ci: Add support for dist-loongarch64-linux

[heiher]

We are preparing to promote loongarch64-unknown-linux-gnu to Tier 2, and one of the tasks is to add CI support. We are currently in the process of upgrading the dependencies for the build tools, and before this is completed, we would like to request comments. Thanks

Progress

• Update linux-raw-sys to 0.3.2 (Update linux-raw-sys to 0.3.2 cargo#11998)
• Update openssl-src to 111.25.3+1.1.1t (Update openssl-src to 111.25.3+1.1.1t cargo#12005)
• bootstrap: Update linux-raw-sys to 0.3.2 (bootstrap: Update linux-raw-sys to 0.3.2 #110516)
• Update linux-raw-sys to 0.3.4 (Update linux-raw-sys to 0.3.4 #110518)
• Update cargo (Update cargo #110834)
• linux_like: Add missing constants for loongarch64 (linux_like: Add missing constants for loongarch64 libc#3237)
• Release 0.2.143 (Release 0.2.143 libc#3236)
• Update libc to 0.2.144 (Update libc to 0.2.144 cargo#12098)
• Update cargo (Update cargo #111456)

Tier 2 with host tools MCP: rust-lang/compiler-team#518

[rustbot]

r? @Mark-Simulacrum

(rustbot has picked a reviewer for you, use r? to override)

[heiher]

@bors try

[bors]

@heiher: 🔑 Insufficient privileges: not in try users

[heiher]

@Mark-Simulacrum Ready for review, ci test result: https://github.com/loongarch-rs/rust/actions/runs/4804319116

[cuviper]

Note that try only runs dist-x86_64-linux unless you (temporarily) edit the CI config for that job.

My crosstool refactoring in #110865 was just approved, so it would be great if you followed suit here. I see that you're using a git snapshot for loongarch support, but apart from that you should be able to work like the others.

[heiher]

Hello @cuviper, Thanks for your comments.

Note that try only runs dist-x86_64-linux unless you (temporarily) edit the CI config for that job.

You're right, I did make a temporary change to the CI configuration for that job.

• Commits: https://github.com/loongarch-rs/rust/commits/auto
• Action results: https://github.com/loongarch-rs/rust/actions

My crosstool refactoring in #110865 was just approved, so it would be great if you followed suit here. I see that you're using a git snapshot for loongarch support, but apart from that you should be able to work like the others.

Great job! Thank you for pushing the update for crosstool-ng. However, support for LoongArch is expected to be available only in the next release version (1.26). Therefore, it may be a good idea for us to use the git snapshot version for now and make the change once version 1.26 is released. This way, we shouldn't miss out on Rust 1.71.

I am grateful for your insightful comments and suggestions. ❤️

[cuviper]

@heiher Yes, it's fine that you need to use a pre-release version of crosstool-ng. I meant that you should try to follow the rest of the refactoring that I implemented: committing a defconfig rather than a full config, and using the new common scripts/crosstool-ng-build.sh instead of individual scripts like your build-toolchains.sh.

[bors]

☔ The latest upstream changes (presumably #110865) made this pull request unmergeable. Please resolve the merge conflicts.

[heiher]

@cuviper It has been rebased to the mainline and passed the tests. Does it look good to you please?

[xen0n]

This is much easier to review now, thanks for the rebasing! Only one question from me below.

[xen0n]

Are we able to raise this to GCC 13.1.0 now that it's released?

[heiher]

Certainly, although GCC 13 has been released, the upgrade for crosstool-ng is still in progress. We would like to catch up with the Rust 1.71 release, and since LoongArch CI is not dependent on GCC 13, I propose that we upgrade after crosstool-ng is ready, possibly through a new PR.

[heiher]

If crosstool-ng is ready earlier than expected, I would be happy to include it in this PR.

[xen0n]

Ah yes. I forgot we privately talked about this a while ago.

[cuviper]

I'm going to leave the formal review to Mark. I'm not sure if the target should be built in CI at all until it has approval per the target tier policy.

[cuviper]

We don't need GDB, so all 4 of these can be removed.

[cuviper]

On all other targets, I left out versions for things like ISL and ZSTD so they could just use their defaults. Are there particular needs here?

[cuviper]

We don't really need logging, but I guess it doesn't hurt either way.

Please do add the mirror config like the other targets though:

CT_USE_MIRROR=y
CT_MIRROR_BASE_URL="https://ci-mirrors.rust-lang.org/rustc"

(We don't have everything mirrored, but AIUI this will let our mirror take precedence when we do add things.)

[heiher]

I'm going to leave the formal review to Mark. I'm not sure if the target should be built in CI at all until it has approval per the target tier policy.

Thanks for your feedback. We are moving forward, and there is a resolved MCP that for LoongArch to Tier2: rust-lang/compiler-team#518

[Mark-Simulacrum]

Infra discussed and agreed to move forward with adding the builder.

@bors r+

[bors]

📌 Commit 5f173e9 has been approved by Mark-Simulacrum

It is now in the queue for this repository.

[bors]

⌛ Testing commit 5f173e9 with merge 4f80088941583158b9f62e8d6a60b94a087329d9...

[bors]

💥 Test timed out

[xen0n]

The dist-x86_64-apple build seems to be flaking out.

[rust-log-analyzer]

A job failed! Check out the build log: (web) (plain)

Click to see the possible cause of the failure (guessed by this bot)

[heiher]

@bors retry

This PR doesn't seem bad, retry again.

[bors]

@heiher: 🔑 Insufficient privileges: not in try users

[workingjubilee]

@bors retry

[bors]

⌛ Testing commit 5f173e9 with merge cda5bec...

[klensy]

rsync used somewhere? Can't see any matches here.

[heiher]

It is used by crosstool-ng git version.

[bors]

☀️ Test successful - checks-actions
Approved by: Mark-Simulacrum
Pushing cda5bec to master...

[rust-timer]

Finished benchmarking commit (cda5bec): 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.

	mean	range	count
Regressions ❌ (primary)	-	-	0
Regressions ❌ (secondary)	2.9%	[2.8%, 3.0%]	2
Improvements ✅ (primary)	-	-	0
Improvements ✅ (secondary)	-1.6%	[-1.6%, -1.6%]	1
All ❌✅ (primary)	-	-	0

Cycles

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

Binary size

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

Bootstrap: 644.363s -> 644.28s (-0.01%)