qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Stefan Hajnoczi <stefanha@redhat.com>
To: qemu-devel@nongnu.org
Cc: qemu-block@nongnu.org, Stefan Hajnoczi <stefanha@redhat.com>,
	hibriansong@gmail.com, Kevin Wolf <kwolf@redhat.com>,
	Hanna Czenczek <hreitz@redhat.com>
Subject: [RFC 00/11] aio: add the aio_add_sqe() io_uring API
Date: Wed, 28 May 2025 15:09:05 -0400	[thread overview]
Message-ID: <20250528190916.35864-1-stefanha@redhat.com> (raw)

This patch series contains io_uring improvements:

1. Support the glib event loop in fdmon-io_uring.
   - aio-posix: fix polling mode with fdmon-io_uring
   - aio-posix: keep polling enabled with fdmon-io_uring.c
   - tests/unit: skip test-nested-aio-poll with io_uring
   - aio-posix: integrate fdmon into glib event loop

2. Enable fdmon-io_uring on hosts where io_uring is available at runtime.
   Otherwise continue using ppoll(2) or epoll(7).
   - aio: remove aio_context_use_g_source()

3. Add the new aio_add_sqe() API for submitting io_uring requests in the QEMU
   event loop.
   - aio: free AioContext when aio_context_new() fails
   - aio: add errp argument to aio_context_setup()
   - aio-posix: gracefully handle io_uring_queue_init() failure
   - aio-posix: add aio_add_sqe() API for user-defined io_uring requests
   - aio-posix: avoid EventNotifier for cqe_handler_bh

4. Use aio_add_sqe() in block/io_uring.c instead of creating a dedicated
   io_uring context for --blockdev aio=io_uring. This simplifies the code,
   reduces the number of file descriptors, and demonstrates the aio_add_sqe()
   API.
   - block/io_uring: use aio_add_sqe()

The highlight is aio_add_sqe(), which is needed for the FUSE-over-io_uring
Google Summer of Code project and other future QEMU features that natively use
Linux io_uring functionality.

I'm not happy with performance yet. This is why I've marked the series as
Request For Comments:
rw        bs iodepth aio    iothread before after  diff
randread  4k       1 native        0  76281 79707  +4.5%
randread  4k      64 native        0 255078 247293 -3.1%
randwrite 4k       1 native        0 132706 123337 -7.1%
randwrite 4k      64 native        0 275589 245192 -11%
randread  4k       1 io_uring      0  75284 78023  +3.5%
randread  4k      64 io_uring      0 254637 248222 -2.5%
randwrite 4k       1 io_uring      0 126519 128641 +1.7%
randwrite 4k      64 io_uring      0 258967 249266 -3.7%
randread  4k       1 native        1  90557 88436  -2.3%
randread  4k      64 native        1 290673 280456 -3.5%
randwrite 4k       1 native        1 183015 169106 -7.6%
randwrite 4k      64 native        1 281316 280078 -0.4%
randread  4k       1 io_uring      1  92479 86983  -5.9%
randread  4k      64 io_uring      1 304229 257730 -15.3%
randwrite 4k       1 io_uring      1 183983 157425 -14.4%
randwrite 4k      64 io_uring      1 299979 264156 -11.9%

Overall the performance decreases, so I need to continue profiling the
iodepth=64 cases with aio=native and aio=io_uring.

This series replaces the following older series that were held off from merging
until the QEMU 10.1 development window opened and the performance results were
collected:
- "[PATCH 0/3] [RESEND] block: unify block and fdmon io_uring"
- "[PATCH 0/4] aio-posix: integrate fdmon into glib event loop"

Stefan Hajnoczi (11):
  aio-posix: fix polling mode with fdmon-io_uring
  aio-posix: keep polling enabled with fdmon-io_uring.c
  tests/unit: skip test-nested-aio-poll with io_uring
  aio-posix: integrate fdmon into glib event loop
  aio: remove aio_context_use_g_source()
  aio: free AioContext when aio_context_new() fails
  aio: add errp argument to aio_context_setup()
  aio-posix: gracefully handle io_uring_queue_init() failure
  aio-posix: add aio_add_sqe() API for user-defined io_uring requests
  aio-posix: avoid EventNotifier for cqe_handler_bh
  block/io_uring: use aio_add_sqe()

 meson.build                       |   2 +-
 include/block/aio.h               | 134 +++++++-
 include/block/raw-aio.h           |   5 -
 util/aio-posix.h                  |  18 +-
 block/file-posix.c                |  38 +--
 block/io_uring.c                  | 489 +++++++-----------------------
 stubs/io_uring.c                  |  32 --
 tests/unit/test-aio.c             |   7 +-
 tests/unit/test-nested-aio-poll.c |  13 +-
 util/aio-posix.c                  | 134 ++++----
 util/aio-win32.c                  |   6 +-
 util/async.c                      |  53 +---
 util/fdmon-epoll.c                |  52 +++-
 util/fdmon-io_uring.c             | 218 ++++++++++---
 util/fdmon-poll.c                 |  88 +++++-
 block/trace-events                |  12 +-
 stubs/meson.build                 |   3 -
 util/trace-events                 |   4 +
 18 files changed, 668 insertions(+), 640 deletions(-)
 delete mode 100644 stubs/io_uring.c

-- 
2.49.0



             reply	other threads:[~2025-05-28 19:12 UTC|newest]

Thread overview: 33+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-05-28 19:09 Stefan Hajnoczi [this message]
2025-05-28 19:09 ` [RFC 01/11] aio-posix: fix polling mode with fdmon-io_uring Stefan Hajnoczi
2025-05-28 20:29   ` Eric Blake
2025-05-28 19:09 ` [RFC 02/11] aio-posix: keep polling enabled with fdmon-io_uring.c Stefan Hajnoczi
2025-05-28 20:34   ` Eric Blake
2025-05-28 19:09 ` [RFC 03/11] tests/unit: skip test-nested-aio-poll with io_uring Stefan Hajnoczi
2025-05-28 20:40   ` Eric Blake
2025-05-28 19:09 ` [RFC 04/11] aio-posix: integrate fdmon into glib event loop Stefan Hajnoczi
2025-05-28 21:01   ` Eric Blake
2025-05-28 19:09 ` [RFC 05/11] aio: remove aio_context_use_g_source() Stefan Hajnoczi
2025-05-28 21:02   ` Eric Blake
2025-05-28 19:09 ` [RFC 06/11] aio: free AioContext when aio_context_new() fails Stefan Hajnoczi
2025-05-28 21:06   ` Eric Blake
2025-06-05 17:49     ` Stefan Hajnoczi
2025-05-28 19:09 ` [RFC 07/11] aio: add errp argument to aio_context_setup() Stefan Hajnoczi
2025-05-28 21:07   ` Eric Blake
2025-05-28 19:09 ` [RFC 08/11] aio-posix: gracefully handle io_uring_queue_init() failure Stefan Hajnoczi
2025-05-28 22:12   ` Eric Blake
2025-05-29 15:38     ` Stefan Hajnoczi
2025-06-03  6:05       ` Markus Armbruster
2025-06-03 18:48         ` Stefan Hajnoczi
2025-06-02 12:26   ` Brian
2025-06-02 20:20     ` Stefan Hajnoczi
2025-06-02 22:37       ` Brian
2025-05-28 19:09 ` [RFC 09/11] aio-posix: add aio_add_sqe() API for user-defined io_uring requests Stefan Hajnoczi
2025-05-28 22:15   ` Eric Blake
2025-05-29 20:02   ` Eric Blake
2025-06-05 17:52     ` Stefan Hajnoczi
2025-05-28 19:09 ` [RFC 10/11] aio-posix: avoid EventNotifier for cqe_handler_bh Stefan Hajnoczi
2025-05-29 20:09   ` Eric Blake
2025-05-28 19:09 ` [RFC 11/11] block/io_uring: use aio_add_sqe() Stefan Hajnoczi
2025-05-29 21:11   ` Eric Blake
2025-06-05 18:40     ` Stefan Hajnoczi

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=20250528190916.35864-1-stefanha@redhat.com \
    --to=stefanha@redhat.com \
    --cc=hibriansong@gmail.com \
    --cc=hreitz@redhat.com \
    --cc=kwolf@redhat.com \
    --cc=qemu-block@nongnu.org \
    --cc=qemu-devel@nongnu.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).