intrinsic-test: run by default by xonx4l · Pull Request #159011 · rust-lang/rust

xonx4l · 2026-07-09T12:43:48Z

This PR makes intrinsic-test run by default in x test and raise sample-percentage to 100.

rustbot · 2026-07-09T12:43:53Z

Kobzol is not on the review rotation at the moment.
They may take a while to respond.

Kobzol · 2026-07-09T18:13:23Z

You'll have to rebless bootstrap tests, IIRC it was just cargo insta review in the bootstrap directory after running ./x test bootstrap. Then we can see the times on CI. How do the times differ for you locally?

xonx4l · 2026-07-10T11:54:10Z

@Kobzol locally running ./x test --stage 2 library/stdarch/crates/intrinsic-test took 48m 25s . Which built the Stage 1 compiler, the standard library, the Stage 2 compiler and reached the execution phase of intrinsic-test but got error due to a cc version mismatch . Thank you!

Kobzol

Thank you! I still have yet to prepare my env to actually run this locally 🙈

View changes since this review

Kobzol · 2026-07-11T13:10:22Z

 /// and a Rust Cargo workspace. Then runs `cargo test` on that workspace
 /// which compiles both versions and compares their outputs on random inputs.
 #[derive(Debug, Clone, PartialEq, Eq, Hash)]
 pub struct IntrinsicTest {


Let's add a comment here that explains the required dependencies on the various OSes. For now, we should say that on x64, it requires a recent GCC (>15?), and the SDE emulator to run.

OK, will add.

Kobzol · 2026-07-11T13:10:35Z

            cmd.arg("--skip").arg(skip);
        }
-        cmd.arg("--sample-percentage").arg("10");
+        cmd.arg("--sample-percentage").arg("100");


Let's do this in a separate PR, it seems that the tests are a lot slower with this sample rate.

Ok, makes sense . Thank you!

Kobzol · 2026-07-14T21:15:53Z

        run.path("library/stdarch/crates/intrinsic-test")
    }

+    fn is_default_step(_builder: &Builder<'_>) -> bool {


Please also move test::IntrinsicTest to the very end in get_step_descriptions.

View changes since the review

rustbot · 2026-07-16T13:10:17Z

This PR modifies src/bootstrap/src/core/config.

If appropriate, please update CONFIG_CHANGE_HISTORY in src/bootstrap/src/utils/change_tracker.rs.

Kobzol · 2026-07-16T20:30:37Z

    pub change_id: Option<ChangeId>,
    pub bypass_bootstrap_lock: bool,
    pub ccache: Option<String>,
+    pub sde: Option<PathBuf>,


Please also add a Info level entry to change_tracker.rs, notifying users of this new config option, introduced in #159011. Thanks!

View changes since the review

Kobzol · 2026-07-16T20:32:20Z

+                    .config
+                    .paths
+                    .iter()
+                    .any(|p| p.to_string_lossy().contains("intrinsic-test"));


I think that we can do an exact match, because should_run will be changed to the intrinsic-test alias anyway.

I would also prefer if the build SDE presence and the explicit path checking was moved to make_run, since we already check the targets there.

View changes since the review

rustbot · 2026-07-19T05:21:26Z

Warning

If you are changing how CI LLVM is built or linked, make sure to bump
src/bootstrap/download-ci-llvm-stamp.

cc @jieyouxu

xonx4l · 2026-07-21T17:37:19Z

@Kobzol I'm a bit confused here different CI jobs (x86_64 , aarch64) seem to want different edits to the same snapshot, so blessing for one breaks the other. Could you help point me the right way? Thanks!

Kobzol · 2026-07-22T06:24:47Z

Ah, this is annoying, the test machinery on non-x64/aarch64 Linux skips the test altogether, so it changes the diff. And we cannot just easily pin a specific target in tests, because then some of the steps will fail.

Ok, sorry, I will have to revert one of my previous advices. Let's move the check for x64/aarch64 from make_run to run:

fn run(self, builder: &Builder<'_>) {
    let host = self.host;
    if !host.contains("aarch64-unknown-linux") && !host.contains("x86_64-unknown-linux") {
        builder.info(&format!("Skipping intrinsic-test, as it is not available for {host}"));
        return;
    }

You can run the tests locally using ./x test bootstrap.

xonx4l · 2026-07-22T07:06:04Z

Hmm ok, got it . Thank you!

rustbot · 2026-07-22T13:04:16Z

This PR was rebased onto a different main commit. Here's a range-diff highlighting what actually changed.

Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers.

xonx4l · 2026-07-22T14:03:27Z

Ah, this is annoying, the test machinery on non-x64/aarch64 Linux skips the test altogether, so it changes the diff. And we cannot just easily pin a specific target in tests, because then some of the steps will fail.

