Re: [PATCH 0/2] mm: Two small fixes for recent syzbot reports

[Andrew Morton]

On Thu, 9 Apr 2020 15:00:20 +0200 Dmitry Vyukov <dvyukov@xxxxxxxxxx> wrote:

> On Thu, Apr 9, 2020 at 1:49 PM Matthew Wilcox <willy@xxxxxxxxxxxxx> wrote:
> >
> > On Wed, Apr 08, 2020 at 05:47:32PM -0700, Andrew Morton wrote:
> > > On Tue,  7 Apr 2020 21:40:08 -0400 Peter Xu <peterx@xxxxxxxxxx> wrote:
> > >
> > > > The two patches should fix below syzbot reports:
> > > >
> > > >   BUG: unable to handle kernel paging request in kernel_get_mempolicy
> > > >   https://lore.kernel.org/lkml/0000000000002b25f105a2a3434d@xxxxxxxxxx/
> > > >
> > > >   WARNING: bad unlock balance in __get_user_pages_remote
> > > >   https://lore.kernel.org/lkml/00000000000005c65d05a2b90e70@xxxxxxxxxx/
> > >
> > > (Is there an email address for the syzbot operators?)
> >
> > I'd suggest syzkaller-bugs@xxxxxxxxxxxxxxxx (added to the Cc).
> 
> syzkaller@xxxxxxxxxxxxxxxx is a better one.
> syzkaller-bugs@xxxxxxxxxxxxxxxx plays more of an LKML role.
> 
> > But there's a deeper problem in that we don't have anywhere to stash
> > that kind of information in the kernel tree right now.  Perhaps a special
> > entry in the MAINTAINERS file for bot operators?  Or one entry per bot?
> 
> I don't mind adding syzkaller. Some time ago I wanted to contact
> KernelCI, CKI, LKFT, 0-day owners, finding relevant lists wasn't
> impossible, but for some it was hard.
> 
> For syzkaller it would be:
> 
> https://github.com/google/syzkaller/issues for bugs/feature requests.
> syzkaller@xxxxxxxxxxxxxxxx for discussions.

OK, thanks.  A MAINTAINERS entry would be great.

Could I please direct attention back to my original question regarding
the problems we've recently discovered in 4426e945df58 ("mm/gup: allow
VM_FAULT_RETRY for multiple times") and 71335f37c5e8 ("mm/gup: allow to
react to fatal signals")?

> sysbot does test linux-next, yet these patches sat in linux-next for a
> month without a peep, but all hell broke loose when they hit Linus's
> tree.  How could this have happened?
> 
> Possibly I've been carrying a later patch which fixed all this up, but
> I'm not seeing anything like that.  Nothing at all against mm/gup.c.