drop deprecated value `if-available` for `download-ci-llvm` option
It's been 5 months since we deprecated this. It should be fine to drop its support now.
Abort a process when FD ownership is violated
When an owned FD has already been closed before it's dropped that means something else touched an FD in ways it is not allowed to. At that point things can already be arbitrarily bad, e.g. clobbered mmaps. Recovery is not possible.
All we can do is hasten the fire.
Unlike the previous attempt in #124130 this shouldn't suffer from the possibility that FUSE filesystems can return arbitrary errors.
Bootstrap: Check validity of `--target` and `--host` triples before starting a build
Resolves#122128
As described in the issue, validating the `target` and `host` triples would save a lot of time before actually starting a build. This would also check for custom targets by looking for a valid JSON spec if the specified target does not exist in the [supported](42825768b1/compiler/rustc_target/src/spec/mod.rs (L1401-L1689)) list of targets.
Unconditionally call `really_init` on GNU/Linux
This makes miri not diverge in behavior, it fixes running Rust linux-gnu binaries on musl with gcompat, it fixes dlopen edge-cases that cranelift somehow hits, etc.
Fixes#124126
thou hast gazed into this abyss with me:
r? ``@ChrisDenton``
bootstrap: Document `struct Builder` and its fields
I'm exploring the code of bootstrap and had a bit of a hard time understanding exactly what `Builder` is for at first. I decided to help document it and its field to help future explorers.
bootstrap: Describe build_steps modules
One of my preferred ways to understand source code is to start with its API. This implies the code is documented reasonably accurately, even if it is a private API. The description of one of these modules had not been updated since 2015 and so was both terse and confusing, so I rewrote it. Then I noticed many others went unremarked, so I offered some remarks.
`x vendor`
This PR implements `x vendor` on bootstrap; enabling dependency vendoring without the need for developers to have `cargo` installed on their system (previously, we suggested running `cargo vendor ...` but now we can accomplish the same task with `x vendor`).
In addition, fixes#112391 problem.
Rollup of 7 pull requests
Successful merges:
- #124370 (Fix substitution parts having a shifted underline in some cases)
- #124394 (Fix ICE on invalid const param types)
- #124425 (Do not ICE on invalid consts when walking mono-reachable blocks)
- #124434 (Remove lazycell and once_cell from compiletest dependencies)
- #124437 (doc: Make the `mod.rs` in the comment point to the correct location)
- #124443 (Elaborate in comment about `statx` probe)
- #124445 (bootstrap: Change `global(true)` to `global = true` for flags for consistency)
r? `@ghost`
`@rustbot` modify labels: rollup
bootstrap: Change `global(true)` to `global = true` for flags for consistency
All other arg properties use the `prop = value` style, which makes it slightly annoying to use the `prop(value)` only style for `global`. Change to `prop = value` also for `global` for consistency.
Remove lazycell and once_cell from compiletest dependencies
Use the standard library `OnceLock` instead of third-party equivalents. A macro is used for the regexes to make their initialization less unwieldy.
Do not ICE on invalid consts when walking mono-reachable blocks
The `bug!` here was written under the logic of "this condition is impossible, right?" except that of course, if the compiler is given code that results in an compile error, then the situation is possible.
So now we just direct errors into the already-existing path for when we can't do a mono-time optimization.
Fix ICE on invalid const param types
Fixes ICE #123863 which occurs because the const param has a type which is not a `bool`, `char` or an integral type.
The ICEing code path begins here in `typeck_with_fallback`: cb3752d20e/compiler/rustc_hir_typeck/src/lib.rs (L167)
The `fallback` invokes the `type_of` query and that eventually ends up calling `ct_infer` from the lowering code over here:
cb3752d20e/compiler/rustc_hir_analysis/src/hir_ty_lowering/mod.rs (L561) and `ct_infer` ICEs at this location: cb3752d20e/compiler/rustc_hir_analysis/src/collect.rs (L392)
To fix the ICE it I'm triggering a `span_delayed_bug` before we hit `ct_infer` if the type of the const param is not one of the supported types
### Edit
On `@lcnr's` suggestion I've changed the approach to not let `ReStatic` region hit the `bug!` in `ct_infer` instead of triggering a `span_delayed_bug`.
