Keep a list of ports for the node. When the node goes away, clear the
port links to the node. Handle the case where the port no longer has a
node.
This avoids a crash when, for example, the node permission is removed
and the port points to the now freed node_data.
Fixes#3708
Check for errors when loading the geometry instead of silently failing.
The points need to be given in the user locale and so might fail to
parse when given in JSON format.
Format the geometry nicely when loading the module.
Some AirPlay devices will announce their IPv4 addresses
over IPv6 mDNS if both are available, so the determined
IP version was not reliable.
The prop is not used by module-raop-sink, so its
removal should be safe.
As part of the setup for IRQ based scheduling, a period event
was installed. Not only is a timer based polling unecessary for
IRQ scheduling, depending on the state of the system, the timer
could fire far enough from the IRQ, causing alsa wakeup events
with no data in the ring buffer. Pipewire would identify these
events as an "early wakeup", adding an extra quantum of time
to the next_time estimate, skewing the clock and causing issues
with apps that depend on precise timing.
Update the started and ready state after we suspend/pause the node so
that we don't complain if scheduling happens between setting the fields
and actually stopping the follower.
Also only complain when the scheduling happens when the node is not
ready. It is possible that the node is scheduled before we manage to set
the started field.
Move the driver and warned bits after the int field in the struct so
that they are placed in separate memory.
Otherwise, a write from the data thread might race with a write from the
main thread and leave the bits in the wrong state.
This reverts commit 49cdb468c2.
We should not do this, the nsec field should be relatable to the clock
monotonic time. If we use the estimated time, without actually using it
as a timer, we might end up with a wakeup time in the future compared to
the MONOTONIC clock time.
Instead, you can use the estimated current time simply by subtracting
the rate corrected duration from the next_nsec. This is really only
useful for some selected use cases (like in the JACK library).
This fixes some issues where in pro-audio mode, a client would try to
compare the current MONOTONIC time to nsec and find that it is in the
past.
This commit was done in an attempt to fix#3657 but it turned out the
real problem was something else.
We should not use nsec from the clock as the current_usec because
current_usec is supposed to be an idealized time when the wakeup would
have happened and nsec is when it actually happened.
Instead use next_nsec and subtract the rate corrected period from it to
get the idealized current_usec.
We can still use nsec to calculate the elapsed time since the wakeup and
be sure that it is always in the past.
The Pro Link 1 replies with Audio-latency=0, patch that up to
1500ms to make it work again.
Previously it configured 1500ms as the default latency but that seems
unnecessary in the usual case.
Fixes#3698
Make sure that NULL params don't cause -EINVAL but ignore them.
Don't add empty param objects. this makes it possible to clear all previous
params by setting an empty object.
We always need to add the Props param because it contains the
debug.aec.wav-path key, which is always available, even when the AEC
implementation has no properties.
Also add the debug.aec.wav-path PropInfo.
We can reuse the same code for setting input and output stream
properties. This also makes it possible to set the debug.aec.wav-path
on all streams, like the properties say.
Don't have separate input route for A2DP and HFP, as it is generally not
necessary.
When in A2DP mode when there's also HFP possible, emit the input route
in SPA_PARAM_Route, even though there is no corresponding input node
emitted.
The host may then emit a loopback microphone node, and switch profiles
according to its status. Having the input route available at all times
allows to retain changes to volume settings made when there is no real
input node.
For params that don't emit change events, the param enumeration does not
start core sync, and its enumeration may be incomplete if a previous
core sync completes first.
Fix by always starting a core sync if we are updating params.
This fixes dev_info.active_port being sporadically SPA_ID_INVALID if
there is event pressure, which causes do_set_volume randomly set the
volume on the node instead of setting it on the device. This can be
triggered e.g. by shaking the volume slider in Pavucontrol with mouse.
Letter codes instead of numbers for log levels are allowed in
PIPEWIRE_DEBUG.
Support them also for log.level in context properties and metadata.
Add a few alias letter codes to harmonize them with what WirePlumber 0.5
supports.
We delay the audio a bit to keep packet intervals equal, which keeps
some data in buffers.
In theory the calculation keeps one buffer free, but it doesn't
explicitly keep "extra" buffer space so in theory might flush too late
and next process() might not have free buffers. However, as we encode
next packet right away this shouldn't really occur...
Try to keep one extra spare buffer free so that the flush time is
certainly early enough.
Tabs and spaces are mixed for indentation in the pipewiresink. Replace
all tabs with 8 spaces although indentation is 2 spaces, since that
looks like the intended indent.
