From: Stefan Hajnoczi <stefanha@redhat.com>
To: Emanuele Giuseppe Esposito <eesposit@redhat.com>
Cc: Kevin Wolf <kwolf@redhat.com>, Fam Zheng <fam@euphon.net>,
Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>,
qemu-block@nongnu.org, qemu-devel@nongnu.org,
Hanna Reitz <hreitz@redhat.com>,
Paolo Bonzini <pbonzini@redhat.com>, John Snow <jsnow@redhat.com>
Subject: Re: [RFC PATCH v2 1/8] aio_wait_kick: add missing memory barrier
Date: Sat, 30 Apr 2022 06:21:56 +0100 [thread overview]
Message-ID: <YmzHdG9a4F+ZjJg7@stefanha-x1.localdomain> (raw)
In-Reply-To: <d8ac2a96-4508-9cc3-9ca8-762851997103@redhat.com>
[-- Attachment #1: Type: text/plain, Size: 1887 bytes --]
On Fri, Apr 29, 2022 at 10:06:33AM +0200, Emanuele Giuseppe Esposito wrote:
>
>
> Am 28/04/2022 um 13:09 schrieb Stefan Hajnoczi:
> > On Tue, Apr 26, 2022 at 04:51:07AM -0400, Emanuele Giuseppe Esposito wrote:
> >> It seems that aio_wait_kick always required a memory barrier
> >> or atomic operation in the caller, but almost nobody actually
> >> took care of doing it.
> >>
> >> Let's put the barrier in the function instead.
> >>
> >> Signed-off-by: Emanuele Giuseppe Esposito <eesposit@redhat.com>
> >> ---
> >> util/aio-wait.c | 3 ++-
> >> 1 file changed, 2 insertions(+), 1 deletion(-)
> >>
> >> diff --git a/util/aio-wait.c b/util/aio-wait.c
> >> index bdb3d3af22..c0a343ac87 100644
> >> --- a/util/aio-wait.c
> >> +++ b/util/aio-wait.c
> >> @@ -35,7 +35,8 @@ static void dummy_bh_cb(void *opaque)
> >>
> >> void aio_wait_kick(void)
> >> {
> >> - /* The barrier (or an atomic op) is in the caller. */
> >> + smp_mb();
> >
> > What is the purpose of the barrier and what does it pair with?
> >
> > I guess we want to make sure that all stores before aio_wait_kick() are
> > visible to the other thread's AIO_WAIT_WHILE() cond expression. that
> > would require smp_wmb(). I'm not sure why it's a full smp_mb() barrier.
>
> I think we need the full smp_mb barrier because we have a read
> afterwards (num_readers) and we want to ensure ordering also for that.
>
> Regarding pairing, yes you are right. I need to also add a smp_mb()
> between the write(num_waiters) and read(condition) in AIO_WAIT_WHILE,
> otherwise it won't work properly.
>
> So we basically have
>
> Caller:
> write(condition)
> aio_wait_kick()
> smp_mb()
> read(num_writers)
>
> AIO_WAIT_WHILE:
> write(num_writers)
> read(condition)
That makes sense to me, thank you! Please include the explanation in the
comments.
Stefan
[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]
next prev parent reply other threads:[~2022-04-30 5:23 UTC|newest]
Thread overview: 67+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-04-26 8:51 [RFC PATCH v2 0/8] Removal of AioContext lock, bs->parents and ->children: new rwlock Emanuele Giuseppe Esposito
2022-04-26 8:51 ` [RFC PATCH v2 1/8] aio_wait_kick: add missing memory barrier Emanuele Giuseppe Esposito
2022-04-28 11:09 ` Stefan Hajnoczi
2022-04-29 8:06 ` Emanuele Giuseppe Esposito
2022-04-30 5:21 ` Stefan Hajnoczi [this message]
2022-04-29 8:12 ` Paolo Bonzini
2022-04-26 8:51 ` [RFC PATCH v2 2/8] coroutine-lock: release lock when restarting all coroutines Emanuele Giuseppe Esposito
