[Qemu-devel] dsound: warning: Voice is already playing

[Qemu-devel] dsound: warning: Voice is already playing

[Robert Riebisch]

Hi!

I'm getting "dsound: warning: Voice is already playing", when starting
QEMU (VM still at POST screen stage). I built my own Win32 version of
QEMU SVN r6626 using MinGW.

I used --audio-card-list="ac97 es1370 sb16 cs4231a adlib gus",
--audio-drv-list="dsound sdl fmod", and "-soundhw all".

Robert Riebisch
-- 
BTTR Software
http://www.bttr-software.de/

Re: [Qemu-devel] dsound: warning: Voice is already playing

[malc]

On Mon, 16 Feb 2009, Robert Riebisch wrote:

> Hi!
> 
> I'm getting "dsound: warning: Voice is already playing", when starting
> QEMU (VM still at POST screen stage). I built my own Win32 version of
> QEMU SVN r6626 using MinGW.

What is the exact sequence of events that leads to it?
 
> I used --audio-card-list="ac97 es1370 sb16 cs4231a adlib gus",
> --audio-drv-list="dsound sdl fmod", and "-soundhw all".
> 
> Robert Riebisch
> 

-- 
mailto:av1474@comtv.ru

Re: [Qemu-devel] dsound: warning: Voice is already playing

[Robert Riebisch]

malc wrote:

>> I'm getting "dsound: warning: Voice is already playing", when starting
>> QEMU (VM still at POST screen stage). I built my own Win32 version of
>> QEMU SVN r6626 using MinGW.
> 
> What is the exact sequence of events that leads to it?

What do you mean by "sequence"?

Robert Riebisch
-- 
BTTR Software
http://www.bttr-software.de/

Re: [Qemu-devel] dsound: warning: Voice is already playing

[malc]

On Mon, 16 Feb 2009, Robert Riebisch wrote:

> malc wrote:
> 
> >> I'm getting "dsound: warning: Voice is already playing", when starting
> >> QEMU (VM still at POST screen stage). I built my own Win32 version of
> >> QEMU SVN r6626 using MinGW.
> > 
> > What is the exact sequence of events that leads to it?
> 
> What do you mean by "sequence"?

Step by step actions after which the following message is seen, including
all command line parameters.

-- 
mailto:av1474@comtv.ru

Re: [Qemu-devel] dsound: warning: Voice is already playing

[Robert Riebisch]

malc wrote:

>> > What is the exact sequence of events that leads to it?
>> 
>> What do you mean by "sequence"?
> 
> Step by step actions after which the following message is seen, including
> all command line parameters.

Build QEMU using:
MSYS-1.0.10.exe
msysDTK-1.0.1.exe +
binutils-2.17.50-20060824-1.tar.gz
gcc-core-3.4.5-20060117-3.tar.gz
gcc-g++-3.4.5-20060117-3.tar.gz
mingw32-make-3.81-20080326-2.tar.gz
mingw-runtime-3.14.tar.gz
mingw-utils-0.3.tar.gz
w32api-3.13-mingw32-dev.tar.gz +
fmodapi375win.zip

./configure --prefix="C:/Programme/qemu-trunk"
--target-list="i386-softmmu" --audio-card-list="ac97 es1370 sb16 cs4231a
adlib gus" --audio-drv-list="dsound sdl fmod"
--fmod-inc="$FMOD_PATH/api/inc"
--fmod-lib="$FMOD_PATH/api/lib/libfmod.a" && make && make install

Run QEMU with "C:\Programme\qemu-trunk\qemu.exe -fda empty.img -soundhw
all -localtime -no-kqemu" on Windows 2000 Professional SP4.

I narrowed it down to GUS emulation, because this warning also appears,
when I just use "-soundhw gus" (not "all").

Robert Riebisch
-- 
BTTR Software
http://www.bttr-software.de/

Re: [Qemu-devel] dsound: warning: Voice is already playing

[malc]

On Tue, 17 Feb 2009, Robert Riebisch wrote:

> malc wrote:
> 
> >> > What is the exact sequence of events that leads to it?
> >> 
> >> What do you mean by "sequence"?
> > 
> > Step by step actions after which the following message is seen, including
> > all command line parameters.
> 

[..snip..]

> 
> I narrowed it down to GUS emulation, because this warning also appears,
> when I just use "-soundhw gus" (not "all").
> 

It's actually fallout from r6402:
"Rework vm_state_change notifiers (Jan Kiszka)"

GUS, unlike all other cards emulated by QEMU, is mixing sound at all times
and as such it calls AUD_set_active_out during intialization stage which
happens before vm_start is called and when it does audio vm change state 
notification callback is invoked which tries to enable underlying audio
subsystem's voice, but it is already enabled and in case of dsound will,
rightfully, complain.

