redesign stage 0 std follow-ups part2
Fixes three bugs:
1. `x check` fails when run on rustdoc without `download-rustc` enabled. (1st commit)
2. `x check` fails when run on the compiler with `download-rustc` enabled. (2nd commit)
3. `x test library` fails with `download-rustc` enabled. (3rd commit)
Fixesrust-lang/rust#142018 (case 1)
Fixes https://github.com/rust-lang/rust/issues/141983 (case 3)
In StdLink::run we subsequently recursively copy the initial sysroot
lib directory into the stage0-sysroot lib directory. If the initial
sysroot is a toolchain that includes the `rust-src` component (in
lib/rustlib/src/rust), if we add this symlink, that recursive copy
will overwrite the repo sources with the toolchain's sources.
Use ccache for stage0 tool builds
Now after the stage0 redesign, we can actually start ccaching the build of the compiler itself. We can also cache the bootstrap tools, since these are also built with the stage0 compiler.
Stage0 compiler builds are now being cached: https://github.com/rust-lang/rust/actions/runs/15397246267#summary-43321151192 (`..bootstrap::core::build_steps::compile::Rustc 483.10s 40.41s -91.6%`). It's not a gigantic win everywhere, but it should help. It seems to make the Linux jobs ~10 minute faster. It should be especially useful on PR builds after https://github.com/rust-lang/rust/pull/141948.
r? `@jieyouxu`
try-job: `x86_64-gnu-llvm-19*`
try-job: `x86_64-msvc*`
try-job: `x86_64-apple*`
try-job: `dist-x86_64-linux`
tools-aux ci runner: also cross-test doctests in Miri
Miri now supports running doctests across different targets. Let's use that to run the std doctests on aarch64-apple-darwin, i686-pc-windows-msvc.
try-job: x86_64-gnu-aux
Fix TLS model on bootstrap for cygwin
There aren't other targets that both use emutls and enable `has_thread_local`, so cygwin triggers this bug first.
r? mati865
See: https://github.com/rust-lang/rust/pull/141719#issuecomment-2925445263
``@jeremyd2019`` Could you check if this PR fixes the issue? I just found my pre-built stage-0 rustc was too old to build the current rustc :(
Exclude `CARGO_HOME` from `generate-copyright` in-tree determination
On Ferrocene, we noticed that in our releases the out-of-tree notices were not being included. When `x.py run generate-copyright` was ran on local development machines, it worked fine.
After some investigations ``@tshepang`` and I determined that the problem was that the cargo registry (located in `CARGO_HOME`) started with the source directory on CI jobs, and was being excluded by this line:
15825b7161/src/tools/generate-copyright/src/cargo_metadata.rs (L85-L88)
In Ferrocene's `run.sh` we set `CARGO_HOME` to be `build/cargo-home`: 96a45dd9a1/ferrocene/ci/run.sh (L34-L46) which caused this issue.
This PR passes the `CARGO_HOME` variable to the `generate-copyright` tool and expands the consideration of in-tree-ness to be aware of `CARGO_HOME`. It is an upstreaming of https://github.com/ferrocene/ferrocene/pull/1491.
## Testing
Run `CARGO_HOME=build/cargo-home ./x.py run generate-copyright` on `master`, then check `build/host/doc/COPYRIGHT` and look for out of tree dependencies (at the bottom).
Then, try running the same command in this branch.
bootstrap: Remove `bin_root` from `PATH`
It's not currently load bearing in typical setups on typical targets.
Basically, if it passes the standard CI, then we can remove it, I think.
If someone later reports this breaking the build, then we can address it in a more fine grained way.
E.g. add it to `PATH` last and not first, only on specific targets, and only if specific files are not already found in `PATH` directories.
Fixes [#t-infra/bootstrap > Build broken in MSYS2 @ 💬](https://rust-lang.zulipchat.com/#narrow/channel/326414-t-infra.2Fbootstrap/topic/Build.20broken.20in.20MSYS2/near/520709527)
On stage 0, rather than compiling std utilize the one from the stage0 sysroot
as stage 0 should represent the snapshot version not the compiled one.
Signed-off-by: onur-ozkan <work@onurozkan.dev>
resolve stage0 sysroot from rustc
Similar to https://github.com/rust-lang/rust/pull/141729, instead of manually navigating directories based on stage0 rustc, use `--print sysroot` to get the sysroot directly. This also works when using the bootstrap `rustc` shim.
r? Kobzol
Refactor the two-phase check for impls and impl items
Refactor the two-phase dead code check to make the logic clearer and simpler:
1. adding assoc fn and impl into `unsolved_items` directly during the initial construction of the worklist
2. converge the logic of checking whether assoc fn and impl are used to `item_should_be_checked`, and the item is considered used only when its corresponding trait and Self adt are used
This PR only refactors as much as possible to avoid affecting the original functions. However, due to the adjustment of the order of checks, the test results are slightly different, but overall, there is no regression problem
Fixesrust-lang/rust#127911Fixesrust-lang/rust#128839
Extracted from https://github.com/rust-lang/rust/pull/128637.
r? petrochenkov
try-job: dist-aarch64-linux
Leaving stage0 target-libdir resolution to rustc. This should also fix the issue with
hard-coding `$sysroot/lib` which fails on systems that use `$sysroot/lib64` or `$sysroot/lib32`.
Signed-off-by: onur-ozkan <work@onurozkan.dev>
Instead of manually navigating directories based on stage0 rustc, use `--print sysroot`
to get the sysroot directly. This also works when using the bootstrap `rustc` shim.
Signed-off-by: onur-ozkan <work@onurozkan.dev>
Do not get proc_macro from the sysroot in rustc
With the stage0 refactor the proc_macro version found in the sysroot will no longer always match the proc_macro version that proc-macros get compiled with by the rustc executable that uses this proc_macro. This will cause problems as soon as the ABI of the bridge gets changed to implement new features or change the way existing features work.
To fix this, this commit changes rustc crates to depend directly on the local version of proc_macro which will also be used in the sysroot that rustc will build.
Building with _LIBUNWIND_IS_NATIVE_ONLY disables code for cross-architecture unwinding
it is disabled by default in LLVM [1], replicate the cmake behavior in bootstrap process
It also enables some additional code that handles PAC-specific unwind info
it helps compiling with the -mbranch-protection=pac or -mbranch-protection=standard flags
This fixes build with clang/musl on aarch64
[1] 85624c5de3
Signed-off-by: Khem Raj <raj.khem@gmail.com>
