable to provide a `DefId`, particularly places involved with trait solving and normalization. It might be worth investigating adding some new wrapper type to pass this around everywhere but I think this might be acceptable for now.
This pr also has the effect of reducing the size of `EarlyParamRegion` from 16 bytes -> 8 bytes. I wouldn't expect this to have any direct performance improvement however, other variants of `RegionKind` over `8` bytes are all because they contain a `BoundRegionKind` which is, as far as I know, mostly there for diagnostics. If we're ever able to remove this it would shrink the `RegionKind` type from `24` bytes to `12` (and with clever bit packing we might be able to get it to `8` bytes). I am curious what the performance impact would be of removing interning of `Region`'s if we ever manage to shrink `RegionKind` that much.
Sidenote: by removing the `DefId` the `Debug` output for `Region` has gotten significantly nicer. As an example see this opaque type debug print before vs after this PR:
`Opaque(DefId(0:13 ~ impl_trait_captures[aeb9]::foo::{opaque#0}), [DefId(0:9 ~ impl_trait_captures[aeb9]::foo::'a)_'a/#0, T, DefId(0:9 ~ impl_trait_captures[aeb9]::foo::'a)_'a/#0])`
`Opaque(DefId(0:13 ~ impl_trait_captures[aeb9]::foo::{opaque#0}), ['a/#0, T, 'a/#0])`
r? `@compiler-errors` (I would like someone who understands the opaque type setup to atleast review the type system commit, but the rest is likely reviewable by anyone)
Rollup of 8 pull requests
Successful merges:
- #124048 (Support C23's Variadics Without a Named Parameter)
- #125046 (Only allow immutable statics with #[linkage])
- #125466 (Don't continue probing for method if in suggestion and autoderef hits ambiguity)
- #125469 (Don't skip out of inner const when looking for body for suggestion)
- #125544 (Also mention my-self for other check-cfg docs changes)
- #125559 (Simplify the `unchecked_sh[lr]` ub-checks a bit)
- #125566 (Notify T-rustdoc for beta-accepted and stable-accepted too)
- #125582 (Avoid a `FieldIdx::from_usize` in InstSimplify)
r? `@ghost`
`@rustbot` modify labels: rollup
Notify T-rustdoc for beta-accepted and stable-accepted too
Otherwise, it's unclear when the nomination label is removed whether the backport was accepted, thus nomination removed, or if the backport was rejected, thus nomination removed.
r? ````@GuillaumeGomez````
Also mention my-self for other check-cfg docs changes
This PR adds a mention for my-self for the recently added `src/doc/rustc/src/check-cfg` directory.
*I had to add a second mention just for the directory since [`Path::starts_with`](https://doc.rust-lang.org/std/path/struct.Path.html#method.starts_with) as used by [triagebot](48f29f351c/src/handlers/mentions.rs (L69)), matches on path components and so can never return true for a file and directory at the same time.*
Don't skip out of inner const when looking for body for suggestion
Self-explanatory title, I'll point out the important logic in an inline comment.
Fixes#125370
Don't continue probing for method if in suggestion and autoderef hits ambiguity
The title is somewhat self-explanatory. When we hit ambiguity in method autoderef steps, we previously would continue to probe for methods if we were giving a suggestion. This seems useless, and causes an ICE when we are not able to unify the receiver later on in confirmation.
Fixes#125432
Support C23's Variadics Without a Named Parameter
Fixes#123773
This PR removes the static check that disallowed extern functions
with ellipsis (varargs) as the only parameter since this is now
valid in C23.
This will not break any existing code as mentioned in the proposal
document: https://www.open-std.org/jtc1/sc22/wg14/www/docs/n2975.pdf.
Also, adds a doc comment for `check_decl_cvariadic_pos()` and
fixes the name of the function (`varadic` -> `variadic`).
Try to not reinstall tools in mingw CI
Reinstalling the tools seems prone to failure (e.g. [latest](https://github.com/rust-lang/rust/pull/125529#issuecomment-2130919307)) and is more work. It also seems unnecessary as CI actually uses a vendored tarball for builds.
cc `@mati865`
completely refactor how we manage blocking and unblocking threads
This hides a lot of invariants from the implementation of the synchronization primitives, and makes sure we never have to release or acquire a vector clock on another thread but the active one.
