mbox series

[v3,0/7] ALSA: aloop: Support sound timer as clock source instead of jiffies

Message ID 20191111110846.18223-1-andrew_gabbasov@mentor.com (mailing list archive)
Headers show
Series ALSA: aloop: Support sound timer as clock source instead of jiffies | expand

Message

Gabbasov, Andrew Nov. 11, 2019, 11:08 a.m. UTC
This patch set is an updated version of patches by Timo Wischer:
https://mailman.alsa-project.org/pipermail/alsa-devel/2019-March/146871.html

This patch set is required for forwarding audio data between a HW sound
card and an aloop device without the usage of an asynchronous sample rate
converter.

Most of sound and timers related code is kept the same as in previous set.
The code, related to snd_pcm_link() functionality and its using for
timer source setting, is removed (as rejected earlier). The changes in this
update are mainly related to the parameters handling and some cleanup.

The timer source can be initially selected by "timer_source" kernel module
parameter. It is supposed to have the following format:
    [<pref>:](<card name>|<card idx>)[{.,}<dev idx>[{.,}<subdev idx>]]
For example: "hw:I82801AAICH.1.0", or "1.1", or just "I82801AAICH".
(Prefix is ignored, just allowed here to be able to use the strings,
the user got used to).
Although the parsing function recognizes both '.' and ',' as a separator,
module parameters handling routines use ',' to separate parameters for
different module instances (array elements), so we have to use '.'
to separate device and subdevice numbers from the card name or number
in module parameters.
Empty string indicates using jiffies as a timer source.

Besides "static" selection of timer source at module load time,
it is possible to dynamically change it via sound "info" interface
(using "/proc/asound/<card>/timer_source" file in read-write mode.
The contents of this file is used as a timer source string for
a particular loopback card, e.g. "hw:0,0,0" (and here ',' can be used
as a separator).

The timer source string value can be changed at any time, but it is
latched by PCM substream open callback "loopback_open()" (the first
one for a particular cable). At this point it is actually used,
that is the string is parsed, and the timer is looked up and opened.
This seems to be a good trade-off between flexibility of updates and
synchronizations or racing complexity.

The timer source is set for a loopback card (the same as initial setting
by module parameter), but every cable uses the value, current at the moment
of opening. Theoretically, it's possible to set the timer source for each
cable independently (via separate files), but it would be inconsistent
with the initial setting via module parameters on a per-card basis.

v2:
https://mailman.alsa-project.org/pipermail/alsa-devel/2019-November/157961.html

v3:
- Change sound card lookup to use snd_card_ref() and avoid direct access
  to sound cards array
- Squash commits on returning error codes for timer start and stop
- Some locking related fixes
- Some code cleanup


Andrew Gabbasov (1):
  ALSA: aloop: Support runtime change of snd_timer via info interface

Timo Wischer (6):
  ALSA: aloop: Describe units of variables
  ALSA: aloop: Support return of error code for timer start and stop
  ALSA: aloop: Use callback functions for timer specific implementations
  ALSA: aloop: Rename all jiffies timer specific functions
  ALSA: aloop: Move CABLE_VALID_BOTH to the top of file
  ALSA: aloop: Support selection of snd_timer instead of jiffies

 sound/drivers/aloop.c | 672 +++++++++++++++++++++++++++++++++++++++---
 1 file changed, 637 insertions(+), 35 deletions(-)

Comments

Takashi Iwai Nov. 14, 2019, 5:10 p.m. UTC | #1
On Mon, 11 Nov 2019 12:08:39 +0100,
Andrew Gabbasov wrote:
> 
> This patch set is an updated version of patches by Timo Wischer:
> https://mailman.alsa-project.org/pipermail/alsa-devel/2019-March/146871.html
> 
> This patch set is required for forwarding audio data between a HW sound
> card and an aloop device without the usage of an asynchronous sample rate
> converter.
> 
> Most of sound and timers related code is kept the same as in previous set.
> The code, related to snd_pcm_link() functionality and its using for
> timer source setting, is removed (as rejected earlier). The changes in this
> update are mainly related to the parameters handling and some cleanup.
> 
> The timer source can be initially selected by "timer_source" kernel module
> parameter. It is supposed to have the following format:
>     [<pref>:](<card name>|<card idx>)[{.,}<dev idx>[{.,}<subdev idx>]]
> For example: "hw:I82801AAICH.1.0", or "1.1", or just "I82801AAICH".
> (Prefix is ignored, just allowed here to be able to use the strings,
> the user got used to).
> Although the parsing function recognizes both '.' and ',' as a separator,
> module parameters handling routines use ',' to separate parameters for
> different module instances (array elements), so we have to use '.'
> to separate device and subdevice numbers from the card name or number
> in module parameters.
> Empty string indicates using jiffies as a timer source.
> 
> Besides "static" selection of timer source at module load time,
> it is possible to dynamically change it via sound "info" interface
> (using "/proc/asound/<card>/timer_source" file in read-write mode.
> The contents of this file is used as a timer source string for
> a particular loopback card, e.g. "hw:0,0,0" (and here ',' can be used
> as a separator).
> 
> The timer source string value can be changed at any time, but it is
> latched by PCM substream open callback "loopback_open()" (the first
> one for a particular cable). At this point it is actually used,
> that is the string is parsed, and the timer is looked up and opened.
> This seems to be a good trade-off between flexibility of updates and
> synchronizations or racing complexity.
> 
> The timer source is set for a loopback card (the same as initial setting
> by module parameter), but every cable uses the value, current at the moment
> of opening. Theoretically, it's possible to set the timer source for each
> cable independently (via separate files), but it would be inconsistent
> with the initial setting via module parameters on a per-card basis.
> 
> v2:
> https://mailman.alsa-project.org/pipermail/alsa-devel/2019-November/157961.html
> 
> v3:
> - Change sound card lookup to use snd_card_ref() and avoid direct access
>   to sound cards array
> - Squash commits on returning error codes for timer start and stop
> - Some locking related fixes
> - Some code cleanup

