[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <6188031e-8af2-5f67-fe79-79bfb7ad4344@kernel.dk>
Date: Thu, 23 Feb 2023 19:29:32 -0700
From: Jens Axboe <axboe@...nel.dk>
To: Ziyang Zhang <ZiyangZhang@...ux.alibaba.com>
Cc: io-uring@...r.kernel.org, linux-kernel@...r.kernel.org,
asml.silence@...il.com
Subject: Re: [PATCH tools/io_uring] tools/io_uring: correctly set "ret" for
sq_poll case
On 2/23/23 7:25 PM, Ziyang Zhang wrote:
> On 2023/2/23 11:46, Jens Axboe wrote:
>> On 2/21/23 12:37?AM, Ziyang Zhang wrote:
>>> For sq_poll case, "ret" is not initialized or cleared/set. In this way,
>>> output of this test program is incorrect and we can not even stop this
>>> program by pressing CTRL-C.
>>>
>>> Reset "ret" to zero in each submission/completion round, and assign
>>> "ret" to "this_reap".
>>
>> Can you check if this issue also exists in the fio copy of this, which
>> is t/io_uring.c in:
>>
>> git://git.kernel.dk/fio
>>
>> The copy in the kernel is pretty outdated at this point, and should
>> probably get removed. But if the bug is in the above main version, then
>> we should fix it there and then ponder if we want to remove the one in
>> the kernel or just get it updated to match the upstream version.
>>
>
> Hi Jens,
>
> I have checked t/io_uring.c and the code is correct with sq_poll.
OK good, I'll attempt a sync with the kernel copy...
--
Jens Axboe
Powered by blists - more mailing lists