Re: A potential deadlock in sockhash map

[John Fastabend <john.fastabend@gmail.com>]

Hsin-Wei Hung wrote:
> Hi,
> 
> Just a quick update. I can still trigger the lockdep warning on bpf
> tree (5b7c4cabbb65).
> 
> Thanks,
> Hsin-Wei

Thanks, I'll take a look.

> 
> On Fri, Feb 24, 2023 at 9:58 AM Hsin-Wei Hung <hsinweih@uci.edu> wrote:
> >
> > Hi Jakub,
> >
> > Thanks for following up. Sorry that I did not receive the previous reply.
> >
> > The latest version I tested is 5.19 (3d7cb6b04c3f) and we can reproduce the
> > issue with the BPF PoC included. Since we modified Syzkaller, we do not
> > have a Syzkaller reproducer.
> >
> > I will follow John's suggestion to test the latest kernel and bpf
> > tree. I will follow
> > up if the issue still reproduces.
> >
> > Thanks,
> > Hsin-Wei
> >
> >
> >
> >
> > On Fri, Feb 24, 2023 at 8:51 AM Jakub Sitnicki <jakub@cloudflare.com> wrote:
> > >
> > > Hi,
> > >
> > > On Mon, Feb 20, 2023 at 07:39 AM -06, Hsin-Wei Hung wrote:
> > > > I think my previous report got blocked since it contained HTML
> > > > subparts so I am sending it again. Our bpf runtime fuzzer (a
> > > > customized syzkaller) triggered a lockdep warning in the bpf subsystem
> > > > indicating a potential deadlock. We are able to trigger this bug on
> > > > v5.15.25 and v5.19. The following code is a BPF PoC, and the lockdep
> > > > warning is attached at the end.
> > >
> > > Not sure if you've seen John's reply to the previous report:
> > >
> > > https://urldefense.com/v3/__https://lore.kernel.org/all/63dddcc92fc31_6bb15208e9@john.notmuch/__;!!CzAuKJ42GuquVTTmVmPViYEvSg!PU-LFxMnx4b-GmTXGI0hYjBiq8vkwrFrlf_b0N5uzy8do5kPFiNcuZJbby-19TtOH2rJoY9UwOvzFArd$
> > >
> > > Are you also fuzzing any newer kernel versions? Or was v5.19 the latest?
> > >
> > > Did syzkaller find a reproducer?
> > >
> > > Thanks,
> > > Jakub