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] io_uring: fix ctx refcounting in io_submit_sqes()
To:     Pavel Begunkov <asml.silence@gmail.com>, io-uring@vger.kernel.org,
        linux-kernel@vger.kernel.org
References: <8b53ce4539784423b493fdbfae9bd4c720b24d2a.1586120916.git.asml.silence@gmail.com>
 <331eb009-a8c3-98c7-4cec-d91a821f22be@gmail.com>
From:   Jens Axboe <axboe@kernel.dk>
Message-ID: <3d0b7db2-59ae-b33c-694e-87d3d659c25e@kernel.dk>
Date:   Sun, 5 Apr 2020 15:25:24 -0700
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101
 Thunderbird/68.4.1
MIME-Version: 1.0
In-Reply-To: <331eb009-a8c3-98c7-4cec-d91a821f22be@gmail.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 4/5/20 3:13 PM, Pavel Begunkov wrote:
> On 06/04/2020 00:08, Pavel Begunkov wrote:
>> If io_get_req() fails, it drops a ref. Then, awhile keeping @submitted
>> unmodified, io_submit_sqes() breaks the loop and puts @nr - @submitted
>> refs. For each submitted req a ref is dropped in io_put_req() and
>> friends. So, for @nr taken refs there will be
>> (@nr - @submitted + @submitted + 1) dropped.
>>
>> Remove ctx refcounting from io_get_req(), that at the same time makes
>> it clearer.
> 
> It seems, nobody hit OOM, so it stayed unnoticed. And neither did I.
> It could be a good idea to do fault-injection for testing.

Actually think we just hit this, was testing with memcached (as per fixes
posted recently), and a bug on the user side ended up with 196G of slab
and running into OOM off request allocation.

But yes, would be nice to have specific fault injection testing to
avoid finding these in prod testing.

-- 
Jens Axboe