Ok, sorry, I will have to revert one of my previous advices. Let's move the check for x64/aarch64 from make_run to run:
fn run(self, builder: &Builder<'_>) {
    let host = self.host;
    if !host.contains("aarch64-unknown-linux") && !host.contains("x86_64-unknown-linux") {
        builder.info(&format!("Skipping intrinsic-test, as it is not available for {host}"));
        return;
    }
You can run the tests locally using ./x test bootstrap.

@Kobzol Applied the change and ./x test bootstrap passes locally but --bless produces no snapshot changes at all. Thank you!

Kobzol · 2026-07-22T19:28:38Z

The target check still seems to be in make_run :) (let's push what you have, so I can check)

Kobzol · 2026-07-23T07:30:40Z

Ah, I see what is the issue now. The test step is not represented in the blessed output at all. We are looking at the build step of the IntrinsicTest tool, and that only happens on x64 and aarch64. Let's just skip the whole step in tests. I pushed a fix.

xonx4l · 2026-07-23T10:17:18Z

OK, Thank you!

Kobzol · 2026-07-23T10:41:22Z

@bors squash msg="Run intrinsic-test bootstrap step by default"

Co-authored-by: Jakub Beránek <berykubik@gmail.com>

rust-bors · 2026-07-23T10:41:47Z

🔨 15 commits were squashed into c19d4a0.

Kobzol · 2026-07-23T10:42:12Z

Let's try again :)

@bors r+

rust-bors · 2026-07-23T10:42:15Z

📌 Commit c19d4a0 has been approved by Kobzol

It is now in the queue for this repository.

rust-bors · 2026-07-23T14:00:20Z

☀️ Test successful - CI
Approved by: Kobzol
Duration: 3h 16m 33s
Pushing 165cce8 to main...

github-actions · 2026-07-23T14:03:53Z

What is this?

This is an experimental post-merge analysis report that shows differences in test outcomes between the merged PR and its parent PR.

Comparing 390279b (parent) -> 165cce8 (this PR)

Test differences

Show 4 test diffs

4 doctest diffs were found. These are ignored, as they are noisy.

Test dashboard

Run

cargo run --manifest-path src/ci/citool/Cargo.toml -- \
    test-dashboard 165cce8d820b229af8f6a8226cf0b910b57600ff --output-dir test-dashboard

And then open test-dashboard/index.html in your browser to see an overview of all executed tests.

Job duration changes

x86_64-gnu-llvm-21-2: 1h 3m -> 1h 27m (+37.3%)
i686-msvc-2: 1h 30m -> 2h 4m (+37.0%)
x86_64-gnu-llvm-21-1: 59m 37s -> 38m 1s (-36.2%)
dist-armv7-linux: 1h 32m -> 1h 3m (-30.7%)
x86_64-gnu-aux: 2h 46m -> 1h 58m (-29.0%)
i686-gnu-1: 2h 20m -> 1h 40m (-28.6%)
x86_64-gnu-distcheck: 1h 44m -> 2h 11m (+24.9%)
x86_64-gnu-gcc-core-tests: 10m 56s -> 8m 13s (-24.8%)
dist-powerpc64le-linux-gnu: 1h 14m -> 1h 32m (+24.8%)
dist-various-2: 44m 49s -> 33m 50s (-24.5%)

How to interpret the job duration changes?

Job durations can vary a lot, based on the actual runner instance
that executed the job, system noise, invalidated caches, etc. The table above is provided
mostly for t-infra members, for simpler debugging of potential CI slow-downs.

rust-timer · 2026-07-23T14:41:52Z

Finished benchmarking commit (165cce8): comparison URL.

Overall result: no relevant changes - no action needed

@rustbot label: -perf-regression

Instruction count

This perf run didn't have relevant results for this metric.

Max RSS (memory usage)

Results (primary 5.1%, secondary 2.8%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	5.1%	[5.1%, 5.1%]	1
Regressions ❌ (secondary)	2.8%	[2.8%, 2.8%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	5.1%	[5.1%, 5.1%]	1

Cycles

Results (secondary 3.9%)

A less reliable metric. May be of interest, but not used to determine the overall result above.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	3.9%	[3.5%, 4.3%]	2
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-	-	0

Binary size

This perf run didn't have relevant results for this metric.

Bootstrap: 486.724s -> 485.902s (-0.17%)
Artifact size: 387.61 MiB -> 387.66 MiB (0.01%)

rustbot assigned Kobzol Jul 9, 2026

rustbot added A-testsuite Area: The testsuite used to check the correctness of rustc S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. T-bootstrap Relevant to the bootstrap subteam: Rust's build system (x.py and src/bootstrap) labels Jul 9, 2026