stdarch subtree update#146669

a4lg · 2025-09-17T08:19:22Z

Subtree update of stdarch to rust-lang/stdarch@9f12c1a.

Created using https://github.com/rust-lang/josh-sync.

The immediate here encodes both the rounding mode (in the low bits) and the scale (in the high bits). Make sure the scale is non-zero.

Fix CI after LLVM upgrade

Add testing for Arm64EC Windows

wasm: use `{read, write}_unaligned` methods

…g-add-sub use `simd_saturating_{add, sub}` on neon

`s390x` llvm 21 improvements

It's historically been flaky and is no longer needed now that std_detect has been moved out of this repository.

Remove FreeBSD CI

use `llvm.roundeven` on arm

use `qemu-user` instead of `qemu-user-static` for loongarch CI

They don't need full "Zbc" extension but only its subset: the "Zbkc" extension. Since the compiler implies `zbkc` from `zbc`, it's safe to use `#[target_feature(enable = "zbkc")]`.

…ments RISC-V: Lower requirements of `clmul` and `clmulh`

Make Clippy happy (batch 2025-09)

This commit performs various improvements (better register allocation, less register clobbering on the worst case and better readability) of RISC-V inline assembly use cases. Note that it does not change the `p` module (which defines the "P" extension draft instructions but very likely to change). 1. Use `lateout` as possible. Unlike `out(reg)` and `in(reg)` pair, `lateout(reg)` and `in(reg)` can share the same register because they state that the late-output register is written after all the reads are performed. It can improve register allocation. 2. Add `preserves_flags` option as possible. While RISC-V doesn't have _regular_ condition codes, RISC-V inline assembly in the Rust language assumes that some registers (mainly vector state registers) may be overwritten by default. By adding `preserves_flags` to the intrinsics corresponding instructions without overwriting them, it can minimize register clobbering on the worst case. 3. Use trailing semicolon. As `asm!` declares an action and it doesn't return a value by itself, it would be better to have trailing semicolon to denote that an `asm!` call is effectively a statement. 4. Make most of `asm!` calls multi-lined. `rustfmt` makes some simple (yet long) `asm!` calls multi-lined but it does not perform formatting of complex `asm!` calls with inputs and/or outputs. To keep consistency, it makes most of the `asm!` calls multi-lined.

Remove big-endian swizzles from `vreinterpret`

…improvements RISC-V: Improvements of inline assembly uses

rustbot · 2025-09-17T08:19:26Z

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

cc @Amanieu, @folkertdev, @sayantn

a4lg · 2025-09-17T08:25:33Z

It's the first time I submit a subtree update PR and I'm not sure whether I'm doing this properly (though I used rustc-josh-sync tool).

My personal motivation for submitting this is a minor fix rust-lang/stdarch#1918 (but it's going to be a prerequisite of larger experiments that require changing both rust and stdarch at the same time).

folkertdev · 2025-09-17T09:40:23Z

I suspect you just need to pick a more recent base commit for CI to work? Otherwise this looks alright to me.

a4lg · 2025-09-17T10:16:47Z

Hmm, I changed rust-version to ba4b6437b301b37a0e1f8f19df74d7d9c832d263 (current latest master; through (a) updating rust-version file and making an update commit or (b) using different rust-version file and the --rust-version-path option) and got:

Updating/installing josh-proxy binary...
+ cd "[...]/stdarch" && "cargo" "install" "--locked" "--git" "https://github.com/josh-project/josh" "--tag" "r24.10.04" "josh-proxy"
+ cd "[...]/stdarch" && "git" "status" "--untracked-files=no" "--porcelain"
josh up and running
Preparing https://github.com/a4lg/rust (base: ba4b6437b301b37a0e1f8f19df74d7d9c832d263)...
+ cd "[...]/rust" && "git" "fetch" "https://github.com/a4lg/rust" "stdarch-sync-20250917-1"
+ cd "[...]/rust" && "git" "fetch" "https://github.com/rust-lang/rust" "ba4b6437b301b37a0e1f8f19df74d7d9c832d263"
+ cd "[...]/rust" && "git" "push" "https://github.com/a4lg/rust" "ba4b6437b301b37a0e1f8f19df74d7d9c832d263:refs/heads/stdarch-sync-20250917-1"

Pushing changes...
+ cd "[...]/stdarch" && "git" "push" "http://localhost:42042/a4lg/rust.git:/library/stdarch.git" "HEAD:stdarch-sync-20250917-1"
Error: cannot perform push

