linux-kernel - Re: [PATCH v3] io_uring: don't wait when under-submitting

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Mon, 16 Dec 2019 19:47:23 +0300
From:   Pavel Begunkov <asml.silence@...il.com>
To:     Jens Axboe <axboe@...nel.dk>, io-uring@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH v3] io_uring: don't wait when under-submitting

On 16/12/2019 00:33, Jens Axboe wrote:
> On 12/15/19 8:48 AM, Pavel Begunkov wrote:
>> On 15/12/2019 08:42, Jens Axboe wrote:
>>> On 12/14/19 11:43 AM, Jens Axboe wrote:
>>>> On 12/14/19 7:53 AM, Pavel Begunkov wrote:
>>>>> There is no reliable way to submit and wait in a single syscall, as
>>>>> io_submit_sqes() may under-consume sqes (in case of an early error).
>>>>> Then it will wait for not-yet-submitted requests, deadlocking the user
>>>>> in most cases.
>>>>>
>>>>> In such cases adjust min_complete, so it won't wait for more than
>>>>> what have been submitted in the current call to io_uring_enter(). It
>>>>> may be less than totally in-flight including previous submissions,
>>>>> but this shouldn't do harm and up to a user.
>>>>
>>>> Thanks, applied.
>>>
>>> This causes a behavioral change where if you ask to submit 1 but
>>> there's nothing in the SQ ring, then you would get 0 before. Now
>>> you get -EAGAIN. This doesn't make a lot of sense, since there's no
>>> point in retrying as that won't change anything.
>>>
>>> Can we please just do something like the one I sent, instead of trying
>>> to over-complicate it?
>>>
>>
>> Ok, when I get to a compiler.
> 
> Great, thanks. BTW, I noticed when a regression test failed.
> 

Yeah, I properly tested only the first one. Clearly, not as easy as
I thought, and there were more to consider.

I sent the next version, but that's odd basically taking your code.
Probably, it would have been easier for you to just commit it yourself.


-- 
Pavel Begunkov



Download attachment "signature.asc" of type "application/pgp-signature" (834 bytes)