2014-9-11 下午1:28 於 "Alexander E. Patrakov" patrakov@gmail.com 寫道：
11.09.2014 09:49, Raymond Yau wrote:
On my desktop PC, on snd-hda-intel with analog outputs for S16LE
stereo, the granularity is 32 bytes (= 8 samples), and I get the pointer granularity of 64 bytes (=16 samples) over HDMI. The minimum period size is 32 samples in both cases.
Do you mean hda-Intel does not support arbritray period size when you say the granularity is 32 bytes ?
However the granularity of the emulated hda sound card inside any VM depend on the vm and the backend audio system and sound card
The precise meaning is defined here:
http://mailman.alsa-project.org/pipermail/alsa-devel/2014-May/076475.html
Of course the VM result would be different.
http://git.kernel.org/cgit/virt/kvm/qemu-kvm.git/plain/hw/hda-audio.c
/* audio in/out widget */ case AC_VERB_SET_CHANNEL_STREAMID: st = a->st + node->stindex; if (st->node == NULL) { goto fail; } hda_audio_set_running(st, false); st->stream = (payload >> 4) & 0x0f; st->channel = payload & 0x0f; dprint(a, 2, "%s: stream %d, channel %d\n", st->node->name, st->stream, st->channel); hda_audio_set_running(st, a->running_real[st->output * 16 + st->stream]); hda_codec_response(hda, true, 0); break;
It seem some vm just emulate start/stop of playback/capture , it does not update any hda controller registers, this mean that it won't support rewind