2022-04-26 14:59 ` Paolo Bonzini
2022-04-28 11:21 ` Stefan Hajnoczi
2022-04-28 22:14 ` Paolo Bonzini
2022-04-29 9:35 ` Emanuele Giuseppe Esposito
2022-04-26 8:51 ` [RFC PATCH v2 3/8] block: introduce a lock to protect graph operations Emanuele Giuseppe Esposito
2022-04-26 15:00 ` Paolo Bonzini
2022-04-28 13:45 ` Stefan Hajnoczi
2022-04-29 8:37 ` Emanuele Giuseppe Esposito
2022-04-30 5:48 ` Stefan Hajnoczi
2022-05-02 7:54 ` Emanuele Giuseppe Esposito
2022-05-03 10:50 ` Stefan Hajnoczi
2022-04-26 8:51 ` [RFC PATCH v2 4/8] async: register/unregister aiocontext in graph lock list Emanuele Giuseppe Esposito
2022-04-28 13:46 ` Stefan Hajnoczi
2022-04-28 22:19 ` Paolo Bonzini
2022-04-29 8:37 ` Emanuele Giuseppe Esposito
2022-04-26 8:51 ` [RFC PATCH v2 5/8] block.c: wrlock in bdrv_replace_child_noperm Emanuele Giuseppe Esposito
2022-04-26 15:07 ` Paolo Bonzini
2022-04-28 13:55 ` Stefan Hajnoczi
2022-04-29 8:41 ` Emanuele Giuseppe Esposito
2022-04-26 8:51 ` [RFC PATCH v2 6/8] block: assert that graph read and writes are performed correctly Emanuele Giuseppe Esposito
2022-04-28 14:43 ` Stefan Hajnoczi
2022-04-26 8:51 ` [RFC PATCH v2 7/8] graph-lock: implement WITH_GRAPH_RDLOCK_GUARD and GRAPH_RDLOCK_GUARD macros Emanuele Giuseppe Esposito
2022-04-28 15:00 ` Stefan Hajnoczi
2022-04-26 8:51 ` [RFC PATCH v2 8/8] mirror: protect drains in coroutine with rdlock Emanuele Giuseppe Esposito
2022-04-27 6:55 ` [RFC PATCH v2 0/8] Removal of AioContext lock, bs->parents and ->children: new rwlock Emanuele Giuseppe Esposito
2022-04-28 10:45 ` Stefan Hajnoczi
2022-04-28 21:56 ` Emanuele Giuseppe Esposito
2022-04-30 5:17 ` Stefan Hajnoczi
2022-05-02 8:02 ` Emanuele Giuseppe Esposito
2022-05-02 13:15 ` Paolo Bonzini
2022-05-03 8:24 ` Kevin Wolf
2022-05-03 11:04 ` Stefan Hajnoczi
2022-04-28 10:34 ` Stefan Hajnoczi
2022-04-29 8:06 ` Emanuele Giuseppe Esposito
2022-05-04 13:39 ` Stefan Hajnoczi
2022-05-17 10:59 ` Stefan Hajnoczi
2022-05-18 12:28 ` Emanuele Giuseppe Esposito
2022-05-18 12:43 ` Paolo Bonzini
2022-05-18 14:57 ` Stefan Hajnoczi
2022-05-18 16:14 ` Kevin Wolf
2022-05-19 11:27 ` Stefan Hajnoczi
2022-05-19 12:52 ` Kevin Wolf
2022-05-22 15:06 ` Stefan Hajnoczi
2022-05-23 8:48 ` Emanuele Giuseppe Esposito
2022-05-23 13:15 ` Stefan Hajnoczi
2022-05-23 13:54 ` Emanuele Giuseppe Esposito
2022-05-23 13:02 ` Kevin Wolf
2022-05-23 15:13 ` Stefan Hajnoczi
2022-05-23 16:04 ` Kevin Wolf
2022-05-23 16:45 ` Stefan Hajnoczi
2022-05-24 7:55 ` Paolo Bonzini
2022-05-24 8:08 ` Stefan Hajnoczi
2022-05-24 9:17 ` Paolo Bonzini
2022-05-24 10:20 ` Stefan Hajnoczi
2022-05-24 17:25 ` Paolo Bonzini
2022-05-24 10:36 ` Kevin Wolf
2022-05-25 7:41 ` Paolo Bonzini
2022-05-18 14:27 ` Stefan Hajnoczi
2022-05-24 12:10 ` Kevin Wolf
2022-05-25 8:27 ` Emanuele Giuseppe Esposito
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=YmzHdG9a4F+ZjJg7@stefanha-x1.localdomain \
--to=stefanha@redhat.com \
--cc=eesposit@redhat.com \
--cc=fam@euphon.net \
--cc=hreitz@redhat.com \
--cc=jsnow@redhat.com \
--cc=kwolf@redhat.com \
--cc=pbonzini@redhat.com \
--cc=qemu-block@nongnu.org \
--cc=qemu-devel@nongnu.org \
--cc=vsementsov@virtuozzo.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).