Re: [net, neigh] 7482e3841d: BUG:unable_to_handle_page_fault_for_address

[Philip Li <philip.li@intel.com>]

[-- Attachment #1: Type: text/plain, Size: 2596 bytes --]

On Thu, Nov 18, 2021 at 09:47:51AM +0100, Daniel Borkmann wrote:
> Hi Oliver,
> 
> On 11/18/21 7:23 AM, Oliver Sang wrote:
> > On Fri, Nov 05, 2021 at 12:21:20PM +0100, Daniel Borkmann wrote:
> > > Hey Oliver,
> > > 
> > > > > To reproduce:
> > > > > 
> > > > >           # build kernel
> > > > >      cd linux
> > > > >      cp config-5.15.0-rc4-01167-g7482e3841d52 .config
> > > > >      make HOSTCC=gcc-9 CC=gcc-9 ARCH=i386 olddefconfig prepare modules_prepare bzImage
> > > > > 
> > > > >           git clone https://github.com/intel/lkp-tests.git
> > > > >           cd lkp-tests
> > > > >           bin/lkp qemu -k <bzImage> job-script # job-script is attached in this email
> > > > > 
> > > > >           # if come across any failure that blocks the test,
> > > > >           # please remove ~/.lkp and /lkp dir to run from a clean state.
> > > 
> > > I ran the test multiple times, but it seems I cannot reproduce. Would be good if you could
> > > double check below whether it looks / passes as expected or if I'm doing something wrong:
> > 
> > sorry about this, we tried above reproducer locally but failed to reproduce
> > either. we are under investigation of this issue.
> 
> Ok, thanks for looking into it!
> 
> > at the same time, we double confirmed by queuing more jobs to our auto-test
> > framework, it still reproduced in a persistent way. Just FYI
> > 2c611ad97a82b512 7482e3841d520a368426ac19672
> > ---------------- ---------------------------
> >         fail:runs  %reproduction    fail:runs
> >             |             |             |
> >             :59          98%          58:58    dmesg.BUG:unable_to_handle_page_fault_for_address
> >             :59          98%          58:58    dmesg.Kernel_panic-not_syncing:Fatal_exception_in_interrupt
> >             :59          98%          58:58    dmesg.Oops:#[##]
> > 
> 
> Is there any way for lkp to test development branches for getting more debug info out of it?
> Something like syzkaller has with 'syz test' where it mails the results back [0]?
thanks for the suggestion, Daniel. So far, we don't support this yet, but this is
very helpful to provide something similar to syzkaller does. We will consider this
seriously as TODO.

> 
> Thanks a lot,
> Daniel
> 
>   [0] https://github.com/google/syzkaller/blob/master/docs/syzbot.md#testing-patches
> _______________________________________________
> LKP mailing list -- lkp(a)lists.01.org
> To unsubscribe send an email to lkp-leave(a)lists.01.org