Re: WARNING: ODEBUG bug in process_one_req

[Leon Romanovsky <leonro@mellanox.com>]

[-- Attachment #1: Type: text/plain, Size: 2755 bytes --]

On Thu, Mar 22, 2018 at 05:13:58PM +0200, Leon Romanovsky wrote:
> On Thu, Mar 22, 2018 at 03:47:41PM +0100, Dmitry Vyukov wrote:
> > On Thu, Mar 22, 2018 at 3:29 PM, Leon Romanovsky <leonro@mellanox.com> wrote:
> > > On Thu, Mar 22, 2018 at 02:36:55PM +0100, Dmitry Vyukov wrote:
> > >> On Thu, Mar 22, 2018 at 2:20 PM, Leon Romanovsky <leonro@mellanox.com> wrote:
> > >> > On Thu, Mar 22, 2018 at 02:10:21PM +0100, Dmitry Vyukov wrote:
> > >> >> This bug is actively happening several times per day, if such bug is
> > >> >> closed syzbot will open another bug on the next crash.
> > >> >> This bug has a reproducer, why do you think it is invalid?
> > >> >
> > >> > I tried to reproduce it on latest rdma-next and failed, so wanted to
> > >> > start from clean page and see if it is still relevant.
> > >>
> > >> Ah,  I see.
> > >> You can now see the current status on dashboard:
> > >> https://syzkaller.appspot.com/bug?id=d4ac7bfeafac8a3d6d06123e078462ac765415e7
> > >> Yes, it still happens.
> > >>
> > >> syzbot has already reported another incarnation of this bug, you can
> > >> see the link to dashboard in the email:
> > >> https://syzkaller.appspot.com/bug?extid=b8d5eb964e412cfd2678
> > >> and it contains a link to the first version with reproducer.
> > >>
> > >> > Does it still reproduce on latest rdma-next?
> > >>
> > >> syzbot does not test rdma-next.
> > >>
> > >> But can you reproduce it on upstream tree where syzbot hits it? Just
> > >> to make sure, you have ODEBUG enabled, right?
> > >
> > > Thanks for the tip, I added more debug options and returned to try the
> > > repro on clean v4.16-rc1, it looks like I still don't have all needed
> > > config options (SELinux).
> > >
> > >   mount(selinuxfs) failed (errno 2)
> >
> >
> > It seems that you have old syzkaller sources, it now explicitly does
> > not fail on ENOENT:
> >
> >     // selinux mount used to be at /selinux, but then moved to /sys/fs/selinux.
> >     const char* selinux_path = "./syz-tmp/newroot/selinux";
> >     if (mount("/selinux", selinux_path, NULL, mount_flags, NULL)) {
> >         if (errno != ENOENT)
> >             fail("mount(/selinux) failed");
> >         if (mount("/sys/fs/selinux", selinux_path, NULL, mount_flags,
> > NULL) && errno != ENOENT)
> >             fail("mount(/sys/fs/selinux) failed");
> >     }
> >
> >
> > > Can you please cancel my email that closes this bug?
> >
> > There is no such option. The new bug is already created:
> > https://syzkaller.appspot.com/bug?id=6ecd3fdba0501c8843300056375abea0880816d2
> > now if we open the old one, we now have 2 for the same root cause. It
> > will lead to mess.
>
> Still no luck, I'll leave this bug for now.

So, except this bug, everything else is fixed.

Thanks

>
> Thanks



[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]