Caused by:
    Command `cd "[...]/stdarch" && "git" "push" "http://localhost:42042/a4lg/rust.git:/library/stdarch.git" "HEAD:stdarch-sync-20250917-1"` failed with exit code Some(1). STDOUT:
    
    STDERR:
    To http://localhost:42042/a4lg/rust.git:/library/stdarch.git
     ! [rejected]                  HEAD -> stdarch-sync-20250917-1 (fetch first)
    error: failed to push some refs to 'http://localhost:42042/a4lg/rust.git:/library/stdarch.git'
    hint: Updates were rejected because the remote contains work that you do not
    hint: have locally. This is usually caused by another repository pushing to
    hint: the same ref. If you want to integrate the remote changes, use
    hint: 'git pull' before pushing again.
    hint: See the 'Note about fast-forwards' in 'git push --help' for details.

Classic merge conflict, except that I don't understand why is it happening.
Perhaps I need help from experts.

a4lg · 2025-09-17T10:41:31Z

After bisecting, the last succeeding Rust commit was 63f6845 and the first failing one is c23f07d (both only counting first parents) but changes between them seem unrelated to stdarch.

rustbot · 2025-09-17T12:53:36Z

This PR was rebased onto a different master 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.

rustbot · 2025-09-17T12:53:37Z

⚠️ Warning ⚠️

This PR is based on an upstream commit that is older than 28 days.

It's recommended to update your branch according to the rustc-dev-guide.

a4lg · 2025-09-17T14:34:08Z

Wow. Somehow exact unchanged PR passed the CI (possibly CI was flaky?)

folkertdev · 2025-09-17T15:14:56Z

I think it's allright, @Kobzol can you double-check that this looks good and won't cause issues?

a4lg · 2025-09-18T01:40:45Z

I'm not sure whether the history is fully preserved (because removing feature gates occurred independently on both rust and stdarch).

But at least the contents under library/stdarch completely matches to the stdarch repository (the latest version as of the initial submission; rust-lang/stdarch@9f12c1a).

Kobzol · 2025-09-18T07:29:18Z

Umm, it's really suspicious to get a merge conflict on a push when the previous pull was empty (https://github.com/rust-lang/stdarch/actions/runs/17817660757/job/50654001348). Also, the rust-version file should never be manually modified, it should only get modified by pulls. For "additional safety" I would suggest closing this PR and doing a "clean" push starting from stdarch's master in a new PR.

a4lg · 2025-09-18T08:23:10Z

@Kobzol
While investigating the first CI failure, I temporally changed rust-version manually (e.g. on 0fdedcf).
However, rust-version of the current PR (c1242fa) is now intact and exactly the same as the initial, clean result of josh push (somehow, CI failed on the first time but succeeded on the second time).

Note: You can see the changes:

c1242fa (clean josh push: CI failed)
0fdedcf (experimenting with rust-version: merge conflict detected on GitHub's interface)
c1242fa (back to the original; CI somehow succeeded)

a4lg · 2025-09-18T08:43:07Z

@Kobzol Anyway, thanks for warning me about changing rust-version. I was not sure how to handle that file.

a4lg · 2025-09-18T08:46:44Z

Ahh, I see. Additional safety. Okay, I'm closing this and opening the exact same commit (clean josh push) with another PR.

dpaoliello and others added 30 commits August 10, 2025 13:19

Add testing for Arm64EC Windows

f2c0c3d

Drop no longer needed feature gates

f9bc63d

Use intrinsics for some s390x operations

135de7c

Add missing avx512vl target features

4a8b823

Work around selection failure without avx512vl

92f6310

Adjust immediate for vrndscalepd tests

3302e3e

The immediate here encodes both the rounding mode (in the low bits) and the scale (in the high bits). Make sure the scale is non-zero.

