[gurry]

View all comments

This PR adds:

• A new intrinsic called codeview_annotation, which lowers to llvm.codeview.annotation and allows users to emit arbitrary S_ANNOTATION records into PDB files
• A std lib API to call this intrinsic

For the underlying motivation and design details see:

• Intrinsic MCP: Add codeview_annotation intrinsic compiler-team#1026
• API ACP: Add API to call codeview_annotation intrinsic libs-team#868

Tracking issues:

• Intrinsic tracking issue: Tracking issue for MCP 1026: Add codeview_annotation intrinsic #162252
• API tracking issue: Tracking Issue for codeview_annotation API #163964

Implementation details

• Codgen happens only on LLVM backend and only if the target uses PDBs. It is a no-op in all other cases. A compile error will still be emitted in all cases (even on non-LLVM, non-PDB ones) if CodeViewAnnotationArgs::ARGS fails to const eval.
• Had to change the feature gate on core::os module to a neutral core_os in order to add the new module core::os::windows. It was earlier gated on darwin_objc, which was not general enough.

[rustbot]

⚠️ #[miri::intrinsic_fallback_is_spec] must only be used if the function actively checks for all UB cases,
and explores the possible non-determinism of the intrinsic.

cc @rust-lang/miri

Any special-casing of Miri in the standard library requires review.

cc @rust-lang/miri

rustc_codegen_gcc is developed in its own repository. If possible, consider making this change to rust-lang/rustc_codegen_gcc instead.

cc @antoyo, @GuillaumeGomez

miri is developed in its own repository. If the Miri part of this change can be broken out, consider making this change to rust-lang/miri instead. However, if Miri needs adjusting for rustc changes, just ignore this message.

cc @rust-lang/miri

rustc_codegen_cranelift is developed in its own repository. If possible, consider making this change to rust-lang/rustc_codegen_cranelift instead.

cc @bjorn3

Some changes occurred to the intrinsics. Make sure the CTFE / Miri interpreter
gets adapted for the changes, if necessary.

cc @rust-lang/miri, @RalfJung, @oli-obk, @lcnr

[rustbot]

r? @folkertdev

rustbot has assigned @folkertdev.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

Why was this reviewer chosen?

The reviewer was selected based on:

• Owners of files modified in this PR: compiler
• compiler expanded to 77 candidates
• Random selection from 17 candidates

[gurry]

⚠️ #[miri::intrinsic_fallback_is_spec] must only be used if the function actively checks for all UB cases, and explores the possible non-determinism of the intrinsic.

The function does not caues any UB or non-determinism (it returns () and has no run-time side effects) so the fallback should be fine.

rustc_codegen_gcc is developed in its own repository. If possible, consider making this change to rust-lang/rustc_codegen_gcc instead.

cc @antoyo, @GuillaumeGomez

...
rustc_codegen_cranelift is developed in its own repository. If possible, consider making this change to rust-lang/rustc_codegen_cranelift instead.

I have now removed the changes in rustc_codegen_gcc and rustc_codegen_cranelift. I had earlier added explicit no-op handling for the intrinic in these backends, but I think the fallback function is enough.

[folkertdev]

Some small things, @RalfJung does the const eval stuff look OK to you otherwise?

View changes since this review

[RalfJung]

On the const-eval / Miri side there's just the fallback body and that seems reasonable to me.

[rustbot]

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.

[folkertdev]

Looks good, thanks!

@bors r+

View changes since this review

[rust-bors]

📌 Commit 9d8da0d has been approved by folkertdev

It is now in the queue for this repository.

[rust-bors]

☀️ Test successful - CI
Approved by: folkertdev
Duration: 2h 58m 8s
Pushing f4d7ecf to main...

[github-actions]

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 5d8fec0 (parent) -> f4d7ecf (this PR)

Test differences

Show 116 test diffs

Stage 1

• [codegen] tests/codegen-llvm/intrinsics/codeview-annotation-non-pdb-targets.rs: [missing] -> pass (J2)
• [codegen] tests/codegen-llvm/intrinsics/codeview-annotation.rs#OPT0: [missing] -> ignore (only executed when the target environment is msvc) (J2)
• [codegen] tests/codegen-llvm/intrinsics/codeview-annotation.rs#OPT3: [missing] -> ignore (only executed when the target environment is msvc) (J2)
• [ui] tests/ui/feature-gates/feature-gate-codeview_annotation.rs: [missing] -> ignore (only executed when the operating system is windows) (J4)
• [ui] tests/ui/intrinsics/codeview-annotation-bad-args.rs: [missing] -> pass (J4)
• [ui] tests/ui/intrinsics/codeview-annotation.rs: [missing] -> ignore (only executed when the operating system is windows) (J4)
• [ui (polonius)] tests/ui/feature-gates/feature-gate-codeview_annotation.rs: [missing] -> ignore (only executed when the operating system is windows) (J7)
• [ui (polonius)] tests/ui/intrinsics/codeview-annotation-bad-args.rs: [missing] -> pass (J7)
• [ui (polonius)] tests/ui/intrinsics/codeview-annotation.rs: [missing] -> ignore (only executed when the operating system is windows) (J7)

