audio/avrcp: Always update transport volume regardless of player

Volume is a special property that not only exists on players but also on the transport (see org.bluez.MediaTransport1). A player is not attached when the controller does not support FEATURE_CATEGORY_1, which is common on headphones without media browsing capabilities.

On such audio devices (headphones, in-ears and the like) Absolute Volume is not available unless an external player is registered (org.bluez.Media1.RegisterPlayer) and the device sends a volume event back after that to set a2dp->volume in transport.c to a valid value (causing volume_exists to finally return true).

This [1] mail thread denoting the same issue has a solution to at least request capabilities from the controller, but the proposed player object is not created on category 2 devices. Any notifications received on AVRCP_EVENT_VOLUME_CHANGED (avrcp_volume_changed) that is subsequently registered, or handling the result of avrcp_set_volume in avrcp_handle_set_volume will be ignored unless said player is present.

This issue is not addressed by adding a fake player but instead dealing with the fact that volume is "special" and available on the transport regardless of the existence of a player. This is confirmed in avrcp_get_capabilities_resp as well which requires a player to register any event except AVRCP_EVENT_VOLUME_CHANGED.

The applied solution moves media_transport_update_device_volume out of the player and into avrcp_volume_changed/avrcp_handle_set_volume where it is unconditionally called. These functions are the only users of avrcp_player->set_volume.

Note that the volume member of media_player is never used which seems a result of updating from org.bluez.MediaPlayer1 to org.mpris.MediaPlayer2.Player in 15e421737ccc4696ed567edcc24d178aedb47854, where the volume property [2] is left out.

[1]: https://marc.info/?l=linux-bluetooth&m=145337574806153
[2]: https://specifications.freedesktop.org/mpris-spec/2.2/Player_Interface.html#Property:Volume

---
Hi,

This is is a resend from an earlier mail that didn't comply with the contributor guidelines. Seeing that the topic of AVRCP volume is brought up recently it is about time to repair it and hereby send it again.

I still have an incomplete patch lying around that synchronizes Volume on org.mpris.MediaPlayer2.Player back when this patch was written 6 months ago. It'll require some time to get back in to it and finalize it, let me know if that's desired.

- Marijn Suijten

 profiles/audio/avrcp.c | 12 ++++++++----
 profiles/audio/media.c | 16 ----------------
 2 files changed, 8 insertions(+), 20 deletions(-)

profiles/audio/avrcp.c +8 -4

index e2428250e..8370c8a44 100644
@@ -3625,12 +3625,13 @@ static void avrcp_volume_changed(struct avrcp *session,
        struct avrcp_player *player = target_get_player(session);
        uint8_t volume;
 
-       if (!player)
-               return;
-
        volume = pdu->params[1] & 0x7F;
 
-       player->cb->set_volume(volume, session->dev, player->user_data);
+       /* Always update the transport volume, which is separate from the player */
+       media_transport_update_device_volume(session->dev, volume);
+
+       if (player)
+               player->cb->set_volume(volume, session->dev, player->user_data);
 }
 
 static void avrcp_status_changed(struct avrcp *session,

@@ -4378,6 +4379,9 @@ static gboolean avrcp_handle_set_volume(struct avctp *conn, uint8_t code,
 
        volume = pdu->params[0] & 0x7F;
 
+       /* Always update the transport volume, which is separate from the player */
+       media_transport_update_device_volume(session->dev, volume);
+
        if (player != NULL)
                player->cb->set_volume(volume, session->dev, player->user_data);
 

profiles/audio/media.c +0 -17

index 993ecb3b3..a0173fdd4 100644
@@ -1202,27 +1202,11 @@ static uint32_t get_duration(void *user_data)
 static void set_volume(uint8_t volume, struct btd_device *dev, void *user_data)
 {
        struct media_player *mp = user_data;
-       GSList *l;
 
        if (mp->volume == volume)
                return;
 
        mp->volume = volume;
-
-       for (l = mp->adapter->endpoints; l; l = l->next) {
-               struct media_endpoint *endpoint = l->data;
-               struct media_transport *transport;
-
-               /* Volume is A2DP only */
-               if (endpoint->sep == NULL)
-                       continue;
-
-               transport = find_device_transport(endpoint, dev);
-               if (transport == NULL)
-                       continue;
-
-               media_transport_update_volume(transport, volume);
-       }
 }
 
 static bool media_player_send(struct media_player *mp, const char *name)
--

2.27.0

Hi Marijn,

Quoted 30 lines · Marijn Suijten

Volume is a special property that not only exists on players but also on the transport (see org.bluez.MediaTransport1). A player is not attached when the controller does not support FEATURE_CATEGORY_1, which is common on headphones without media browsing capabilities.

