Re: possible deadlock in xfrm_policy_delete

On Thu, Sep 24, 2020 at 09:30:03AM +0200, peterz@infradead.org wrote:

On Thu, Sep 24, 2020 at 06:44:12AM +0200, Dmitry Vyukov wrote:

quoted

On Thu, Sep 24, 2020 at 6:36 AM Herbert Xu [off-list ref] wrote:

quoted

quoted

 (k-slock-AF_INET6){+.-.}-{2:2}

That's a seqlock.

Are you sure? Line 2503 in net/ipv4/tcp.c says

	bh_lock_sock(sk);

Did that tree you're testing include 267580db047e ("seqlock: Unbreak
lockdep") ?

According to

https://syzkaller.appspot.com/bug?extid=c32502fd255cb3a44048

the git tree used was

https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/tree/include?id=5fa35f247b563a7893f3f68f19d00ace2ccf3dff

so no it doesn't contain that patch.

Hopefully this would address those seqlock backtraces.  But what
about this particular one which is caused by the socket lock?

Thanks,
-- 
Email: Herbert Xu [off-list ref]
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt