qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Dmitry Monakhov <dmonakhov@openvz.org>
To: Paolo Bonzini <pbonzini@redhat.com>,
	"Denis V. Lunev" <den@odin.com>,
	Stefan Hajnoczi <stefanha@gmail.com>
Cc: Dmitry Monakhov <dmonakhov@odin.com>,
	Stefan Hajnoczi <stefanha@redhat.com>,
	qemu-block@nongnu.org, qemu-devel@nongnu.org
Subject: Re: [Qemu-devel] [PATCH v5 0/2] block: enforce minimal 4096 alignment in qemu_blockalign
Date: Mon, 01 Jun 2015 14:16:32 +0300	[thread overview]
Message-ID: <87382b8yr3.fsf@openvz.org> (raw)
In-Reply-To: <556C36BC.1000409@redhat.com>


Paolo Bonzini <pbonzini@redhat.com> writes:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA256
>
>
>
> On 01/06/2015 12:34, Dmitry Monakhov wrote:
>> Yes. Improvement is not huge, but it can be detected for old qemu 
>> unpatched kernel: 728 MiB/sec ± 20Mb patched kernel  : 748 MiB/sec
>> ± 10Mb
>
> Ok, so about 3-4%.  What does the blktrace look like with a patched
> kernel, and what is the performance with old QEMU?
new qemu looks perfect
259,0   31       62     0.711100207 10768  D  WS 29364992 + 256 [qemu-io]
259,0   31       63     0.711128098 10768  Q  WS 29365248 + 256 [qemu-io]
259,0   31       64     0.711128509 10768  G  WS 29365248 + 256 [qemu-io]
259,0   31       65     0.711130463 10768  D  WS 29365248 + 256 [qemu-io]
259,0   31       66     0.711134222 10768  C  WS 29364480 + 256 [0]
259,0   31       67     0.711158858 10768  Q  WS 29365504 + 256 [qemu-io]
259,0   31       68     0.711159226 10768  G  WS 29365504 + 256 [qemu-io]
259,0   31       69     0.711161104 10768  D  WS 29365504 + 256 [qemu-io]
259,0   31       70     0.711171863 10768  C  WS 29364736 + 256 [0]

old qemu on patched kernel stil issues two bios Max_sz-1page + 1page, such bios are not
merged and completed separately  
259,0   31      385     0.719283423 10729  Q  WS 29376775 + 248 [qemu-io]
259,0   31      386     0.719283694 10729  G  WS 29376775 + 248 [qemu-io]
259,0   31      387     0.719285909 10729  D  WS 29376775 + 248 [qemu-io]
259,0   31      388     0.719287600 10729  Q  WS 29377023 + 8 [qemu-io]
259,0   31      389     0.719287810 10729  G  WS 29377023 + 8 [qemu-io]
259,0   31      390     0.719288666 10729  D  WS 29377023 + 8 [qemu-io]
259,0   31      391     0.719315193 10729  Q  WS 29377031 + 248 [qemu-io]
259,0   31      392     0.719315400 10729  G  WS 29377031 + 248 [qemu-io]
259,0   31      393     0.719317411 10729  D  WS 29377031 + 248 [qemu-io]
259,0   31      394     0.719319179 10729  Q  WS 29377279 + 8 [qemu-io]
259,0   31      395     0.719319366 10729  G  WS 29377279 + 8 [qemu-io]
259,0   31      396     0.719320014 10729  D  WS 29377279 + 8 [qemu-io]
.....
259,0   31      916     0.722132787 10729  C  WS 29377031 + 248 [0]
....
259,0   31      995     0.722549469 10729  C  WS 29376775 + 248 [0]
....
259,0   31     4260     0.748148953 10729  C  WS 29377023 + 8 [0]
.....
259,0   31     4274     0.748217568 10729  C  WS 29377279 + 8 [0]
I'm not quite sure why each bio dispatched to device that soon, AFAIU plug/unplug
scope whole do_direct_IO() loop, So bio it looks like bio submit to request
path should be modified.
>
> Paolo
>
>> IMHO patch is not sufficient. We have to correct bio construction
>> inside direct-io code. I'll be back with the patch.

  reply	other threads:[~2015-06-01 11:19 UTC|newest]

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-05-04 13:42 [Qemu-devel] [PATCH v5 0/2] block: enforce minimal 4096 alignment in qemu_blockalign Denis V. Lunev
2015-05-04 13:42 ` [Qemu-devel] [PATCH 1/2] block: minimal bounce buffer alignment Denis V. Lunev
2015-05-04 13:42 ` [Qemu-devel] [PATCH 2/2] block: align bounce buffers to page Denis V. Lunev
2015-05-11 14:54   ` [Qemu-devel] [Qemu-block] " Stefan Hajnoczi
2015-05-11 15:32     ` Eric Blake
2015-05-11 15:40       ` Denis V. Lunev
2015-05-11 15:08 ` [Qemu-devel] [Qemu-block] [PATCH v5 0/2] block: enforce minimal 4096 alignment in qemu_blockalign Stefan Hajnoczi
2015-05-11 16:07   ` Denis V. Lunev
2015-05-11 16:38     ` Denis V. Lunev
2015-05-11 16:47     ` Denis V. Lunev
2015-05-12 10:01       ` Stefan Hajnoczi
2015-05-12 10:19         ` Denis V. Lunev
2015-05-12 10:46           ` Paolo Bonzini
2015-05-13 15:43             ` Stefan Hajnoczi
2015-05-13 16:46               ` Denis V. Lunev
2015-05-29 16:43                 ` [Qemu-devel] " Paolo Bonzini
2015-06-01 10:34                   ` Dmitry Monakhov
2015-06-01 10:41                     ` Paolo Bonzini
2015-06-01 11:16                       ` Dmitry Monakhov [this message]
2015-06-01 11:26                         ` Paolo Bonzini
2015-06-01 11:57                           ` Dmitry Monakhov
2015-05-14  9:13               ` [Qemu-devel] [Qemu-block] " Paolo Bonzini

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=87382b8yr3.fsf@openvz.org \
    --to=dmonakhov@openvz.org \
    --cc=den@odin.com \
    --cc=dmonakhov@odin.com \
    --cc=pbonzini@redhat.com \
    --cc=qemu-block@nongnu.org \
    --cc=qemu-devel@nongnu.org \
    --cc=stefanha@gmail.com \
    --cc=stefanha@redhat.com \
    /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).