[Spice-devel] [spice (stable)] Resending volume value on migration

[Frediano Ziglio]

> 
> Hey,
> 
> On Tue, Mar 07, 2017 at 10:42:09AM -0500, Frediano Ziglio wrote:
> > > Hi,
> > > 
> > > On Tue, Feb 28, 2017 at 11:18:15AM +0100, Victor Toso wrote:
> > > > Hi,
> > > >
> > > > On Tue, Feb 28, 2017 at 04:56:54AM -0500, Frediano Ziglio wrote:
> > > > > > 
> > > > > > From: Victor Toso <me at victortoso.com>
> > > > > > 
> > > > > > Hi,
> > > > > > 
> > > > > > This patch is actually an improvement of an old patch [0]. It fixes
> > > > > > rhbz#1425443
> > > > > > here. I'm sending this for stable branch only as we might have a
> > > > > > better
> > > > > > way
> > > > > > to
> > > > > > do it on the current master as a lot has been improved.
> > > > > > 
> > > > > > [0]
> > > > > > https://lists.freedesktop.org/archives/spice-devel/2015-April/019605.html
> > > > > > [1] https://bugzilla.redhat.com/show_bug.cgi?id=1425443
> > > > > > 
> > > > > > Let me know if this seems good enough for stable branch.
> > > > > > 
> > > > > > Victor Toso (1):
> > > > > >   sound: do not change volume or mute state on migration
> > > > > > 
> > > > > >  server/snd_worker.c | 12 ++++++++++++
> > > > > >  1 file changed, 12 insertions(+)
> > > > > > 
> > > > >
> > > > > Patch looks good.
> > > > > Didn't manage to test (I never managed to try a migration, my fault).
> > > > > What do you mean by stable branch? 0.13.3 or 0.12.8 ?
> > > > 
> > > > 0.12 branch which has the stable releases, 0.12.8 being the last one!
> > > > In git master, this volume-jumps are much improved in my tests but I
> > > > didn't manage to try migration with upstream yet.
> > > >
> > > > Cheers,
> > > >         toso
> > > 
> > > Just to confirm. I just tested spice from git master in the same
> > > environment of rhbz#1425443 and I did not see any volume-jump.
> > > 
> > > Sounds good to have this in the stable branch only?
> > > 
> > >         toso
> > > 
> > 
> > This quite surprise me. I really don't understand how we manage to
> > fix this? Did we fix as regression? Looks like sometimes a regression
> > could be even a good thing :-)
> > Mumble... maybe the old was sending mute+volume while now we just send
> > volume? Or something like that? Can you see if there is a difference
> > in the messages sent by 0.12 and 0.13?
> >
> > Frediano
> 
> I'm glad that you asked :)
> 
> On my first tests I was only paying attention to the volume bar in the
> client and in the guest. After migration I would see it move in the
> client with 0.12.4 but not with 0.13.3.
> 

I assume you are using the same exact Qemu/libvirt/rhv infrastructure,
just changing spice-server.

> Might be a regression, I've filed:
> https://bugs.freedesktop.org/show_bug.cgi?id=100136
> 
> In the logs that I've filed, looking into the client you can see that
> with 0.12.4 we got:
> 
> spice-session.c:1806 main-1:0: migration: all channel migrated, success
> ...
> spice-session.c:1743 no migration in progress
> ...
> spice-channel.c:2330 playback-5:0: unfreeze channel
> ...
> spice-channel.c:2642 playback-5:0: Coroutine exit playback-5:0
> ...
> /* Bellow is the bug that I'm trying to fix with this patch */
> spice-pulse.c:730 record volume changed 65536
> spice-pulse.c:730 record volume changed 65536
> spice-gtk-session.c:203 inputs-3:0: client_modifiers:0x2, guest_modifiers:0x0
> spice-pulse.c:622 playback volume changed 65536
> spice-pulse.c:622 playback volume changed 65536
> ...
> channel-playback.c:371 playback-5:0: playback_handle_start: fmt 1 channels 2
> freq 44100 time 243933442
> 

There is something I don't understand from this log.
Seems that the volume is sent after the migration is completed which seems
nothing wrong to me. But from the original patch it seems that the server
still thinks is migrating.

> --
> 
> Now, the same is not true with 0.13.3+ (upstream) as after `coroutine
> exit playback` we don't receive neither the volume messages nor the
> 

the ? :)

> Cheers,
>         toso
> 

Frediano