Re: An example of a much more impactful way of doing file system-specific fuzzing

[Mark Rutland]

Hi Dmitry,

I hope the below doesn't sound like a criticism; I am a *huge* fan of
syzkaller.

On Tue, Jun 19, 2018 at 06:18:07PM +0200, 'Dmitry Vyukov' via syzkaller wrote:
> On Mon, Jun 18, 2018 at 7:15 AM, Theodore Y. Ts'o <tytso@xxxxxxx> wrote:
> 3. You say "not actionable reports without reproducers", but you can
> find hundreds of fixed bugs without reproducers at [1] and [2]. Fix
> ratio for bugs without reproducers is 66% which is not significantly
> lower than 76% for bugs with reproducers. A person without expertise
> in a particular subsystem (me) can't know if a bug is actionable by an
> expert in the subsystem (you) without first reporting this bug.

I think it might be possible to make this a bit easier, without any
manual effort per-bug.

For comparison, when the LKP kernel test robot reports a bug, it
provides a script to reproduce the issue in a VM, such that the
developer need only provide a kernel. The script launches the VM with
the right options, providing a filesystem if necessary, etc.

This is a little more actionale, since the developer need not expend any
effort trying to reproduce the correct envinronment, which can be
especially tricky for bugs that don't have a C reproducer.

Would it be possible for syzbot to do something similar?

Thanks,
Mark.
 
> [1] https://syzkaller.appspot.com/?fixed=upstream
> [2] https://github.com/google/syzkaller/blob/master/docs/linux/found_bugs.md