loopback_pos_update() can be called in the timer callback, thus the lock
held should be irq-safe.  Otherwise you'll get AB/BA deadlock together
with substream->self_group.lock.
Reported-and-tested-by: NKnut Petersen <Knut_Petersen@t-online.de>
Cc: <stable@kernel.org>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

Signed-off-by: NJaroslav Kysela <perex@perex.cz>

On SMP machines, the cable->running update must be atomic, otherwise
stream is not started correctly sometimes.
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

Show some useful runtime information using procfs.
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

In some circumstances (the rate shift value was changed), the irq_pos
value may be higher than the fraction value in the timer start function.
Check for it.

Also, to avoid value overflow, decrease maximum period size.
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

This patch fixes the hw_params restrictions when first (or playback) stream
sets the final hardware parameters. Also, fix the hw_params checking
in the trigger callback.
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

Simple coding fix.
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

In a special case, some old samples are left in the capture ring buffer.
Fix it.
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

Obvious copy-and-paste error.
Signed-off-by: NJaroslav Kysela <perex@perex.cz>

The snd-aloop module allows redirecting of the PCM playback in the
kernel back to the user space using the standard ALSA PCM capture API.

The module also allows time synchronization with another timing source
and notifications of playback stream parameter changes.
Signed-off-by: NJaroslav Kysela <perex@perex.cz>