Less syscalls for the `copy_file_range` probe
If it's obvious from the actual syscall results themselves that the syscall is supported or unsupported, don't do an extra syscall with an invalid file descriptor.
CC #122052
fix(opt-dist): respect existing config.toml
This is another step toward making opt-dist work in sandboxed environments. See also <https://github.com/rust-lang/rust/pull/125465>.
opt-dist verifies the final built rustc against a subset of rustc test
suite. However it overwrote the pre-existing `config.toml` [^1],
and that results in ./vendor/ directory removed [^2].
Instead of overwriting, this patch use `--set <config-value>` to
override paths to rustc / cargo / llvm-config.
[^1]: 606afbb617/src/tools/opt-dist/src/tests.rs (L62-L77)
[^2]: 8679004993/src/bootstrap/bootstrap.py (L1057)
tidy: stop special-casing tests/ui entry limit
It is genuinely more annoying to have this error, now that this value is below the general `ENTRY_LIMIT` cap, when one is trying to clean out tests from tests/ui! This code has served its purpose well, let it rest now rather than force it to continue haunting us.
Avoid clone when constructing runnable label.
I stumbled across this when reading this code. This seems like an unnecessary allocation (though likely small?)
Use correct format for setting environment variables when debugging with cpptools
The RA VSCode extension uses an incorrect format for the environment variables in the `launch.json` when debugging with the C/C++ Extension. This extension uses a different format than CodeLLDB or NativeDebug, which means that the environment variables are not actually set for the debuggee.
What it currently looks like:
```json
"env": {
"NAME": "VALUE"
}
```
What the C/C++ extension expects:
```json
"environment": [
{ "name": "NAME", "value": "VALUE" }
]
```
For reference: https://code.visualstudio.com/docs/cpp/launch-json-reference#_environment
Turn remaining non-structural-const-in-pattern lints into hard errors
This completes the implementation of https://github.com/rust-lang/rust/issues/120362 by turning our remaining future-compat lints into hard errors: indirect_structural_match and pointer_structural_match.
They have been future-compat lints for a while (indirect_structural_match for many years, pointer_structural_match since Rust 1.75 (released Dec 28, 2023)), and have shown up in dependency breakage reports since Rust 1.78 (just released on May 2, 2024). I don't expect a lot of code will still depend on them, but we will of course do a crater run.
A lot of cleanup is now possible in const_to_pat, but that is deferred to a later PR.
Fixes https://github.com/rust-lang/rust/issues/70861
Otherwise, it's unclear when the nomination label is removed whether the
backport was accepted, thus nomination removed, or if the backport was
rejected, thus nomination removed.
Panic if `PathBuf::set_extension` would add a path separator
This is likely never intended and potentially a security vulnerability if it happens.
I'd guess that it's mostly literal strings that are passed to this function in practice, so I'm guessing this doesn't break anyone.
CC #125060
This is another step toward making opt-dist work in sandboxed environments
opt-dist verifies the final built rustc against a subset of rustc test
suite. However it overwrote the pre-existing `config.toml` [^1],
and that results in ./vendor/ directory removed [^2].
Instead of overwriting, this patch use `--set <config-value>` to
override paths to rustc / cargo / llvm-config.
[^1]: 606afbb617/src/tools/opt-dist/src/tests.rs (L62-L77)
[^2]: 8679004993/src/bootstrap/bootstrap.py (L1057)
Rollup of 7 pull requests
Successful merges:
- #121377 (Stabilize `LazyCell` and `LazyLock`)
- #122986 (Fix c_char on AIX)
- #123803 (Fix `VecDeque::shrink_to` UB when `handle_alloc_error` unwinds.)
- #124080 (Some unstable changes to where opaque types get defined)
- #124667 (Stabilize `div_duration`)
- #125472 (tidy: validate LLVM component names in tests)
- #125523 (Exit the process a short time after entering our ctrl-c handler)
r? `@ghost`
`@rustbot` modify labels: rollup
tidy: validate LLVM component names in tests
LLVM component names are not immediately obvious (they usually omit any suffixes on the target arch name), and if they're incorrect, the test will silently never run.
This happened [here](https://github.com/rust-lang/rust/pull/125220#discussion_r1612626002), and it would be nice to prevent it.
