Do not copy .rmeta files into the sysroot of the build compiler during check of rustc/std #144252

Kobzol · 2025-07-21T11:08:00Z

Before, when bootstrap did a check build of rustc stage N (with a build compiler that was stage N-1), it automatically copied the resulting .rmeta artifacts into the sysroot of the stage N-1 build compiler, so that stage N rustc_private tools such as miri could be compiled using the stage N-1 build compiler. This has a number of issues:

It was done unconditionally, even if no rustc_private tools were actually built.
If we did a check and a build of the same stage compiler in the same bootstrap invocation, the generated rmeta and rlib files could clash. This is also why you can see that check::Std actually doesn't copy the artifacts anymore (which forced us to build std instead of just checking it in a bunch of Check steps).
It was polluting the sysroot of the build compiler. This is especially annoying for the stage 0 compiler, because we are forced to create an artificial sysroot for it, so that we can copy new stuff into it.
It was very implicit in bootstrap.

Based on suggestions by @cuviper and @bjorn3, I tried to change how this behaves. Instead of copying the rmeta artifacts into the sysroot of the build compiler (from where they would be loaded implicitly), they are now stored in a separate transient bootstrap build directory, and they are then explicitly passed only when checking rustc_private tools using the -L flag. The flags are passed out-of-band through our rustc wrapper, to avoid invalidating the build cache. This is the first commit.

The second commit does the same for std. For a few months, we used to build std instead of just checking it when doing a cross-compile check of something that required std, this now fixes it. There is still the previous ordering requirement though, that check::Std has to be executed as the last check step, or rather nothing that requires checked std should be executed after it, because it will run into rmeta/rlib duplications (

rust/src/bootstrap/src/core/builder/mod.rs

Line 1066 in 4fa90ef

check::Std,

). I tried to fix in this PR, but it quickly runs into the fact that building things currently copies rlib artifacts into the build compiler sysroot. I want to fix that as one of the next steps. After we get rid of all the copies (or rather, we only do the copies for dist/stage2+ and do not copy anything into the stage0 compiler's sysroot), we could hopefully finally get rid of stage0-sysroot.

Based on my local tests, this seems to be working fine. If it works on CI, and we don't run into other issues after merging it, I'd like to do the same also for rlib artifacts generated during x build.

r? @jieyouxu

rustbot · 2025-07-21T11:08:05Z

jieyouxu is currently at their maximum review capacity.
They may take a while to respond.

bjorn3 · 2025-07-21T11:44:08Z

nit: s/hcekc/check in the PR title

Kobzol · 2025-07-21T11:45:15Z

I will also need to do this for the Clippy steps.

Kobzol · 2025-07-21T13:35:27Z

Uhh, x clippy ci ran clippy using in-tree clippy, not stage 0 clippy. I will have to go through the clippy steps first and make that more compatible with what x check does first.

This helps to avoid polluting the sysroot of the build compiler.

Kobzol · 2025-08-16T10:39:17Z

This should be ready for a review now.

jieyouxu · 2025-08-16T10:54:11Z

aur naur

jieyouxu · 2025-08-18T00:47:11Z

@rustbot author

Kobzol · 2025-08-18T08:46:54Z

Turns out that overwriting the same environment variable twice is not a great idea...

Kobzol · 2025-08-18T09:42:13Z

@rustbot ready

jieyouxu

Looks good in general, a question

jieyouxu · 2025-08-18T11:34:02Z

src/bootstrap/src/bin/rustc.rs

+    // Here we pass additional paths that essentially act as a sysroot.
+    // These are used to load rustc crates (e.g. `extern crate rustc_ast;`)
+    // for rustc_private tools, so that we do not have to copy them into the
+    // actual sysroot of the compiler that builds the tool.
+    if let Ok(dirs) = env::var("RUSTC_ADDITIONAL_SYSROOT_PATHS") {
+        for dir in dirs.split(",") {
+            cmd.arg(format!("-L{dir}"));
+        }
+    }


Remark: in theory, this can break if dir name has a comma. In practice, simply don't do that.

If someone comes with that use-case, we can always switch to some different encoding, e.g. a zero byte.

jieyouxu · 2025-08-18T11:35:37Z

src/bootstrap/src/core/build_steps/check.rs

+        run_cargo(
+            builder,
+            cargo,
+            builder.config.free_args.clone(),
+            &check_stamp,
+            vec![],
+            true,
+            false,
+        );


Remark: this call-site has a few too many args, but one of the long-standing cleanup issues.