On such audio devices (headphones, in-ears and the like) Absolute Volume is not available unless an external player is registered (org.bluez.Media1.RegisterPlayer) and the device sends a volume event back after that to set a2dp->volume in transport.c to a valid value (causing volume_exists to finally return true).

This [1] mail thread denoting the same issue has a solution to at least request capabilities from the controller, but the proposed player object is not created on category 2 devices. Any notifications received on AVRCP_EVENT_VOLUME_CHANGED (avrcp_volume_changed) that is subsequently registered, or handling the result of avrcp_set_volume in avrcp_handle_set_volume will be ignored unless said player is present.

This issue is not addressed by adding a fake player but instead dealing with the fact that volume is "special" and available on the transport regardless of the existence of a player. This is confirmed in avrcp_get_capabilities_resp as well which requires a player to register any event except AVRCP_EVENT_VOLUME_CHANGED.

The applied solution moves media_transport_update_device_volume out of the player and into avrcp_volume_changed/avrcp_handle_set_volume where it is unconditionally called. These functions are the only users of avrcp_player->set_volume.

Quoted 5 lines

Note that the volume member of media_player is never used which seems a result of updating from org.bluez.MediaPlayer1 to org.mpris.MediaPlayer2.Player in 15e421737ccc4696ed567edcc24d178aedb47854, where the volume property [2] is left out.

This is actually on purpose since the volume notification indicates the volume at the device side we don't want to change the player volume as well as it would most likely result in duplicating the volume change both at sink and source, note that the likes of handling the transport like pulseaudio would already notify the device volume level change without actually applying any change to the volume locally.

Quoted 84 lines

[1]: https://marc.info/?l=linux-bluetooth&m=145337574806153
[2]: https://specifications.freedesktop.org/mpris-spec/2.2/Player_Interface.html#Property:Volume

---
Hi,

This is is a resend from an earlier mail that didn't comply with the contributor guidelines. Seeing that the topic of AVRCP volume is brought up recently it is about time to repair it and hereby send it again.

I still have an incomplete patch lying around that synchronizes Volume on org.mpris.MediaPlayer2.Player back when this patch was written 6 months ago. It'll require some time to get back in to it and finalize it, let me know if that's desired.

- Marijn Suijten

 profiles/audio/avrcp.c | 12 ++++++++----
 profiles/audio/media.c | 16 ----------------
 2 files changed, 8 insertions(+), 20 deletions(-)

profiles/audio/avrcp.c +8 -4

index e2428250e..8370c8a44 100644
@@ -3625,12 +3625,13 @@ static void avrcp_volume_changed(struct avrcp *session,
        struct avrcp_player *player = target_get_player(session);
        uint8_t volume;
 
-       if (!player)
-               return;
-
        volume = pdu->params[1] & 0x7F;
 
-       player->cb->set_volume(volume, session->dev, player->user_data);
+       /* Always update the transport volume, which is separate from the player */
+       media_transport_update_device_volume(session->dev, volume);
+
+       if (player)
+               player->cb->set_volume(volume, session->dev, player->user_data);
 }
 
 static void avrcp_status_changed(struct avrcp *session,

@@ -4378,6 +4379,9 @@ static gboolean avrcp_handle_set_volume(struct avctp *conn, uint8_t code,
 
        volume = pdu->params[0] & 0x7F;
 
+       /* Always update the transport volume, which is separate from the player */
+       media_transport_update_device_volume(session->dev, volume);
+
        if (player != NULL)
                player->cb->set_volume(volume, session->dev, player->user_data);
 

profiles/audio/media.c +0 -16

index 993ecb3b3..a0173fdd4 100644
@@ -1202,27 +1202,11 @@ static uint32_t get_duration(void *user_data)
 static void set_volume(uint8_t volume, struct btd_device *dev, void *user_data)
 {
        struct media_player *mp = user_data;
-       GSList *l;
 
        if (mp->volume == volume)
                return;
 
        mp->volume = volume;
-
-       for (l = mp->adapter->endpoints; l; l = l->next) {
-               struct media_endpoint *endpoint = l->data;
-               struct media_transport *transport;
-
-               /* Volume is A2DP only */
-               if (endpoint->sep == NULL)
-                       continue;
-
-               transport = find_device_transport(endpoint, dev);
-               if (transport == NULL)
-                       continue;
-
-               media_transport_update_volume(transport, volume);
-       }
 }
 

Quoted 4 lines

profiles/audio/media.c +0 -1

 static bool media_player_send(struct media_player *mp, const char *name)
--

2.27.0

Hi Marijn,

Quoted 47 lines · Luiz Augusto von Dentz

Hi Marijn,

Quoted 30 lines · Marijn Suijten

Volume is a special property that not only exists on players but also on the transport (see org.bluez.MediaTransport1). A player is not attached when the controller does not support FEATURE_CATEGORY_1, which is common on headphones without media browsing capabilities.