The patch won't work with the latest ALSA timer code found in my
for-next branch due to the API changes.  Essentially you need to
rewrite as:
	timeri = snd_timer_instance_new(...);
	if (!timeri)
		no_memory_error;
	timeri->flags |= ...;
	timeri->ccallback = ...;
	....
	err = snd_timer_open(timeri, ....);
	if (err < 0)
		error;


thanks,

Takashi
Gabbasov, Andrew Nov. 20, 2019, 12:15 p.m. UTC | #2
Hello Takashi,

> -----Original Message-----
> From: Takashi Iwai [mailto:tiwai@suse.de]
> Sent: Thursday, November 14, 2019 8:11 PM
> To: Gabbasov, Andrew
> Cc: alsa-devel@alsa-project.org; linux-kernel@vger.kernel.org; Jaroslav
> Kysela; Takashi Iwai; Timo Wischer
> Subject: Re: [PATCH v3 0/7] ALSA: aloop: Support sound timer as clock
> source instead of jiffies
> 
> On Mon, 11 Nov 2019 12:08:39 +0100,
> Andrew Gabbasov wrote:
> >
> > This patch set is an updated version of patches by Timo Wischer:
> > https://mailman.alsa-project.org/pipermail/alsa-devel/2019-
> March/146871.html
> >
> > This patch set is required for forwarding audio data between a HW sound
> > card and an aloop device without the usage of an asynchronous sample
rate
> > converter.
> >
> > Most of sound and timers related code is kept the same as in previous
> set.
> > The code, related to snd_pcm_link() functionality and its using for
> > timer source setting, is removed (as rejected earlier). The changes in
> this
> > update are mainly related to the parameters handling and some cleanup.
> >
> > The timer source can be initially selected by "timer_source" kernel
> module
> > parameter. It is supposed to have the following format:
> >     [<pref>:](<card name>|<card idx>)[{.,}<dev idx>[{.,}<subdev idx>]]
> > For example: "hw:I82801AAICH.1.0", or "1.1", or just "I82801AAICH".
> > (Prefix is ignored, just allowed here to be able to use the strings,
> > the user got used to).
> > Although the parsing function recognizes both '.' and ',' as a
separator,
> > module parameters handling routines use ',' to separate parameters for
> > different module instances (array elements), so we have to use '.'
> > to separate device and subdevice numbers from the card name or number
> > in module parameters.
> > Empty string indicates using jiffies as a timer source.
> >
> > Besides "static" selection of timer source at module load time,
> > it is possible to dynamically change it via sound "info" interface
> > (using "/proc/asound/<card>/timer_source" file in read-write mode.
> > The contents of this file is used as a timer source string for
> > a particular loopback card, e.g. "hw:0,0,0" (and here ',' can be used
> > as a separator).
> >
> > The timer source string value can be changed at any time, but it is
> > latched by PCM substream open callback "loopback_open()" (the first
> > one for a particular cable). At this point it is actually used,
> > that is the string is parsed, and the timer is looked up and opened.
> > This seems to be a good trade-off between flexibility of updates and
> > synchronizations or racing complexity.
> >
> > The timer source is set for a loopback card (the same as initial setting
> > by module parameter), but every cable uses the value, current at the
> moment
> > of opening. Theoretically, it's possible to set the timer source for
each
> > cable independently (via separate files), but it would be inconsistent
> > with the initial setting via module parameters on a per-card basis.
> >
> > v2:
> > https://mailman.alsa-project.org/pipermail/alsa-devel/2019-
> November/157961.html
> >
> > v3:
> > - Change sound card lookup to use snd_card_ref() and avoid direct access
> >   to sound cards array
> > - Squash commits on returning error codes for timer start and stop
> > - Some locking related fixes
> > - Some code cleanup
> 
> The patch won't work with the latest ALSA timer code found in my
> for-next branch due to the API changes.  Essentially you need to
> rewrite as:
> 	timeri = snd_timer_instance_new(...);
> 	if (!timeri)
> 		no_memory_error;
> 	timeri->flags |= ...;
> 	timeri->ccallback = ...;
> 	....
> 	err = snd_timer_open(timeri, ....);
> 	if (err < 0)
> 		error;
> 

The updated patch set version 4 was sent to mailing list:
https://mailman.alsa-project.org/pipermail/alsa-devel/2019-November/158896.h
tml

It is based on the linux-next repo master branch as of Nov 19, 2019.

Thanks!

Best regards,
Andrew