bootstrap: always propagate `CARGO_TARGET_{host}_LINKER`
We were already setting `CARGO_TARGET_{target}_LINKER` when there is a
setting in `bootstrap.toml`, and when the host and target are the same,
this is also used for build scripts and proc-macros.
However, the host value wasn't set when building for any other target,
and Cargo would see that as a fingerprint change for those build
artifacts, rebuilding them.
If we always set the `CARGO_TARGET_{host}_LINKER`, then those build
scripts will keep a consistent Cargo fingerprint, so they'll remain
cached no matter how we're alternating targets.
Fix `SourceFile::normalized_byte_pos`
This method was broken by 258ace6, which changed `self.normalized_pos` to use relative offsets however this method continued to compare against an absolute offset.
Also adds a regression test for the issue that this method was originally introduced to fix.
Closesrust-lang/rust#149568
Fixes regression of rust-lang/rust#110885
r? cjgillot (as author of the breaking commit)
Introduce helper `ty::Generics::has_own_self`
The pattern `generics.has_self && generics.parent.is_none()` only occurs 5 times in rustc+rustdoc at the time of writing but I keep getting reminded/annoyed that there doesn't exist a nice wrapper fn that abstracts it & immediately clarifies the intent. Most recently that happened when working on my open PR RUST-129543 in which I add yet another occurrence of it ([via](ae8c0a5a46/compiler/rustc_hir_analysis/src/collect/resolve_bound_vars.rs (L1771-L1773))).
For context, `generics.has_self` indicates that there is a `Self` type parameter (at position 0) in the "oldest" / "root" generic parameter list in the chain of generic parameter lists (rephrased: it's true if the chain *as a whole* has a `Self` type parameter). `has_own_self` on the other hand indicates that the `own_params` contain a `Self` type parameter which is sometimes needed for offsetting the own parameters or arguments.
Convert to inline diagnostics in `rustc_lint`
For https://github.com/rust-lang/rust/issues/151366
r? @jdonszelmann
This was a big one! Second to last one, `rustc_parse` is the last one and even bigger :P
Linker-plugin-based LTO: give an explanation how to use linker-plugin-lto with full LTO
Closesrust-lang/rust#138910
The existing linker-plugin-based LTO documentation does not describe the correct usage of full LTO. Specifically, when invoking `rustc` with full LTO, the `-C lto` flag must be passed in addition to `-C linker-plugin-lto`.
Also, this PR documents the use of full LTO when linking Rust with Fortran. Unfortunately, LLVM `flang` does not currently support ThinLTO, so full LTO is the only viable option in this case.
Toolchain combinations were slightly updated.
TODO:
- [x] check swiftc compiler. Almost unusable.
- [x] check how std lib is actually compiled
- [x] add note about LLD and bitcode
- [x] report bug to LLVM: https://github.com/llvm/llvm-project/issues/179800
<details>
<summary>Swiftc is unusable</summary>
https://www.swift.org/install/ gave me LLVM-17. During playing with swift main + rust static library, LLVM-23 removed main :D
```console
# thin LTO Rust:
rustc --crate-type=staticlib -Clinker-plugin-lto -Copt-level=3 ./ftn.rs
# full LTO swift:
swiftc -static libftn.a main.swift -lto=llvm-full -O -use-ld=/tmp/test/llvm-project/install/bin/ld.lld -Xlinker --gc-sections -Xlinker --as-needed -o sr
./sr
> ftn() returned: 77
# thin LTO swift:
swiftc -static libftn.a main.swift -lto=llvm-thin -O -use-ld=/tmp/test/llvm-project/install/bin/ld.lld -Xlinker --gc-sections -Xlinker --as-needed -o sr
./sr
> No output
```
</details>
Add avr_target_feature
This adds the following unstable target features (tracking issue: https://github.com/rust-lang/rust/issues/146889):
- The following two are particularly important for properly supporting inline assembly:
- `tinyencoding`: AVR has devices that reduce the number of registers, similar to RISC-V's RV32E. This feature is necessary to support inline assembly in such devices. (see also https://github.com/rust-lang/rust/pull/146901)
- `lowbytefirst`: AVR's memory access is per 8-bit, and when writing 16-bit ports, the bytes must be written in a specific order. This order depends on devices, making this feature necessary to write proper inline assembly for such use cases. (see also 2a528760bf)
- The followings help recognizing whether specific instructions are available:
- `addsubiw`
- `break`
- `eijmpcall`
- `elpm`
- `elpmx`
- `ijmpcall`
- `jmpcall`
- `lpm`
- `lpmx`
- `movw`
- `mul`
- `rmw`
- `spm`
- `spmx`
Of these, all except `addsubiw`, `break`, `ijmpcall`, `lpm`, `rmw`, `spm`, and `spmx` have [corresponding conditional codes in avr-libc](https://github.com/search?q=repo%3Aavrdudes%2Favr-libc+%2F__AVR_HAVE_%2F&type=code&p=1). LLVM also has `des` feature, but I excluded it from this PR because [DES](https://en.wikipedia.org/wiki/Data_Encryption_Standard) is insecure.
