qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Kevin Wolf <kwolf@redhat.com>
To: John Snow <jsnow@redhat.com>
Cc: qemu-block@nongnu.org, pkrempa@redhat.com, jtc@redhat.com,
	qemu-devel@nongnu.org
Subject: Re: [Qemu-devel] [RFC v4 15/21] blockjobs: add prepare callback
Date: Wed, 28 Feb 2018 18:04:54 +0100	[thread overview]
Message-ID: <20180228170454.GI4855@localhost.localdomain> (raw)
In-Reply-To: <20180223235142.21501-16-jsnow@redhat.com>

Am 24.02.2018 um 00:51 hat John Snow geschrieben:
> Some jobs upon finalization may need to perform some work that can
> still fail. If these jobs are part of a transaction, it's important
> that these callbacks fail the entire transaction.
> 
> We allow for a new callback in addition to commit/abort/clean that
> allows us the opportunity to have fairly late-breaking failures
> in the transactional process.
> 
> The expected flow is:
> 
> - All jobs in a transaction converge to the WAITING state
>   (added in a forthcoming commit)
> - All jobs prepare to call either commit/abort
> - If any job fails, is canceled, or fails preparation, all jobs
>   call their .abort callback.
> - All jobs enter the PENDING state, awaiting manual intervention
>   (also added in a forthcoming commit)
> - block-job-finalize is issued by the user/management layer
> - All jobs call their commit callbacks.
> 
> Signed-off-by: John Snow <jsnow@redhat.com>

You almost made me believe the scary thought that we need transactional
graph modifications, but after writing half of the reply, I think it's
just that your order here is wrong.

So .prepare is the last thing in the whole process that is allowed to
fail. Graph manipulations such as bdrv_replace_node() can fail. Graph
manipulations can also only be made in response to block-job-finalize
because the management layer must be aware of them. Take them together
and you have a problem.

Didn't we already establish earlier that .prepare/.commit/.abort must be
called together and cannot be separated by waiting for a QMP command
because of locking and things?

So if you go to PENDING first, then wait for block-job-finalize and only
then call .prepare/.commit/.abort, we should be okay for both problems.

And taking a look at the final state, that seems to be what you do, so
in the end, it's probably just the commit message that needs a fix.

>  blockjob.c                   | 34 +++++++++++++++++++++++++++++++---
>  include/block/blockjob_int.h | 10 ++++++++++
>  2 files changed, 41 insertions(+), 3 deletions(-)
> 
> diff --git a/blockjob.c b/blockjob.c
> index 8f02c03880..1c010ec100 100644
> --- a/blockjob.c
> +++ b/blockjob.c
> @@ -394,6 +394,18 @@ static void block_job_update_rc(BlockJob *job)
>      }
>  }
>  
> +static int block_job_prepare(BlockJob *job)
> +{
> +    if (job->ret) {
> +        goto out;
> +    }
> +    if (job->driver->prepare) {
> +        job->ret = job->driver->prepare(job);
> +    }
> + out:
> +    return job->ret;
> +}

Why not just if (job->ret == 0 && job->driver->prepare) and save the
goto?

Kevin

  parent reply	other threads:[~2018-02-28 17:05 UTC|newest]

