Skip to contentSkip to conversation

Audio: misc fixes for "Audio 20190821 patches"

PATCH Superseded Kővágó, Zoltán 2 participants 9 messages Aug 26, 2019

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#1 lore raw

PATCH 1/3 audio: omitting audiodev= parameter is only deprecated

In reply to Kővágó, Zoltán

Unfortunately, changes introduced in af2041ed2d "audio: audiodev= parameters no longer optional when -audiodev present" breaks backward compatibility. This patch changes the error into a deprecation warning.

---
 qemu-deprecated.texi | 7 +++++++
 audio/audio.c        | 8 ++++----
 2 files changed, 11 insertions(+), 4 deletions(-)
qemu-deprecated.texi +7 -0
index 00a4b6f350..9d74a1cfc0 100644
@@ -72,6 +72,13 @@ backend settings instead of environment variables.  To ease migration to
 the new format, the ``-audiodev-help'' option can be used to convert
 the current values of the environment variables to ``-audiodev'' options.
 
+@subsection Creating sound card devices and vnc without audiodev= property (since 4.2)
+
+When not using the deprecated legacy audio config, each sound card
+should specify an @code{audiodev=} property.  Additionally, when using
+vnc, you should specify an @code{audiodev=} propery if you plan to
+transmit audio through the VNC protocol.
+
 @subsection -mon ...,control=readline,pretty=on|off (since 4.1)
 
 The @code{pretty=on|off} switch has no effect for HMP monitors, but is
audio/audio.c +4 -4
index 7d715332c9..e13addf922 100644
@@ -1412,8 +1412,9 @@ static AudioState *audio_init(Audiodev *dev, const char *name)
         drvname = AudiodevDriver_str(dev->driver);
     } else if (!QTAILQ_EMPTY(&audio_states)) {
         if (!legacy_config) {
-            dolog("You must specify an audiodev= for the device %s\n", name);
-            exit(1);
+            dolog("Device %s: audiodev default parameter is deprecated, please "
+                  "specify audiodev=%s\n", name,
+                  QTAILQ_FIRST(&audio_states)->dev->id);
         }
         return QTAILQ_FIRST(&audio_states);
     } else {
@@ -1548,8 +1549,7 @@ CaptureVoiceOut *AUD_add_capture(
 
     if (!s) {
         if (!legacy_config) {
-            dolog("You must specify audiodev when trying to capture\n");
-            return NULL;
+            dolog("Capturing without setting an audiodev is deprecated\n");
         }
         s = audio_init(NULL, NULL);
     }
Signature · 1 line
2.22.0
#0 lore raw

cover Audio: misc fixes for "Audio 20190821 patches"

Hi,

This series fixes two problems reported by Maxim Levitsky in relation of my multiple audio backend patch series, and a small feature request.

Unfortunately I don't really use PulseAudio nowadays, so I haven't tested it other than making sure it compiles and connects to pa.

Regards,
Zoltan

Kővágó, Zoltán (3):
audio: omitting audiodev= parameter is only deprecated
audio: paaudio: fix client name
audio: paaudio: ability to specify stream name

 qemu-deprecated.texi | 7 +++++++
 qapi/audio.json      | 6 ++++++
 audio/audio.c        | 8 ++++----
 audio/paaudio.c      | 6 +++---
 4 files changed, 20 insertions(+), 7 deletions(-)
Signature · 1 line
2.22.0
#3 lore raw

PATCH 2/3 audio: paaudio: fix client name

pa_context_new expects a client name, not a server socket path.

---
 audio/paaudio.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
audio/paaudio.c +1 -1
index bfef9acaad..777b8e4718 100644
@@ -866,7 +866,7 @@ static void *qpa_conn_init(const char *server)
     }
 
     c->context = pa_context_new(pa_threaded_mainloop_get_api(c->mainloop),
-                                server);
+                                "qemu");
     if (!c->context) {
         goto fail;
     }
Signature · 1 line
2.22.0
#7 lore raw

PATCH 3/3 audio: paaudio: ability to specify stream name

In reply to Kővágó, Zoltán

This can be used to identify stream in tools like pavucontrol when one creates multiple -audiodevs or runs multiple qemu instances.

---
 qapi/audio.json | 6 ++++++
 audio/paaudio.c | 4 ++--
 2 files changed, 8 insertions(+), 2 deletions(-)
qapi/audio.json +6 -0
index 9fefdf5186..a433b3c9d7 100644
@@ -206,6 +206,11 @@
 #
 # @name: name of the sink/source to use
 #
+# @stream-name: name of the PulseAudio stream created by qemu.  Can be
+#               used to identify the stream in PulseAudio when you
+#               create multiple PulseAudio devices or run multiple qemu
+#               instances (default "qemu", since 4.2)
+#
 # @latency: latency you want PulseAudio to achieve in microseconds
 #           (default 15000)
 #
@@ -215,6 +220,7 @@
   'base': 'AudiodevPerDirectionOptions',
   'data': {
     '*name': 'str',
+    '*stream-name': 'str',
     '*latency': 'uint32' } }
 
 ##
audio/paaudio.c +2 -2
index 777b8e4718..827f442b6e 100644
@@ -562,7 +562,7 @@ static int qpa_init_out(HWVoiceOut *hw, struct audsettings *as,
 
     pa->stream = qpa_simple_new (
         c,
-        "qemu",
+        ppdo->has_stream_name ? ppdo->stream_name : "qemu",
         PA_STREAM_PLAYBACK,
         ppdo->has_name ? ppdo->name : NULL,
         &ss,
@@ -630,7 +630,7 @@ static int qpa_init_in(HWVoiceIn *hw, struct audsettings *as, void *drv_opaque)
 
     pa->stream = qpa_simple_new (
         c,
-        "qemu",
+        ppdo->has_stream_name ? ppdo->stream_name : "qemu",
         PA_STREAM_RECORD,
         ppdo->has_name ? ppdo->name : NULL,
         &ss,
Signature · 1 line
2.22.0
Reviewed-by on v1 3/3
#8 lore raw

Re: [Qemu-devel] [PATCH 3/3] audio: paaudio: ability to specify stream name

Quoted 47 lines · Zoltán

This can be used to identify stream in tools like pavucontrol when one creates multiple -audiodevs or runs multiple qemu instances.

---
 qapi/audio.json | 6 ++++++
 audio/paaudio.c | 4 ++--
 2 files changed, 8 insertions(+), 2 deletions(-)
qapi/audio.json +6 -0
index 9fefdf5186..a433b3c9d7 100644
@@ -206,6 +206,11 @@
 #
 # @name: name of the sink/source to use
 #
+# @stream-name: name of the PulseAudio stream created by qemu.  Can be
+#               used to identify the stream in PulseAudio when you
+#               create multiple PulseAudio devices or run multiple qemu
+#               instances (default "qemu", since 4.2)
+#
 # @latency: latency you want PulseAudio to achieve in microseconds
 #           (default 15000)
 #
@@ -215,6 +220,7 @@
   'base': 'AudiodevPerDirectionOptions',
   'data': {
     '*name': 'str',
+    '*stream-name': 'str',
     '*latency': 'uint32' } }
 
 ##
audio/paaudio.c +1 -1
index 777b8e4718..827f442b6e 100644
@@ -562,7 +562,7 @@ static int qpa_init_out(HWVoiceOut *hw, struct audsettings *as,
 
     pa->stream = qpa_simple_new (
         c,
-        "qemu",
+        ppdo->has_stream_name ? ppdo->stream_name : "qemu",
         PA_STREAM_PLAYBACK,
         ppdo->has_name ? ppdo->name : NULL,
         &ss,

audio/paaudio.c static int qpa_init_in(HWVoiceIn *hw, struct audsettings *as, void *drv_opaque) · line 636

Quoted 8 lines
audio/paaudio.c +1 -1
@@ -630,7 +630,7 @@ static int qpa_init_in(HWVoiceIn *hw, struct audsettings *as, void *drv_opaque)
 
     pa->stream = qpa_simple_new (
         c,
-        "qemu",
+        ppdo->has_stream_name ? ppdo->stream_name : "qemu",
         PA_STREAM_RECORD,
         ppdo->has_name ? ppdo->name : NULL,
         &ss,

Tested and works.
Code also seems to be OK.

Best regards,
Maxim Levitsky

Reviewed-by on v1 2/3
#4 lore raw

Re: [Qemu-devel] [PATCH 2/3] audio: paaudio: fix client name

In reply to Kővágó, Zoltán
Quoted 12 lines · Zoltán

pa_context_new expects a client name, not a server socket path.

---
 audio/paaudio.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
audio/paaudio.c +0 -0
index bfef9acaad..777b8e4718 100644

audio/paaudio.c static void *qpa_conn_init(const char *server) · line 872

Quoted 8 lines
audio/paaudio.c +1 -1
@@ -866,7 +866,7 @@ static void *qpa_conn_init(const char *server)
     }
 
     c->context = pa_context_new(pa_threaded_mainloop_get_api(c->mainloop),
-                                server);
+                                "qemu");
     if (!c->context) {
         goto fail;
     }

Also tested, and this works.

May I suggest though to make this configurable as well, for the sake of usability since gnome sound settings show only the client name, and it is per each sound card.
Although on the other thing the client name is qemu.

Best regards,
Maxim Levitsky

#2 lore raw

Re: [Qemu-devel] [PATCH 1/3] audio: omitting audiodev= parameter is only deprecated

In reply to Kővágó, Zoltán
Quoted 45 lines · Zoltán

Unfortunately, changes introduced in af2041ed2d "audio: audiodev= parameters no longer optional when -audiodev present" breaks backward compatibility. This patch changes the error into a deprecation warning.

---
 qemu-deprecated.texi | 7 +++++++
 audio/audio.c        | 8 ++++----
 2 files changed, 11 insertions(+), 4 deletions(-)
qemu-deprecated.texi +7 -0
index 00a4b6f350..9d74a1cfc0 100644
@@ -72,6 +72,13 @@ backend settings instead of environment variables.  To ease migration to
 the new format, the ``-audiodev-help'' option can be used to convert
 the current values of the environment variables to ``-audiodev'' options.
 
