Re: general protection fault in lockdep_invariant_state (2)

[Peter Zijlstra]

On Wed, Dec 06, 2017 at 04:54:17PM +0100, Dmitry Vyukov wrote:
> On Wed, Nov 29, 2017 at 1:00 PM, Dmitry Vyukov <dvyukov@xxxxxxxxxx> wrote:
> > On Mon, Nov 20, 2017 at 2:28 PM, Dmitry Vyukov <dvyukov@xxxxxxxxxx> wrote:
> >>>> > R13: 0000000000402260 R14: 00000000004022f0 R15: 0000000000000000
> >>>> > kasan: CONFIG_KASAN_INLINE enabled
> >>>> > kasan: GPF could be caused by NULL-ptr deref or user memory access
> >>>> > general protection fault: 0000 [#1] SMP KASAN
> >>>> > Dumping ftrace buffer:
> >>>> >    (ftrace buffer empty)
> >>>> > Modules linked in:
> >>>> > CPU: 3 PID: 2982 Comm: syzkaller521009 Not tainted 4.14.0-rc7-next-20171103+
> >>>> > #10
> >>>> > Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
> >>>> > task: ffff88003d6b25c0 task.stack: ffff88003abd0000
> >>>> > RIP: 0010:invalidate_xhlock kernel/locking/lockdep.c:4719 [inline]
> >>>> > RIP: 0010:lockdep_invariant_state+0xd6/0x120 kernel/locking/lockdep.c:4793
> >>>
> >>>
> >>> Does this help?
> >>>
> >>> diff --git a/kernel/locking/lockdep.c b/kernel/locking/lockdep.c
> >>> index db933d063bfc..da2904418599 100644
> >>> --- a/kernel/locking/lockdep.c
> >>> +++ b/kernel/locking/lockdep.c
> >>> @@ -4793,7 +4793,8 @@ void lockdep_invariant_state(bool force)
> >>>          * Verify the former, enforce the latter.
> >>>          */
> >>>         WARN_ON_ONCE(!force && current->lockdep_depth);
> >>> -       invalidate_xhlock(&xhlock(current->xhlock_idx));
> >>> +       if (current->xhlocks)
> >>> +               invalidate_xhlock(&xhlock(current->xhlock_idx));
> >>>  }
> >>>
> >>>  static int cross_lock(struct lockdep_map *lock)
> >>
> >>
> >> Hi Peter,
> >>
> >> We are not able to test all fixes for all bugs, but we are rolling out
> >> a patch testing feature for syzbot:
> >>
> >> https://github.com/google/syzkaller/blob/master/docs/syzbot.md#communication-with-syzbot
> >>
> >> Please give it a try!
> >
> >
> > Peter, ping.
> > This still happens. Please either submit the fix if you already tested
> > it, or test using syzbot and submit.
> 
> 
> Peter, do you want me to mail your patch?

Urgh, sorry, got stuck doing KAISER muck. I'll go write up a proper
patch.