[adamgemmell]

ICX was bumped to 2026.2 on this image which broke the path lookup. The "latest" symlink exists in this package to mitigate this.

r? @folkertdev

cc: @Jamesbarford

[folkertdev]

https://triage.rust-lang.org/gha-logs/rust-lang/stdarch/119751826143

The is_powerpc64_feature_detected macro was merged into is_powerpc_feature_detected. Resolving that needs a sync which is blocked on ppcf128 giving us some trouble. So, maybe we should just make that ppc macro change here to unblock things?

[folkertdev]

ah, we really need a sync first. But that will fail because of the icx thing. Fun.

[adamgemmell]

riscvgc-unknown-linux-gnu fails with

error[E0425]: cannot find value SYS_futex in crate libc
--> /rust/lib/rustlib/src/rust/library/std/src/sys/sync/futex/unix.rs:64:31
|
64 | libc::SYS_futex,
| ^^^^^^^^^ not found in libc
|

I think the upstream riscv build is just broken. rust-lang/rust#163732 bumps libc to 0.2.191, which removes SYS_futex, but std still references it . The build-std step fails, nothing a bump can do I think

ah, we really need a sync first. But that will fail because of the icx thing. Fun.

I'm going to stick the icx fix on the sync branch and see if that helps

[folkertdev]

oh, right, this was already reported as rust-lang/rust#164694

[folkertdev]

I'm going to stick the icx fix on the sync branch and see if that helps

Not really I think, ppcf128 is the real issue there. Due to how this crate works, the lang item ends up being defined twice. I'm trying to fix that in rust-lang/rust#164692

[folkertdev]

Maybe just make that riscv32 job non-blocking then?