Re: [PATCH 3/3] io_uring: use kmem_cache to alloc sqe

From: Jens Axboe <axboe@kernel.dk>
To: Zhengyuan Liu <liuzhengyuan@kylinos.cn>
Cc: linux-block@vger.kernel.org
Subject: Re: [PATCH 3/3] io_uring: use kmem_cache to alloc sqe
Date: Mon, 15 Jul 2019 07:52:50 -0600	[thread overview]
Message-ID: <6059ea46-b87f-e8e9-5e41-47b77f4c4c60@kernel.dk> (raw)
In-Reply-To: <5d2c1314.1c69fb81.ecfb1.bf96SMTPIN_ADDED_BROKEN@mx.google.com>

On 7/14/19 11:45 PM, Zhengyuan Liu wrote:
> 
> On 7/15/19 11:51 AM, Jens Axboe wrote:
>> On Jul 14, 2019, at 9:38 PM, Zhengyuan Liu <liuzhengyuan@kylinos.cn> wrote:
>>>
>>>> On 7/14/19 5:44 AM, Jens Axboe wrote:
>>>>> On 7/12/19 10:54 PM, Zhengyuan Liu wrote:
>>>>> As we introduced three lists(async, defer, link), there could been
>>>>> many sqe allocation. A natural idea is using kmem_cache to satisfy
>>>>> the allocation just like io_kiocb does.
>>>> A change like this needs to come with some performance numbers
>>>> or utilization numbers showing the benefit. I have considered
>>>> doing this before, but just never got around to testing if it's
>>>> worth while or not.
>>>> Have you?
>>> I only did some simple testing with fio. The benefit was deeply depend on the IO  scenarios. For random and direct IO , it appears to be nearly no seq copying, but for buffered sequential rw, it appears to be more than 60% copying compared to original submition.
>> Right, which is great as it’s then working as designed! But my
>> question was, for that sequential case, what kind of speed up (or
>> reduction in overhead) do you see from allocating the unit out of
>> slab vs kmalloc? There has to be a win there for the change to be
>> worthwhile.
>
> Thanks for your comments  Jens. No speed up indeed in overhead from my
> testing.

Then I suggest we just drop this change, it only makes sense if there's
a win.

-- 
Jens Axboe