- Report future-incompatible warning (https://github.com/rust-lang/rust/issues/116344) for -C target-feature=-sram and -C target-cpu=<device_without_sram> cases because SRAM is minimum requirement for non-assembly language in both avr-gcc and LLVM.
- See https://github.com/rust-lang/rust/pull/146900#issuecomment-3323558005 for details.
LLVM also has `smallstack`, `wrappingrjmp`, and `memmappedregs` features, but I skipped them because they didn't seem to belong to either of the above categories, but I might have missed something.
(The feature names are match with [definitions in LLVM](https://github.com/llvm/llvm-project/blob/llvmorg-21.1.0/llvm/lib/Target/AVR/AVRDevices.td).)
cc @Patryk27 @Rahix
r? workingjubilee
@rustbot label +O-AVR +A-target-feature
Use relative paths for std links in rustc-docs
Fixesrust-lang/rust#151496
This adds `--extern-html-root-url` for each `STD_PUBLIC_CRATES` entry pointing to `../`, so that `rustc-docs` links to the locally installed `rust-docs` via relative paths.
r? @GuillaumeGomez @rami3l
Dont strip const blocks in array lengths
r? oli-obk
mGCA now handles const blocks by *always* handling them during `lower_expr_to_const_arg_direct` instead of *sometimes* stripping them out at parse time. This is just generally a lot clearer/nicer but also means parsing isn't lossy which is just straight up wrong.
We now use `MgcaDisambiguation::Direct` for const blocks because we "directly" represent a const block as `hir::ConstArgKind::Anon` :> The only time that an anon const for const generics uses `MgcaDisambiguation::AnonConst` is for unbraced literals.
Once we properly support literals in `hir::ConstArgKind` (see rust-lang/rust#152139rust-lang/rust#152001) then `MgcaDisambiguation` can be renamed to `AnonConstKind` with `TypeSystem` and `NonTypeSystem` variants. We can also get rid of `mgca_direct_lit_hack`. I expect this to be a very nice cleanup :)
Fixesrust-lang/rustfmt#6788
The diff relating to passing spans around is to avoid a bunch of mGCA diagnostics changing from `const {}` to `{}`. I'm not entirely sure why this was happening.
cc @rust-lang/rustfmt
How do I run the tests in the rustfmt repo from here? `x test rustfmt` only seems to run like 100 tests and doesn't result in a `target/issue-6788.rs` getting created. I've verified locally that this formats correctly though
bootstrap: Remove `ShouldRun::paths`
Split out from https://github.com/rust-lang/rust/pull/151930. I've copied my comment in https://github.com/rust-lang/rust/pull/151930#discussion_r2750409129 into the commit description.
r? @Zalathar cc @Mark-Simulacrum @Kobzol
---
Some history about `paths()`. The original intent @Mark-Simulacrum had
when he introduced PathSet in f104b12059, to my knowledge, was that multiple paths
could be aliases for the same step. That's what rustdoc is doing; both
paths for rustdoc run exactly the same Step, regardless of whether one
or both are present.
That never really caught on. To my knowledge, rustdoc is the only usage
of paths() there's ever been.
Later, in rust-lang/rust#95503, I repurposed PathSet to mean "each crate in this set
should be passed to Step::make_run in RunConfig". That was not the
previous meaning.
Rustdoc never looks at run.paths in make_run, so it's safe to just treat
it as an alias, like elsewhere in bootstrap. Same for all the other tool
steps.
Always use Xcode-provided Clang in macOS CI
Most of our macOS CI runners use the Xcode-provided Clang. `dist-apple-various` still downloads Clang from LLVM's prebuilt sources, which is a bit problematic, because we're still using the Xcode-provided _headers_.
As a concrete example, using Xcode's Clang is required by https://github.com/rust-lang/rust/pull/152013, as otherwise updating to Xcode 26 headers fails with an obscure `fatal error: 'net/route.h' file not found`, see [this build log](https://github.com/rust-lang/rust/pull/152013#issuecomment-3838801617). I suspect this is because building the new headers isn't officially supported with the older LLVM/Clang 15.
This PR removes that functionality, so that we instead always build with the Xcode-provided Clang. This is effectively the same as setting `USE_XCODE_CLANG=1` on `dist-apple-various` as well, but I thought I'd clean things up while I'm at it.
