Fix wrong directories in host_libdir.

[jnbr]

This fixes a regression from #7482 where the sysroot_target_libdir leaks into the host libdir. This can cause problems when the dynamic linker does not ignore the target libraries but tries to load them instead. This happens for example when building on x86_64-musl for aarch64-musl.

[rust-highfive]

Thanks for the pull request, and welcome! The Rust team is excited to review your changes, and you should hear from @Eh2406 (or someone else) soon.

If any changes to this PR are deemed necessary, please add them as extra commits. This ensures that the reviewer can see what has changed since they last reviewed the code. Due to the way GitHub handles out-of-date commits, this should also make it reasonably obvious what issues have or haven't been addressed. Large or tricky changes may require several passes of review and changes.

Please see the contribution instructions for more information.

[ehuss]

Hi! Thanks for the PR!

I'm a bit confused because I expect sysroot_host_libdir is the same no matter what the target kind is. Can you explain more about the issue you are having?

[jnbr]

Can you explain more about the issue you are having?

Sure. Here is what I try to do:

To cross compile a hello world on a x86_64-musl host system for aarch64-musl, I run the folowing commands.

export RUSTFLAGS="--sysroot=../aarch64-musl-chroot/usr/"
export CARGO_TARGET_AARCH64_UNKNOWN_LINUX_MUSL_LINKER=aarch64-linux-musl-gcc
~/src/cargo/target/debug/cargo build --target=aarch64-unknown-linux-musl

This used to work, but since #7482, rustc (called by cargo) fails to run, the dynamic linker shows this error message:

Error relocating ../aarch64-musl-chroot/usr/lib/libgcc_s.so.1: unsupported relocation type 1026

Setting a break point in Compilation::new and inspecting sysroot_host_libdir, shows that it contains a target libdir, strictly speaking, the one set via RUSTFLAGS="--sysroot=../aarch64-musl-chroot/usr/":

(gdb) p bcx.info(default_kind).sysroot_host_libdir
$1 = std::path::PathBuf {inner: "../aarch64-musl-chroot/usr/lib"}

So the actual problems is either me using --sysroot in the wrong way, or – in my opinion – that it is included in sysroot_host_dir.

[ehuss]

Ah, I see, using --sysroot is a critical part here.

This change seems fine to me. I'm wondering if a test might be possible, but it seems difficult. I'll leave final review to @alexcrichton who wrote this.

[alexcrichton]

@bors: r+

[bors]

📌 Commit f917e16 has been approved by alexcrichton

[bors]

⌛ Testing commit f917e16 with merge f6449ba...

[bors]

☀️ Test successful - checks-azure
Approved by: alexcrichton
Pushing f6449ba to master...