linux-kernel - Re: [PATCH v3 1/1] io_uring: optimise sqe-to-req flags translation

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] [day] [month] [year] [list]

Date:   Sun, 19 Jan 2020 10:36:43 -0700
From:   Jens Axboe <axboe@...nel.dk>
To:     Pavel Begunkov <asml.silence@...il.com>, io-uring@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH v3 1/1] io_uring: optimise sqe-to-req flags translation

On 1/19/20 12:46 AM, Pavel Begunkov wrote:
> On 18/01/2020 23:46, Jens Axboe wrote:
>> On 1/18/20 10:22 AM, Pavel Begunkov wrote:
>>> For each IOSQE_* flag there is a corresponding REQ_F_* flag. And there
>>> is a repetitive pattern of their translation:
>>> e.g. if (sqe->flags & SQE_FLAG*) req->flags |= REQ_F_FLAG*
>>>
>>> Use same numeric values/bits for them and copy instead of manual
>>> handling.
> 
> I wonder, why this isn't a common practice around the kernel. E.g. I'm
> looking at iocb_flags() and kiocb_set_rw_flags(), and their one by one
> flags copying is just wasteful.

If I were to guess, I'd assume that it's due to continually adding flags
one at the time. For the first flag, it's not a big deal. If you end up
with a handful or more, it's clearly much better to have them occupy the
same space and avoid lots of branches checking and setting matching
flags.

You should send in patches for IOCB flags.

-- 
Jens Axboe