alsa-project/alsa-lib issue #454 was opened from dbrignoli:
This issue affects an out of tree `snd-asihpi` driver which uses a different h/w pointer update implementation compared to the in-tree `snd-asihpi` driver. In particular, this version of the driver uses the `.ack` callback to manage h/w pointer update. Driver source is available for download here https://www.audioscience.com/internet/download/drivers/released/v4/20/47/hpklinux_4.20.47-g51a97a1a2.tar.gz.
Using the following `.asoundrc` (with or without `slowptr true`):
``` pcm.testPlugDmix { type plug slave.pcm "dmixer" }
pcm.dmixer { type dmix ipc_key 1024 slave { pcm "hw:1,0" channels 2 format S32_LE period_time 0 period_size 1024 buffer_size 4096 rate 44100 } bindings { 0 0 1 1 } # slowptr true } ```
and running aplay with `-D testPlugDmix` the trace looks like this:
``` 1) | snd_pcm_ioctl [snd_pcm]() { 1) | snd_pcm_common_ioctl [snd_pcm]() { 1) | snd_pcm_sync_ptr [snd_pcm]() { 1) 0.388 us | pcm_lib_apply_appl_ptr [snd_pcm](); /* = 0x0 */ 1) 1.155 us | } /* snd_pcm_sync_ptr [snd_pcm] = 0x0 */ 1) 1.697 us | } /* snd_pcm_common_ioctl [snd_pcm] = 0x0 */ 1) 2.055 us | } /* snd_pcm_ioctl [snd_pcm] = 0x0 */ ```
Running aplay with `-D hw:1` instead the trace looks like this:
``` 0) | snd_pcm_ioctl [snd_pcm]() { 0) | snd_pcm_common_ioctl [snd_pcm]() { 0) | __snd_pcm_lib_xfer [snd_pcm]() { 0) 0.333 us | snd_pcm_stream_lock_irq [snd_pcm](); /* = 0x0 */ 0) 0.302 us | snd_pcm_stream_unlock_irq [snd_pcm](); /* = 0x0 */ 0) | interleaved_copy [snd_pcm]() { 0) | do_transfer [snd_pcm]() { 0) 4.608 us | default_write_copy [snd_pcm](); /* = 0x0 */ 0) 5.249 us | } /* do_transfer [snd_pcm] = 0x0 */ 0) 8.835 us | } /* interleaved_copy [snd_pcm] = 0x0 */ 0) 0.311 us | snd_dma_buffer_sync [snd_pcm](); /* = 0x0 */ 0) 0.299 us | snd_pcm_stream_lock_irq [snd_pcm](); /* = 0x0 */ 0) | pcm_lib_apply_appl_ptr [snd_pcm]() { 0) | snd_card_asihpi_playback_ack [snd_asihpi]() { 0) | /* asihpi_stream_xfer: pcmC1D0p/sub0 : 4096 */ 0) | /* asihpi_stream_state: pcmC1D0p/sub0 in .ack: ALSA: aptr 4096, hptr 0, avail 12288; HPI size 16384, host 4096 (4096), dsp 0 (0, 0), fifo 0, avail 12288, xfer 0 */ 0) 1.376 us | } /* snd_card_asihpi_playback_ack [snd_asihpi] = 0x0 */ 0) 1.999 us | } /* pcm_lib_apply_appl_ptr [snd_pcm] = 0x0 */ 0) 0.364 us | snd_pcm_update_state [snd_pcm](); /* = 0x0 */ 0) 0.299 us | snd_pcm_stream_unlock_irq [snd_pcm](); /* = 0x0 */ 0) + 15.415 us | } /* __snd_pcm_lib_xfer [snd_pcm] = 0x1000 */ 0) + 16.252 us | } /* snd_pcm_common_ioctl [snd_pcm] = 0x0 */ 0) + 16.920 us | } /* snd_pcm_ioctl [snd_pcm] = 0x0 */ ```
With dmix the driver never sees changes to `appl_ptr` and the `.ack` callback is never called. The driver sets the `SNDRV_PCM_INFO_SYNC_APPLPTR` flag and and I have tried adding `SNDRV_PCM_INFO_EXPLICIT_SYNC` but it makes no difference. Using `dsnoop` on the capture side shows the same symptoms.
Is this expected with dmix? Thank you.
Issue URL : https://github.com/alsa-project/alsa-lib/issues/454 Repository URL: https://github.com/alsa-project/alsa-lib