From: "Pavel Dovgalyuk" <Pavel.Dovgaluk@ispras.ru>
To: 'Kevin Wolf' <kwolf@redhat.com>
Cc: edgar.iglesias@xilinx.com, peter.maydell@linaro.org,
igor.rubinov@gmail.com, mark.burton@greensocs.com,
real@ispras.ru, hines@cert.org, qemu-devel@nongnu.org,
maria.klimushenkova@ispras.ru, stefanha@redhat.com,
pbonzini@redhat.com, batuzovk@ispras.ru, alex.bennee@linaro.org,
fred.konrad@greensocs.com
Subject: Re: [Qemu-devel] [PATCH 3/3] replay: introduce block devices record/replay
Date: Tue, 9 Feb 2016 14:52:59 +0300 [thread overview]
Message-ID: <001201d16330$6ce67e40$46b37ac0$@Dovgaluk@ispras.ru> (raw)
In-Reply-To: <20160209102739.GB8554@noname.redhat.com>
> From: Kevin Wolf [mailto:kwolf@redhat.com]
> Am 09.02.2016 um 06:55 hat Pavel Dovgalyuk geschrieben:
> > This patch introduces a set of functions that implement recording
> > and replaying of block devices' operations. These functions form a thin
> > layer between blk_aio_ functions and replay subsystem.
> > All asynchronous block requests are added to the queue to be processed
> > at deterministically invoked record/replay checkpoints.
> > Queue is flushed at checkpoints and information about processed requests
> > is recorded to the log. In replay phase the queue is matched with
> > events read from the log. Therefore block devices requests are processed
> > deterministically.
> >
> > Signed-off-by: Pavel Dovgalyuk <pavel.dovgaluk@ispras.ru>
>
> This series doesn't seem to apply to current master, so it's somewhat
> hard to look at the end result. I can see just from patches, though,
> that this will need some more discussion.
Thank you for the response.
I forgot to rebase these patches, but there are minor problems with applying them.
>
> Just picking one example of how you convert blk_* functions:
>
> > -BlockAIOCB *blk_aio_write_zeroes(BlockBackend *blk, int64_t sector_num,
> > - int nb_sectors, BdrvRequestFlags flags,
> > - BlockCompletionFunc *cb, void *opaque)
> > +BlockAIOCB *blk_aio_write_zeroes_impl(BlockBackend *blk, int64_t sector_num,
> > + int nb_sectors, BdrvRequestFlags flags,
> > + BlockCompletionFunc *cb, void *opaque)
> > {
> > int ret = blk_check_request(blk, sector_num, nb_sectors);
> > if (ret < 0) {
> > @@ -673,6 +674,13 @@ BlockAIOCB *blk_aio_write_zeroes(BlockBackend *blk, int64_t sector_num,
> > cb, opaque);
> > }
> >
> > +BlockAIOCB *blk_aio_write_zeroes(BlockBackend *blk, int64_t sector_num,
> > + int nb_sectors, BdrvRequestFlags flags,
> > + BlockCompletionFunc *cb, void *opaque)
> > +{
> > + return replay_aio_write_zeroes(blk, sector_num, nb_sectors, flags, cb, opaque);
> > +}
> > +
>
> > +BlockAIOCB *replay_aio_write_zeroes(BlockBackend *blk, int64_t sector_num,
> > + int nb_sectors, BdrvRequestFlags flags,
> > + BlockCompletionFunc *cb, void *opaque)
> > +{
> > + if (replay_mode == REPLAY_MODE_NONE) {
> > + return blk_aio_write_zeroes_impl(blk, sector_num, nb_sectors, flags, cb, opaque);
> > + } else {
> > + ReplayAIOCB *acb = replay_aio_create(REPLAY_ASYNC_EVENT_BLOCK_WRITE_ZEROES,
> > + blk, cb, opaque);
> > + acb->req.sector = sector_num;
> > + acb->req.nb_sectors = nb_sectors;
> > + acb->req.flags = flags;
> > + replay_add_event(REPLAY_ASYNC_EVENT_BLOCK_WRITE_ZEROES, acb, NULL, 0);
> > +
> > + return &acb->common;
> > + }
> > +}
>
> I think it's obvious that adding two functions to the call chain which
> do nothing in the common case is a bit ugly. If we did this for every
> feature that could possibly be enabled, we'd end up with two-kilometer
> stack traces.
>
> So definitely don't call into replay.c, which just calls back in 99.9%
> of the cases, but if anything, do the check in block-backends.c.
This way seems to be better.
