Three tasks have been implemented here.
Add a new `download-ci-llvm = if-unchange` option and enable if by
default for `profile = codegen`.
Include all build artifacts by traversing the llvm-project build output,
Keep the downloadable llvm the same state as if you have just run a full
source build.
After selecting the codegen profile during ./x.py setup, the submodule
will be automatically downloaded.
enable parallel rustc front end in nightly builds
Refers to the [MCP](https://github.com/rust-lang/compiler-team/issues/681), this pr does:
1. Enable the parallel front end in nightly builds, and keep the default number of threads as 1. Then users can use the parallel rustc front end via -Z threads=n option.
2. Set it up to serial front end for beta/stable builds via bootstrap.
3. Switch over the alt builders from parallel rustc to serial, so we have artifacts without parallel to test against the artifacts with parallel.
r? `@oli-obk`
cc `@cjgillot` `@nnethercote` `@bjorn3` `@Kobzol`
improve compiler&tool documenting and re-enable cranelift on CI
First commit addresses the linking issue with compiler crates. Second one ensures that compiler crates are linked correctly (with later commits we added this check for tools as well), allowing us to detect these hard-to-catch bugs on CI. Following three commits cherry-picked from #117328 to re-enable the Cranelift backend on CI.
More info: https://github.com/rust-lang/rust/issues/117430
cc `@bjorn3` `@RalfJung`
bootstrap/setup: create hooks directory if non-existing
When running `./x setup` on a local repository I chose to install a `pre-push` git hook, but this happened:
```shell
Would you like to install the git hook?: [y/N] y
error: could not create hook .git/hooks/pre-push: do you already have the git hook installed?
No such file or directory (os error 2)
thread 'main' panicked at src/core/build_steps/setup.rs:462:9:
install_git_hook_maybe(&config) failed with No such file or directory (os error 2)
```
This was caused because the `.git/hooks` directory did not exist in my local repository. Creating it manually and re-running the command works fine.
This PR tests for the above directory and if it does not exist then it _tries_ to create it before hard linking the pre-push hook - we use `fs::create_dir()` and disregard the result (ie. it could fail if the directory was created in the meantime) and proceed to call `fs::hard_link()` all the same.
This was a leftover from when inline asm wasn't universally supported by
the cranelift backend yet. It would cause the optimized inline asm
intrinsics to be avoided for the standard library distributed with
rustup now that we build the cranelift backend by default on nightly for
several tier 1 targets.
This reverts commit 1dfb6b162b, reversing
changes made to bcb5798dd8.
That commit broke generating nightly rustc docs. Revert it until we can figure out how to have both, cranelift and docs.
improve and fix `x install`
Fix: Write access check of `prefix` and `sysconfdir` when DESTDIR is present.
Improvement: Instead of repeatedly reading `DESTDIR` within each `fn prepare_dir` usage, read it once and pass it to the `fn prepare_dir`.
Fixes#117203
Fix: Write access check of `prefix` and `sysconfdir`
when DESTDIR is present.
Improvement: Instead of repeatedly reading `DESTDIR` within
each `fn prepare_dir` usage, read it once and pass it to
the `fn prepare_dir`.
Signed-off-by: onur-ozkan <work@onurozkan.dev>
pass `CODEGEN_BACKENDS` to docker
The backends to build are now defined in the `CODEGEN_BACKENDS` env var. It's correctly set in CI, but wasn't passed to docker, hence cg_clif wasn't actually built in #81746.
r? `@Kobzol:` I locally tried `CODEGEN_BACKENDS="cranelift" DEPLOY=1 src/ci/docker/run.sh dist-x86_64-linux` and this change was enough for `ci/run.sh` to read the env var.
So I'll try as-is and we'll see.
add bootstrap flag `--skip-stage0-validation`
This change introduces the `--skip-stage0-validation` flag, which permits the use of any desired version of the stage0 compiler without verifying its version.
Additionally, stage0 compiler validation check is reverted(#115103) to its default enabled state.
Helps to #115065
r? Mark-Simulacrum
change default output mode of `BootstrapCommand`
`SuppressOnSuccess` on `BootstrapCommand` is a problematic default mode as it affects the logs during the bootstrapping (as shown in the screenshot below). The default behavior should be to print everything unless we explicitly modify the behavior within build steps.
Fixes#117315
cc `@Kobzol`
Distribute cg_clif as rustup component on the nightly channel
This makes it possible to use cg_clif using:
```bash
$ rustup component add rustc-codegen-cranelift-preview --toolchain nightly
$ RUSTFLAGS="-Zcodegen-backend=cranelift" cargo +nightly build
```
cc https://github.com/rust-lang/compiler-team/issues/405.
r? `@Mark-Simulacrum`
Avoid unnecessary builds/rebuilds of `rust-demangler`
This is a combination of two loosely-related changes:
- Don't build `rust-demangler` as a dependency of `tests/run-make`, because after #112300 none of the remaining run-make tests actually use it. (If future run-make tests ever do need the demangler, it'll be easy to add it back.)
- For `tests/run-coverage`, build the demangler with the stage 0 compiler instead of the current-stage compiler. This avoids having to uselessly rebuild the demangler after modifying and rebuilding the compiler itself.
This avoids needlessly building cg_clif for other targets and makes it
easier for the dist code to determine if it should distribute cg_clif as
component.
The demangler was only needed by coverage tests, but those tests were migrated
into their own custom test mode in #112300.
This avoids having to build the demangler just for run-make tests. It will
still be built as needed by run-coverage tests or for other purposes.
Centralize command running in boostrap (part one)
This PR tries to consolidate the various `run, try_run, run_quiet, run_quiet_delaying_failure, run_delaying_failure` etc. methods on `Builder`. This PR only touches command execution which doesn't produce output that would be later read by bootstrap, and it also only refactors spawning of commands that happens after a builder is created (commands executed during download & git submodule checkout are left as-is, for now).
The `run_cmd` method is quite meaty, but I expect that it will be changing rapidly soon, so I considered it easy to kept everything in a single method, and only after things settle down a bit, then maybe again split it up a bit.
I still kept the original shortcut methods like `run_quiet_delaying_failure`, but they now only delegate to `run_cmd`. I tried to keep the original behavior (or as close to it as possible) for all the various commands, but it is a giant mess, so there may be some deviations. Notably, `cmd.output()` is now always called, instead of just `status()`, which was called previously in some situations.
Apart from the refactored methods, there is also `Config::try_run`, `check_run`, methods that run commands that produce output, oh my… that's left for follow-up PRs :)
The driving goal of this (and following) refactors is to centralize command execution in bootstrap on a single place, to make command mocking feasible.
r? `@onur-ozkan`