Yeah, it would be great to have a builder pattern for Cargo invocations instead of this function.

src/bootstrap/src/core/build_steps/check.rs

jieyouxu · 2025-08-18T11:37:05Z

src/bootstrap/src/core/build_steps/check.rs

+    fn configure_cargo(&self, cargo: &mut Cargo) {
+        cargo.append_to_env(
+            "RUSTC_ADDITIONAL_SYSROOT_PATHS",
+            format!("{},{}", self.host_dir.display(), self.target_dir.display()),


Suggestion: I'd .to_str().unwrap() here, we should just fail if it's not UTF-8 IMO. We already assume that in so many places.

src/bootstrap/src/core/build_steps/check.rs

jieyouxu · 2025-08-18T11:56:01Z

src/bootstrap/src/core/builder/cargo.rs

+        let key = key.as_ref();
+        if let Some((_, Some(previous_value))) = self.command.get_envs().find(|(k, _)| *k == key) {
+            let mut combined: OsString = previous_value.to_os_string();
+            combined.push(delimiter.as_ref());


Note: this might be a little bit of a footgun, because PATH delimiter on Windows is ; but on Unix it's :.

Or wait, is this only intended to be used for ENV=THING_1,THING_2?

It's intended to be used for whatever you want 😆 For paths specifically we already have env::join_paths. This is more for use-cases where I have an arbitrary list that I want to incrementally build, where the list elements are separated e.g. with spaces or commas.

jieyouxu · 2025-08-18T12:00:54Z

@rustbot author

Kobzol · 2025-08-18T20:13:15Z

@rustbot ready

jieyouxu · 2025-08-18T22:56:30Z

Thanks
@bors r+ rollup

bors · 2025-08-18T22:56:33Z

📌 Commit 533ecdb has been approved by jieyouxu

It is now in the queue for this repository.

Do not copy .rmeta files into the sysroot of the build compiler during check of rustc/std Before, when bootstrap did a check build of rustc stage N (with a build compiler that was stage N-1), it automatically copied the resulting `.rmeta` artifacts into the sysroot of the stage N-1 build compiler, so that stage N `rustc_private` tools such as `miri` could be compiled using the stage N-1 build compiler. This has a number of issues: - It was done unconditionally, even if no `rustc_private` tools were actually built. - If we did a check and a build of the same stage compiler in the same bootstrap invocation, the generated rmeta and rlib files could clash. This is also why you can see that `check::Std` actually doesn't copy the artifacts anymore (which forced us to build std instead of just checking it in a bunch of `Check` steps). - It was polluting the sysroot of the build compiler. This is especially annoying for the stage 0 compiler, because we are forced to create an artificial sysroot for it, so that we can copy new stuff into it. - It was very implicit in bootstrap. Based on suggestions by `@cuviper` and `@bjorn3,` I tried to change how this behaves. Instead of copying the rmeta artifacts into the sysroot of the build compiler (from where they would be loaded implicitly), they are now stored in a separate transient bootstrap build directory, and they are then explicitly passed *only* when checking `rustc_private` tools using the `-L` flag. The flags are passed out-of-band through our rustc wrapper, to avoid invalidating the build cache. This is the first commit. The second commit does the same for std. For a few months, we used to build std instead of just checking it when doing a cross-compile check of something that required std, this now fixes it. There is still the previous ordering requirement though, that `check::Std` has to be executed as the last check step, or rather nothing that requires checked std should be executed *after* it, because it will run into rmeta/rlib duplications (https://github.com/rust-lang/rust/blob/4fa90ef7996f891f7f1e126411e5d75afe64accf/src/bootstrap/src/core/builder/mod.rs#L1066). I tried to fix in this PR, but it quickly runs into the fact that building things currently copies *rlib* artifacts into the build compiler sysroot. I want to fix that as one of the next steps. After we get rid of all the copies (or rather, we only do the copies for dist/stage2+ and do not copy anything into the stage0 compiler's sysroot), we could hopefully finally get rid of `stage0-sysroot`. Based on my local tests, this seems to be working fine. If it works on CI, and we don't run into other issues after merging it, I'd like to do the same also for rlib artifacts generated during `x build`. r? `@jieyouxu`

