For unclear reasons, the extra voice was set up with half the buffer size instead of the period size. Commit 27ae958cf6 ("emu10k1 driver - add multichannel device hw:x,3 [2-8/8]") mentions half-loop interrupts, so maybe this was an artifact of an earlier iteration of the patch. While at it, also fix periods_min of the regular playback - one period makes just no sense. Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@xxxxxx> --- sound/pci/emu10k1/emupcm.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/sound/pci/emu10k1/emupcm.c b/sound/pci/emu10k1/emupcm.c index 0572dfb80943..2764e7867b33 100644 --- a/sound/pci/emu10k1/emupcm.c +++ b/sound/pci/emu10k1/emupcm.c @@ -429,15 +429,16 @@ static int snd_emu10k1_efx_playback_prepare(struct snd_pcm_substream *substream) struct snd_pcm_runtime *runtime = substream->runtime; struct snd_emu10k1_pcm *epcm = runtime->private_data; unsigned int start_addr; - unsigned int channel_size; + unsigned int extra_size, channel_size; int i; start_addr = epcm->start_addr >> 1; // 16-bit voices + extra_size = runtime->period_size; channel_size = runtime->buffer_size; snd_emu10k1_pcm_init_extra_voice(emu, epcm->extra, true, - start_addr, start_addr + (channel_size / 2)); + start_addr, start_addr + extra_size); epcm->ccca_start_addr = start_addr; for (i = 0; i < NUM_EFX_PLAYBACK; i++) { @@ -465,7 +466,7 @@ static const struct snd_pcm_hardware snd_emu10k1_efx_playback = .buffer_bytes_max = (128*1024), .period_bytes_max = (128*1024), .periods_min = 2, - .periods_max = 2, + .periods_max = 1024, .fifo_size = 0, }; @@ -925,7 +926,7 @@ static const struct snd_pcm_hardware snd_emu10k1_playback = .channels_max = 2, .buffer_bytes_max = (128*1024), .period_bytes_max = (128*1024), - .periods_min = 1, + .periods_min = 2, .periods_max = 1024, .fifo_size = 0, }; -- 2.40.0.152.g15d061e6df