[Spice-devel] [PATCH spice-gtk 1/5] audio: emit stop when the channels are reset

[Victor Toso]

Hi,

On Thu, May 07, 2015 at 02:00:24PM +0200, Marc-André Lureau wrote:
> Hi Victor,
>
> It turns out that a "stop" message is sent by spice server during
> migration. Qemu audio_vm_change_state_handler() will disable all audio
> devices on finish migrate.  See backtrace:
> #0  0x00007fffee057cd8 in spice_server_playback_stop (sin=0x555556bb44f8)
> at snd_worker.c:1055
> #1  0x0000555555736240 in line_out_ctl (hw=0x555556bb4470, cmd=<optimized
> out>)    at audio/spiceaudio.c:219
> #2  0x000055555572e874 in audio_vm_change_state_handler
> (opaque=0x555555dccfc0 <glob_audio_state>, running=<opti
> mized out>, state=<optimized out>) at audio/audio.c:1766
> #3  0x0000555555718d0f in vm_state_notify (running=running at entry=0,
> state=state at entry=RUN_STATE_FINISH_MIGRATE)
> at vl.c:1581
> #4  0x000055555562937a in vm_stop (state=RUN_STATE_FINISH_MIGRATE) at
> /usr/src/debug/qemu-2.2.0/cpus.c:613
> #5  0x000055555562937a in vm_stop (state=RUN_STATE_FINISH_MIGRATE) at
> /usr/src/debug/qemu-2.2.0/cpus.c:1300
> #6  0x0000555555704c3b in migration_thread (opaque=0x555555d0ea00
> <current_migration>) at migration.c:610

Good to know, thanks for explaining!

> Playback is re-started on destination. Tbh, I don't think this is a big
> issue as the seamless-migration switch time is usually << 1s. Although you
> can notice a small audio glitch in some cases.
>
> As explained earlier, this patch doesn't create or add another stop signal,
> so this should clear your concerns about it.

Indeed. The series seems fine, so ACK by me as well.

Best,