Update to LLVM 18.1.0 rc 4

[nikic]

Fixes #120819.
Fixes #121180.
Fixes #121239.
Fixes #121367.

[rustbot]

r? @cuviper

rustbot has assigned @cuviper.
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

[rustbot]

⚠️ Warning ⚠️

• These commits modify submodules.

[cuviper]

@bors r+ rollup

[bors]

📌 Commit 17a7f3c has been approved by cuviper

It is now in the queue for this repository.

[cuviper]

@bors rollup=never

[bors]

⌛ Testing commit 17a7f3c with merge 729a664...

[bors]

💔 Test failed - checks-actions

[nikic]

Probably related to the libLLVM soname change in llvm/llvm-project@235306b.

[rustbot]

This PR changes how LLVM is built. Consider updating src/bootstrap/download-ci-llvm-stamp.

[nikic]

@bors try

[bors]

⌛ Trying commit 87481e3 with merge da675ab...

[bors]

💔 Test failed - checks-actions

[nikic]

@bors try

[bors]

⌛ Trying commit bb1d57b with merge 6793580...

[cuviper]

Aw, I was hoping that the binary size actually did get better...
(IIRC, BOLT was essentially doubling .text with its rewrite.)

[Kobzol]

https://perf.rust-lang.org/compare.html?start=6cbf0926d54c80ea6d15df333be9281f65bbeb36&end=5e826decbc9c72668d6e1e8e7116a68c86f877c9&stat=instructions%3Au&tab=artifact-size it would have to be infinitely better, since the libLLVM.so file simply disappeared from the artifact size metrics 😁

[cuviper]

I don't think not distributing the symlink is possible, so what I did now is to limit it to just the rust-dev component (aka ci-llvm). That's the only place that should need it. All the "normal" dist tarballs continue to be symlink-free.

I can see that the symlink survives into the rust-dev tarball, but it looks like its install.sh is dereferencing that, because the installed copy gets duplicated as a whole file.

[nikic]

I don't think not distributing the symlink is possible, so what I did now is to limit it to just the rust-dev component (aka ci-llvm). That's the only place that should need it. All the "normal" dist tarballs continue to be symlink-free.

I can see that the symlink survives into the rust-dev tarball, but it looks like its install.sh is dereferencing that, because the installed copy gets duplicated as a whole file.

Does the behavior of install.sh matter though? I do get the symlink when using the try build in download-ci-llvm, which I believe is the only thing this component should get used for.

[cuviper]

Hmm, I was thinking of it as a rustup component, but I guess we don't actually distribute it that way. (and rustup doesn't invoke install.sh either.)

@bors r+ rollup=never

[bors]

📌 Commit 1a652fa has been approved by cuviper

It is now in the queue for this repository.

[bors]

⌛ Testing commit 1a652fa with merge e612d07...

[bors]

☀️ Test successful - checks-actions
Approved by: cuviper
Pushing e612d07 to master...

[cuviper]

All good -- with that merged, a new ./x build does extract rust-dev with that symlink preserved!

[rust-timer]

Finished benchmarking commit (e612d07): comparison URL.

Overall result: ❌✅ regressions and improvements - ACTION NEEDED

Next Steps: If you can justify the regressions found in this perf run, please indicate this with @rustbot label: +perf-regression-triaged along with sufficient written justification. If you cannot justify the regressions please open an issue or create a new PR that fixes the regressions, add a comment linking to the newly created issue or PR, and then add the perf-regression-triaged label to this PR.

@rustbot label: +perf-regression
cc @rust-lang/wg-compiler-performance

Instruction count

This is a highly reliable metric that was used to determine the overall result at the top of this comment.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	1.7%	[1.7%, 1.7%]	1
Improvements ✅ (primary)	-2.7%	[-2.7%, -2.7%]	1
Improvements ✅ (secondary)	-	-	0
All ❌✅ (primary)	-2.7%	[-2.7%, -2.7%]	1

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.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	4.4%	[3.9%, 4.8%]	2
Improvements ✅ (primary)	-1.0%	[-1.0%, -1.0%]	1
Improvements ✅ (secondary)	-3.4%	[-4.7%, -2.0%]	2
All ❌✅ (primary)	-1.0%	[-1.0%, -1.0%]	1

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.

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

Binary size

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

Bootstrap: 650.452s -> 652.171s (0.26%)
Artifact size: 311.17 MiB -> 175.40 MiB (-43.63%)

[RalfJung]

Something seems to still be wrong with the new soname: #121889

[lqd]

syn-opt still looks currently noisy, and the coercions change looks noisy as well, albeit with a bigger amplitude than recently: it is a few merges later trending down close to prior to this PR, via some steps that didn't change compilation (a cargo update).

@rustbot label: +perf-regression-triaged