I didn't find the reason why we did this in first place, maybe because the Xcode Clang at the time was too out of date to build LLVM? It doesn't seem to be a problem anymore though.
r? t-infra
Update wasi-sdk used in CI/releases
This is similar to prior updates such as rust-lang/rust#149037 in that this is just updating a URL. This update though has some technical updates accompanying it as well, however:
* The `wasm32-wasip2` target no longer uses APIs from WASIp1 on this target, even for startup. This means that the final binary no longer has an "adapter" which can help making instantiation of a component a bit more lean.
* In rust-lang/rust#147572 libstd was updated to use wasi-libc more often on the `wasm32-wasip2` target. This uncovered a number of bugs in wasi-libc such as rust-lang/rust#149864, rust-lang/rust#150291, and rust-lang/rust#151016. These are all fixed in wasi-sdk-30 so the workarounds in the standard library are all removed.
Overall this is not expected to have any sort of major impact on users of WASI targets. Instead it's expected to be a normal routine update to keep the wheels greased and oiled.
Some history about `paths()`. The original intent Mark-Simulacrum had
when he introduced PathSet, to my knowledge, was that multiple paths
could be aliases for the same step. That's what rustdoc is doing; both
paths for rustdoc run exactly the same Step, regardless of whether one
or both are present.
That never really caught on. To my knowledge, rustdoc is the only usage
of paths() there's ever been.
Later, in 95503, I repurposed PathSet to mean "each crate in this set
should be passed to Step::make_run in RunConfig". That was not the
previous meaning.
Rustdoc never looks at run.paths in make_run, so it's safe to just treat
it as an alias, like elsewhere in bootstrap. Same for all the other tool
steps.
Co-authored-by: Tshepang Mbambo <hopsi@tuta.io>
update openmp/offload builds to LLVM 22, Part 1
This part of the update of OpenMP/Offload to LLVM 22 does not depend on a backport in the future LLVM release candidate 3. Let's therefore merge it first, to unblock existing users/developers.
r? @jieyouxu
This is similar to prior updates such as 149037 in that this is just
updating a URL. This update though has some technical updates
accompanying it as well, however:
* The `wasm32-wasip2` target no longer uses APIs from WASIp1 on this
target, even for startup. This means that the final binary no longer
has an "adapter" which can help making instantiation of a component a
bit more lean.
* In 147572 libstd was updated to use wasi-libc more often on the
`wasm32-wasip2` target. This uncovered a number of bugs in
wasi-libc such as 149864, 150291, and 151016. These are all fixed in
wasi-sdk-30 so the workarounds in the standard library are all
removed.
Overall this is not expected to have any sort of major impact on users
of WASI targets. Instead it's expected to be a normal routine update to
keep the wheels greased and oiled.
bootstrap: exclude hexagon-unknown-qurt from llvm-libunwind default
Hexagon Linux targets (hexagon-unknown-linux-musl) use in-tree llvm-libunwind for stack unwinding. However, hexagon-unknown-qurt uses libc_eh from the Hexagon SDK instead.
Port depgraph testing attributes to parser
Tracking issue: rust-lang/rust#131229
Ports `#[rustc_clean]`, `#[rustc_if_this_changed]` and `#[rustc_then_this_would_need]` attributes.
Removes references to `rustc_dirty` as that attribute was folded into `rustc_clean` some time ago and rename some code accordingly.
r? JonathanBrouwer
Return `ExitCode` from `rustc_driver::main` instead of calling `process::exit`
This makes rustc simply return an exit code from main rather than calling `std::process::exit` with an exit code. This means that drops run normally and the process exits cleanly. This is similar to what happens when an ICE occurs (due to being a panic that's caught by std's `lang_start`).
Also instead of hard coding success and failure codes this uses `ExitCode::SUCCESS` and `ExitCode::FAILURE`, which in turn effectively uses `libc::EXIT_SUCCESS` and `libc::EXIT_FAILURE` (via std). These are `0` and `1` respectively for all currently supported host platforms so it doesn't actually change the exit code.
Use fewer intermediate functions for short backtraces in queries
If we make sure that `compute_fn` in the query's vtable is actually named `__rust_begin_short_backtrace`, we can avoid the need for some additional intermediate functions and stack frames.
This is similar to how the `get_query_incr` and `get_query_non_incr` functions are actually named `__rust_end_short_backtrace`.
---
Before/after comparison: https://github.com/rust-lang/rust/pull/151739#issuecomment-3815432527
---
- Earlier draft of this PR: https://github.com/rust-lang/rust/pull/151719
- Introduction of this backtrace-trimming: https://github.com/rust-lang/rust/pull/108938