Thread overview: 98+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-02-23 23:51 [Qemu-devel] [RFC v4 00/21] blockjobs: add explicit job management John Snow
2018-02-23 23:51 ` [Qemu-devel] [RFC v4 01/21] blockjobs: fix set-speed kick John Snow
2018-02-27 16:32   ` Eric Blake
2018-02-27 17:18   ` Kevin Wolf
2018-02-23 23:51 ` [Qemu-devel] [RFC v4 02/21] blockjobs: model single jobs as transactions John Snow
2018-02-27 16:36   ` Eric Blake
2018-02-27 17:18   ` Kevin Wolf
2018-02-23 23:51 ` [Qemu-devel] [RFC v4 03/21] blockjobs: add manual property John Snow
2018-02-27 16:39   ` Eric Blake
2018-02-27 17:18   ` Kevin Wolf
2018-02-23 23:51 ` [Qemu-devel] [RFC v4 04/21] blockjobs: add status enum John Snow
2018-02-27 16:44   ` Eric Blake
2018-02-27 17:19   ` Kevin Wolf
2018-02-23 23:51 ` [Qemu-devel] [RFC v4 05/21] blockjobs: add state transition table John Snow
2018-02-27 16:27   ` Kevin Wolf
2018-02-27 16:45     ` John Snow
2018-02-27 17:08       ` Kevin Wolf
2018-02-27 18:58         ` John Snow
2018-02-27 18:58   ` Eric Blake
2018-02-27 19:06     ` John Snow
2018-02-27 19:22     ` John Snow
2018-02-23 23:51 ` [Qemu-devel] [RFC v4 06/21] iotests: add pause_wait John Snow
2018-02-27 17:19   ` Kevin Wolf
2018-02-27 19:01   ` Eric Blake
2018-02-23 23:51 ` [Qemu-devel] [RFC v4 07/21] blockjobs: add block_job_verb permission table John Snow
2018-02-27 17:19   ` Kevin Wolf
2018-02-27 19:25   ` Eric Blake
2018-02-27 19:38     ` John Snow
2018-02-23 23:51 ` [Qemu-devel] [RFC v4 08/21] blockjobs: add ABORTING state John Snow
2018-02-27 19:34   ` Eric Blake
2018-02-28 14:54   ` Kevin Wolf
2018-02-28 19:26     ` John Snow
2018-02-23 23:51 ` [Qemu-devel] [RFC v4 09/21] blockjobs: add CONCLUDED state John Snow
2018-02-27 19:38   ` Eric Blake
2018-02-27 19:44     ` John Snow
2018-02-28 15:37   ` Kevin Wolf
2018-02-28 19:29     ` John Snow
2018-02-23 23:51 ` [Qemu-devel] [RFC v4 10/21] blockjobs: add NULL state John Snow
2018-02-27 19:41   ` Eric Blake
2018-02-28 15:42   ` Kevin Wolf
2018-02-28 20:04     ` John Snow
2018-02-23 23:51 ` [Qemu-devel] [RFC v4 11/21] blockjobs: add block_job_dismiss John Snow
2018-02-27 19:44   ` Eric Blake
2018-02-28 15:53   ` Kevin Wolf
2018-02-28 20:35     ` John Snow
2018-02-23 23:51 ` [Qemu-devel] [RFC v4 12/21] blockjobs: ensure abort is called for cancelled jobs John Snow
2018-02-27 19:49   ` Eric Blake
2018-02-27 20:43     ` John Snow
2018-02-28 16:05   ` Kevin Wolf
2018-02-23 23:51 ` [Qemu-devel] [RFC v4 13/21] blockjobs: add commit, abort, clean helpers John Snow
2018-02-27 19:50   ` Eric Blake
2018-02-28 16:07   ` Kevin Wolf
2018-02-23 23:51 ` [Qemu-devel] [RFC v4 14/21] blockjobs: add block_job_txn_apply function John Snow
2018-02-27 19:52   ` Eric Blake
2018-02-28 16:32   ` Kevin Wolf
2018-02-23 23:51 ` [Qemu-devel] [RFC v4 15/21] blockjobs: add prepare callback John Snow
2018-02-27 19:56   ` Eric Blake
2018-02-27 20:45     ` John Snow
2018-02-28 17:04   ` Kevin Wolf [this message]
2018-03-07  3:19     ` John Snow
2018-02-23 23:51 ` [Qemu-devel] [RFC v4 16/21] blockjobs: add waiting status John Snow
2018-02-27 20:00   ` Eric Blake
2018-02-27 20:50     ` John Snow
2018-02-28 17:46       ` Kevin Wolf
2018-02-23 23:51 ` [Qemu-devel] [RFC v4 17/21] blockjobs: add PENDING status and event John Snow
2018-02-27 20:05   ` Eric Blake
2018-02-27 20:54     ` John Snow
2018-02-28 17:55   ` Kevin Wolf
2018-02-23 23:51 ` [Qemu-devel] [RFC v4 18/21] blockjobs: add block-job-finalize John Snow
2018-02-27 20:13   ` Eric Blake
2018-02-28 18:15   ` Kevin Wolf
2018-02-28 19:14     ` John Snow
2018-03-01 10:01       ` Kevin Wolf
2018-03-01 19:24         ` John Snow
2018-02-23 23:51 ` [Qemu-devel] [RFC v4 19/21] blockjobs: Expose manual property John Snow
2018-02-27 20:16   ` Eric Blake
2018-02-27 20:42     ` John Snow
2018-02-27 21:57     ` John Snow
2018-02-27 22:27       ` Eric Blake
2018-02-28 18:23       ` Kevin Wolf
2018-02-28 19:19         ` John Snow
2018-02-28 18:25   ` Kevin Wolf
2018-02-28 19:20     ` John Snow
2018-02-28 19:27       ` [Qemu-devel] [Qemu-block] " Eric Blake
2018-02-23 23:51 ` [Qemu-devel] [RFC v4 20/21] iotests: test manual job dismissal John Snow
2018-02-27 20:21   ` Eric Blake
2018-02-27 20:41     ` John Snow
2018-02-23 23:51 ` [Qemu-devel] [RFC v4 21/21] blockjobs: add manual_mgmt option to transactions John Snow
2018-02-27 20:24   ` Eric Blake
2018-02-28 18:29     ` Kevin Wolf
2018-02-28 19:24       ` John Snow
2018-03-01 10:10         ` Kevin Wolf
2018-02-24  0:30 ` [Qemu-devel] [RFC v4 00/21] blockjobs: add explicit job management no-reply
2018-02-24 14:31 ` no-reply
2018-02-27 21:01   ` John Snow
2018-02-28 18:32     ` Kevin Wolf
2018-02-25 23:25 ` no-reply
2018-02-27 20:58   ` 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=20180228170454.GI4855@localhost.localdomain \
    --to=kwolf@redhat.com \
    --cc=jsnow@redhat.com \
    --cc=jtc@redhat.com \
    --cc=pkrempa@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).