When we get buffers from the driver, check if we have at least as many
as we requested. If we have more, that's ok, we will simply not queue
them.
Previously we would ignore the input number of buffers and use the
allocated amount to fill up the buffer array, which might be too small
and then we crash.
Just like the optional build, make all field parsing optional. This
will leave the fields with their default values if they are not parsed
from the param.
We can then remove our custom functions and use the generic ones in
various places.
Add a new seq field in the param-info struct. Users can use this
field to keep track of pending param updates.
Store the latest seq number of the param update in the seq field. Remove
all params that don't match the sequence number because they are too
old. This avoids duplicate old params in pw-dump output.
Rework the pulseaudio manager with this same method.
Set up the adapter and then set the started field to true before we
Start the follower and converter. It is possible that the follower will
start processing immediately so the started state needs to be set
correctly.
Similarly, first perform the Pause and Suspend calls on the follower
and converter before clearing our buffers and format so that things
are stopped properly.
See #2764
AVDTP in principle allows 62 endpoints, but in practice it appears some
devices (Samsung Galaxy Buds2 Pro, Redmi Buds 3 Lite, probably others)
fail to connect A2DP when the number is somewhere above 24. A2DP
connection works when initiated from the Central, but not when the
device itself does it, so these devices are not fully broken. We should
reduce the number of registered A2DP endpoints to avoid running into
problems with such broken devices.
Some of our source codecs are the same actual codec with the same
configuration, and don't need separate source endpoints.
Allow codecs to not have a registered endpoint (fill_caps == NULL), and
tolerate codecs with the same endpoint name. In codec switch, keep
track separately which of the codecs with the same endpoint name the
local endpoint is currently associated with.
Pick the first format in the list of supported as something closest
to 640x480 instead of the lowest possible resolution.
Applications that don't suggest a default size will then get something
more sensible than a poster frame.
Make it possible for a aec-plugin to provide props/params
through module-echo-cancel and make it possible for a
aec-plugin to listen and react to these props/params getting updated.
Add method enum_props() that is used under setup_streams() to
get props from aec_plugin.
Add get_params() and set_params() for updating and getting
aec-plugin params under the input/output_param_changed calls
and getting params under setup_streams().
The graph cycle goes: driver timeout -> process output nodes -> process
driver node. Hence, driver should produce buffers in the timeout,
otherwise there's one quantum extra latency.
Make the bluez5 media/sco sources as drivers put a buffer to io before
indicating ready, and as follower do it in process. Also make checks if
io == NULL, and don't set io->status to HAVE_DATA unless there really is
a buffer ready.
Introduce `unregister_media_endpoint()` to unregister the specified
media endpoint object from dbus. Moreover, move the logic that decides
whether or not a particular codec should be registered for a given
direction into `endpoint_should_be_registered()` and use that in
both `(un)register_media_endpoint()`.
Driver timeouts need to be started/stopped when we switch from follower
to driver or vice versa.
The BT sources fail to do this, so fix it. Sinks already do it right.
Use separate timers for driving graph and for flushing, since they don't
have the same period.
Flushing is done based on the time positions of the next sample to be
written, so it will stay in sync with the graph. Because writing too
much data to SCO sockets generally only causes the device to skip ahead,
we don't need to handle the case where the writing has been lagging.
This fixes simultaneous playback to both ALSA and SCO sinks from the
same graph, with SCO as driver, which previously produces broken sound
(e.g. with pw-play --latency 512, linked to the two sinks) ALSA nodes
require regular driver intervals, which was not true previously.
Bigger buffer allows for more fluctuation in transmission rate without
sound glitches.
It doesn't matter much for latency, as under normal conditions we are
not producing data faster than the BT adapter can transmit, so the
buffer generally is almost always empty or full, and in the latter case
we have to reduce the bitrate.
Do not start the playback device until there is data to play. Otherwise
time consuming configuration of other nodes (such as setting hw params
of a capture device) may be done after playback has been started, which
may cause xrun.
The parsing functions expect float values in the default locale so use
the spa_dtoa function to generate such a float.
Fixes setting params with floating point values when the locale is not
the default locale.
Clang 15 at least with my build configuration emits warnings about
function prototypes that lack argument types. Most notably this
happens with functions that take no arguments which in compiler view
equates to the void type i.e. void f(void) instead of void f(). As I
understand, this will become an error in some future Clang release,
so might as well fix it now.
Since these were discovered not by a linter but by the actual compiler
for my particular build configuration, some f() may have escaped for
now. But at least it's enough to build PipeWire with most optional
features enabled even when -Werror=strict-prototypes is enabled.
For anyone else wanting to have a go at this, these can be upgraded
from warnings to errors by adding -Werror=strict-prototypes to the
custom CFLAGS which probably works with GCC, too, but has only been
done with Clang 15.0.2.
Finally my editor automatically stripped trailing spaces upon saving
the modified files. I assume it's probably not worth keeping those
invisible bytes around but this may have slightly dubious implications
as it did also turn indented empty lines of JACK license header into
regular empty lines.
Signed-off-by: Niklāvs Koļesņikovs <89q1r14hd@relay.firefox.com>
The codec may need to take into account endpoint preferred QoS values,
so parse this information and pass it to the codec's get_qos.
All the QoS struct values need to be set, as otherwise BlueZ may pass
uninitialized zero values over the air, which devices can reject. This
does not apply to CIG/CIS ids, which are automatically allocated by
default.
Fixes connecting to nrf5340 devkit, which requires a valid TargetLatency
value.
Commit 947ee152d3 ("bluez: check dbus service before enumerating objects")
introduced a check that would esure that GetManagedObjects()
is not called if the "org.bluez" name has no owner. This
was done to avoid auto-starting the bluez daemon.
However, this is not needed because it is possible to set on
a per-message basis whether or not the destination service
should be automatically started by the dbus daemon.
When we were using the resampler and then bypass it when the configured
rate == 1.0, we create a pop because we don't process the queued data in
the resampler anymore.
Avoid this by keeping the resampler active as soon as the rate property
is set on the audioconvert. The resampler itself will use a more
efficient copy method in that case anyway and it is expected that the
rate will change again later when we need to reactivate the resampler.
