Previously we just sent packets when ready, which especially during startup
meant a largish number of packets got sent rapidly. This seemed to cause audio
glitches. With this change the rate is adapted to follow ACKs from the device,
which is more in line with Chromium's way, though still far from the same.
Also added capability to resend packets when a nACK is received.
Also some wip attempts at sending coverart as VP8 video.
For unknown reasons some Chromecast devices disconnect causing a TLS error if
we don't OFFER a video stream. Seems actually sending video isn't required to
fix the issue.
With this commit auto reconnection per default will only be done for ATV4s and
HomePods. Reconnection is not always desirable, for instance if the device cuts
the connection because it is busy with something else, ref. issue #934.
The commit also adds an option to override auto reconnection, thus either
enabling it for other devices or disabling it for affected devices.
Include ALSA's device name in the ALSA modules 'info' logging to help
identify sound devices as seen by the system for assisting config setup
Many configs use ALSA's hw ids to refer to device but ALSA can also use
device names:
laudio: Available ALSA playback mixer(s) on hw:0 CARD=Intel (HDA Intel): 'Master' 'Headphone' 'Speaker' 'PCM' 'Mic' 'Beep'
laudio: Available ALSA playback mixer(s) on hw:1 CARD=E30 (E30): 'E30 '
From the example above can use these ALSA names interchangably:
'hw:0' and 'hw:Intel'
'hw:1' and 'hw:E30'
Before, if a user never verified the device, we would have a device->session
even though the device was not streaming and was in a failed state.
This solution should be more clean and in line with the overall principle that
we only have a session when communicating with the device.
Also includes a bit of code refactoring.
outputs_authorize() has two issues, one that the caller can't specify device
(problem if there are two devices waiting for verification), the other that it
didn't offer a standard callback, so difficult to catch failure/success.
Moves speaker selection, volume handling and startup to outputs.c, plus adds
the ability to "resurrect" a speaker that disconnects.
The purpose of moving the code is to concentrate device handling in one place.
Also changes how we deal with speaker selection. The player will now generally
not alter a user selection, even if the device fails. The purpose of this is to
maintain selection both if the device briefly fails, and if the user switches
off the device (we stop playback) and later turns it on + starts new playback.
Instead of using OPTIONS we use SET_PARAMETER with progress metadata to avoid
disconnects from Apple TVs, Homepods and possibly also Airport Expresses.
The player will write 24 bit samples using 3 bytes, not 4, so the appropriate
sample format is SND_PCM_FORMAT_S24_3LE, not SND_PCM_FORMAT_S24_LE.
For extra protection we also use snd_pcm_bytes_to_frames() instead of BTOS(),
because that way we can be more certain that the buffer is not too short for
snd_pcm_writei().
Since it is unknown how to do real sync on Chromecast, this commit instead adds
a primitive delay to the stream, so that it is at least somewhat closer to
Airplay/local audio.
Also some cleanup of unused stuff.
Fixes bugs which were due to incorrect handling of unsigned integer wrap-around:
1. Calling packet_resend() with seqnum + len greater than UINT16_MAX => infinite loop
2. Calling rtp_packet_get() with session->seqnum - seqnum greater than pktbuf_next => wrong packet
