Properly position position mixed audio data

 - Audio data was just being popped to the "front" of the mix buffer, so
   instead it now properly pops into the correct position in the mix
   buffer (proper mixing still needs to be implemented)

libobs/media-io/audio-io.c

@@ -111,10 +111,26 @@ static inline uint64_t min_uint64(uint64_t a, uint64_t b)
 }
 
 static inline void mix_audio_line(struct audio_output *audio,
-		struct audio_line *line, size_t size)
+		struct audio_line *line, size_t size, uint64_t timestamp)
 {
 	/* TODO: this just overwrites, handle actual mixing */
-	circlebuf_pop_front(&line->buffer, audio->mix_buffer.array, size);
+	if (!line->buffer.size) {
+		if (!line->alive)
+			audio_output_removeline(audio, line);
+		return;
+	}
+
+	size_t time_offset = convert_to_sample_offset(audio,
+			line->base_timestamp - timestamp);
+	if (time_offset > size)
+		return;
+
+	size -= time_offset;
+
+	size_t pop_size = min_uint64(size, line->buffer.size);
+	circlebuf_pop_front(&line->buffer,
+			audio->mix_buffer.array + time_offset,
+			pop_size);
 }
 
 static void mix_audio_lines(struct audio_output *audio, uint64_t audio_time,
@@ -122,7 +138,7 @@ static void mix_audio_lines(struct audio_output *audio, uint64_t audio_time,
 {
 	struct audio_line *line = audio->first_line;
 	uint64_t time_offset = audio_time - prev_time;
-	uint64_t byte_offset = convert_to_sample_offset(audio, time_offset);
+	size_t byte_offset = convert_to_sample_offset(audio, time_offset);
 
 	da_resize(audio->mix_buffer, byte_offset);
 	memset(audio->mix_buffer.array, 0, byte_offset);
@@ -132,16 +148,12 @@ static void mix_audio_lines(struct audio_output *audio, uint64_t audio_time,
 
 		if (line->buffer.size && line->base_timestamp < prev_time) {
 			clear_excess_audio_data(line,
-					line->base_timestamp - prev_time);
+					prev_time - line->base_timestamp);
 			line->base_timestamp = prev_time;
 		}
 
-		size_t pop_size = min_uint64(byte_offset, line->buffer.size);
-		if (pop_size)
-			mix_audio_line(audio, line, pop_size);
-		else if (!line->alive)
-			audio_output_removeline(audio, line);
-
+		mix_audio_line(audio, line, byte_offset, prev_time);
+		line->base_timestamp = audio_time;
 		line = next;
 	}
 

libobs/obs-source.c

@@ -334,7 +334,7 @@ static void source_output_audio_line(obs_source_t source,
 
 	if (!source->timing_set) {
 		source->timing_set    = true;
-		source->timing_adjust = in.timestamp - os_gettime_ns();
+		source->timing_adjust = os_gettime_ns() - in.timestamp;
 
 		/* detects 'directly' set timestamps as long as they're within
 		 * a certain threshold */