linux-block.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Jens Axboe <axboe@kernel.dk>
To: Linus Torvalds <torvalds@linux-foundation.org>
Cc: "linux-block@vger.kernel.org" <linux-block@vger.kernel.org>,
	"linux-aio@kvack.org" <linux-aio@kvack.org>
Subject: Re: [GIT PULL] Support for the io_uring IO interface
Date: Sun, 3 Mar 2019 20:23:32 -0700	[thread overview]
Message-ID: <4d04fb7d-f3f8-cf5b-fe10-d6eb8a59f5cf@kernel.dk> (raw)
In-Reply-To: <CAHk-=wgf6WfxnskG06zRGN=+6u5xyRdAb2GP0mT7J6f5xk+3MA@mail.gmail.com>

On 3/3/19 7:34 PM, Linus Torvalds wrote:
> On Sun, Mar 3, 2019 at 5:53 PM Jens Axboe <axboe@kernel.dk> wrote:
>>
>> This pull request adds support for a new IO interface, io_uring.
>> io_uring allows an application to communicate with the kernel through
>> two rings, the submission queue (SQ) and completion queue (CQ) ring.
>> This allows for very efficient handling of IOs, see the v5 posting for
>> some basic numbers:
> 
> Having just looked at the bugs in fs/aio.c and not finding a lot of
> users of *that*, and absolutely zero actual test-cases, I'm actually
> less than thrilled to see a new interface with the same behavior.
> 
> So I would want to see some test programs that actually use this.
> Perhaps as part of tools/testing/?

I think there's a few things to take into account here. Nothing
(basically) uses fs/aio.c. Obviously that's true for io_uring as well
right now, but the hope is that that is NOT going to be true going
forward as it works for any type of IO, instead of being just O_DIRECT.

That said, there are test programs. As mentioned in the pull, there's a
fio IO engine that uses it. In the fio repo, there's also a standalone
test program, t/io_uring.c.

Outside of that, the liburing repo mentioned does have some actual test
cases, in the test/ directory. Some are unit tests, some are just random
test cases.

So it's not like there are NO test cases. Not a huge amount, but most of
them have been organically grown as bugs have shown up.

Now that you know there are some, how do you want to proceed? I can
write some more unit like tests and migrate the ones I do have to
tools/testing/?

-- 
Jens Axboe


  reply	other threads:[~2019-03-04  3:23 UTC|newest]

Thread overview: 15+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-03-04  1:53 [GIT PULL] Support for the io_uring IO interface Jens Axboe
2019-03-04  2:34 ` Linus Torvalds
2019-03-04  3:23   ` Jens Axboe [this message]
  -- strict thread matches above, loose matches on Subject: below --
2019-03-06 16:13 Jens Axboe
2019-03-06 20:05 ` Jens Axboe
2019-03-08 22:55   ` Linus Torvalds
2019-03-08 23:36     ` Linus Torvalds
2019-03-08 23:52       ` Linus Torvalds
2019-03-09  0:08         ` Linus Torvalds
2019-03-09  4:25           ` Jens Axboe
2019-03-09  4:33             ` Linus Torvalds
2019-03-09  4:39               ` Jens Axboe
2019-03-09  5:41                 ` Linus Torvalds
2019-03-12 21:52                   ` Jens Axboe
2019-03-09  3:03       ` Al Viro

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=4d04fb7d-f3f8-cf5b-fe10-d6eb8a59f5cf@kernel.dk \
    --to=axboe@kernel.dk \
    --cc=linux-aio@kvack.org \
    --cc=linux-block@vger.kernel.org \
    --cc=torvalds@linux-foundation.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).