 Mike Brady
|
bfc7b9b774
|
lengthen some mutex timeouts to allow valgrind to work okay on a Pi 3B. Make operations that open the mixer uncancellable until it's closed again to stop a few small memory leaks when a player is cancelled.
|
2019-02-01 16:19:27 +00:00 |
|
|
Mike Brady
|
0c5521488c
|
add a special mixer-only mutex, do make open/close mixer stuff safer and probably quicker, not waiting for the main alsa mutex.
|
2019-02-01 12:38:45 +00:00 |
|
|
Mike Brady
|
2c0250b32a
|
Don't do a hardware mute when doing a flush -- if disable_standby is true, the mode goes to connected and then if player goes back to play immediately, you might hear some of the sound still moving through the output buffer. This happens due to a bug in IOS player.
|
2019-01-31 17:11:33 +00:00 |
|
|
Mike Brady
|
60487c86b1
|
Make mute work only if not diconnected. Clewan up some debug messages.
|
2019-01-31 13:35:11 +00:00 |
|
|
Mike Brady
|
555bb97a39
|
Rebuild the hardware mute code -- clear up the logic. Needs to check if disconnected though.
|
2019-01-31 09:01:35 +00:00 |
|
|
Mike Brady
|
fd08511f1f
|
Make it work with deferred initialisation of the alsa device -- deferred to first use. Investigate hardware mute again.
|
2019-01-30 17:43:15 +00:00 |
|
|
Mike Brady
|
c5ff0dfed0
|
Rebuild the volume control calculations. Implement software mute. Try to delay alsa initialisation as long as possible.
|
2019-01-30 09:24:10 +00:00 |
|
|
Mike Brady
|
aa5698fce2
|
Remove the preflight call - don't seem to need it.
|
2019-01-27 15:53:36 +00:00 |
|
|
Mike Brady
|
02955ecabd
|
bring up to date with unstable branch
|
2019-01-27 14:25:02 +00:00 |
|
|
Mike Brady
|
56bf886206
|
gracefully handle situation when the output device is already in use; release it properly when finished.
|
2019-01-27 14:19:39 +00:00 |
|
|
Mike Brady
|
c807513422
|
quieten a few debug messages, bring a standard one forward.
|
2019-01-24 16:33:50 +00:00 |
|
|
Mike Brady
|
b02e5d9f20
|
Use returned codes from do_open() to control the state machine -- don't go to abm_connected unless, eh, connected.
|
2019-01-24 09:14:41 +00:00 |
|
|
Mike Brady
|
9fb37e4293
|
Reorganise code to make it respond dynamically to changes in settings.
|
2019-01-23 22:22:27 +00:00 |
|
|
Mike Brady
|
bd81e733cd
|
Introduce a simple state machine into the alsa backend to keep track of the connected state of it. Needed to deal properly with disabling and enabling standby dynamically.
|
2019-01-23 16:44:02 +00:00 |
|
|
Mike Brady
|
2e4428538d
|
Add an "active" mode that can time out after a period -- 10 seconds by default -- after play ends.
|
2019-01-21 21:43:55 +00:00 |
|
|
Mike Brady
|
020a69dfb9
|
remove the cause of a CLANG warning
|
2019-01-18 16:27:29 +00:00 |
|
|
Mike Brady
|
fbd8377519
|
Always add dither if no hardware mixer and volume control is not ignored.
|
2019-01-10 16:57:40 +00:00 |
|
|
Mike Brady
|
c958d506b1
|
reduce thresholds towards a buffer time of 0.2 seconds rather than the 0.25 seconds before.
|
2019-01-09 14:15:05 +00:00 |
|
|
Mike Brady
|
2a9d46afed
|
backend silence threshold should be 0.060 not 0.6!
|
2019-01-09 10:10:04 +00:00 |
|
|
Mike Brady
|
e9623faa44
|
Go for a 0.250 second buffer length.
|
2019-01-08 17:28:38 +00:00 |
|
|
Mike Brady
|
56f4b124bd
|
Adjust thresholds and timings to suit pi zero.
|
2019-01-08 13:06:46 +00:00 |
|
|
Mike Brady
|
caf06c5aba
|
Add the idea of a preflight message to tell the backend that frames will be coming...
|
2019-01-07 17:34:50 +00:00 |
|
|
Mike Brady
|
4613ee6bcb
|
Make buffer size threashold 100 ms instead of 50 ms. Remove some surplus snd_pcm_prepare calls.
|
2019-01-07 16:00:58 +00:00 |
|
|
Mike Brady
|
0e37f7e78d
|
Only look for snd_pcm_status_get_driver_htstamp is alsa version 1.1 or later...
|
2019-01-04 21:52:36 +00:00 |
|
|
Mike Brady
|
6b42abab72
|
Use driver htstamp rather than regular one. Use untimed_delay so that a regualr delay sets the standoff timer for silence.
