The internal API in bluetooth-util should not use the const qualifier
for operations involving a device object. After all, the structure
contains many pointers and thus the const qualifier provides no real
protection.
Instead of repeatedly asking the discovery API to find a transport given
our transport path, let's hold a pointer to the transport and make sure
we remove the reference when the hook is fired reporting that the
transport has been removed. This makes the code easier to follow and
slightly more efficient.
The recently added hook can be used to detect that the transport being
used has been removed. In this case, the profile needs to be set to off.
Additionally, the change fixes a significant problem: without this
transition, the transport could be destroyed while the hook slots (i.e.
nrec_changed_slot) were still set. This led to a double free of these
objects in stop_thread().
The internal API in bluetooth-util should not use the const qualifier
for operations modifying the transport object. This is specially useful
in order to use the available hooks.
If profile could not be successfully initialized, the card should be
set to PROFILE_OFF automatically. If sinks or sources exist, they need
to be destroyed, therefore stop_thread() is called.
Remove stream moving policies from module-bluetooth-device. It is not
clear if such policies are needed at all and in case yes, they should be
implemented in module-bluetooth-policy.
If the card is being set to off profile, it is not necessary to check
if the device exists. This could potentially happen during shutdown,
immediately before the module is unloaded.
It might happen that a PropertyChanged signal is received but the
corresponding card profile has not been created, leading to an assertion
failure in filter_cb() due to inexistent ports. This can happen if BlueZ
misbehaves, or also if the UUIDs are reported later on (i.e. during
pairing discovery). In any case, the signal should just be ignored.
When PA is doing gateway role, let module-suspend-on-idle resume the
audio stream automatically. This will work until the user (or the remote
side, which we also consider user-initiated) suspend the stream
manually.
Card profile hfgw should be no different from the rest, and thus no
internal policy inside module-bluetooth-device should decide to switch
to its profile automatically.
This should be handled by policy modules.
Handle the Playing->Connected transition gracefully by releasing the
transport and setting the sink and sources as suspended. This is
necessary since the IO thread might not encounter a HUP always.
Until today, setting the card to some profile resulted in a transport
acquisition, leading to audio stream setup. This is generally not very
interesting and even undesireable for HFGW use-cases, where the
Gateway role (the remote end) would typically request the SCO link.
Nevertheless, there is no safe way to implement such check without race
conditions, since the BlueZ's state can change between the state report
and the call to Acquire(). The chances for this to reproduce are quite
low though, since interface state changes are relatively slow.
This race condition requires that BlueZ's API is extended in order to
perform the operation atomically, which has already been discussed and
ack-ed in the BlueZ mailing list.
Note that this patch does not introduce a new race condition, since it
already existed before (the PropertyChanged->Acquire race condition,
affecting HFGW use-cases). It is just more explicit now.
If the acquisition of the transport fails, the profile should still be
set. In this case the audio is not actually streaming, so the sink and
source will be created but left suspended.
If the transport needs to be acquired later, for example because the
user wants to route the audio the remote device, the suspend flag should
have to be changed.
Use the port availability flag to expose whether a certain profile is
connected and whether it's doing actual audio streaming.
The proposed mapping is the following:
- Profile disconnected: port is unavailable
- Profile is connected (but not streaming/playing): availability unknown
- Profile is streaming/playing: port is available
The availability-unknown is specially interesting: it involves that if
the sink/source exists (corresponding card profile set), it is currently
in suspended state.
For example, for SCO cases (HFGW or HSP), this means the SCO is down. A
policy module would typically not change this, unless someone is really
trying to use the sink/source. This situation would be nicely handled by
module-suspend-on-idle, which would automatically connect SCO.
On the other hand, if the user wants to control the status of the SCO,
it will still be possible by resuming the sink or source (suspend=0).
This works out-of-the-box since most UIs would show to the user ports
whose availability is unknown.
The configuration of the transport that depends on the MTU should be
performed every time the transport has been acquired, since the
parameters depend on what the Media API provides. This requires to
update the parameters of the sinks and sources as well.