Do not copy .rmeta files into the sysroot of the build compiler during check of rustc/std Before, when bootstrap did a check build of rustc stage N (with a build compiler that was stage N-1), it automatically copied the resulting `.rmeta` artifacts into the sysroot of the stage N-1 build compiler, so that stage N `rustc_private` tools such as `miri` could be compiled using the stage N-1 build compiler. This has a number of issues: - It was done unconditionally, even if no `rustc_private` tools were actually built. - If we did a check and a build of the same stage compiler in the same bootstrap invocation, the generated rmeta and rlib files could clash. This is also why you can see that `check::Std` actually doesn't copy the artifacts anymore (which forced us to build std instead of just checking it in a bunch of `Check` steps). - It was polluting the sysroot of the build compiler. This is especially annoying for the stage 0 compiler, because we are forced to create an artificial sysroot for it, so that we can copy new stuff into it. - It was very implicit in bootstrap. Based on suggestions by ``@cuviper`` and ``@bjorn3,`` I tried to change how this behaves. Instead of copying the rmeta artifacts into the sysroot of the build compiler (from where they would be loaded implicitly), they are now stored in a separate transient bootstrap build directory, and they are then explicitly passed *only* when checking `rustc_private` tools using the `-L` flag. The flags are passed out-of-band through our rustc wrapper, to avoid invalidating the build cache. This is the first commit. The second commit does the same for std. For a few months, we used to build std instead of just checking it when doing a cross-compile check of something that required std, this now fixes it. There is still the previous ordering requirement though, that `check::Std` has to be executed as the last check step, or rather nothing that requires checked std should be executed *after* it, because it will run into rmeta/rlib duplications (https://github.com/rust-lang/rust/blob/4fa90ef7996f891f7f1e126411e5d75afe64accf/src/bootstrap/src/core/builder/mod.rs#L1066). I tried to fix in this PR, but it quickly runs into the fact that building things currently copies *rlib* artifacts into the build compiler sysroot. I want to fix that as one of the next steps. After we get rid of all the copies (or rather, we only do the copies for dist/stage2+ and do not copy anything into the stage0 compiler's sysroot), we could hopefully finally get rid of `stage0-sysroot`. Based on my local tests, this seems to be working fine. If it works on CI, and we don't run into other issues after merging it, I'd like to do the same also for rlib artifacts generated during `x build`. r? ``@jieyouxu``

Fix uplifting in `Assemble` step In rust-lang#145310, I removed [this line](https://github.com/rust-lang/rust/pull/145310/files#diff-5a1e05f2688d271039171a547d407d0c8a96715ee64d35562fc76b4c9a874303L2109), which adjusted the stage of the build compiler if an uplift has happened. This broke stage3+ uplifted rustc builds (rust-lang#145534). I could swear I tested this in the PR, but somehow I missed it. Instead of keeping the original returned stage, I made it more explicit by returning the actually used `build_compiler` from the `Rustc` step, and then use that in the `Assemble` step. The changes to `RustcLink` were needed to fix `ui-fulldeps`, which apparently build a stage3 rustc, because I haven't fixed the test steps yet 😅 Hopefully we might be able to remove `RustcLink` if the approach from rust-lang#144252 will work. Should fix rust-lang#145534. r? `@jieyouxu`

Fix uplifting in `Assemble` step In rust-lang#145310, I removed [this line](https://github.com/rust-lang/rust/pull/145310/files#diff-5a1e05f2688d271039171a547d407d0c8a96715ee64d35562fc76b4c9a874303L2109), which adjusted the stage of the build compiler if an uplift has happened. This broke stage3+ uplifted rustc builds (rust-lang#145534). I could swear I tested this in the PR, but somehow I missed it. Instead of keeping the original returned stage, I made it more explicit by returning the actually used `build_compiler` from the `Rustc` step, and then use that in the `Assemble` step. The changes to `RustcLink` were needed to fix `ui-fulldeps`, which apparently build a stage3 rustc, because I haven't fixed the test steps yet 😅 Hopefully we might be able to remove `RustcLink` if the approach from rust-lang#144252 will work. Should fix rust-lang#145534. r? ``@jieyouxu``

