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 v3] skbuff: fix a data race in skb_queue_len()
To:     "Jason A. Donenfeld" <Jason@zx2c4.com>, eric.dumazet@gmail.com
Cc:     cai@lca.pw, netdev@vger.kernel.org, linux-kernel@vger.kernel.org,
        Marco Elver <elver@google.com>
References: <1580841629-7102-1-git-send-email-cai@lca.pw>
 <20200206163844.GA432041@zx2c4.com>
From:   Eric Dumazet <eric.dumazet@gmail.com>
Message-ID: <453212cf-8987-9f05-ceae-42a4fc3b0876@gmail.com>
Date:   Thu, 6 Feb 2020 09:10:34 -0800
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101
 Thunderbird/68.2.2
MIME-Version: 1.0
In-Reply-To: <20200206163844.GA432041@zx2c4.com>
Content-Type: text/plain; charset=utf-8
Content-Language: en-US
Content-Transfer-Encoding: 7bit
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk


On 2/6/20 8:38 AM, Jason A. Donenfeld wrote:
> Hi Eric,
> 
> On Tue, Feb 04, 2020 at 01:40:29PM -0500, Qian Cai wrote:
>> -	list->qlen--;
>> +	WRITE_ONCE(list->qlen, list->qlen - 1);
> 
> Sorry I'm a bit late to the party here, but this immediately jumped out.
> This generates worse code with a bigger race in some sense:
> 
> list->qlen-- is:
> 
>    0:   83 6f 10 01             subl   $0x1,0x10(%rdi)
> 
> whereas WRITE_ONCE(list->qlen, list->qlen - 1) is:
> 
>    0:   8b 47 10                mov    0x10(%rdi),%eax
>    3:   83 e8 01                sub    $0x1,%eax
>    6:   89 47 10                mov    %eax,0x10(%rdi)
> 
> Are you sure that's what we want?
> 
> Jason
> 


Unfortunately we do not have ADD_ONCE() or something like that.

Sure, on x86 we could get much better code generation.

If we agree a READ_ONCE() was needed at the read side,
then a WRITE_ONCE() is needed as well on write sides.

If we believe load-tearing and/or write-tearing must not ever happen,
then we must document this.