This patch moves this code into a new function that will be called
when the stream is starting (setup_stream), from the IO thread.
This makes the code more robust, since the existing multiple calls to
bt_transport_acquire() do not rely on setup_bt() being able to acquire
the transport.
There should be one port per sink/source so a dummy set_port callback
will be enough.
Adding this callback avoid the "operation not implemented" error
message and additionally makes the module work nicely with
module-switch-on-port-available.
The transport might have disapeared exactly before acquiring, so we
should avoid an assertion failure, in this case inside the function
pa_bluetooth_discovery_get_by_path().
The HFGW source should be consistent with the sink by not setting the
"phone" intended role.
Even though setting this role seems to make sense strictly speaking, the
rest of the codebase doesn't handle this well. Therefore, the audio
coming from a Bluetooth phone can be routed back to the same device.
Make code more readable by introducing the helper function
bt_transport_is_acquired(). This also adds assertions to check whether
the internal state is consistent.
Property bluetooth.protocol did make a distinction between A2DP sink and
source roles but on the contrary did not separate HFP roles (headset vs
gateway). For consistency, they should both behave similarly.
This automatically fixes another incosistency: the HFGW (or HSP) sink
was set to bluetooth.protocol="sco", while the source was set to "hsp".
There is no use for this distinction, since the protocol (including the
role) is the same.
Profile a2dp_source, just like any other card profile, should have
state guards when the profile is being changed. If the BlueZ interface
is not connected, the profile should be set to "off".
This simplifies the code a lot, in favour of the D-Bus Media interface
in BlueZ. The old socket-based IPC mechanism has been deprecated and is
about to be removed soon.
In practice there is always at least one profile, and I
don't think there will ever be cards without profiles.
Therefore, I added assertions to pa_card_new() stating that
the card new data must always contain at least one profile.
Now a lot of code can be simplified, because it's guaranteed
that the profiles hashmap and the active_profile field are
always non-NULL.
PropertyChanged signal of org.BlueZ.MediaTransport is processed in
pa_bluetooth_transport_parse_property() which updates t->nrec.
This is called by :
- First by filter_cb() of bluetooth-util.c
- Then by filter_cb() of module-bluetooth-device.c which retrieve value
of t->nrec before calling parse function, then it checks if t->nrec
has changed before updating bluetooth.nrec property.
As t->nrec has alreday been changed during first process, property
update is never performed.
This patch creates a new hook in pa_bluetooth_transport called
PA_BLUETOOTH_TRANSPORT_HOOK_NREC_CHANGED.
The hook is fired by bluetooth-util.c when the transport's NREC
property changes.
module-bluetooth-device.c won't listen the PropertyChanged signal of
MediaTransport anymore. Instead, it will use the hook in
pa_bluetooth_transport to get a notification when the NREC property
changes, and update the sink or source proplist accordingly.
const qualifier for returned pointer of
pa_bluetooth_discovery_get_transport() is removed.
The bluetooth device should have ports so we can attach a latency to
the ports.
Every profile (a2dp, hsp...) has his own set of ports depending on the
number of sinks and sources it provides.
When a Bluetooth headset is connected only to HFP profile (not connected
to A2DP) and host streams to it, a crash occurs if host disconnects.
When HFP disconnects, audio thread will fail on POLLHUP then generate
a message to set PA profile to Off before ending.
If this message is managed before PA unload bluetooth device module,
all works fine.
But, if this message is managed during module unload, this finish by
re-entrance in release code (stop_thread) and a crash.
This fix prevents to process profile change when module is unloading.
Allow module-bluetooth-device to listens to HandsfreeGateway state
changes using DBUS signals. When an handsfree connects, module-bluetooth-device
is loaded and goes to playing state. When the handsfree disconnect audio,
the card profile is set to "off". If the headset connects audio again after
that, the card profile should switch to "hfgw" again to match state of audio
connection.
