ld.lld: error: unknown argument '--fix-cortex-a53-835769' when trying to compile rust on aarch64

[bschnei]

I'm encountering the error in the title when trying to compile rust with lld 20.1.8 as packaged by Arch on aarch64. I'm using gcc 15.2.1 compiled with --enable-fix-cortex-a53-835769 and --enable-fix-cortex-a53-843419 options.

The --fix-cortex-a53-835769 flag seems to be getting injected at some point in rust's build process and there doesn't appear to be a straightforward way to override it nor the usage of lld instead of another linker that supports this flag.

Per https://lld.llvm.org/, lld is meant to be a drop-in replacement for GNU ld. Also it appears that the work to try to support the fix was actually merged quite some time ago. But perhaps the option was simply never added to lld?

Let me know if there is any other information I can provide. Thanks!

[MaskRay]

I think your compiler passes --fix-cortex-a53-835769 to lld, a workaround for some early versions of Cortex A53. https://sourceware.org/pipermail/binutils/2014-October/086323.html The compile also implements the workaround. The linker option is probably to provide a little more coverage but it is 2025 and I am not sure there is any value in implementing this ancient workaround in the linker... In addition, the option probably scans the whole text sections and causes slowdown for all aarch64 links.

[bschnei]

Hi @MaskRay, thank you for considering this!

While I still think it's fair to call this an "issue" given the way lld is advertised, I can understand not wanting to fix and am fine with you closing/tagging it as such. At least it will be clear to others who may come across the same problem that they need to figure out a workaround or propose a solution.

With respect to building rust specifically, the workaround is to avoid using lld when boostrapping rust. This is actually the default value as seen here.

[MaskRay]

You might want to check whether fix-cortex-a53-835769 is a cargo cult and should be eliminated...

Per lld.llvm.org, lld is meant to be a drop-in replacement for GNU ld.

While I still think it's fair to call this an "issue" given the way lld is advertised, [...]

https://maskray.me/blog/2020-12-19-lld-and-gnu-linker-incompatibilities

Subtitle: Is ld.lld a drop-in replacement for GNU ld?

The motivation for this article was someone challenging the "drop-in replacement" claim on LLD's website (the discussion was about Linux-like ELF toolchain):

LLD is a linker from the LLVM project that is a drop-in replacement for system linkers and runs much faster than them. It also provides features that are useful for toolchain developers.

99.9% pieces of software work with ld.lld without a change. Some linker script applications may need an adaption (such adaption is oftentimes due to brittle assumptions: asking too much from GNU ld's behavior which should be fixed anyway). So I defended for this claim.