Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Subject: Re: [PATCH] net: scm: Fix a possible sleep-in-atomic-context bug in
 scm_fp_copy()
To:     David Miller <davem@davemloft.net>
Cc:     ktkhai@virtuozzo.com, viro@zeniv.linux.org.uk, adobriyan@gmail.com,
        dvlasenk@redhat.com, xiyou.wangcong@gmail.com,
        netdev@vger.kernel.org, linux-kernel@vger.kernel.org
References: <20180901100026.16956-1-baijiaju1990@gmail.com>
 <20180902.160144.542360312136980090.davem@davemloft.net>
From:   Jia-Ju Bai <baijiaju1990@gmail.com>
Message-ID: <0b682dc4-9c94-027d-dbab-25fdb15f48a7@gmail.com>
Date:   Mon, 3 Sep 2018 09:43:49 +0800
User-Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:52.0) Gecko/20100101
 Thunderbird/52.2.0
MIME-Version: 1.0
In-Reply-To: <20180902.160144.542360312136980090.davem@davemloft.net>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 7bit
Content-Language: en-US
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Thanks for your reply.


On 2018/9/3 7:01, David Miller wrote:
> From: Jia-Ju Bai <baijiaju1990@gmail.com>
> Date: Sat,  1 Sep 2018 18:00:26 +0800
>
>> The kernel module may sleep with holding a spinlock.
>>
>> The function call paths (from bottom to top) in Linux-4.16 are:
>>
>> [FUNC] kmalloc(GFP_KERNEL)
>> net/core/scm.c, 85: kmalloc in scm_fp_copy
>> net/core/scm.c, 161: scm_fp_copy in __scm_send
>> ./include/net/scm.h, 88: __scm_send in scm_send
>> net/unix/af_unix.c, 1600: scm_send in maybe_init_creds
>> net/unix/af_unix.c, 1983: maybe_init_creds in unix_stream_sendpage
>> net/unix/af_unix.c, 1973: spin_lock in unix_stream_sendpage
> Please, do a full analysis of the code for these changes you are
> submitting.
>
> Read maybe_init_creds(), it sets msg.msg_controllen to zero.
>
> 	struct msghdr msg = { .msg_controllen = 0 };
>
> When that is zero, __scm__send() is never called.

Oh, I did not notice this, sorry...

> static __inline__ int scm_send(struct socket *sock, struct msghdr *msg,
> 			       struct scm_cookie *scm, bool forcecreds)
> {
>   ...
> 	if (msg->msg_controllen <= 0)
> 		return 0;
> 	return __scm_send(sock, msg, scm);
>
> If this bug existed, sleeping in atomic warnings would be triggering
> all the time and people would report that.

Sorry for this false positive.
I will check the code more carefully before submitting my patches.


Best wishes,
Jia-Ju Bai