Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Subject: Re: KASAN: use-after-free Read in userfaultfd_release (2)
To:     Al Viro <viro@zeniv.linux.org.uk>, Hillf Danton <hdanton@sina.com>
Cc:     syzbot <syzbot+75867c44841cb6373570@syzkaller.appspotmail.com>,
        linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org,
        syzkaller-bugs@googlegroups.com,
        Markus Elfring <Markus.Elfring@web.de>
References: <0000000000001bbb6705aa49635a@google.com>
 <20200713084512.10416-1-hdanton@sina.com>
 <20200720160059.GO2786714@ZenIV.linux.org.uk>
From:   Daniel Colascione <dancol@dancol.org>
Message-ID: <1f69c0ab-5791-974f-8bc0-3997ab1d61ea@dancol.org>
Date:   Mon, 20 Jul 2020 16:46:00 -0700
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101
 Thunderbird/68.10.0
MIME-Version: 1.0
In-Reply-To: <20200720160059.GO2786714@ZenIV.linux.org.uk>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On 7/20/20 9:00 AM, Al Viro wrote:
> On Mon, Jul 13, 2020 at 04:45:12PM +0800, Hillf Danton wrote:
> 
>> Bridge the gap between slab free and the fput in task work wrt
>> file's private data.
> 
> No.  This
> 
>> @@ -2048,6 +2055,7 @@ SYSCALL_DEFINE1(userfaultfd, int, flags)
>>   
>>   	fd = get_unused_fd_flags(O_RDONLY | O_CLOEXEC);
>>   	if (fd < 0) {
>> +		file->private_data = NULL;
>>   		fput(file);
>>   		goto out;
>>   	}
>>
> 
> is fundamentally wrong; you really shouldn't take over the cleanups
> if you ever do fput().

Yep. I don't recall how the O_CLOEXEC got in there: that's indeed wrong, 
and probably the result of patch-editing butchery. As for the exit 
cleanup: yes, that's a bug. I was trying to keep the exit paths 
together. We could fix it forward (which seems simple enough) or re-submit.