Fix substitution parts having a shifted underline in some cases
If two suggestions parts are side by side, the underline's offset:
(WIP PR as an example, not yet pushed)
```
error: expected a pattern, found an expression
--> ./main.rs:4:9
|
4 | 1 + 2 => 3
| ^^^^^ arbitrary expressions are not allowed in patterns
|
help: check the value in an arm guard
|
4 | n if n == 1 + 2 => 3
| ~ +++++++++++++
```
The emitter didn't take into account that the string had shrunk/grown if two substitution parts were side-by-side (surprisingly, there was only one case in the ui testsuite.)
```
help: check the value in an arm guard
|
4 | n if n == 1 + 2 => 3
| ~ +++++++++++++
```
``@rustbot`` label +A-suggestion-diagnostics
miri core/alloc tests: do not test a 2nd target
check-aux seems to be one of the slowest runners since we started running standard library tests in Miri on it. So maybe it'd be better to reduce test coverage a bit by not doing cross-target testing of core and alloc? I don't recall finding target-specific issues in these libraries ever (and we still have the extra test coverage via our [out-of-tree nightly tests](https://github.com/rust-lang/miri-test-libstd)). This gives us more buffer to deal with the fact that the number of tests we run will only grow over time.
Cc `@rust-lang/miri` `@rust-lang/infra`
This makes miri not diverge in behavior, it fixes running Rust linux-gnu
binaries on musl with gcompat, it fixes dlopen edge-cases that cranelift
somehow hits, etc.
All other arg properties use the `prop = value` style, which makes it
slightly annoying to use the `prop(value)` style for `global`. Change to
`prop = value` also for `global` for consistency.
thread_local: be excruciatingly explicit in dtor code
Use raw pointers to accomplish internal mutability, and clearly split references where applicable. This reduces the likelihood that any of these parts are misunderstood, either by humans or the compiler's optimizations.
Fixes#124317
r? ``@joboet``
ast: Generalize item kind visiting
And avoid duplicating logic for visiting `Item`s with different kinds (regular, associated, foreign).
The diff is better viewed with whitespace ignored.
Update cargo
9 commits in c9392675917adc2edab269eea27c222b5359c637..b60a1555155111e962018007a6d0ef85207db463
2024-04-23 19:35:19 +0000 to 2024-04-26 16:37:29 +0000
- fix(toml): Remove underscore field support in 2024 (rust-lang/cargo#13804)
- fix: emit 1.77 syntax error only when msrv is incompatible (rust-lang/cargo#13808)
- docs(ref): Index differences between virtual / real manifests (rust-lang/cargo#13794)
- refactor(toml): extract dependency-to-source-id to function (rust-lang/cargo#13802)
- Add where lint was set (rust-lang/cargo#13801)
- fix(toml): Don't double-warn when underscore is used in workspace dep (rust-lang/cargo#13800)
- fix(toml): Be more forceful with underscore/dash redundancy (rust-lang/cargo#13798)
- Fix warning suppression for config.toml vs config compat symlinks (rust-lang/cargo#13793)
- Cleanup linting system (rust-lang/cargo#13797)
r? ghost
PathBuf: replace transmuting by accessor functions
The existing `repr(transparent)` was anyway insufficient as `OsString` was not `repr(transparent)`. And furthermore, on Windows it was blatantly wrong as `OsString` wraps `Wtf8Buf` which is a `repr(Rust)` type with 2 fields:
51a7396ad3/library/std/src/sys_common/wtf8.rs (L131-L146)
So let's just be honest about what happens and add accessor methods that make this abstraction-breaking act of PathBuf visible on the APIs that it pierces through.
Fixes https://github.com/rust-lang/rust/issues/124409
Port run-make `--print=native-static-libs` to rmake.rs
This PR port the run-make `--print=native-static-libs` test to rmake.rs
The dedup was really awful in the `Makefile`, I'm glad to finally have a proper dedup detection for this.
Related to https://github.com/rust-lang/rust/issues/121876
r? `@jieyouxu`
resolve: Remove two cases of misleading macro call visiting
Macro calls are ephemeral, they should not add anything to the definition tree, even if their AST could contains something with identity.
Thankfully, macro call AST cannot contain anything like that, so these walks are just noops.
In majority of other places in def_collector / build_reduced_graph they are already not visited.
(Also, a minor match reformatting is included.)