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: [PATCH net-next] net: sock: add the case if sk is NULL
To:     yajun.deng@linux.dev, Jakub Kicinski <kuba@kernel.org>
Cc:     davem@davemloft.net, netdev@vger.kernel.org,
        linux-kernel@vger.kernel.org
References: <20210806061136.54e6926e@kicinski-fedora-pc1c0hjn.dhcp.thefacebook.com>
 <20210806063815.21541-1-yajun.deng@linux.dev>
 <489e6f1ce9f8de6fd8765d82e1e47827@linux.dev>
From:   Eric Dumazet <eric.dumazet@gmail.com>
Message-ID: <79e7c9a8-526c-ae9c-8273-d1d4d6170b69@gmail.com>
Date:   Mon, 9 Aug 2021 11:34:31 +0200
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101
 Thunderbird/78.12.0
MIME-Version: 1.0
In-Reply-To: <489e6f1ce9f8de6fd8765d82e1e47827@linux.dev>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Precedence: bulk


On 8/9/21 8:12 AM, yajun.deng@linux.dev wrote:
> August 6, 2021 9:11 PM, "Jakub Kicinski" <kuba@kernel.org> wrote:
> 
>> On Fri, 6 Aug 2021 14:38:15 +0800 Yajun Deng wrote:
>>
>>> Add the case if sk is NULL in sock_{put, hold},
>>> The caller is free to use it.
>>>
>>> Signed-off-by: Yajun Deng <yajun.deng@linux.dev>
>>
>> The obvious complaint about this patch (and your previous netdev patch)
>> is that you're spraying branches everywhere in the code. Sure, it may
> 
> Sorry for that, I'll be more normative in later submission.
>> be okay for free(), given how expensive of an operation that is but
>> is having refcounting functions accept NULL really the best practice?
>>
>> Can you give us examples in the kernel where that's the case?
> 
> 0   include/net/neighbour.h         neigh_clone()
> 1   include/linux/cgroup.h          get_cgroup_ns() and put_cgroup_ns()  (This is very similar to my submission)
> 2   include/linux/ipc_namespace.h   get_ipc_ns()
> 3   include/linux/posix_acl.h       posix_acl_dup()
> 4   include/linux/pid.h             get_pid()
> 5   include/linux/user_namespace.h  get_user_ns()
> 

These helpers might be called with NULL pointers by design.

sock_put() and sock_hold() are never called with NULL.

Same for put_page() and hundreds of other functions.

By factorizing a conditional in the function, hoping to remove one in few callers,
we add more conditional branches (and increase code size)