The illumos linker does not support --strip-debug

[citrus-it]

When building and testing rust 1.64.0 on illumos, we saw a large number of failing tests associated with:

   = note: ld: fatal: unrecognized option '--strip-debug'
           ld: fatal: use the -z help option for usage information
           collect2: error: ld returned 1 exit status

The illumos linker does not support the --strip-debug option (although it does support --strip-all).

[rust-highfive]

r? @nagisa

(rust-highfive has picked a reviewer for you, use r? to override)

[nagisa]

I believe the intended solution here would be to invoke an external tool after linking that could do this, similar to the handling for macOS as introduced in #82037.

You probably could just reuse the code path there, for the time being. That way when somebody gets to fixing the way we represent linkers in rustc, illumos will get automatically handled the mostly “right” way alongside with macOS (though you should still keep the comment to the effect that illumos does support --strip-all and could be improved to use that too)

[citrus-it]

Thanks for taking a look at this. I've pushed a new commit which uses the linker's --strip-all but invokes an external strip utility for the --strip-debug case. After talking to people who know more about linkers than I do, they advised that it is best to let the linker do this when possible, hence the hybrid approach.
It may be that the illumos linker can be extended in the future to support --strip-debug but until then this seems like the best approach, and I'd like people to be able to successfully run the tests on illumos again.

[nagisa]

@bors r+ Seems OK.

[bors]

📌 Commit f8a3dc4 has been approved by nagisa

It is now in the queue for this repository.

[JohnTitor]

This needs to run rustfmt (somehow CI didn't run on this PR): #102865 (comment)
@bors r-

[citrus-it]

I've run nightly rustfmt on compiler/rustc_codegen_ssa/src/back/link.rs and pushed the branch. It fails to run on linker.rs, the first error is:

build% ~/.rustup/toolchains/nightly-x86_64-unknown-illumos/bin/rustfmt compiler/rustc_codegen_ssa/src/back/linker.rs
error: expected identifier, found keyword `let`
   --> /data/omnios-build/omniosorg/rust/compiler/rustc_codegen_ssa/src/back/linker.rs:676:17
    |
675 |             let res: io::Result<()> = try {
    |                                       --- while parsing this struct
676 |                 let mut f = BufWriter::new(File::create(&path)?);
    |                 ^^^ expected identifier, found keyword

[lqd]

Don't run your local rustfmt manually, it won't be the correct version and you'll miss parameters (e.g. the edition), causing the errors above.

Use ./x.py fmt instead. There's more info in the dev guide here.

[citrus-it]

Don't run your local rustfmt manually, it won't be the correct version and you'll miss parameters (e.g. the edition), causing the errors above.

Use ./x.py fmt instead. There's more info in the dev guide here.

Thanks, I've run that, it exited cleanly and made no changes.

build:rust:illumos-strip-debug% ./x.py fmt
Building rustbuild
    Finished dev [unoptimized] target(s) in 0.09s
Build completed successfully in 0:00:03
build:rust:illumos-strip-debug% git diff
build:rust:illumos-strip-debug%

[lqd]

That means the formatting changes you pushed earlier fixed all the formatting issues (and linker.rs was already correctly formatted: the failures you were seeing there were solely due to the missing edition parameter, not actual errors preventing this PR from landing).

So the branch is now correctly formatted. Tidy passes, CI is green.

[nagisa]

@bors r+

[bors]

📌 Commit f3deac2 has been approved by nagisa

It is now in the queue for this repository.

[bors]

⌛ Testing commit f3deac2 with merge a24a020...

[bors]

☀️ Test successful - checks-actions
Approved by: nagisa
Pushing a24a020 to master...

[rust-timer]

Finished benchmarking commit (a24a020): 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

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean1	range	count2
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	-	-	0
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-3.3%	[-4.3%, -2.2%]	5
All ❌✅ (primary)	-	-	0

Cycles

Results

This is a less reliable metric that may be of interest but was not used to determine the overall result at the top of this comment.

	mean1	range	count2
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	2.6%	[2.6%, 2.6%]	1
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-	-	0

Footnotes

1. the arithmetic mean of the percent change ↩ ↩²

2. number of relevant changes ↩ ↩²