[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CANp29Y5Bq1fo8pFWGxKT9NTNW_-C6vf=A_hUKG2BJey5JRrO1g@mail.gmail.com>
Date: Fri, 1 Sep 2023 13:07:20 +0200
From: Aleksandr Nogikh <nogikh@...gle.com>
To: Andy Shevchenko <andriy.shevchenko@...ux.intel.com>
Cc: syzbot <syzbot+d2c31705e468a347e6db@...kaller.appspotmail.com>,
gregkh@...uxfoundation.org, linux-kernel@...r.kernel.org,
rafael@...nel.org, syzkaller-bugs@...glegroups.com
Subject: Re: [syzbot] [kernel?] general protection fault in netdev_register_kobject
On Thu, Aug 31, 2023 at 6:33 PM Andy Shevchenko
<andriy.shevchenko@...ux.intel.com> wrote:
>
> On Thu, Aug 31, 2023 at 05:53:29PM +0200, Aleksandr Nogikh wrote:
> > On Thu, Aug 31, 2023 at 5:39 PM Andy Shevchenko
> > <andriy.shevchenko@...ux.intel.com> wrote:
> > > On Thu, Aug 31, 2023 at 04:15:23PM +0200, Aleksandr Nogikh wrote:
> > > > #syz dup: general protection fault in nfc_register_device
> > >
> > > Thanks!
> > >
> > > Maybe you can improve it to
> > > 1) drop those prefixes by itself;
> >
> > Yeah, I've just thought about it too when I was correcting the command.
> >
> > > 2) explain better in the quick help below what "exact" means.
> > >
> > > Also is it possible to test just patches without repository?
> > > Can you add this feature when you simply submit a message ID
> > > to it and it uses lore.kernel.org for the patch or series?
> >
> > Thanks, these are good suggestions!
> > The first one is easy to do (we actually already support that
> > internally, so only a small change is needed), the second one is a bit
> > trickier, but I'll look into that.
FTR I've just sent two PRs:
1) Support `#syz test` without arguments -- in that case, syzbot
checks out the latest commit on the fuzzed branch.
2) Parse all kinds of email titles in `#syz dup`.
I'll deploy them soon.
>
> JFYI:
> 20230828145824.3895288-1-andriy.shevchenko@...ux.intel.com
>
> Can be used for testing.
> So I like to write
>
> #syz test patch: 20230828145824.3895288-1-andriy.shevchenko@...ux.intel.com
Thanks for the idea!
I've filed https://github.com/google/syzkaller/issues/4182
The biggest problem is to figure out the command syntax that'd also
let us specify the tree on which to apply the patch. Or to restrict
the patch testing to either apply the patch from lore on top of the
fuzzed tree or to accept repo/branch/patch explicitly.
>
> The responses can be (among others?):
> 1/ tested and test results;
> 2/ not found.
>
> --
> With Best Regards,
> Andy Shevchenko
>
>
Powered by blists - more mailing lists