| Age | Commit message (Collapse) | Author | Lines |
|---|
|
The flag propagates cargo configs to `rustc-perf --cargo-config`,
which is particularly useful when the environment is air-gapped,
and you want to use the default set of training crates vendored
in the rustc-src tarball.
|
|
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]: https://github.com/rust-lang/rust/blob/606afbb617a2949a4e35c4b0258ff94c980b9451/src/tools/opt-dist/src/tests.rs#L62-L77
[^2]: https://github.com/rust-lang/rust/blob/8679004993f08807289911d9f400f4ac4391d2bc/src/bootstrap/bootstrap.py#L1057
|
|
This replaces the hardcoded rustc-perf commit and ad-hoc downloading and
unpacking of its zipped source with defaulting to use the new rustc-perf
submodule.
While it would be nice to make `opt-dist` able to initialize the
submodule automatically when pointing to a Rust checkout _other_ than
the one opt-dist was built in, that would require a bigger refactor that
moved `update_submodule`, from bootstrap, into build_helper.
Regardless, I imagine it must be quite rare to use `opt-dist` with a
checkout that is neither from a rust-src tarball (which will contain the
submodule), nor the checkout opt-dist itself was built (bootstrap will
update the submodule when opt-dist is built).
|
|
|
|
xz crate consist of simple reexport of xz2 crate. Why? Idk.
|
|
compile-flags to repro an ice and add README
|
|
reproduced
|
|
[BOLT] Use CDSort and CDSplit
CDSort and CDSplit are the most recent versions of function ordering and function splitting algorithms with some improvements over the previous baseline (ext-tsp and two-way splitting).
|
|
|
|
Modify opt-dist logic for finding LLVM artifacts
This is the `rustc` side of fixing https://github.com/rust-lang/rust/pull/121395#issuecomment-1973572885.
|
|
|
|
LLVM now includes the minor version in the soname, and also changed
the names of shared object files. libLLVM-18.so is now a symlink to
libLLVM.so.18.1. We need to make some changes to support this:
First, we need to run the installed llvm-config binary, rather
than the one from the build directory. This is because the symlink
does not exist in the build directory, but llvm-config requires it.
This looks like an LLVM bug to me, but it's probably a good idea to
use the installed version anyway.
Second, when installing LLVM into the libdir, we need to install
the target of the symlink, ans this is what will get loaded at
runtime.
However, the rust-dev component in particular also needs to
distribute the symlink itself, as download-ci-llvm will end up
invoking llvm-config, which requires the symlink to exist. The
symlink is not shipped in other components.
|
|
|
|
|
|
|
|
|
|
CDSort and CDSplit are the most recent versions of function ordering and function splitting algorithms with some improvements over the previous baseline (ext-tsp and two-way splitting).
|
|
|
|
Update rustc-perf version
Needed to unblock https://github.com/rust-lang/rust/pull/116033.
The commit first needs to be uploaded to our mirrors.
r? `@Mark-Simulacrum`
|
|
|
|
|
|
Using the new cargo caused issues when a backwards-incompatible change was made to cargo.
|
|
|
|
|
|
|
|
|
|
summary
|
|
bootstrap major change detection implementation
The use of `changelog-seen` and `bootstrap/CHANGELOG.md` has not been functional in any way for many years. We often do major/breaking changes but never update the changelog file or the `changelog-seen`. This is an alternative method for tracking major or breaking changes and informing developers when such changes occur.
Example output when bootstrap detects a major change:
|
|
Signed-off-by: onur-ozkan <work@onurozkan.dev>
|
|
|
|
|
|
|
|
This makes it easier to build a PGO/BOLT optimized `rustc` locally, outside of CI.
|
|
Before it was not deleting non-empty directories.
|
|
|
|
|
|
|
|
This should help resolving spurious network errors. It also increases the timeout for the archive download.
|
|
This avoids issues with the GitHub /archive/ links which can be somewhat
unreliable and are currently causing CI issues.
|
|
fix(bootstrap): rename exclude flag to skip 🐛
fixes #111894
|
|
|
|
This makes it clearer that the LLVM is the host one (it doesn't necessarily have to be downloaded).
|
|
|
|
Use the correct `llvm-profdata` binary in `opt-dist`
|
|
Resolve correct archive version name in `opt-dist`
Should resolve the master part of https://github.com/rust-lang/rust/issues/113784.
r? `@Mark-Simulacrum`
|
|
archive
|
|
|
|
Build the first LLVM without LTO in try builds
Currently, we perform three LLVM builds in the Linux x64 dist builder, which is used for `try` builds:
1) "Normal" LLVM - takes ~5s to compile thanks to `sccache`, but ~8 minutes to link because of ThinLTO
2) PGO instrumented LLVM - same timings as 1)
3) PGO optimized LLVM - takes about 20 minutes to build
When I tried to disable LTO for build 1), it suddenly takes only about a minute to build, because the linking step is much faster. The first LLVM doesn't really need LTO all that much. Without it, it will be a bit slower to build `rustc` in two subsequent steps, but it seems that the ~7 minutes saved on linking it do win that back.
Btw, we can't use the host LLVM for build 1), because this LLVM then builds `rustc` in PGO instrumented mode, and we need the same compiler when later PGO optimizing `rustc`. And we want to use our in-house LLVM for that I think.
|
|
|
|
|
