On 09/25/2018 04:31 PM, Mark Asselstine wrote:
On Tuesday, September 25, 2018 5:20:26 AM EDT Michal Privoznik
wrote:
> On 09/24/2018 05:11 PM, Mark Asselstine wrote:
Here is my read, again I have to caveat that this code is rather new to me.
The object poisoning in virObjectUnref() would prevent a double free as the
second call would fail VIR_OBJECT_NOTVALID(). However, you are still correct
as the mon object would potentially be destroyed and cleaned up before the
close callback is run and thus when the the close callback is eventually run
it would be operating on an old, no longer valid object.
So good catch on your part, thanks.
Yep, makes sense. I've ACKed and pushed this.
Congratulations on your first libvirt contribution. Usually it is just a
typo fix but you fixed a deadlock. Impressive.
Michal