On 12/20/2013 02:36 PM, Jiri Denemark wrote:
Generally, every API that is going to begin a job should do that
before
fetching data from vm->def. However, qemuDomainGetBlockInfo does not
know whether it will have to start a job or not before checking vm->def.
To avoid using disk alias that might have been freed while we were
waiting for a job, we use its copy. In case the disk was removed in the
meantime, we will fail with "cannot find statistics for device '...'"
error message.
Signed-off-by: Jiri Denemark <jdenemar(a)redhat.com>
---
src/qemu/qemu_driver.c | 17 ++++++++++++-----
1 file changed, 12 insertions(+), 5 deletions(-)
This fixes the potential for a crash, but isn't this related to the same
problem that John has been working on? That is, when one thread is
migrating a domain and the other is getting block info, we have a race
where qemu will quit before we get a chance to probe for the block info.
https://www.redhat.com/archives/libvir-list/2013-December/msg00984.html
Would changing when we grab the job (and grabbing it unconditionally,
even if we end up not needing to query) help solve both issues?
--
Eric Blake eblake redhat com +1-919-301-3266
Libvirt virtualization library
http://libvirt.org