svn commit: r310180 - head/sys/net

[Alan Somers]

On Fri, Dec 16, 2016 at 3:45 PM, Mateusz Guzik <mjguzik at gmail.com> wrote:
> On Fri, Dec 16, 2016 at 10:39:31PM +0000, Alan Somers wrote:
>> Author: asomers
>> Date: Fri Dec 16 22:39:30 2016
>> New Revision: 310180
>> URL: https://svnweb.freebsd.org/changeset/base/310180
>>
>> Log:
>>   Fix panic during lagg destruction with simultaneous status check
>>
>>   If you run "ifconfig lagg0 destroy" and "ifconfig lagg0" at the same time a
>>   page fault may result. The first process will destroy ifp->if_lagg in
>>   lagg_clone_destroy (called by if_clone_destroy). Then the second process
>>   will observe that ifp->if_lagg is NULL at the top of lagg_port_ioctl and
>>   goto fallback: where it will promptly dereference ifp->if_lagg anyway.
>>
>>   The solution is to repeat the NULL check for ifp->if_lagg
>>
>
> I don't understand how this solves the problem. What prevents the object
> from getting freed after the pointer got NULLified? That is, it seems
> the patch turns a null pointer deref into a use-after-free.
>
> There seems to be a refcounting issue here.
>
> That said, I only did cursory reading.

What object are you talking about?  The problem solved by this commit
is the null-pointer dereference in the thread calling lagg_port_ioctl,
for example "ifconfig lagg0".