Received-SPF: pass (google.com: domain of linux-kernel+bounces-75549-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.48.161 as permitted sender) client-ip=147.75.48.161;
Date: Wed, 21 Feb 2024 13:29:16 -0800
From: Jakub Kicinski <kuba@kernel.org>
To: Thomas Gleixner <tglx@linutronix.de>
Cc: syzbot <syzbot+039399a9b96297ddedca@syzkaller.appspotmail.com>,
 linux-kernel@vger.kernel.org, syzkaller-bugs@googlegroups.com,
 netdev@vger.kernel.org, Paolo Abeni <pabeni@redhat.com>, "David S. Miller"
 <davem@davemloft.net>
Subject: Re: [syzbot] [kernel?] BUG: unable to handle kernel NULL pointer
 dereference in hrtimer_active
Message-ID: <20240221132916.6c7c4219@kernel.org>
In-Reply-To: <875xytjnio.ffs@tglx>
References: <00000000000014671906112cb2ef@google.com>
	<875xytjnio.ffs@tglx>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit

On Tue, 13 Feb 2024 00:01:03 +0100 Thomas Gleixner wrote:
> So something in that syzbot test case manages to tear down a napi
> context which has not yet been fully initialized. While the rest of
> napi_disable() does not care much as long as neither NAPIF_STATE_SCHED
> nor NAPIF_STATE_NPSVC are set in napi->state, hrtimer_cancel() pretty
> much cares as demonstrated by the NULL pointer dereference.
> 
> While it would be trivial to harden the hrtimer code for the case that a
> non-initialized hrtimer is canceled, I wonder whether this invocation of
> napi_disable() is harmless (aside of the hrtimer issue) or if there are
> some hidden subtle issues with that.

Thanks for the forward, I stared at it for a bit and I can see one way
to make veth disable unregistered NAPI. I'll send a fix shortly.