Stage 2

• [codegen] tests/codegen-llvm/intrinsics/codeview-annotation-non-pdb-targets.rs: [missing] -> pass (J0)
• [codegen] tests/codegen-llvm/intrinsics/codeview-annotation.rs#OPT0: [missing] -> ignore (only executed when the target environment is msvc) (J0)
• [codegen] tests/codegen-llvm/intrinsics/codeview-annotation.rs#OPT3: [missing] -> ignore (only executed when the target environment is msvc) (J0)
• [ui] tests/ui/feature-gates/feature-gate-codeview_annotation.rs: [missing] -> ignore (only executed when the operating system is windows) (J1)
• [ui] tests/ui/intrinsics/codeview-annotation.rs: [missing] -> ignore (only executed when the operating system is windows) (J1)
• [ui] tests/ui/feature-gates/feature-gate-codeview_annotation.rs: [missing] -> pass (J3)
• [ui] tests/ui/intrinsics/codeview-annotation.rs: [missing] -> pass (J3)
• [codegen] tests/codegen-llvm/intrinsics/codeview-annotation-non-pdb-targets.rs: [missing] -> ignore (ignored when the target environment is msvc) (J5)
• [codegen] tests/codegen-llvm/intrinsics/codeview-annotation.rs#OPT0: [missing] -> pass (J5)
• [codegen] tests/codegen-llvm/intrinsics/codeview-annotation.rs#OPT3: [missing] -> pass (J5)
• [ui] tests/ui/intrinsics/codeview-annotation-bad-args.rs: [missing] -> pass (J6)

Additionally, 96 doctest diffs were found. These are ignored, as they are noisy.

Job group index

• J0: test-aarch64-apple-1, test-aarch64-apple-macos-26-1, test-aarch64-gnu, test-aarch64-gnu-llvm-22-1, test-arm-android, test-armhf-gnu, test-i686-gnu-1, test-i686-gnu-nopt-1, test-various, test-x86_64-gnu, test-x86_64-gnu-llvm-22-2, test-x86_64-gnu-nopt, test-x86_64-gnu-stable, test-x86_64-mingw-1
• J1: test-aarch64-apple-1, test-aarch64-apple-macos-26-1, test-aarch64-gnu, test-aarch64-gnu-llvm-22-1, test-arm-android, test-armhf-gnu, test-i686-gnu-1, test-i686-gnu-nopt-1, test-various, test-x86_64-gnu, test-x86_64-gnu-debug, test-x86_64-gnu-gcc, test-x86_64-gnu-llvm-22, test-x86_64-gnu-llvm-22-2, test-x86_64-gnu-nopt, test-x86_64-gnu-parallel-frontend, test-x86_64-gnu-stable
• J2: test-pr-check-2, test-x86_64-gnu-llvm-22-3
• J3: test-aarch64-msvc-1, test-i686-msvc, test-x86_64-mingw-1, test-x86_64-msvc-1
• J4: test-x86_64-gnu-llvm-22-3, test-x86_64-gnu-next-trait-solver-polonius
• J5: test-aarch64-msvc-1, test-i686-msvc, test-x86_64-msvc-1
• J6: test-aarch64-apple-1, test-aarch64-apple-macos-26-1, test-aarch64-gnu, test-aarch64-gnu-llvm-22-1, test-aarch64-msvc-1, test-arm-android, test-armhf-gnu, test-i686-gnu-1, test-i686-gnu-nopt-1, test-i686-msvc, test-various, test-x86_64-gnu, test-x86_64-gnu-debug, test-x86_64-gnu-gcc, test-x86_64-gnu-llvm-22, test-x86_64-gnu-llvm-22-2, test-x86_64-gnu-nopt, test-x86_64-gnu-parallel-frontend, test-x86_64-gnu-stable, test-x86_64-mingw-1, test-x86_64-msvc-1
• J7: test-x86_64-gnu-next-trait-solver-polonius

Test dashboard

Run

cargo run --manifest-path src/ci/citool/Cargo.toml -- \
    test-dashboard f4d7ecfb0efc8330cef8f8020f192920d8721911 --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

1. dist-loongarch64-linux: 5h 44m -> 1h 19m (-77.0%)
2. dist-loongarch64-musl: 5h 40m -> 1h 31m (-73.0%)
3. test-i686-gnu-nopt-1: 1h 31m -> 2h 24m (+58.5%)
4. test-x86_64-msvc-ext2: 1h 9m -> 1h 50m (+58.3%)
5. dist-arm-linux-gnueabi: 1h 4m -> 1h 41m (+57.9%)
6. test-x86_64-gnu: 1h 42m -> 2h 36m (+53.2%)
7. dist-s390x-linux: 1h 3m -> 1h 33m (+47.8%)
8. dist-x86_64-solaris: 1h 10m -> 1h 43m (+47.6%)
9. dist-powerpc64le-linux-musl: 1h 8m -> 1h 38m (+44.5%)
10. dist-i686-linux: 1h 20m -> 1h 53m (+42.2%)

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]

Finished benchmarking commit (f4d7ecf): 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 (secondary 4.2%)

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)	4.2%	[4.2%, 4.2%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-	-	0

Cycles

Results (secondary 2.0%)

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)	2.0%	[2.0%, 2.0%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-	-	0

Binary size

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

Bootstrap: 490.438s -> 497.38s (1.42%)
Artifact size: 406.97 MiB -> 407.06 MiB (0.02%)