Re: [PATCH] umh: fix UAF when the process is being killed

From: Luis Chamberlain <mcgrof@kernel.org>
To: Schspa Shi <schspa@gmail.com>
Cc: mingo@redhat.com, peterz@infradead.org, juri.lelli@redhat.com,
	vincent.guittot@linaro.org, dietmar.eggemann@arm.com,
	rostedt@goodmis.org, bsegall@google.com, mgorman@suse.de,
	bristot@redhat.com, vschneid@redhat.com,
	linux-kernel@vger.kernel.org,
	syzbot+10d19d528d9755d9af22@syzkaller.appspotmail.com,
	syzbot+70d5d5d83d03db2c813d@syzkaller.appspotmail.com,
	syzbot+83cb0411d0fcf0a30fc1@syzkaller.appspotmail.com
Subject: Re: [PATCH] umh: fix UAF when the process is being killed
Date: Fri, 23 Dec 2022 07:01:48 -0800	[thread overview]
Message-ID: <Y6XC3Du9pFKQFNkt@bombadil.infradead.org> (raw)
In-Reply-To: <m21qor3afi.fsf@gmail.com>

On Thu, Dec 22, 2022 at 08:09:38PM +0800, Schspa Shi wrote:
> 
> Attaching the full test program in case anyone wants to add some
> comments.

Good stuff.

That looks like a kernel sefltest. So you can just add it as an
initial selftest for completion so lib/test_completion.c and extend
lib/Kconfig.debug for a new kconfig symbol for it, and then just add
a script on tools/testing/selftets/completion/ with a simple makefile
which references a script which just calls modprobe. You can look at
tools/testing/selftests/kmod/ for an example.

But I still think you may want an SmPL Coccinelle grammer patch to hunt
down other users with this pattern. The beneefit is that then you can
use the same Coccinelle patch to also then *fix* the issue in other
places.

The current uaf on umh is not something I'm terribly concerned to be
exploited in the wild. I don't think other use cases would be easier,
but, all this work would close the gap completely.

Thanks for doing this.

  Luis