On such audio devices (headphones, in-ears and the like) Absolute Volume is not available unless an external player is registered (org.bluez.Media1.RegisterPlayer) and the device sends a volume event back after that to set a2dp->volume in transport.c to a valid value (causing volume_exists to finally return true).

This [1] mail thread denoting the same issue has a solution to at least request capabilities from the controller, but the proposed player object is not created on category 2 devices. Any notifications received on AVRCP_EVENT_VOLUME_CHANGED (avrcp_volume_changed) that is subsequently registered, or handling the result of avrcp_set_volume in avrcp_handle_set_volume will be ignored unless said player is present.

This issue is not addressed by adding a fake player but instead dealing with the fact that volume is "special" and available on the transport regardless of the existence of a player. This is confirmed in avrcp_get_capabilities_resp as well which requires a player to register any event except AVRCP_EVENT_VOLUME_CHANGED.

The applied solution moves media_transport_update_device_volume out of the player and into avrcp_volume_changed/avrcp_handle_set_volume where it is unconditionally called. These functions are the only users of avrcp_player->set_volume.

Quoted 5 lines

Note that the volume member of media_player is never used which seems a result of updating from org.bluez.MediaPlayer1 to org.mpris.MediaPlayer2.Player in 15e421737ccc4696ed567edcc24d178aedb47854, where the volume property [2] is left out.

This is actually on purpose since the volume notification indicates the volume at the device side we don't want to change the player volume as well as it would most likely result in duplicating the volume change both at sink and source, note that the likes of handling the transport like pulseaudio would already notify the device volume level change without actually applying any change to the volume locally.

Quoted 84 lines

Quoted 84 lines

[1]: https://marc.info/?l=linux-bluetooth&m=145337574806153
[2]: https://specifications.freedesktop.org/mpris-spec/2.2/Player_Interface.html#Property:Volume

---
Hi,

This is is a resend from an earlier mail that didn't comply with the contributor guidelines. Seeing that the topic of AVRCP volume is brought up recently it is about time to repair it and hereby send it again.

I still have an incomplete patch lying around that synchronizes Volume on org.mpris.MediaPlayer2.Player back when this patch was written 6 months ago. It'll require some time to get back in to it and finalize it, let me know if that's desired.

- Marijn Suijten

 profiles/audio/avrcp.c | 12 ++++++++----
 profiles/audio/media.c | 16 ----------------
 2 files changed, 8 insertions(+), 20 deletions(-)

profiles/audio/avrcp.c +8 -4

index e2428250e..8370c8a44 100644
@@ -3625,12 +3625,13 @@ static void avrcp_volume_changed(struct avrcp *session,
        struct avrcp_player *player = target_get_player(session);
        uint8_t volume;
 
-       if (!player)
-               return;
-
        volume = pdu->params[1] & 0x7F;
 
-       player->cb->set_volume(volume, session->dev, player->user_data);
+       /* Always update the transport volume, which is separate from the player */
+       media_transport_update_device_volume(session->dev, volume);
+
+       if (player)
+               player->cb->set_volume(volume, session->dev, player->user_data);
 }
 
 static void avrcp_status_changed(struct avrcp *session,

@@ -4378,6 +4379,9 @@ static gboolean avrcp_handle_set_volume(struct avctp *conn, uint8_t code,
 
        volume = pdu->params[0] & 0x7F;
 
+       /* Always update the transport volume, which is separate from the player */
+       media_transport_update_device_volume(session->dev, volume);
+
        if (player != NULL)
                player->cb->set_volume(volume, session->dev, player->user_data);
 

profiles/audio/media.c +0 -16

index 993ecb3b3..a0173fdd4 100644
@@ -1202,27 +1202,11 @@ static uint32_t get_duration(void *user_data)
 static void set_volume(uint8_t volume, struct btd_device *dev, void *user_data)
 {
        struct media_player *mp = user_data;
-       GSList *l;
 
        if (mp->volume == volume)
                return;
 
        mp->volume = volume;
-
-       for (l = mp->adapter->endpoints; l; l = l->next) {
-               struct media_endpoint *endpoint = l->data;
-               struct media_transport *transport;
-
-               /* Volume is A2DP only */
-               if (endpoint->sep == NULL)
-                       continue;
-
-               transport = find_device_transport(endpoint, dev);
-               if (transport == NULL)
-                       continue;
-
-               media_transport_update_volume(transport, volume);
-       }
 }
 

Quoted 4 lines

Quoted 4 lines

profiles/audio/media.c +0 -1

 static bool media_player_send(struct media_player *mp, const char *name)
--

2.27.0

Applied, Thanks. Note that I did fix some coding style problems.

--
Luiz Augusto von Dentz