Rollup of 19 pull requests Successful merges: - #140956 (`impl PartialEq<{str,String}> for {Path,PathBuf}`) - #141744 (Stabilize `ip_from`) - #142681 (Remove the `#[no_sanitize]` attribute in favor of `#[sanitize(xyz = "on|off")]`) - #142871 (Trivial improve doc for transpose ) - #144252 (Do not copy .rmeta files into the sysroot of the build compiler during check of rustc/std) - #144476 (rustdoc-search: search backend with partitioned suffix tree) - #144567 (Fix RISC-V Test Failures in ./x test for Multiple Codegen Cases) - #144804 (Don't warn on never to any `as` casts as unreachable) - #144960 ([RTE-513] Ignore sleep_until test on SGX) - #145013 (overhaul `&mut` suggestions in borrowck errors) - #145041 (rework GAT borrowck limitation error) - #145243 (take attr style into account in diagnostics) - #145405 (cleanup: use run_in_tmpdir in run-make/rustdoc-scrape-examples-paths) - #145432 (cg_llvm: Small cleanups to `owned_target_machine`) - #145484 (Remove `LlvmArchiveBuilder` and supporting code/bindings) - #145557 (Fix uplifting in `Assemble` step) - #145563 (Remove the `From` derive macro from prelude) - #145565 (Improve context of bootstrap errors in CI) - #145584 (interpret: avoid forcing all integer newtypes into memory during clear_provenance) Failed merges: - #145359 (Fix bug where `rustdoc-js` tester would not pick the right `search.js` file if there is more than one) - #145573 (Add an experimental unsafe(force_target_feature) attribute.) r? `@ghost` `@rustbot` modify labels: rollup

Rollup merge of #145557 - Kobzol:rustc-link-fix, r=jieyouxu Fix uplifting in `Assemble` step In #145310, I removed [this line](https://github.com/rust-lang/rust/pull/145310/files#diff-5a1e05f2688d271039171a547d407d0c8a96715ee64d35562fc76b4c9a874303L2109), which adjusted the stage of the build compiler if an uplift has happened. This broke stage3+ uplifted rustc builds (#145534). I could swear I tested this in the PR, but somehow I missed it. Instead of keeping the original returned stage, I made it more explicit by returning the actually used `build_compiler` from the `Rustc` step, and then use that in the `Assemble` step. The changes to `RustcLink` were needed to fix `ui-fulldeps`, which apparently build a stage3 rustc, because I haven't fixed the test steps yet 😅 Hopefully we might be able to remove `RustcLink` if the approach from #144252 will work. Should fix #145534. r? ``@jieyouxu``

Rollup merge of #144252 - Kobzol:rmeta-sysroot, r=jieyouxu Do not copy .rmeta files into the sysroot of the build compiler during check of rustc/std Before, when bootstrap did a check build of rustc stage N (with a build compiler that was stage N-1), it automatically copied the resulting `.rmeta` artifacts into the sysroot of the stage N-1 build compiler, so that stage N `rustc_private` tools such as `miri` could be compiled using the stage N-1 build compiler. This has a number of issues: - It was done unconditionally, even if no `rustc_private` tools were actually built. - If we did a check and a build of the same stage compiler in the same bootstrap invocation, the generated rmeta and rlib files could clash. This is also why you can see that `check::Std` actually doesn't copy the artifacts anymore (which forced us to build std instead of just checking it in a bunch of `Check` steps). - It was polluting the sysroot of the build compiler. This is especially annoying for the stage 0 compiler, because we are forced to create an artificial sysroot for it, so that we can copy new stuff into it. - It was very implicit in bootstrap. Based on suggestions by ```@cuviper``` and ```@bjorn3,``` I tried to change how this behaves. Instead of copying the rmeta artifacts into the sysroot of the build compiler (from where they would be loaded implicitly), they are now stored in a separate transient bootstrap build directory, and they are then explicitly passed *only* when checking `rustc_private` tools using the `-L` flag. The flags are passed out-of-band through our rustc wrapper, to avoid invalidating the build cache. This is the first commit. The second commit does the same for std. For a few months, we used to build std instead of just checking it when doing a cross-compile check of something that required std, this now fixes it. There is still the previous ordering requirement though, that `check::Std` has to be executed as the last check step, or rather nothing that requires checked std should be executed *after* it, because it will run into rmeta/rlib duplications (https://github.com/rust-lang/rust/blob/4fa90ef7996f891f7f1e126411e5d75afe64accf/src/bootstrap/src/core/builder/mod.rs#L1066). I tried to fix in this PR, but it quickly runs into the fact that building things currently copies *rlib* artifacts into the build compiler sysroot. I want to fix that as one of the next steps. After we get rid of all the copies (or rather, we only do the copies for dist/stage2+ and do not copy anything into the stage0 compiler's sysroot), we could hopefully finally get rid of `stage0-sysroot`. Based on my local tests, this seems to be working fine. If it works on CI, and we don't run into other issues after merging it, I'd like to do the same also for rlib artifacts generated during `x build`. r? ```@jieyouxu```

Zalathar · 2025-08-19T10:14:52Z

Bors hasn't noticed that this was merged.

