https://bugzilla.redhat.com/show_bug.cgi?id=892079
With current code, if user calls virDomainPMSuspendForDuration()
followed by virDomainDestroy(), the former API checks for qemu agent
presence, which will evaluate as true (if agent is configured). While
talking to qemu agent, the qemu driver is unlocked, so the latter API
starts executing. However, if machine dies meanwhile, libvirtd gets
EOF on the agent socket and qemuProcessHandleAgentEOF() is called. The
handler clears reference to qemu agent while the destroy API already
holding a reference to it. This leads to NULL dereferencing later in
the code. Therefore, the agent pointer should be set to NULL only if
we are the exclusive owner of it.
---
There's a reproducer in the BZ. It doesn't have to be a windows guest,
I was able to reproduce with F17 guest as well.
src/qemu/qemu_process.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/src/qemu/qemu_process.c b/src/qemu/qemu_process.c
index 938c17e..320c0c6 100644
--- a/src/qemu/qemu_process.c
+++ b/src/qemu/qemu_process.c
@@ -133,7 +133,8 @@ qemuProcessHandleAgentEOF(qemuAgentPtr agent,
virDomainObjLock(vm);
priv = vm->privateData;
- if (priv->agent == agent)
+ if (priv->agent == agent &&
+ !virObjectUnref(priv->agent))
priv->agent = NULL;
virDomainObjUnlock(vm);
--
1.8.0.2