pushsamplesreaquestbuff overflow

[AsheX1]

i dont know if this is and emu problem or settings as i find 0 settings to change anything about video at all

[sverx]

I'm getting a ton of these errors too... but I'm not concerned about the messages, mainly about the audio being totally broken, at least on my Linux machine. 🤔

I'm trying to understand if I can fix it. It seems like there's not enough space in the buffer to put newly rendered audio data, but even creating a much bigger audio buffer it doesn't solve at all, the buffer gets all used quickly and audio stutters.

#define SOUND_BUFFERS_SAMPLE_COUNT (16384)

But according to the code, when there's not enough space in the buffer to write more data, it just means it's pretty full of data already, so the audio renderer should just not produce any more audio data and come back later. But then audio breaks, which likely means the buffer underruns, even if the Audio Debug window shows plenty of data in the buffer. I'm very puzzled.

[sverx]

I suspect for some reason the replay of the audio uses less data than what the audio engine renders, or the audio engine renders too much data compared to what it needed. I had set #define SOUND_BUFFERS_SAMPLE_COUNT (65536) and the audio becomes totally fine for a few seconds, then the clearly visible effect is that the buffer becomes full of audio data that still need to be played and the emulator wants to push more.

So I would say it's either generating more than 44100 samples/sec or it's reproducing less than that.

[sverx]

The audio coming from the application is definitely at 44100 samples/sec according to pactl list sink-inputs

Sink Input #62
	Driver: protocol-native.c
	Owner Module: 12
	Client: 160
	Sink: 3
	Sample Specification: s16le 2ch 44100Hz
	Channel Map: front-left,front-right
	Format: pcm, format.sample_format = "\"s16le\""  format.rate = "44100"  format.channels = "2"  format.channel_map = "\"front-left,front-right\""
	Corked: no
	Mute: no
	Volume: front-left: 65536 / 100% / 0.00 dB,   front-right: 65536 / 100% / 0.00 dB
	        balance 0.00
	Buffer Latency: 13990 usec
	Sink Latency: 23848 usec
	Resample method: n/a
	Properties:
		media.name = "Allegro Voice"
		application.name = "meka"
		native-protocol.peer = "UNIX socket client"
		native-protocol.version = "33"
		application.process.id = "12699"
		application.process.user = "sverx"
		application.process.host = "redacted"
		application.process.binary = "meka"
		application.language = "C"
		window.x11.display = ":0"
		application.process.machine_id = "939972095cf1459c8b22cc608eff85da"
		application.process.session_id = "c2"
		module-stream-restore.id = "sink-input-by-application-name:meka"