kcat/openal-soft
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
5,450 Commits 1 Branch 0 Tags
Commit Graph

548 Commits

Author SHA1 Message Date
Chris Robinson
2b5fdad2c8 Free the global ALSA config after opening capture devices too 2012-09-02 14:26:46 -07:00
Lauri Kasanen
a6f9ede849 alsa: Free the global config cache to remove a lot of reported leaks 
Signed-off-by: Lauri Kasanen <curaga@operamail.com>
 2012-09-02 14:16:23 -07:00
Chris Robinson
f2fa5e7c88 Print the error when reporting snd_pcm_delay failure 2012-08-27 22:52:37 -07:00
Chris Robinson
f5e0500df4 Add a macro to help make a 64-bit value 2012-08-19 22:31:55 -07:00
Chris Robinson
37924887e8 Use a default method to provide a dummy latency 2012-08-19 12:31:59 -07:00
Chris Robinson
0db7c4131b Hold the device lock with PulseAudio between mixing and writing 2012-08-18 17:20:20 -07:00
Chris Robinson
a56a974149 Hold the device lock with ALSA between mixing and writing 2012-08-18 17:18:53 -07:00
Chris Robinson
af7f3768fa Use non-blocking access for ALSA playback 2012-08-18 17:01:56 -07:00
Chris Robinson
0865db564f Move the device lock into the backend function table 
For backend-specific implementations: this should hold the audio mixer loop for
playback devices, and provide recursive mutex behavior.
 2012-08-18 15:58:04 -07:00
Chris Robinson
ac4fc4026c Add a device method to retrieve the active latency 
This is effectively the time until the next update will be heard by the user,
or the closest approximation thereof, in nanoseconds.
 2012-08-17 13:38:52 -07:00
Chris Robinson
147fd9de4e Avoid dynamic handling of libsndio 
This makes it safer and easier to handle as OpenBSD updates it
 2012-08-16 10:12:50 -07:00
Chris Robinson
f42f655ea9 Don't unlock the pulseaudio mainloop when mixing 
Lock contention could cause a lengthy delay between mixing and writing the
audio. There shouldn't be any risk of dead-locking as the device lock should
never be held when the mainloop gets locked.
 2012-08-12 03:52:38 -07:00
Chris Robinson
8001d86751 Add missing include 2012-06-19 22:15:19 -07:00
Chris Robinson
acca3e5a21 Use a proper method to wait for a PulseAudio operation 2012-06-19 10:37:09 -07:00
Chris Robinson
ac84d95c6e Silence some GCC warnings 2012-06-17 22:17:03 -07:00
Chris Robinson
785f52aa29 Only try float32 with DSound when specifically requested 
Some hardware drivers seem to have a bug where a buffer descriptor with a
32-bit float sample type ends up creating a non-float32 buffer.
 2012-06-17 04:13:01 -07:00
Chris Robinson
051195a991 Avoid failing if ALSA can't set a buffer or period time near to our request 
These functions should not fail, but there seem to be bugs that cause them to
sometimes. We're flexible enough to handle almost anything it throws back at us
though, so anything is better than nothing here.
 2012-06-16 23:02:59 -07:00
Chris Robinson
90ca6675c2 Allow PulseAudio to spawn a server by default 
Since only one backend is used at a time now, the issues with device ownership
aren't really prevelant anymore. An application that tries to open all
enumerated devices won't run into the problem of spawning a server and end up
preventing other backend devices from working.
 2012-06-16 22:37:45 -07:00
Chris Robinson
0db9730a50 Add an option to allow PulseAudio to move streams 
This could be made the default, if there's a safe way to update the AL device
specifier.
 2012-06-16 22:32:19 -07:00
Chris Robinson
46840ee787 Add DEVPKEY_Device_FriendlyName declaration for mingw-w64 compatibility 2012-06-14 18:10:10 -07:00
Chris Robinson
259e4f5b38 Rename AllDevice -> AllDevices 2012-05-09 16:28:16 -07:00
Chris Robinson
e44229c74e Clamp the ALSA period time between min and max if setting near fails 2012-05-08 01:23:53 -07:00
Chris Robinson
92827cc467 Clamp the ALSA buffer time between min and max if setting near fails 
It seems there's some bugs with snd_pcm_hw_params_set_buffer_time_near that
cause it to return "invalid argument" despite a valid nearby buffer time being
available. Clamping the buffer time seems to mitigate the issue a bit.
 2012-05-07 11:42:59 -07:00
Chris Robinson
dd3d5bc2d4 Look for libsndio.so.2 instead of libsndio.so 2012-05-05 16:19:51 -07:00
Chris Robinson
648464a2da Check the right playback device file with OSS 2012-04-22 05:06:44 -07:00
Chris Robinson
511b4790d0 Remove hungarian notation from dsound and mmdevapi 2012-04-19 23:42:02 -07:00
Chris Robinson
cff0f938e5 Remove hungarian notation from the winmm backend 2012-04-19 23:22:11 -07:00
Chris Robinson
5418519637 Remove some more hungarian notation 2012-04-19 23:00:58 -07:00
Chris Robinson
b023dbe7b8 Remove hungarian notation from the device and context structs 2012-04-19 22:50:11 -07:00
Chris Robinson
a08080bd82 Avoid modifying the ALCdevice in portaudio's open method 2012-03-13 22:18:51 -07:00
Chris Robinson
415bf58a06 Store the portaudio stream parameters with the device 2012-03-13 21:52:47 -07:00
Chris Robinson
3b165809cc Minor coreaudio fix, and make sure to set the device name 2012-03-13 13:48:42 -07:00
Chris Robinson
5c59e6e704 Try to ensure the full mmdevapi buffer is used 2012-03-13 01:39:57 -07:00
Chris Robinson
ef2f237276 Try to ensure at least 2 mmdevapi updates 2012-03-13 01:35:31 -07:00
Chris Robinson
0af14c7a11 Try to find a multiple of mmdevapi's period size nearest to the wanted update size 2012-03-13 01:32:15 -07:00
Chris Robinson
663b43f4f1 Attempted fix for the coreaudio backend 2012-03-12 21:58:40 -07:00
Chris Robinson
dddacc25a5 Store a duplicate of the mmdevapi device ID 2012-03-06 13:15:55 -08:00
Chris Robinson
2ba2da467c Add a comment explaining the pulseaudio prop_filter 2012-03-06 12:01:15 -08:00
Chris Robinson
f2b64fd111 Update the ALCdevice in winmm's reset method instead of open 2012-03-06 10:14:05 -08:00
Chris Robinson
3131104595 Filter out a couple pulseaudio stream properties by setting them to 0-length data 
It seems Qt/Phonon will globally set media.role and phonon.streamid properties,
causing them to apply even to our streams if they happen to be loaded in the
same process. This is particularly a problem because media.role gets set to
"event", preventing the streams from showing in pavucontrol.
 2012-03-05 13:00:51 -08:00
Chris Robinson
f64bdc4bd4 Returned pa_operations can be NULL 2012-03-05 08:37:50 -08:00
Chris Robinson
fe6e532c87 Use a separate backend callback to start playback of the device 
This allows us to properly update the ALCdevice and its resources with the new
parameters before starting playback, instead of expecting the mixer to block
and wait after it has begun.

