Re: general protection fault in propagate_entity_cfs_rq

[Willem de Bruijn]

On Thu, Nov 7, 2019 at 9:58 AM Willem de Bruijn
<willemdebruijn.kernel@xxxxxxxxx> wrote:
>
> On Thu, Nov 7, 2019 at 8:42 AM syzbot
> <syzbot+2e37f794f31be5667a88@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote:
> >
> > syzbot suspects this bug was fixed by commit:
> >
> > commit bab2c80e5a6c855657482eac9e97f5f3eedb509a
> > Author: Willem de Bruijn <willemb@xxxxxxxxxx>
> > Date:   Wed Jul 11 16:00:44 2018 +0000
> >
> >      nsh: set mac len based on inner packet
> >
> > bisection log:  https://syzkaller.appspot.com/x/bisect.txt?x=170cc89c600000
> > start commit:   6fd06660 Merge branch 'bpf-arm-jit-improvements'
> > git tree:       bpf-next
> > kernel config:  https://syzkaller.appspot.com/x/.config?x=a501a01deaf0fe9
> > dashboard link: https://syzkaller.appspot.com/bug?extid=2e37f794f31be5667a88
> > syz repro:      https://syzkaller.appspot.com/x/repro.syz?x=1014db94400000
> > C reproducer:   https://syzkaller.appspot.com/x/repro.c?x=11f81e78400000
> >
> > If the result looks correct, please mark the bug fixed by replying with:
> >
> > #syz fix: nsh: set mac len based on inner packet
>
> #syz fix: nsh: set mac len based on inner packet

The stack traces in both the bisection log and my manual run, when
running the linked reproducer, differ from the one in the dashboard.
Those more obviously include nsh functions. The trace in the dashboard
does not and sees a GPF in propagate_entity_cfs_rq, which does not
immediately appear related. That said, it is reported only once over a
year ago, so probably still preferable to close. A new report will be
opened if incorrect.