+@subsection Creating sound card devices and vnc without audiodev= property (since 4.2)
+
+When not using the deprecated legacy audio config, each sound card
+should specify an @code{audiodev=} property.  Additionally, when using
+vnc, you should specify an @code{audiodev=} propery if you plan to
+transmit audio through the VNC protocol.
+
 @subsection -mon ...,control=readline,pretty=on|off (since 4.1)
 
 The @code{pretty=on|off} switch has no effect for HMP monitors, but is
audio/audio.c +3 -2
index 7d715332c9..e13addf922 100644
@@ -1412,8 +1412,9 @@ static AudioState *audio_init(Audiodev *dev, const char *name)
         drvname = AudiodevDriver_str(dev->driver);
     } else if (!QTAILQ_EMPTY(&audio_states)) {
         if (!legacy_config) {
-            dolog("You must specify an audiodev= for the device %s\n", name);
-            exit(1);
+            dolog("Device %s: audiodev default parameter is deprecated, please "
+                  "specify audiodev=%s\n", name,
+                  QTAILQ_FIRST(&audio_states)->dev->id);
         }
         return QTAILQ_FIRST(&audio_states);
     } else {

audio/audio.c CaptureVoiceOut *AUD_add_capture( · line 1555

Quoted 9 lines
audio/audio.c +1 -2
@@ -1548,8 +1549,7 @@ CaptureVoiceOut *AUD_add_capture(
 
     if (!s) {
         if (!legacy_config) {
-            dolog("You must specify audiodev when trying to capture\n");
-            return NULL;
+            dolog("Capturing without setting an audiodev is deprecated\n");
         }
         s = audio_init(NULL, NULL);
     }

This allowed me to boot th VM with single audiodev without specifying the audiodev in the device, but on shutdown qemu crashes with heap corruption sadly.

-audiodev pa,id=snd1,server=/run/user/103992/pulse/native -device ich9-intel-hda,id=sound0,msi=on
-device hda-micro,id=sound0-codec0,bus=sound0.0,cad=0

In qemu output:
free(): invalid next size (fast)

Best regards,
Maxim Levitsky

#5 lore raw

Re: [Qemu-devel] [PATCH 2/3] audio: paaudio: fix client name

In reply to Maxim Levitsky
Quoted 24 lines · Maxim Levitsky
Quoted 12 lines · Zoltán

pa_context_new expects a client name, not a server socket path.

---
 audio/paaudio.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
audio/paaudio.c +0 -0
index bfef9acaad..777b8e4718 100644

audio/paaudio.c static void *qpa_conn_init(const char *server) · line 872

Quoted 8 lines
audio/paaudio.c +1 -1
@@ -866,7 +866,7 @@ static void *qpa_conn_init(const char *server)
     }
 
     c->context = pa_context_new(pa_threaded_mainloop_get_api(c->mainloop),
-                                server);
+                                "qemu");
     if (!c->context) {
         goto fail;
     }

Also tested, and this works.

Quoted 4 lines

May I suggest though to make this configurable as well, for the sake of usability since gnome sound settings show only the client name, and it is per each sound card.
Although on the other thing the client name is qemu.

There is a small problem with that. Currently we only open one connection to pa, even with multiple -audiodevs (they will just create different streams), which means we can only use a single client name per qemu process. Because of that, I wouldn't turn this into an audiodev property. Some other kind of global setting could work, but I'm not sure whether it's worth it or not.

Regards,
Zoltan

Quoted 7 lines

Best regards,
Maxim Levitsky

#6 lore raw

Re: [Qemu-devel] [PATCH 2/3] audio: paaudio: fix client name

Quoted 37 lines · Zoltán Kővágó
Quoted 24 lines · Maxim Levitsky
Quoted 12 lines · Zoltán

pa_context_new expects a client name, not a server socket path.

---
 audio/paaudio.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
audio/paaudio.c +0 -0
index bfef9acaad..777b8e4718 100644

audio/paaudio.c static void *qpa_conn_init(const char *server) · line 872

Quoted 8 lines
audio/paaudio.c +1 -1
@@ -866,7 +866,7 @@ static void *qpa_conn_init(const char *server)
     }
 
     c->context = pa_context_new(pa_threaded_mainloop_get_api(c->mainloop),
-                                server);
+                                "qemu");
     if (!c->context) {
         goto fail;
     }

Also tested, and this works.

Quoted 4 lines

May I suggest though to make this configurable as well, for the sake of usability since gnome sound settings show only the client name, and it is per each sound card.
Although on the other thing the client name is qemu.

There is a small problem with that. Currently we only open one connection to pa, even with multiple -audiodevs (they will just create different streams), which means we can only use a single client name per qemu process. Because of that, I wouldn't turn this into an audiodev property. Some other kind of global setting could work, but I'm not sure whether it's worth it or not.

Quoted 2 lines

Regards,
Zoltan

All right.
We could use the VM name for that though, so that at least multiple VMs would show up as different client.

Best regards,
Maxim Levitsky