error: unable to run `rust-objcopy`: Bad CPU type in executable (os error 86)

[Wyvern]

Host: x86_64-apple-darwin
rustc 1.91.0-nightly (040a98af7 2025-08-20)

Compile with error:
error: unable to run rust-objcopy: Bad CPU type in executable (os error 86)

full output:

error: unable to run `rust-objcopy`: Bad CPU type in executable (os error 86)

error: could not compile `proc-macro2` (build script) due to 1 previous error
warning: build failed, waiting for other jobs to finish...
error: could not compile `libc` (build script) due to 1 previous error

[Wyvern]

on x86_64-apple-darwin host run:

> file .rustup/toolchains/nightly-x86_64-apple-darwin/lib/rustlib/x86_64-apple-darwin/bin/rust-objcopy
rust-objcopy: Mach-O 64-bit executable arm64

looks like rust-objcopy compiled to tier1 arm64 apple target rather than host x86_64.

The problem maybe related to demote x86_64-apple-darwin to tier2.

[Gelbpunkt]

This is a problem on all tier 2 targets. The rust-objcopy executable is linked and compiled for the host that compiled it rather than the target. See also #143415 (comment)

[Wyvern]

interesting, only objcopy got a wrong target compiling.

> file .rustup/toolchains/nightly-x86_64-apple-darwin/lib/rustlib/x86_64-apple-darwin/bin/*
rust-lld:          Mach-O 64-bit executable x86_64
rust-objcopy:      Mach-O 64-bit executable arm64
wasm-component-ld: Mach-O 64-bit executable x86_64

[benalleng]

I can confirm I see the same problem only with rust-objcopy.

file $HOME/.rustup/toolchains/nightly-x86_64-apple-darwin/lib/rustlib/x86_64-apple-darwin/bin/*
/Users/runner/.rustup/toolchains/nightly-x86_64-apple-darwin/lib/rustlib/x86_64-apple-darwin/bin/rust-lld:          Mach-O 64-bit executable x86_64
/Users/runner/.rustup/toolchains/nightly-x86_64-apple-darwin/lib/rustlib/x86_64-apple-darwin/bin/rust-objcopy:      Mach-O 64-bit executable arm64
/Users/runner/.rustup/toolchains/nightly-x86_64-apple-darwin/lib/rustlib/x86_64-apple-darwin/bin/wasm-component-ld: Mach-O 64-bit executable x86_64

[lqd]

Could anyone bisect this issue with https://github.com/rust-lang/cargo-bisect-rustc/?

I'll also cc @shepmaster in case it's related to demoting the target, but without the bisection it could be any other CI change on the builder or a bootstrap issue, so I'll also cc @Kobzol.

BTW, a possible temporary workaround would be setting strip = false in your Cargo.toml, that should remove the need to run rust-objcopy in the first place.

[shepmaster]

bisect this issue

If you bisect this specific issue of rust-objcopy being aarch64 instead of x86_64, I'd expect it to come back to the demotion PR. However, if this is occurring for any cross-compilation (as mentioned above), then you'd have to bisect any (every?) cross-compiled toolchain for a rust-objcopy binary being the wrong architecture.

For example, this shows rustc-objcopy is x86_64 when it should be aarch64 for the aarch64-unknown-linux-musl target:

cd $HOME/.rustup/toolchains/nightly-aarch64-unknown-linux-musl/lib/rustlib/*/bin ; file *

gcc-ld:            directory
rust-lld:          ELF 64-bit LSB executable, ARM aarch64, version 1 (SYSV), dynamically linked, interpreter /lib/ld-musl-aarch64.so.1, with debug_info, not stripped
rust-objcopy:      ELF 64-bit LSB pie executable, x86-64, version 1 (GNU/Linux), dynamically linked, interpreter /lib64/ld-linux-x86-64.so.2, BuildID[sha1]=486eb10c870e12cbb52f742c3322a19cc027b776, for GNU/Linux 3.2.0, not stripped
wasm-component-ld: ELF 64-bit LSB pie executable, ARM aarch64, version 1 (SYSV), dynamically linked, interpreter /lib/ld-musl-aarch64.so.1, stripped

[thaliaarchi]

The regression appears to have happened yesterday:

$ file "$(find ~/.rustup -name rust-objcopy)"
~/.rustup/toolchains/1.85-x86_64-apple-darwin/lib/rustlib/x86_64-apple-darwin/bin/rust-objcopy:               Mach-O 64-bit executable x86_64
~/.rustup/toolchains/1.86-x86_64-apple-darwin/lib/rustlib/x86_64-apple-darwin/bin/rust-objcopy:               Mach-O 64-bit executable x86_64
~/.rustup/toolchains/stable-x86_64-apple-darwin/lib/rustlib/x86_64-apple-darwin/bin/rust-objcopy:             Mach-O 64-bit executable x86_64
~/.rustup/toolchains/beta-x86_64-apple-darwin/lib/rustlib/x86_64-apple-darwin/bin/rust-objcopy:               Mach-O 64-bit executable x86_64
~/.rustup/toolchains/nightly-2025-02-15-x86_64-apple-darwin/lib/rustlib/x86_64-apple-darwin/bin/rust-objcopy: Mach-O 64-bit executable x86_64
~/.rustup/toolchains/nightly-2025-03-20-x86_64-apple-darwin/lib/rustlib/x86_64-apple-darwin/bin/rust-objcopy: Mach-O 64-bit executable x86_64
~/.rustup/toolchains/nightly-2025-03-30-x86_64-apple-darwin/lib/rustlib/x86_64-apple-darwin/bin/rust-objcopy: Mach-O 64-bit executable x86_64
~/.rustup/toolchains/nightly-2025-04-22-x86_64-apple-darwin/lib/rustlib/x86_64-apple-darwin/bin/rust-objcopy: Mach-O 64-bit executable x86_64
~/.rustup/toolchains/nightly-2025-08-20-x86_64-apple-darwin/lib/rustlib/x86_64-apple-darwin/bin/rust-objcopy: Mach-O 64-bit executable x86_64
~/.rustup/toolchains/nightly-2025-08-21-x86_64-apple-darwin/lib/rustlib/x86_64-apple-darwin/bin/rust-objcopy: Mach-O 64-bit executable arm64

[shepmaster]

The regression appears to have happened yesterday

Yes, as stated above, this will look like it occurred due to the Tier 2 demotion in the last few days, but the root problem predates this as it affects cross compiled builds in general.