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;
Message-ID: <9419ed33-5df2-9c4d-ae54-07bde5e3f4ff@kernel.dk>
Date:   Mon, 26 Sep 2022 14:58:56 -0600
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux aarch64; rv:102.0) Gecko/20100101
 Thunderbird/102.2.2
Subject: Re: [PATCH v2 1/3] io_uring: register single issuer task at creation
To:     Pavel Begunkov <asml.silence@gmail.com>,
        Dylan Yudaken <dylany@fb.com>
Cc:     io-uring@vger.kernel.org, linux-kernel@vger.kernel.org,
        kernel-team@fb.com
References: <20220926170927.3309091-1-dylany@fb.com>
 <20220926170927.3309091-2-dylany@fb.com>
 <35d9be6b-89ca-f2a1-ce5f-53e72610db6e@gmail.com>
 <4623be74-d877-9042-f876-09feba2f0587@kernel.dk>
 <3a582199-7ee6-caf7-0314-a8a32a17b980@gmail.com>
Content-Language: en-US
From:   Jens Axboe <axboe@kernel.dk>
In-Reply-To: <3a582199-7ee6-caf7-0314-a8a32a17b980@gmail.com>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Precedence: bulk

On 9/26/22 2:29 PM, Pavel Begunkov wrote:
> On 9/26/22 20:40, Jens Axboe wrote:
>> On 9/26/22 1:12 PM, Pavel Begunkov wrote:
>>> On 9/26/22 18:09, Dylan Yudaken wrote:
>>>> Instead of picking the task from the first submitter task, rather use the
>>>> creator task or in the case of disabled (IORING_SETUP_R_DISABLED) the
>>>> enabling task.
>>>>
>>>> This approach allows a lot of simplification of the logic here. This
>>>> removes init logic from the submission path, which can always be a bit
>>>> confusing, but also removes the need for locking to write (or read) the
>>>> submitter_task.
>>>>
>>>> Users that want to move a ring before submitting can create the ring
>>>> disabled and then enable it on the submitting task.
>>>
>>> I think Dylan briefly mentioned before that it might be a good
>>> idea to task limit registration as well. I can't think of a use
>>> case at the moment but I agree we may find some in the future.
>>>
>>>
>>> diff --git a/io_uring/io_uring.c b/io_uring/io_uring.c
>>> index 242d896c00f3..60a471e43fd9 100644
>>> --- a/io_uring/io_uring.c
>>> +++ b/io_uring/io_uring.c
>>> @@ -3706,6 +3706,9 @@ static int __io_uring_register(struct io_ring_ctx *ctx, unsigned opcode,
>>>       if (WARN_ON_ONCE(percpu_ref_is_dying(&ctx->refs)))
>>>           return -ENXIO;
>>>   +    if (ctx->submitter_task && ctx->submitter_task != current)
>>> +        return -EEXIST;
>>> +
>>>       if (ctx->restricted) {
>>>           if (opcode >= IORING_REGISTER_LAST)
>>>               return -EINVAL;
>>
>> Yes, I don't see any reason why not to enforce this for registration
>> too. Don't think there's currently a need to do so, but it'd be easy
>> to miss once we do add that. Let's queue that up for 6.1?
> 
> 6.1 + stable sounds ok, I don't have an opinion on how to how
> to merge it.

That's the plan. If you can just send it out as a separate commit,
I'll stage it up behind the two others from Dylan.

-- 
Jens Axboe