[PATCH v1] fix build failures due to incorrect format specifier for uint64_t

[Matthieu Longo]

I rebuilt with CC="gcc -m32" CXX="g++ -m32" as advised by Mark, and ran the tests, but got a segfault in ba-aarch64-1-unknown-tag.d

Looking at it closer, the code in _bfd_aarch64_oav2_attr_merge seems to have some formatting issue as well.
I tried to fix it as follows:

info->callbacks->einfo
   (_("%pB: warning: cannot merge unknown tag 'Tag_unknown_%" PRIu64 "'"
      " (=0x%" PRIx32 ") in subsection '%s'.\n"),
    abfd, rhs->tag, rhs->val.uint, subsec->name);

No segfault, but the formatting does not work.
   warning: cannot merge unknown tag 'Tag_unknown_%llu' (=0x4) in subsection '(null)'

It seems to me that this is due to vfinfo() in ldmisc.c not supporting %llu.
The formatting fix does not really make sense without the support of %llu inside vfinfo() even if it does not segfault anymore.
Given that you have changed recently the code of vfinfo to extend the accepted formats, do you think that it is something feasible in the short time we have before branching ?

Matthieu