https://sourceware.org/bugzilla/show_bug.cgi?id=33470
Sam James <sam at gentoo dot org> changed:
What |Removed |Added
----------------------------------------------------------------------------
See Also| |https://sourceware.org/bugz
| |illa/show_bug.cgi?id=28893
CC| |sam at gentoo dot org
--- Comment #1 from Sam James <sam at gentoo dot org> ---
(In reply to Alan Modra from comment #0)
> [...]
> That's the wrong plugin for clang. For my Ubuntu 25.04 system apparently
> the correct option is --plugin /usr/lib/llvm-20/lib/LLVMgold.so. Just how
> we are supposed to figure that out is not clear to me, particularly since
> clang --print-file-name liblto_plugin.so
> /usr/lib/gcc/x86_64-linux-gnu/14/liblto_plugin.so
>
In Gentoo, what we do is:
/usr/x86_64-pc-linux-gnu/binutils-bin/lib/bfd-plugins/LLVMgold.so ->
../../../../lib/llvm/21/lib64/LLVMgold.so
But the configure test indeed just does nothing for me because:
$ clang --print-prog-name liblto_plugin.so
liblto_plugin.so
$ clang --print-file-name liblto_plugin.so
liblto_plugin.so
and then it never passes --plugin, but Clang does that for its own invocations
(because it knows where LLVMgold.so) is. What isn't clear is how ranlib and ar
work for me?
>
> Also another bug, perhaps relevant. The system binutils-2.44 results in
> /usr/bin/nm bfd/bfd.o
> [...]
> At a guess, this is misbehaviour of LLVMgold-16.so, or perhaps all of the
> LLVMgold*.so since the error comes from whichever one is opened last (ie. is
> last in the directory) unless that happens to be LLVMgold-20.so.
.. and we only ever install one version of the plugin.
--
You are receiving this mail because:
You are on the CC list for the bug.
