Comment 6 for bug 1941977

So why did Bluez initiate A2DP even when the headset was the one that established the ACL?

The bluetoothd logs help out here:

Aug 29 00:54:47 RhoPC bluetoothd[4046739]: src/adapter.c:connected_callback() hci0 device FC:58:FA:7F:E5:18 connected eir_len 5
Aug 29 00:54:47 RhoPC bluetoothd[4046739]: src/profile.c:ext_confirm() incoming connect from FC:58:FA:7F:E5:18
Aug 29 00:54:47 RhoPC bluetoothd[4046739]: src/service.c:btd_service_ref() 0x562f6441d710: ref=2
Aug 29 00:54:47 RhoPC bluetoothd[4046739]: src/profile.c:ext_confirm() Headset Voice gateway authorizing connection from FC:58:FA:7F:E5:18
Aug 29 00:54:47 RhoPC bluetoothd[4046739]: src/profile.c:ext_auth() FC:58:FA:7F:E5:18 authorized to connect to Headset Voice gateway
Aug 29 00:54:47 RhoPC bluetoothd[4046739]: src/profile.c:ext_connect() Headset Voice gateway connected to FC:58:FA:7F:E5:18
Aug 29 00:54:47 RhoPC bluetoothd[4046739]: src/service.c:change_state() 0x562f6441d710: device FC:58:FA:7F:E5:18 profile Headset Voice gateway state changed: disconnected -> connecting (0)
Aug 29 00:54:47 RhoPC bluetoothd[4046739]: src/service.c:change_state() 0x562f6441d710: device FC:58:FA:7F:E5:18 profile Headset Voice gateway state changed: connecting -> connected (0)
Aug 29 00:54:47 RhoPC bluetoothd[4046739]: src/device.c:device_profile_connected() Headset Voice gateway Success (0)
Aug 29 00:54:47 RhoPC bluetoothd[4046739]: plugins/policy.c:policy_connect() /org/bluez/hci0/dev_FC_58_FA_7F_E5_18 profile a2dp-sink
Aug 29 00:54:47 RhoPC bluetoothd[4046739]: profiles/audio/a2dp.c:a2dp_sink_connect() path /org/bluez/hci0/dev_FC_58_FA_7F_E5_18

One can easily see at the end there, where something in plugins/policy.c kicks off A2DP connection, apparently as soon as it got word that HFP had connected.

Checking that file, the offending code seems to be this in hs_cb():

case BTD_SERVICE_STATE_CONNECTED:
  /* Check if service initiate the connection then proceed
   * immediately otherwise set timer
   */
  if (old_state == BTD_SERVICE_STATE_CONNECTING)
   policy_connect(data, sink);
  else if (btd_service_get_state(sink) !=
      BTD_SERVICE_STATE_CONNECTED)
   policy_set_sink_timer(data);
  break;

The reasoning is sound though -- if the HFP connection was locally initiated i.e. previously in STATE_CONNECTING, then go ahead and connect A2DP right now. Otherwise, respect right-of-way and set a timer to connect it in around 2s, in case the headset somehow fails to do so.

The problem is, this HFP connection was NOT locally initiated. So why was policy.c informed that it is?