Jan, can you please check that this doesn't break Musicpal or the other
ARM stuff you have written and i have no means to test:

diff --git a/audio/audio.c b/audio/audio.c
index b0a5f3b..a2636cb 100644
--- a/audio/audio.c
+++ b/audio/audio.c
@@ -1127,6 +1127,7 @@ void AUD_set_active_out (SWVoiceOut *sw, int on)
 
     hw = sw->hw;
     if (sw->active != on) {
+        AudioState *s = &glob_audio_state;
         SWVoiceOut *temp_sw;
         SWVoiceCap *sc;
 
@@ -1134,7 +1135,9 @@ void AUD_set_active_out (SWVoiceOut *sw, int on)
             hw->pending_disable = 0;
             if (!hw->enabled) {
                 hw->enabled = 1;
-                hw->pcm_ops->ctl_out (hw, VOICE_ENABLE);
+                if (s->vm_running) {
+                    hw->pcm_ops->ctl_out (hw, VOICE_ENABLE);
+                }
             }
         }
         else {
@@ -1170,12 +1173,15 @@ void AUD_set_active_in (SWVoiceIn *sw, int on)
 
     hw = sw->hw;
     if (sw->active != on) {
+        AudioState *s = &glob_audio_state;
         SWVoiceIn *temp_sw;
 
         if (on) {
             if (!hw->enabled) {
                 hw->enabled = 1;
-                hw->pcm_ops->ctl_in (hw, VOICE_ENABLE);
+                if (s->vm_running) {
+                    hw->pcm_ops->ctl_in (hw, VOICE_ENABLE);
+                }
             }
             sw->total_hw_samples_acquired = hw->total_samples_captured;
         }
@@ -1623,6 +1629,7 @@ static void audio_vm_change_state_handler (void *opaque, int running,
     HWVoiceIn *hwi = NULL;
     int op = running ? VOICE_ENABLE : VOICE_DISABLE;
 
+    s->vm_running = running;
     while ((hwo = audio_pcm_hw_find_any_enabled_out (s, hwo))) {
         hwo->pcm_ops->ctl_out (hwo, op);
     }
diff --git a/audio/audio_int.h b/audio/audio_int.h
index c1da710..fc87458 100644
--- a/audio/audio_int.h
+++ b/audio/audio_int.h
@@ -192,6 +192,7 @@ struct AudioState {
     LIST_HEAD (cap_listhead, CaptureVoiceOut) cap_head;
     int nb_hw_voices_out;
     int nb_hw_voices_in;
+    int vm_running;
 };
 
 extern struct audio_driver no_audio_driver;


-- 
mailto:av1474@comtv.ru

Re: [Qemu-devel] dsound: warning: Voice is already playing

[Jan Kiszka]

[-- Attachment #1: Type: text/plain, Size: 1212 bytes --]

malc wrote:
> On Tue, 17 Feb 2009, Robert Riebisch wrote:
> 
>> malc wrote:
>>
>>>>> What is the exact sequence of events that leads to it?
>>>> What do you mean by "sequence"?
>>> Step by step actions after which the following message is seen, including
>>> all command line parameters.
> 
> [..snip..]
> 
>> I narrowed it down to GUS emulation, because this warning also appears,
>> when I just use "-soundhw gus" (not "all").
>>
> 
> It's actually fallout from r6402:
> "Rework vm_state_change notifiers (Jan Kiszka)"
> 
> GUS, unlike all other cards emulated by QEMU, is mixing sound at all times
> and as such it calls AUD_set_active_out during intialization stage which
> happens before vm_start is called and when it does audio vm change state 
> notification callback is invoked which tries to enable underlying audio
> subsystem's voice, but it is already enabled and in case of dsound will,
> rightfully, complain.
> 
> Jan, can you please check that this doesn't break Musicpal or the other
> ARM stuff you have written and i have no means to test:

Sounds good, at least on the MusicPal (I haven't written the "other
stuff" nor do I have test setups at hand).

Jan


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 257 bytes --]

Re: [Qemu-devel] dsound: warning: Voice is already playing

[malc]

On Wed, 18 Feb 2009, Jan Kiszka wrote:

> malc wrote:

[..snip..]

> > Jan, can you please check that this doesn't break Musicpal or the other
> > ARM stuff you have written and i have no means to test:
> 
> Sounds good, at least on the MusicPal (I haven't written the "other
> stuff" nor do I have test setups at hand).

Thanks.

-- 
mailto:av1474@comtv.ru