The messages are printed for every buffer. Therefore, they should be log
messages. Also add the bufferpool to the message to be able to identify
the bufferpool that handles the buffers.
The buffers are added to the pipewirepool during setup. Therefore, they
should be debug messages. As at it, use the debug helper to print the
string of the buffer type.
The alsa sequencer rate matching was not actually working correctly.
It would compare the previous queue time with the current time and
compare that to the quantum. This would include uncorrected errors from
jitter and would result in the timeouts being scaled in the wrong
direction forever.
Instead, calculate an ideal queue time and compare our current queue
time against that. We then use the correction to scale the timeout or
the next queue time prediction.
Also use the predicted time as the base time for the event timestamps.
this results in less jitter.
Fixes#3657
While switching to the Pro Audio profile from a UCM profile, the active
UCM profile is not disabled because the devices may need a UCM verb to
be set before being used. Then, switching away from the Pro profile to a
UCM profile is done by passing the Off profile as the old profile,
instead of the actually-still-enabled UCM profile.
This means the devices from the old UCM profile are kept enabled when we
try to switch to another UCM profile. And the devices to enable may be
conflicting with the still-enabled devices, which will cause failures.
To avoid this, we need to tell the UCM code to disable the profile when
switching away from it. Doing so disables UCM devices and the UCM verb.
Existing code sets the highest priority UCM verb before probing the Pro
profile, so manually enable the same one here as well, instead of
relying on unclean state from whatever profile that was last active.
Signed-off-by: Alper Nebi Yasak <alpernebiyasak@gmail.com>
The hardware controls we track for volume and mute state can change as
part of enabling or disabling the UCM device. This then triggers a
re-sync of our state with the hardware controls, which can e.g. set our
volume to an unwanted value as the result of enabling the device, or set
a hardware control to a wrong value based on our volume while disabling
the device. So these UCM-triggered changes to enable/disable the device
will at best show up on user interfaces and cause confusion, but maybe
even will push the hardware into an unexpected state.
The volume and mute state we set from user interfaces only make sense
when the devices are enabled. They should not be kept in sync with
hardware for inactive UCM devices [1]. Skip the callbacks for reading
and changing volume and mute state if the UCM device is disabled. This
way, the volume/mute controls for sinks/sources are essentially detached
from the hardware controls until the UCM device is re-enabled.
[1] https://gitlab.freedesktop.org/pulseaudio/pulseaudio/-/merge_requests/772#note_1872757
Link: https://gitlab.freedesktop.org/pulseaudio/pulseaudio/-/merge_requests/772
Signed-off-by: Alper Nebi Yasak <alpernebiyasak@gmail.com>
The pa_alsa_ucm_set_port() function is passed both a mapping context and
a device port, and both of these refer to their respective UCM device.
While switching over to having one port per mapping per UCM device, I
expected both of these to be the same device struct, so added an assert
checking so.
This assertion gets triggered when we have multiple UCM verbs declaring
the same UCM device name. The root cause here is that the ports' UCM
device references are set once while creating the ports for the card, so
they happen to be those of a specific verb and may not match those from
a different UCM verb's profiles' mappings.
Solving the root cause necessitates a larger refactor. What we actually
assume here is that name of the UCM device is same for both the port and
the UCM context, which ends up always true in practice. For now, replace
the assert with a check and error.
Link: https://gitlab.freedesktop.org/pulseaudio/pulseaudio/-/merge_requests/802
Signed-off-by: Alper Nebi Yasak <alpernebiyasak@gmail.com>
Some versions of the ALSA libraries run into a segmentation fault when
we query a UCM device/modifier status without first setting a UCM verb.
It's not a reasonable thing to do anyway, so check for this case and
return an error. Also do the check in other helpers.
Link: https://gitlab.freedesktop.org/pulseaudio/pulseaudio/-/merge_requests/801
Signed-off-by: Alper Nebi Yasak <alpernebiyasak@gmail.com>
The ucm_get_device_property() function adds to each UCM device's
playback_volumes (or capture_volumes) hash map an associated volume
mixer keyed with the UCM verb. These key-value pairs are then iterated
over in various places which assume the key is a profile name. This
assumption is no longer true since we can generate multiple profiles to
use conflicting devices.
A previous commit 45278904167f ("alsa-ucm: Stop conflating profile name
with UCM verb name") fixes some instances of this assumption, but misses
the relation explained above. Fix more instances of misleading
"profile"s where the UCM verb name is actually meant.
Link: 7005eafee3
Fixes: 57295beee1a6 ("alsa-ucm: Stop conflating profile name with UCM verb name")
Signed-off-by: Alper Nebi Yasak <alpernebiyasak@gmail.com>
