Fix build error on sparc64 caused by using the gold linker

[gibmat]

When building on a sparc64 machine running Debian sid with gcc 12.2.0, linking fails with the following error:

collect2: fatal error: ld terminated with signal 10 [Bus error] (build log)

It appears to be caused by using the gold linker, although I'm not exactly sure why. None of the other Linux-based Debian architectures exhibit this issue with the gold linker.

I made the proposed change and tested it on one of the GCC Compile Farm sparc64 machines (gcc202), which was then able to successfully compile lxc.

[lxc-jenkins]

This pull request didn't trigger Jenkins as its author isn't in the allow list.

An organization member must perform one of the following:

• To have this branch tested by Jenkins, use the "ok to test" command.
• To have a one time test done, use the "test this please" command.

Those commands are simple Github comments of the format: "jenkins: COMMAND"

[gibmat]

@stgraber I'm not sure where your comment went, as I got the email notification but when I click the link to reply it doesn't show up. So, I'll quote it here and reply below:

@gibmat so your change is actually breaking our own builds now:
https://launchpadlibrarian.net/645662600/buildlog_snap_ubuntu_focal_ppc64el_lxd-latest-edge_BUILDING.txt.gz

Any idea of what to do to make everyone happy?

The only thing I can think of is that the order of the flags is now slightly different:

-Wl,-z,now -Wl,-fuse-ld=gold -fstack-protector -fstack-protector-strong -Wl,--warn-common

vs

-Wl,-z,now -fstack-protector -fstack-protector-strong -Wl,-fuse-ld=gold -Wl,--warn-common

Other than that there shouldn't be any difference for any architectures other than sparc64. Locally I also did an amd64 build with those changes prior to submitting the pull request, but didn't test other architectures.

I'm not very familiar with meson; is there some sort of list.remove(item) that might be used to remove the gold linker flag conditionally just for sparc64?

[stgraber]

I deleted the comment almost immediately because we were dealing with ppc64el and not sparc64 so couldn't have been your change. In the end, it was a linker bug. A simple rebuild got us past it.

[gibmat]

😌 Phew -- not my fault! 😄