[libvirt] [PATCH] Revert "tests: Don't link mock libraries against libvirt and gnulib"

Andrea Bolognani abologna at redhat.com
Thu Feb 11 08:54:45 UTC 2016

On Wed, 2016-02-10 at 18:09 +0000, Daniel P. Berrange wrote:
> On Wed, Feb 10, 2016 at 06:33:21PM +0100, Andrea Bolognani wrote:
> > This reverts commit 6aa90452aa63cb1e1ffa84ff5f93f5873bf810a0.
>> > Turns out that not linking against libvirt and gnulib is okay for
> > regular Linux (and FreeBSD) builds, but makes mingw very unhappy.
>> >   .../virnetserverclientmock_la-virnetserverclientmock.o:
> >     In function `virNetSocketGetSELinuxContext':
> >     .../virnetserverclientmock.c:61: undefined reference to `rpl_strdup'
> >   .../libvirportallocatormock_la-virportallocatortest.o:
> >     In function `init_syms':
> >     .../virportallocatortest.c:61: undefined reference to `virFileClose'
> > ---
> > Pushed as build breaker.
>> > Bunny ears of shame will be deployed in the morning.
> You should still be able to drop linkage of libvirt.la, just keep
> the gnulib bits.  The gnulib addition is harmless, because it is
> a static library the linker will drop all functions which are not
> actually used by the mock.o, so it will end up being a no-op on
> Linux, and will only pull in rpl_strup on mingw32. We should never
> link libvirt.la to mock libs though.

Looks like it is complaining about virFileClose(), plus
virFilePrint() in the full output, as well. So maybe linking against
libvirt.la is needed after all?

If there's no harm in that, how about defining

    $(GNULIB_LIBS) \

and using it for every mock library, like the existing


Andrea Bolognani
Software Engineer - Virtualization Team

More information about the libvir-list mailing list