> But even this doesn't feel completely right, because block drivers are
> already layered and there is no need to hardcode something optional (and
> rarely used) in the hot code path that could just be another layer.
>
> I assume that you know beforehand if you want to replay something, so
> requiring you to configure your block devices with a replay driver on
> top of the stack seems reasonable enough.
I cannot use block drivers for this. When driver functions are used, QEMU
is already used coroutines (and probably started bottom halves).
Coroutines make execution non-deterministic.
That's why we have to intercept blk_aio_ functions, that are called
deterministically.
Pavel Dovgalyuk
next prev parent reply other threads:[~2016-02-09 11:53 UTC|newest]
Thread overview: 46+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-02-09 5:55 [Qemu-devel] [PATCH 0/3] Deterministic replay extensions Pavel Dovgalyuk
2016-02-09 5:55 ` [Qemu-devel] [PATCH 1/3] replay: character devices Pavel Dovgalyuk
2016-02-09 5:55 ` [Qemu-devel] [PATCH 2/3] replay: introduce new checkpoint for icount warp Pavel Dovgalyuk
2016-02-09 5:55 ` [Qemu-devel] [PATCH 3/3] replay: introduce block devices record/replay Pavel Dovgalyuk
2016-02-09 10:27 ` Kevin Wolf
2016-02-09 11:52 ` Pavel Dovgalyuk [this message]
2016-02-10 11:45 ` Kevin Wolf
2016-02-10 12:05 ` Pavel Dovgalyuk
2016-02-10 12:28 ` Kevin Wolf
2016-02-10 12:51 ` Pavel Dovgalyuk
2016-02-10 13:25 ` Kevin Wolf
2016-02-10 13:33 ` Pavel Dovgalyuk
2016-02-10 13:52 ` Kevin Wolf
2016-02-11 6:05 ` Pavel Dovgalyuk
2016-02-11 9:43 ` Kevin Wolf
2016-02-11 11:00 ` Pavel Dovgalyuk
2016-02-11 12:18 ` Kevin Wolf
2016-02-11 12:24 ` Pavel Dovgalyuk
2016-02-12 8:33 ` Pavel Dovgalyuk
2016-02-12 9:44 ` Kevin Wolf
2016-02-12 13:19 ` Pavel Dovgalyuk
2016-02-12 13:58 ` Kevin Wolf
2016-02-15 8:38 ` Pavel Dovgalyuk
2016-02-15 9:10 ` Kevin Wolf
2016-02-15 9:14 ` Pavel Dovgalyuk
2016-02-15 9:38 ` Kevin Wolf
2016-02-15 11:19 ` Pavel Dovgalyuk
2016-02-15 12:46 ` Kevin Wolf
2016-02-15 13:54 ` Pavel Dovgalyuk
2016-02-15 14:06 ` Kevin Wolf
2016-02-15 14:24 ` Pavel Dovgalyuk
2016-02-15 15:01 ` Kevin Wolf
2016-02-16 6:25 ` Pavel Dovgalyuk
2016-02-16 10:02 ` Kevin Wolf
2016-02-16 11:20 ` Pavel Dovgalyuk
2016-02-16 12:54 ` Kevin Wolf
2016-02-18 9:18 ` Pavel Dovgalyuk
2016-02-20 7:11 ` Pavel Dovgalyuk
2016-02-22 11:06 ` Kevin Wolf
2016-02-24 11:59 ` Pavel Dovgalyuk
2016-02-24 13:14 ` Kevin Wolf
2016-02-25 9:06 ` Pavel Dovgalyuk
2016-02-26 9:01 ` Kevin Wolf
2016-02-29 7:03 ` Pavel Dovgalyuk
2016-02-29 7:54 ` Kevin Wolf
2016-02-15 14:50 ` Pavel Dovgalyuk
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='001201d16330$6ce67e40$46b37ac0$@Dovgaluk@ispras.ru' \
--to=pavel.dovgaluk@ispras.ru \
--cc=alex.bennee@linaro.org \
--cc=batuzovk@ispras.ru \
--cc=edgar.iglesias@xilinx.com \
--cc=fred.konrad@greensocs.com \
--cc=hines@cert.org \
--cc=igor.rubinov@gmail.com \
--cc=kwolf@redhat.com \
--cc=maria.klimushenkova@ispras.ru \
--cc=mark.burton@greensocs.com \
--cc=pbonzini@redhat.com \
--cc=peter.maydell@linaro.org \
--cc=qemu-devel@nongnu.org \
--cc=real@ispras.ru \
--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).