qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Max Reitz <mreitz@redhat.com>
To: John Snow <jsnow@redhat.com>, qemu-block@nongnu.org
Cc: kwolf@redhat.com, jcody@redhat.com, qemu-devel@nongnu.org,
	stefanha@redhat.com
Subject: Re: [Qemu-devel] [PATCH v2 11/13] block/mirror: remove block_job_sleep_ns calls
Date: Wed, 7 Feb 2018 23:46:25 +0100	[thread overview]
Message-ID: <98b08bf5-6ac1-a1df-e92d-046b7975e3de@redhat.com> (raw)
In-Reply-To: <20180119205847.7141-12-jsnow@redhat.com>

[-- Attachment #1: Type: text/plain, Size: 3164 bytes --]

On 2018-01-19 21:58, John Snow wrote:
> We're attempting to slacken the mirror loop in three different places,
> but we can combine these three attempts. Combine the early loop call to
> block_job_pause_point with the two late-loop calls to block_job_sleep_ns.
> 
> When delay_ns is 0 and it has not been SLICE_TIME since the last yield,
> block_job_relax is merely a call to block_job_pause_point, so this should
> be equivalent with the exception that if we have managed to not yield at
> all in the last SLICE_TIME ns, we will now do so.
> 
> I am not sure that condition was possible,
> so this loop should be equivalent.

Well, to me it even sounds like a positive change if it was a change.
We want the job to yield after SLICE_TIME ns, after all, and I don't
think it matters where that happens, exactly.

> 
> Signed-off-by: John Snow <jsnow@redhat.com>
> ---
>  block/mirror.c     | 22 +++++++++++-----------
>  block/trace-events |  2 +-
>  2 files changed, 12 insertions(+), 12 deletions(-)
> 
> diff --git a/block/mirror.c b/block/mirror.c
> index a0e0044de2..192e03694f 100644
> --- a/block/mirror.c
> +++ b/block/mirror.c
> @@ -761,7 +761,7 @@ static void coroutine_fn mirror_run(void *opaque)
>      assert(!s->dbi);
>      s->dbi = bdrv_dirty_iter_new(s->dirty_bitmap);
>      for (;;) {
> -        uint64_t delay_ns = 0;
> +        static uint64_t delay_ns = 0;

Errr.  Are you sure about that?

Now every mirror job in the qeny process will share this single
variable.  Was that your intention?

>          int64_t cnt, delta;
>          bool should_complete;
>  
> @@ -770,9 +770,16 @@ static void coroutine_fn mirror_run(void *opaque)
>              goto immediate_exit;
>          }
>  
> -        block_job_pause_point(&s->common);
> -
>          cnt = bdrv_get_dirty_count(s->dirty_bitmap);
> +
> +        trace_mirror_before_relax(s, cnt, s->synced, delay_ns);
> +        if (block_job_relax(&s->common, delay_ns)) {

See the reply to <a href="HEAD^^">that patch</a>.

> +            if (!s->synced) {
> +                goto immediate_exit;
> +            }
> +        }
> +        delay_ns = 0;
> +
>          /* s->common.offset contains the number of bytes already processed so
>           * far, cnt is the number of dirty bytes remaining and
>           * s->bytes_in_flight is the number of bytes currently being
> @@ -849,15 +856,8 @@ static void coroutine_fn mirror_run(void *opaque)
>          }
>  
>          ret = 0;
> -        trace_mirror_before_sleep(s, cnt, s->synced, delay_ns);
> -        if (!s->synced) {
> -            block_job_sleep_ns(&s->common, delay_ns);
> -            if (block_job_is_cancelled(&s->common)) {
> -                break;
> -            }
> -        } else if (!should_complete) {
> +        if (s->synced && !should_complete) {
>              delay_ns = (s->in_flight == 0 && cnt == 0 ? SLICE_TIME : 0);
> -            block_job_sleep_ns(&s->common, delay_ns);
>          }
>      }

Basic idea looks good to me (apart from the static delay_ns), but, well,
block-job-cancel on a busy job still breaks.

Max


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 512 bytes --]

  reply	other threads:[~2018-02-07 22:46 UTC|newest]

Thread overview: 28+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-01-19 20:58 [Qemu-devel] [PATCH v2 00/13] blockjob: refactor mirror_throttle John Snow
2018-01-19 20:58 ` [Qemu-devel] [PATCH v2 01/13] blockjob: record time of last entrance John Snow
2018-02-07 22:05   ` Max Reitz
2018-01-19 20:58 ` [Qemu-devel] [PATCH v2 02/13] blockjob: consolidate SLICE_TIME definition John Snow
2018-01-19 20:58 ` [Qemu-devel] [PATCH v2 03/13] blockjob: create block_job_relax John Snow
2018-02-07 22:12   ` Max Reitz
2018-01-19 20:58 ` [Qemu-devel] [PATCH v2 04/13] blockjob: allow block_job_throttle to take delay_ns John Snow
2018-02-07 22:17   ` Max Reitz
2018-01-19 20:58 ` [Qemu-devel] [PATCH v2 05/13] block/commit: use block_job_relax John Snow
2018-01-19 20:58 ` [Qemu-devel] [PATCH v2 06/13] block/stream: " John Snow
2018-01-19 20:58 ` [Qemu-devel] [PATCH v2 07/13] block/backup: " John Snow
2018-02-07 22:19   ` Max Reitz
2018-01-19 20:58 ` [Qemu-devel] [PATCH v2 08/13] allow block_job_relax to return -ECANCELED John Snow
2018-02-07 22:27   ` Max Reitz
2018-01-19 20:58 ` [Qemu-devel] [PATCH v2 09/13] block/backup: remove yield_and_check John Snow
2018-02-07 22:33   ` Max Reitz
2018-01-19 20:58 ` [Qemu-devel] [PATCH v2 10/13] block/mirror: condense cancellation and relax calls John Snow
2018-02-07 22:36   ` Max Reitz
2018-01-19 20:58 ` [Qemu-devel] [PATCH v2 11/13] block/mirror: remove block_job_sleep_ns calls John Snow
2018-02-07 22:46   ` Max Reitz [this message]
2018-02-07 22:56     ` Max Reitz
2018-02-09 23:13     ` John Snow
2018-01-19 20:58 ` [Qemu-devel] [PATCH v2 12/13] blockjob: privatize block_job_sleep_ns John Snow
2018-02-07 22:47   ` Max Reitz
2018-01-19 20:58 ` [Qemu-devel] [PATCH v2 13/13] blockjob: remove block_job_pause_point from interface John Snow
2018-02-07 22:50   ` Max Reitz
2018-01-19 21:14 ` [Qemu-devel] [PATCH v2 00/13] blockjob: refactor mirror_throttle no-reply
2018-01-30 20:25 ` John Snow

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=98b08bf5-6ac1-a1df-e92d-046b7975e3de@redhat.com \
    --to=mreitz@redhat.com \
    --cc=jcody@redhat.com \
    --cc=jsnow@redhat.com \
    --cc=kwolf@redhat.com \
    --cc=qemu-block@nongnu.org \
    --cc=qemu-devel@nongnu.org \
    --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).