[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [libvirt] [PATCH 06/10] Qemu Monitor API entry point.



On Tue, Apr 27, 2010 at 05:36:40PM -0400, Chris Lalancette wrote:
> On 04/27/2010 04:25 PM, Daniel P. Berrange wrote:
> >> Causes the build to fail when linking virsh:
> >>
> >>   CCLD   virsh
> >> ../src/.libs/libvirt-qemu.so: undefined reference to `qemuMonitorCommand'
> >> collect2: ld returned 1 exit status
> >>
> >> Indeed, if I nm src/.libs/libvirt-qemu.so:
> >>
> >> [clalance localhost libvirt (qemu_monitor)]$ nm src/.libs/libvirt-qemu.so | grep Monitor
> >>          U qemuMonitorCommand
> >>          U remoteQemuMonitorCommand@@LIBVIRT_PRIVATE_0.8.0
> >> 000010b0 T virDomainQemuMonitorCommand
> >>
> >> Any suggestions about what I'm doing wrong?
> > 
> > Have you re-ordered your patches ? There is a qemuMonitorCommand  defined
> > in src/qemu/qemu_monitor_text.c but that is static. The qemuMonitorCommand
> > you actually want is not added till your next patch. So perhaps your old
> > code was pullin in the wrong definition ?
> > 
> > It'd be good to rename this to reflect the public API name and avoid the
> > clash
> 
> I did not re-order my patches, but you are right, there is a name clash.
> I've now resolved that as you suggested (calling the external function
> qemuDomainMonitorCommand).
> 
> That's not actually the problem, though.  What's happening is that
> src/Makefile.am has this:
> 
> if WITH_QEMU
> if WITH_DRIVER_MODULES
> mod_LTLIBRARIES += libvirt_driver_qemu.la
> else
> noinst_LTLIBRARIES += libvirt_driver_qemu.la
> # Stateful, so linked to daemon instead
> #libvirt_la_LIBADD += libvirt_driver_qemu.la
> 
> i.e. libvirt_driver_qemu.la is not being added to libvirt.la, so that's
> why my linking step in virsh is failing (and *not* failing when linking
> libvirtd).  So I think I need to manually specify libvirt_qemu.la in
> my libvirt_qemu_la_LIBADD to avoid this failure.

Ahh, damn.  This is a result of my suggestion to you not to use the
virDriver struct for dispatching these methods :-( It means that the
main libvirt-qemu.c file gets a direct dependancy on the QEMU driver
code, instead of indirectly via the driver table. So I'm afraid I
was wrong & you should go back to using the driver table for dispatch

Regards,
Daniel
-- 
|: Red Hat, Engineering, London    -o-   http://people.redhat.com/berrange/ :|
|: http://libvirt.org -o- http://virt-manager.org -o- http://deltacloud.org :|
|: http://autobuild.org        -o-         http://search.cpan.org/~danberr/ :|
|: GnuPG: 7D3B9505  -o-   F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]