Add target_feature = "gc" for Wasm

[daxpedda]

This PR adds a new target feature for Wasm targets: GC.

I went ahead and made this insta-stable because it has already passed phase 5 in addition to being implemented and stabilized by all major JS engines.

This doesn't enable the target feature by default!

For context: while this proposal adds a lot of new features to Wasm, they are not accessible through Rust apart from unstable inline ASM. This is largely useful to signal tools to make use of the feature, e.g. wasm-bindgen, where this can be used to make some pretty optimizations.

Companion PR: rust-lang/reference#2114.

r? @alexcrichton

[daxpedda]

Ah, 'tis unfortunate, will have to wait until the next LLVM version.

[alexcrichton]

Looks good to me, thanks! You can update this function to skip the feature on unsupported LLVM versions. Do you have a use case in mind for the insta-stable-ness? I would expect that the GC configuration for wasm-bindgen would be out-of-band rather than via a compiler configuration (e.g. an extra flag to wasm-bindgen itself)

[daxpedda]

Do you have a use case in mind for the insta-stable-ness? I would expect that the GC configuration for wasm-bindgen would be out-of-band rather than via a compiler configuration (e.g. an extra flag to wasm-bindgen itself)

wasm-bindgen has shifted to removing extra CLI configurations in favor of automatic detection, e.g. reference types, multivalue and other proposals. So instead of a --enable-gc flag we want to detect target_feature = "gc".

In the meantime of course wasm-bindgen might just introduce a switch.

[daxpedda]

You can update this function to skip the feature on unsupported LLVM versions.

Done.

[alexcrichton]

What would you think about landing this unstable and having a follow-up PR to switch it to stable? That'd help and get this in without merge conflicts and it can be separated from the FCP required for stabilization too. I'm not actually certain myself which team would need to sign off on this, T-lang or T-compiler, in terms of stabilization

[daxpedda]

What would you think about landing this unstable and having a follow-up PR to switch it to stable? That'd help and get this in without merge conflicts and it can be separated from the FCP required for stabilization too. I'm not actually certain myself which team would need to sign off on this, T-lang or T-compiler, in terms of stabilization

Yeah, why not!

I will open a follow-up PR right away for stabilization so we can get that rolling in the meantime until LLVM 22 is released (a long way off probably).

[alexcrichton]

@bors: r+

[bors]

📌 Commit 4ee01c4 has been approved by alexcrichton

It is now in the queue for this repository.

[bors]

⌛ Testing commit 4ee01c4 with merge 9044e98...

[bors]

☀️ Test successful - checks-actions
Approved by: alexcrichton
Pushing 9044e98 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 806c2a3 (parent) -> 9044e98 (this PR)

Test differences

Show 3 test diffs

3 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 9044e98b66d074e7f88b1d4cea58bb0538f2eda6 --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. pr-check-1: 1559.0s -> 2035.2s (+30.5%)
2. aarch64-apple: 9361.6s -> 12113.6s (+29.4%)
3. x86_64-gnu-miri: 4149.4s -> 5065.6s (+22.1%)
4. x86_64-gnu-tools: 3258.4s -> 3793.7s (+16.4%)
5. dist-x86_64-apple: 8153.6s -> 7035.5s (-13.7%)
6. test-various: 6635.9s -> 7472.1s (+12.6%)
7. x86_64-gnu-llvm-21-3: 6224.2s -> 6968.4s (+12.0%)
8. x86_64-gnu-llvm-21-1: 4162.3s -> 4652.1s (+11.8%)
9. x86_64-gnu-gcc: 3177.9s -> 3544.4s (+11.5%)
10. i686-gnu-2: 5310.1s -> 5891.4s (+10.9%)

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 (9044e98): comparison URL.

Overall result: no relevant changes - no action needed

@rustbot label: -perf-regression

Instruction count

This benchmark run did not return any relevant results for this metric.

Max RSS (memory usage)

Results (primary 6.9%)

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

	mean	range	count
Regressions ❌ (primary)	6.9%	[6.9%, 6.9%]	1
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	6.9%	[6.9%, 6.9%]	1

Cycles

Results (primary -7.1%, secondary -4.8%)

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)	-	-	0
Improvements ✅ (primary)	-7.1%	[-15.9%, -2.6%]	5
Improvements ✅ (secondary)	-4.8%	[-8.4%, -2.4%]	6
All ❌✅ (primary)	-7.1%	[-15.9%, -2.6%]	5

Binary size

This benchmark run did not return any relevant results for this metric.

Bootstrap: 477.964s -> 478.41s (0.09%)
Artifact size: 390.53 MiB -> 390.55 MiB (0.00%)