Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Subject: Re: [PATCH] kernfs: fix potential null-ptr-deref in
 kernfs_path_from_node_locked()
From:   "Leizhen (ThunderTown)" <thunder.leizhen@huawei.com>
To:     Tejun Heo <tj@kernel.org>
CC:     Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        <linux-kernel@vger.kernel.org>
References: <20221123020419.1867-1-thunder.leizhen@huawei.com>
 <Y35Qgw4Q8XYD5Did@slm.duckdns.org>
 <ba083b44-93d5-37c4-380c-8e0249b2333c@huawei.com>
 <ba46adff-3604-9ccf-b1c5-83411f6652d9@huawei.com>
Message-ID: <55041efe-7443-d576-287b-49d1221fced2@huawei.com>
Date:   Thu, 24 Nov 2022 10:52:43 +0800
User-Agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:60.0) Gecko/20100101
 Thunderbird/60.7.0
MIME-Version: 1.0
In-Reply-To: <ba46adff-3604-9ccf-b1c5-83411f6652d9@huawei.com>
Content-Type: text/plain; charset="utf-8"
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Precedence: bulk


On 2022/11/24 10:28, Leizhen (ThunderTown) wrote:
> 
> 
> On 2022/11/24 10:24, Leizhen (ThunderTown) wrote:
>>
>>
>> On 2022/11/24 0:55, Tejun Heo wrote:
>>> On Wed, Nov 23, 2022 at 10:04:19AM +0800, Zhen Lei wrote:
>>>> Ensure that the 'buf' is not empty before strlcpy() uses it.
>>>>
>>>> Commit bbe70e4e4211 ("fs: kernfs: Fix possible null-pointer dereferences
>>>> in kernfs_path_from_node_locked()") first noticed this, but it didn't
>>>> fix it completely.
>>>>
>>>> Fixes: 9f6df573a404 ("kernfs: Add API to generate relative kernfs path")
>>>> Signed-off-by: Zhen Lei <thunder.leizhen@huawei.com>
>>>
>>> I think the right thing to do is removing that if. It makes no sense to call
>>> that function with NULL buf and the fact that nobody reported crashes on
>>> NULL buf indicates that we in fact never do.
>>
>> OK.
>>
>> How about I remove "buf[0] = '\0';" too? It seems to be a useless operation.
>> When 'kn_from' and 'kn_to' have a common ancestor, there must be a path from
>> 'kn_from' to 'kn_to', and strlcpy() always fills in the terminator correctly,
>> even if the buf is too small to save the first path node.
> 
> Sorry, I misanalyzed. The length used by "len < buflen ? buflen - len : 0" may
> be zero.

Ah, my brain is unstable today. The initial value of len is 0. So "buf[0] = '\0';"
can still be safely removed.

> 
>>
>> static void test(void)
>> {
>>         char buf[4];
>>         int i, n, buflen;
>>
>>         buflen = 1;
>>         n = strlcpy(buf, "string", buflen);
>>         for (i = 0; i < buflen; i++)
>>                 printk("%d: %02x\n", i, buf[i]);
>>         printk("n=%d\n\n", n);
>>
>>         buflen = sizeof(buf);
>>         n = strlcpy(buf, "string", buflen);
>>         for (i = 0; i < buflen; i++)
>>                 printk("%d: %02x\n", i, buf[i]);
>>         printk("n=%d\n", n);
>> }
>>
>> Output:
>> [   33.691497] 0: 00
>> [   33.691569] n=6
>>
>> [   33.691595] 0: 73
>> [   33.691622] 1: 74
>> [   33.691630] 2: 72
>> [   33.691637] 3: 00
>> [   33.691650] n=6
>>
>>>
>>> Thanks.
>>>
>>
> 

-- 
Regards,
  Zhen Lei