|
2019-01-04 20:57:18 +00:00 |
|
|
Mike Brady
|
48043a81cc
|
Move stall threshold calculation to init from start, duh.
|
2019-01-04 18:17:30 +00:00 |
|
|
Mike Brady
|
80f15e1f57
|
clang format
|
2019-01-04 17:26:55 +00:00 |
|
|
Mike Brady
|
9552b278b7
|
remove some debugging code from delay().
|
2019-01-04 17:26:00 +00:00 |
|
|
Mike Brady
|
7496364aaa
|
continue to change delay() and play() to reflect use of snd_pcm_recover _after_ a XRUN has occurred -- fix rate calculations etc.
|
2019-01-04 17:20:15 +00:00 |
|
|
Mike Brady
|
459dd51a31
|
Use snd_pcm_recover only _after_ an XRUN or suspend has happened. Don't use it to remove one -- it doesn't work.
|
2019-01-04 13:50:14 +00:00 |
|
|
Mike Brady
|
8e5733e6ae
|
Check before using snd_pcm_prepare in delay()
|
2019-01-04 11:29:18 +00:00 |
|
|
Mike Brady
|
41c8c8b9a1
|
Strengthen the ability to recover from an XRUN by putting in a call to snd_pcm_prepare first.
|
2019-01-03 10:19:44 +00:00 |
|
|
Mike Brady
|
15729ebcd4
|
Fix trivial error in a debug message.
|
2019-01-02 22:40:48 +00:00 |
|
|
Mike Brady
|
8d8b8c05f6
|
new play(), delay(), delay_prep_and_status() routines in place and working, but with the old stuff taken out.
|
2019-01-02 19:54:12 +00:00 |
|
|
Mike Brady
|
59d26918f6
|
new play(), delay(), delay_prep_and_status() routines in place and working, but with the old stuff still around.
|
2019-01-02 19:49:15 +00:00 |
|
|
Mike Brady
|
6f8b35d121
|
Try out a new alsa delay routine. Could be very buggy!
|
2019-01-02 14:43:25 +00:00 |
|
|
Mike Brady
|
fe857037de
|
If output is stalled, don't write any more frames to the output device.
|
2018-12-31 17:46:05 +00:00 |
|
|
Mike Brady
|
fc08572631
|
Modify my_snd_pcm_delay but don't test it yet!
|
2018-12-29 09:34:10 +00:00 |
|
|
Mike Brady
|
919c5fe8b8
|
Add some diagnostics around the state of the DAC during delays.
|
2018-12-28 10:35:13 +00:00 |
|
|
Mike Brady
|
ad6c721d5a
|
Format
|
2018-12-27 14:04:31 +00:00 |
|
|
Mike Brady
|
1828545602
|
Account for duration of samples to be played and use the untimed_play when calculating silence-filling backoff period.
|
2018-12-27 14:01:32 +00:00 |
|
|
Mike Brady
|
284cb9d3fb
|
Return genuine error codes where available in actual_open_alsa_device.
|
2018-12-26 10:41:20 +00:00 |
|
|
Mike Brady
|
70cd4fe81f
|
Further adjustments to the silence thread -- make the sleep time 30 ms, thus if the queue goes below 60 ms of samples, silence will be inserted unless an audio write occurred within the last 30 ms.
|
2018-12-17 17:11:46 +00:00 |
|
|
Mike Brady
|
2830cdf712
|
adjust minimum interval between last audio write and writing silence.
|
2018-12-17 16:50:36 +00:00 |
|
|
Mike Brady
|
a96d28c4e5
|
Need to use an untimed play for the silence buffer filler.
|
2018-12-17 16:44:19 +00:00 |
|
|
Mike Brady
|
cb17fb263a
|
Don't send silence into the alsa device within the scanner's sleep time of the last legitimate write to it. This is to stop the scanner inserting silent frames while the player is flushing fames.
|
2018-12-17 16:31:03 +00:00 |
|
|
Mike Brady
|
57e9619e5b
|
Don't add dither if volume control is set to max.
|
2018-12-17 13:14:40 +00:00 |
|
|
Mike Brady
|
14e867a66b
|
use dither if using software mixer (i.e. no hardware mixer and not ignoring the volume control) -- fix a mistake in accounting for the ignore_volume_control setting.
|
2018-12-17 11:30:23 +00:00 |
|
|
Mike Brady
|
4ba0947ddb
|
Move the keep_dac_busy in to a configuration setting. Always start the keep_dac_busy thread and check if the keep_dac_busy is set. Don't dither is ignore volume control is set.
|
2018-12-15 15:36:44 +00:00 |
|
|
Mike Brady
|
7708ee3cdf
|
Bring in the "disable_standby_mode" feature on alsa to stop some DACs making popping noises. Fix dither generation.
|
2018-12-14 15:28:32 +00:00 |
|