From: Peter Xu <peterx@redhat.com>
To: Wei Wang <wei.w.wang@intel.com>
Cc: qemu-devel@nongnu.org, virtio-dev@lists.oasis-open.org,
mst@redhat.com, quintela@redhat.com, dgilbert@redhat.com,
pbonzini@redhat.com, liliang.opensource@gmail.com,
nilal@redhat.com, riel@redhat.com
Subject: Re: [Qemu-devel] [PATCH v9 5/8] migration/ram.c: add a notifier chain for precopy
Date: Fri, 30 Nov 2018 13:57:36 +0800 [thread overview]
Message-ID: <20181130055736.GA15865@xz-x1> (raw)
In-Reply-To: <5C00C52F.20606@intel.com>
On Fri, Nov 30, 2018 at 01:05:51PM +0800, Wei Wang wrote:
> On 11/29/2018 01:10 PM, Peter Xu wrote:
> > On Thu, Nov 29, 2018 at 11:40:57AM +0800, Wei Wang wrote:
> > I think this precopy notifier callchain is expected to be used only for
> > the precopy mode. Postcopy has its dedicated notifier callchain that
> > users could use.
> >
> > How about changing the migrate_postcopy() check to "ms->start_postcopy":
> >
> > bool migration_postcopy_start(void)
> > {
> > MigrationState *s;
> >
> > s = migrate_get_current();
> >
> > return atomic_read(&s->start_postcopy);
> > }
> >
> >
> > static void precopy_notify(PrecopyNotifyReason reason)
> > {
> > if (migration_postcopy_start())
> > return;
> >
> > notifier_list_notify(&precopy_notifier_list, &reason);
> > }
> >
> > If postcopy started with precopy, the precopy optimization feature
> > could still be used until it switches to the postcopy mode.
> > I'm not sure we can use start_postcopy. It's a variable being set in
> > the QMP handler but it does not mean postcopy has started. I'm afraid
> > there can be race where it's still precopy but the variable is set so
> > event could be missed...
>
> Peter,
> I just found that migration_bitmap_sync is also called by
> ram_postcopy_send_discard_bitmap().
> But we don't expect notifier_list_notify to be called in the postcopy mode.
>
> So, probably we still need the start_postcopy check in notifier_list_notify
> though we have
> the COMPLETE notifier.
I still don't think using start_postcopy is a good idea, as
explained in my previous reply.
Maybe move the notify()s outside migration_bitmap_sync() but only at
the call sites in precopy? Say these three:
ram_init_bitmaps[3508] migration_bitmap_sync(rs);
ram_save_complete[3731] migration_bitmap_sync(rs);
ram_save_pending[3776] migration_bitmap_sync(rs);
Or you can introduce migration_bitmap_sync_precopy() and let precopy
code call that one instead.
PS. I'm a bit unsure on why we need to sync bitmap in ram_init_bitmaps.
I feel like it can be removed...
Regards,
--
Peter Xu
next prev parent reply other threads:[~2018-11-30 5:58 UTC|newest]
Thread overview: 32+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-11-15 10:07 [Qemu-devel] [PATCH v9 0/8] virtio-balloon: free page hint support Wei Wang
2018-11-15 10:07 ` [Qemu-devel] [PATCH v9 1/8] bitmap: fix bitmap_count_one Wei Wang
2018-11-15 10:07 ` [Qemu-devel] [PATCH v9 2/8] bitmap: bitmap_count_one_with_offset Wei Wang
2018-11-15 10:07 ` [Qemu-devel] [PATCH v9 3/8] migration: use bitmap_mutex in migration_bitmap_clear_dirty Wei Wang
2018-11-27 5:40 ` Peter Xu
2018-11-27 6:02 ` Wei Wang
2018-11-27 6:12 ` [Qemu-devel] [virtio-dev] " Wei Wang
2018-11-27 7:41 ` Peter Xu
2018-11-27 10:17 ` Wei Wang
2018-11-15 10:08 ` [Qemu-devel] [PATCH v9 4/8] migration: API to clear bits of guest free pages from the dirty bitmap Wei Wang
2018-11-27 6:06 ` Peter Xu
2018-11-27 6:52 ` Wei Wang
2018-11-27 7:43 ` Peter Xu
2018-11-15 10:08 ` [Qemu-devel] [PATCH v9 5/8] migration/ram.c: add a notifier chain for precopy Wei Wang
2018-11-27 7:38 ` Peter Xu
2018-11-27 10:25 ` Wei Wang
2018-11-28 5:26 ` Peter Xu
2018-11-28 9:01 ` Wei Wang
2018-11-28 9:32 ` Peter Xu
2018-11-29 3:40 ` Wei Wang
2018-11-29 5:10 ` Peter Xu
2018-11-29 5:47 ` Peter Xu
2018-11-29 6:30 ` Wei Wang
2018-11-30 5:05 ` Wei Wang
2018-11-30 5:57 ` Peter Xu [this message]
2018-11-30 7:09 ` Wei Wang
2018-11-15 10:08 ` [Qemu-devel] [PATCH v9 6/8] migration/ram.c: add a function to disable the bulk stage Wei Wang
2018-11-15 10:08 ` [Qemu-devel] [PATCH v9 7/8] migration: move migrate_postcopy() to include/migration/misc.h Wei Wang
2018-11-15 10:08 ` [Qemu-devel] [PATCH v9 8/8] virtio-balloon: VIRTIO_BALLOON_F_FREE_PAGE_HINT Wei Wang
2018-11-15 18:50 ` [Qemu-devel] [PATCH v9 0/8] virtio-balloon: free page hint support no-reply
2018-11-16 1:38 ` Wei Wang
2018-11-27 3:11 ` Wei Wang
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=20181130055736.GA15865@xz-x1 \
--to=peterx@redhat.com \
--cc=dgilbert@redhat.com \
--cc=liliang.opensource@gmail.com \
--cc=mst@redhat.com \
--cc=nilal@redhat.com \
--cc=pbonzini@redhat.com \
--cc=qemu-devel@nongnu.org \
--cc=quintela@redhat.com \
--cc=riel@redhat.com \
--cc=virtio-dev@lists.oasis-open.org \
--cc=wei.w.wang@intel.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).