Merge pull request rust-lang#1897 from nikic/dummy

fe3a9b7

Fix CI after LLVM upgrade

s390x: define unpack_low using core::intrinsics::simd

fa163a1

s390x: add assert_instr for vec_round

c5ec096

s390x: add assert_instr for vec_extend

97d6466

s390x: implement vec_mule using core::intrinsics::simd

1cda88a

wasm: use {read, write}_unaligned methods

d5cb1c4

s390x: implement vec_mulo using core::intrinsics::simd

e1a1b1d

s390x: implement vec_subc_u128 using overflowing_sub

dfa95c6

s390x: implement vec_sld using fshl

e9162f2

s390x: link to a missed optimization

45af206

Merge pull request rust-lang#1899 from dpaoliello/arm64ec

6d74280

Add testing for Arm64EC Windows

Merge pull request rust-lang#1901 from folkertdev/wasm-read-unaligned

0b0c424

wasm: use `{read, write}_unaligned` methods

use simd_saturating_{add, sub} on neon

98bd1d7

Merge pull request rust-lang#1575 from folkertdev/neon-simd-saturatin…

e8b818d

…g-add-sub use `simd_saturating_{add, sub}` on neon

Merge pull request rust-lang#1903 from folkertdev/s390x-llvm-21-fixes

b2189b8

`s390x` llvm 21 improvements

use llvm.roundeven on arm

ae648be

Remove FreeBSD CI

00b27f4

It's historically been flaky and is no longer needed now that std_detect has been moved out of this repository.

Merge pull request rust-lang#1907 from Amanieu/remove-cirrus

6da36e6

Remove FreeBSD CI

Merge pull request rust-lang#1906 from folkertdev/arm-roundeven

bbb222f

use `llvm.roundeven` on arm

Remove some llvm workarounds

18fa6d9

use qemu-user instead of qemu-user-static for loongarch CI

bb3598e

Merge pull request rust-lang#1912 from folkertdev/loongarch-qemu-user

60874ec

use `qemu-user` instead of `qemu-user-static` for loongarch CI

RISC-V: Lower requirements of clmul and clmulh

4679533

They don't need full "Zbc" extension but only its subset: the "Zbkc" extension. Since the compiler implies `zbkc` from `zbc`, it's safe to use `#[target_feature(enable = "zbkc")]`.

Merge pull request rust-lang#1909 from a4lg/riscv-clmul-lower-require…

0a97662

…ments RISC-V: Lower requirements of `clmul` and `clmulh`

folkertdev and others added 4 commits September 12, 2025 12:39

Merge pull request rust-lang#1920 from a4lg/clippy-warnings-2025.09

3338743

Make Clippy happy (batch 2025-09)

Merge pull request rust-lang#1919 from sayantn/fix-vreinterpret

5dd0fdc

Remove big-endian swizzles from `vreinterpret`

Merge pull request rust-lang#1921 from a4lg/riscv-inline-asm-general-…

c1242fa

…improvements RISC-V: Improvements of inline assembly uses

rustbot added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. T-libs Relevant to the library team, which will review and decide on the PR/issue. labels Sep 17, 2025

This comment has been minimized.

Sign in to view

a4lg force-pushed the stdarch-sync-20250917 branch from c1242fa to 0fdedcf Compare September 17, 2025 10:51

This comment has been minimized.

Sign in to view

a4lg force-pushed the stdarch-sync-20250917 branch from 0fdedcf to c1242fa Compare September 17, 2025 12:53

a4lg closed this Sep 18, 2025

rustbot removed the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Sep 18, 2025

This was referenced Sep 18, 2025

stdarch subtree update #146709

Merged

[DRAFT] RISC-V: Use .insn on (Zkne or Zknd) intrinsics rust-lang/stdarch#1917

Closed

Uh oh!

Conversation

a4lg commented Sep 17, 2025

Uh oh!

rustbot commented Sep 17, 2025

Uh oh!

This comment has been minimized.

a4lg commented Sep 17, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

This comment has been minimized.

folkertdev commented Sep 17, 2025

Uh oh!

a4lg commented Sep 17, 2025

Uh oh!

a4lg commented Sep 17, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

This comment has been minimized.

This comment has been minimized.

rustbot commented Sep 17, 2025

Uh oh!

rustbot commented Sep 17, 2025

Uh oh!

a4lg commented Sep 17, 2025

Uh oh!

folkertdev commented Sep 17, 2025

Uh oh!

a4lg commented Sep 18, 2025

Uh oh!

Kobzol commented Sep 18, 2025

Uh oh!

a4lg commented Sep 18, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

a4lg commented Sep 18, 2025

Uh oh!

a4lg commented Sep 18, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

10 participants

a4lg commented Sep 17, 2025 •

edited

Loading

a4lg commented Sep 17, 2025 •

edited

Loading

a4lg commented Sep 18, 2025 •

edited

Loading