[ovs-dev] [PATCH 2/3] vswitchd: Fix segmentation fault with bonded ports.

Ethan Jackson ethan at nicira.com
Fri Jan 7 01:02:15 UTC 2011


Good point.  I sent out a revised patch.

Ethan

On Thu, Jan 6, 2011 at 6:45 PM, Ben Pfaff <blp at nicira.com> wrote:
> We can add or remove ifaces on a port, are the other cases for that
> doing the right thing?  Should iface_destroy() be taking care of
> removing itself from the monitor?
>
> On Thu, Jan 06, 2011 at 06:41:03PM -0600, Ethan Jackson wrote:
>> In non-bonded ports the port->monitor is NULL so
>> netdev_monitor_destroy has no effect.  In the bonded case
>> port->monitor is non-NULL.  Destroying all of the interfaces before
>> removing the monitor was causing the monitor to be in an inconsistent
>> state which was causing it's destruction to segfault.
>>
>> Ethan
>>
>> On Thu, Jan 6, 2011 at 6:37 PM, Ben Pfaff <blp at nicira.com> wrote:
>> > On Thu, Jan 06, 2011 at 04:23:48PM -0800, Ethan Jackson wrote:
>> >> This commit fixes a segmentation fault which could occur when a
>> >> bonded port was destroyed.
>> >
>> > It looks like a harmless change but the faulting code path isn't obvious
>> > to me.
>> >
>




More information about the dev mailing list