Re: [Qemu-devel] [PATCH v2] async: aio_context_new(): Handle event_notifier_init failure

[Chrysostomos Nanakos <cnanakos@grnet.gr>]

On Tue, Sep 16, 2014 at 06:50:38PM +0300, Chrysostomos Nanakos wrote:
> On Tue, Sep 16, 2014 at 05:45:16PM +0200, Paolo Bonzini wrote:
> > Il 16/09/2014 17:43, Chrysostomos Nanakos ha scritto:
> > >> >     error_propagate(errp, local_error);
> > >> >     return;
> > > Just to note that after propagating the error and returning, QEMU fails
> > > silently without printing the error message.
> > 
> > What is your testcase?
> 
> I am starting QEMU with the options below and explicitly set iothread->ctx to
> NULL. Is that ok as a testcase or should I reduce my open files limit to
> produce the error?
> 
> qemu --enable-kvm -smp 2 -m 1024 -object iothread,id=iothread0 -drive file=archipelago:fedora_stable
> -vnc 0.0.0.0:0 -qmp tcp:127.0.0.1:8888,server,nowait
> 
> No error message, at least the propagated one, fails silently.
> 
The segfault is caused when QOM tries to object_unref() the iothread object,
type->instance_finalize(obj) is called which calls the registered
iothread_instance_finalize function. A check there for the iothread->ctx
seems to solve the problem.

I will include the fix in the next patch series.

Regards,
Chrysostomos.