@bors r- retry

rustbot assigned jieyouxu Jul 21, 2025

rustbot added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. T-bootstrap Relevant to the bootstrap subteam: Rust's build system (x.py and src/bootstrap) labels Jul 21, 2025

This comment has been minimized.

Sign in to view

Kobzol changed the title ~~Do not copy .rmeta files into the sysroot of the build compiler during hcekc~~ Do not copy .rmeta files into the sysroot of the build compiler during check Jul 21, 2025

Avoid copying rustc rmeta artifacts into the build compiler sysroot

ddf39ca

This helps to avoid polluting the sysroot of the build compiler.

Kobzol force-pushed the rmeta-sysroot branch from d874cf5 to ddf39ca Compare August 16, 2025 07:56

Only check std in cross-compilation instead of building it

814b8e6

Kobzol changed the title ~~Do not copy .rmeta files into the sysroot of the build compiler during check~~ Do not copy .rmeta files into the sysroot of the build compiler during check of rustc/std Aug 16, 2025

Kobzol marked this pull request as ready for review August 16, 2025 10:39

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. labels Aug 16, 2025

This comment has been minimized.

Sign in to view

rustbot added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Aug 18, 2025

Kobzol mentioned this pull request Aug 18, 2025

Fix uplifting in Assemble step #145557

Merged

Do not overwrite the value of RUSTC_ADDITIONAL_SYSROOT_PATHS

22519d3

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. labels Aug 18, 2025

jieyouxu approved these changes Aug 18, 2025

View reviewed changes

rustbot added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Aug 18, 2025

Assume UTF-8 in sysroot paths

533ecdb

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. labels Aug 18, 2025

bors added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Aug 18, 2025

Zalathar mentioned this pull request Aug 19, 2025

Rollup of 19 pull requests #145589

Merged

bors merged commit b9fdc6b into rust-lang:master Aug 19, 2025
10 checks passed

rustbot added this to the 1.91.0 milestone Aug 19, 2025

Kobzol deleted the rmeta-sysroot branch August 19, 2025 11:51

Do not copy .rmeta files into the sysroot of the build compiler during check of rustc/std #144252

Do not copy .rmeta files into the sysroot of the build compiler during check of rustc/std #144252

Conversation

Kobzol commented Jul 21, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

rustbot commented Jul 21, 2025

Uh oh!

This comment has been minimized.

bjorn3 commented Jul 21, 2025

Uh oh!

Kobzol commented Jul 21, 2025

Uh oh!

Kobzol commented Jul 21, 2025

Uh oh!

Kobzol commented Aug 16, 2025

Uh oh!

This comment has been minimized.

jieyouxu commented Aug 16, 2025

Uh oh!

jieyouxu commented Aug 18, 2025

Uh oh!

Kobzol commented Aug 18, 2025

Uh oh!

Kobzol commented Aug 18, 2025

Uh oh!

jieyouxu left a comment

Choose a reason for hiding this comment

Uh oh!

jieyouxu Aug 18, 2025

Choose a reason for hiding this comment

Uh oh!

Kobzol Aug 18, 2025

Choose a reason for hiding this comment

Uh oh!

jieyouxu Aug 18, 2025

Choose a reason for hiding this comment

Uh oh!

Kobzol Aug 18, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

jieyouxu Aug 18, 2025

Choose a reason for hiding this comment

Uh oh!

Uh oh!

jieyouxu Aug 18, 2025

Choose a reason for hiding this comment

Uh oh!

Kobzol Aug 18, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jieyouxu commented Aug 18, 2025

Uh oh!

Kobzol commented Aug 18, 2025

Uh oh!

jieyouxu commented Aug 18, 2025

Uh oh!

bors commented Aug 18, 2025

Uh oh!

Uh oh!

Zalathar commented Aug 19, 2025

Uh oh!

Uh oh!

Kobzol commented Jul 21, 2025 •

edited

Loading

Kobzol Aug 18, 2025 •

edited

Loading