From mboxrd@z Thu Jan  1 00:00:00 1970
From: Peter Xu <peterx@redhat.com>
Subject: Re: [RFC PATCH 1/3] balloon: Allow nested inhibits
Date: Thu, 19 Jul 2018 12:45:17 +0800
Message-ID: <20180719044517.GB4071@xz-mi>
References: <20180717222721.14019.27548.stgit@gimli.home>
	<20180717224737.14019.3324.stgit@gimli.home>
	<20180718064015.GI20520@xz-mi> <20180718103736.2406cdc8@t450s.home>
Mime-Version: 1.0
Content-Type: text/plain; charset=utf-8
Cc: qemu-devel@nongnu.org, kvm@vger.kernel.org
To: Alex Williamson <alex.williamson@redhat.com>
Return-path: <qemu-devel-bounces+gceq-qemu-devel2=m.gmane.org@nongnu.org>
Content-Disposition: inline
In-Reply-To: <20180718103736.2406cdc8@t450s.home>
List-Unsubscribe: <https://lists.nongnu.org/mailman/options/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=unsubscribe>
List-Archive: <http://lists.nongnu.org/archive/html/qemu-devel/>
List-Post: <mailto:qemu-devel@nongnu.org>
List-Help: <mailto:qemu-devel-request@nongnu.org?subject=help>
List-Subscribe: <https://lists.nongnu.org/mailman/listinfo/qemu-devel>,
	<mailto:qemu-devel-request@nongnu.org?subject=subscribe>
Errors-To: qemu-devel-bounces+gceq-qemu-devel2=m.gmane.org@nongnu.org
Sender: "Qemu-devel"
	<qemu-devel-bounces+gceq-qemu-devel2=m.gmane.org@nongnu.org>
List-Id: kvm.vger.kernel.org

On Wed, Jul 18, 2018 at 10:37:36AM -0600, Alex Williamson wrote:
> On Wed, 18 Jul 2018 14:40:15 +0800
> Peter Xu <peterx@redhat.com> wrote:
> 
> > On Tue, Jul 17, 2018 at 04:47:37PM -0600, Alex Williamson wrote:
> > > A simple true/false internal state does not allow multiple users.  Fix
> > > this within the existing interface by converting to a counter, so long
> > > as the counter is elevated, ballooning is inhibited.
> > > 
> > > Signed-off-by: Alex Williamson <alex.williamson@redhat.com>
> > > ---
> > >  balloon.c |    7 ++++---
> > >  1 file changed, 4 insertions(+), 3 deletions(-)
> > > 
> > > diff --git a/balloon.c b/balloon.c
> > > index 6bf0a9681377..2a6a7e1a22a0 100644
> > > --- a/balloon.c
> > > +++ b/balloon.c
> > > @@ -37,16 +37,17 @@
> > >  static QEMUBalloonEvent *balloon_event_fn;
> > >  static QEMUBalloonStatus *balloon_stat_fn;
> > >  static void *balloon_opaque;
> > > -static bool balloon_inhibited;
> > > +static int balloon_inhibited;
> > >  
> > >  bool qemu_balloon_is_inhibited(void)
> > >  {
> > > -    return balloon_inhibited;
> > > +    return balloon_inhibited > 0;
> > >  }
> > >  
> > >  void qemu_balloon_inhibit(bool state)
> > >  {
> > > -    balloon_inhibited = state;
> > > +    balloon_inhibited += (state ? 1 : -1);
> > > +    assert(balloon_inhibited >= 0);  
> > 
> > Better do it atomically?
> 
> I'd assumed we're protected by the BQL anywhere this is called.  Is
> that not the case?  Generally when I try to add any sort of locking to
> QEMU it's shot down because the code paths are already serialized.

Ah I see. :-)

But I guess current code might call these without BQL.  For example,
qemu_balloon_inhibit() has:

postcopy_ram_listen_thread
  qemu_loadvm_state_main
    loadvm_process_command
      loadvm_postcopy_handle_listen
        postcopy_ram_enable_notify
          qemu_balloon_inhibit

While the whole stack is inside a standalone thread to load QEMU for
postcopy incoming migration rather than the main thread.

Thanks,

-- 
Peter Xu