This also lets us avoid holding the device lock while resetting and starting
the device, which helps prevent lock inversion on some backends (ie, one thread
locking A then B, and another thread locking B then A), ultimately allowing
certain backends to asynchronously update the ALCdevice without risk of lockup.
Capture still has issues here, however.
 2012-03-05 07:11:09 -08:00
Chris Robinson
bc1bb7f90f Print and handle errors from pa_stream_readable_size 2012-03-05 00:11:28 -08:00
Chris Robinson
0f0141b9dd Set prebuf to 0 since we're handling the stream start manually 2012-03-04 09:24:07 -08:00
Chris Robinson
b643f13a6e Make sure the pulse stream is properly started and better handle buffer size changes 2012-03-04 05:30:42 -08:00
Chris Robinson
49433c2074 Slight correction for handling the wave writer backend's timer wrapping 2012-03-03 08:50:15 -08:00
Chris Robinson
a93ff6ef3c Slight correction for handling the null backend's timer wrapping 2012-03-03 08:47:23 -08:00
Chris Robinson
44e10c4f69 Remove some unneeded pulseaudio calls 2012-03-02 09:46:36 -08:00
Chris Robinson
7a832c4522 Don't unload libpulse 
This sometimes spawns esoteric errors in dlclose
 2012-03-02 09:27:30 -08:00
Chris Robinson
7d0faba424 Remove some unneeded markers 2012-03-02 09:08:17 -08:00