Signaling user-space through kcontrol

Signaling user-space through kcontrol

[Patrick Lai]

Hi Mark,

The codec I am working on, beside headset detection, also have other
type of events to be propagated to user space. Per our discussion at
IPC, there is way to signal user-space through kcontrol. Is it through 
calling snd_ctl_notify()? What is the ALSA library API to intercept the 
event? Is there an sample user-space code which gets kcontrol event?

Thanks
Patrick

-- 
Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

Re: Signaling user-space through kcontrol

[Vinod Koul]

On Tue, 2011-09-27 at 23:14 -0700, Patrick Lai wrote:
> Hi Mark,
> 
> The codec I am working on, beside headset detection, also have other
> type of events to be propagated to user space. Per our discussion at
> IPC, there is way to signal user-space through kcontrol. Is it through 
> calling snd_ctl_notify()? What is the ALSA library API to intercept the 
> event? Is there an sample user-space code which gets kcontrol event?
> 
What events do you have in mind?

-- 
~Vinod

Re: Signaling user-space through kcontrol

[Patrick Lai]

On 9/27/2011 11:15 PM, Vinod Koul wrote:
> On Tue, 2011-09-27 at 23:14 -0700, Patrick Lai wrote:
>> Hi Mark,
>>
>> The codec I am working on, beside headset detection, also have other
>> type of events to be propagated to user space. Per our discussion at
>> IPC, there is way to signal user-space through kcontrol. Is it through
>> calling snd_ctl_notify()? What is the ALSA library API to intercept the
>> event? Is there an sample user-space code which gets kcontrol event?
>>
> What events do you have in mind?
>
Main one is detection of short circuiting headphone For this case,
there is no need for extra payload. In case of there is need to pass
back some payload, I presume user-space application can simply call the
get function upon receiving notification.

Thanks
Patrick

-- 
Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

Re: Signaling user-space through kcontrol

[Vinod Koul]

On Tue, 2011-09-27 at 23:47 -0700, Patrick Lai wrote:
> On 9/27/2011 11:15 PM, Vinod Koul wrote:
> > On Tue, 2011-09-27 at 23:14 -0700, Patrick Lai wrote:
> >> Hi Mark,
> >>
> >> The codec I am working on, beside headset detection, also have other
> >> type of events to be propagated to user space. Per our discussion at
> >> IPC, there is way to signal user-space through kcontrol. Is it through
> >> calling snd_ctl_notify()? What is the ALSA library API to intercept the
> >> event? Is there an sample user-space code which gets kcontrol event?
> >>
> > What events do you have in mind?
> >
> Main one is detection of short circuiting headphone For this case,
> there is no need for extra payload. In case of there is need to pass
> back some payload, I presume user-space application can simply call the
> get function upon receiving notification.
And what would userspace do with that information? Apart from logging
this, is there anything useful we can do with such info

-- 
~Vinod

Re: Signaling user-space through kcontrol

[Patrick Lai]

On 9/28/2011 12:03 AM, Vinod Koul wrote:
>> Main one is detection of short circuiting headphone For this case,
>> there is no need for extra payload. In case of there is need to pass
>> back some payload, I presume user-space application can simply call the
>> get function upon receiving notification.
> And what would userspace do with that information? Apart from logging
> this, is there anything useful we can do with such info
>
Well, it's up to user-space application to decide. I suppose
application would not allow audio route to headphone/headset at this
point.

Thanks
Patrick

-- 
Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

Re: Signaling user-space through kcontrol

[Mark Brown]

On Wed, Sep 28, 2011 at 12:26:43AM -0700, Patrick Lai wrote:
> On 9/28/2011 12:03 AM, Vinod Koul wrote:

> >> Main one is detection of short circuiting headphone For this case,
> >> there is no need for extra payload. In case of there is need to pass
> >> back some payload, I presume user-space application can simply call the
> >> get function upon receiving notification.

> > And what would userspace do with that information? Apart from logging
> > this, is there anything useful we can do with such info

> Well, it's up to user-space application to decide. I suppose
> application would not allow audio route to headphone/headset at this
> point.

Something like that specific example should probably be reported via the
snd_jack accessory detection infrastructure which can then offer a
consistent interface between all sound cards with minimal work on the
part of the individual driver.  There will doubtless be other examples
though.

Re: Signaling user-space through kcontrol

[Patrick Lai]

> Something like that specific example should probably be reported via the
> snd_jack accessory detection infrastructure which can then offer a
> consistent interface between all sound cards with minimal work on the
> part of the individual driver.  There will doubtless be other examples
> though.

Agree that should go through consistent interface but not exactly sure
if going through input device interface is appropriate though. Isn't
the same reasoning why it has been decided to create ALSA specific API
for headset detection?

-- 
Sent by an employee of the Qualcomm Innovation Center, Inc.
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum.

Re: Signaling user-space through kcontrol

[Mark Brown]

On Wed, Sep 28, 2011 at 09:27:31AM -0700, Patrick Lai wrote:

> Agree that should go through consistent interface but not exactly sure
> if going through input device interface is appropriate though. Isn't
> the same reasoning why it has been decided to create ALSA specific API
> for headset detection?

That's why I said use the jack API not the input API, if you're open
coding an input interface for jacks in ALSA you're doing something
wrong.  Just use the jack API and let it worry about how to tell the
rest of the system about what's going on - a driver shouldn't have to
worry about the details.

Re: Signaling user-space through kcontrol

[Clemens Ladisch]

Patrick Lai wrote:
> The codec I am working on, beside headset detection, also have other
> type of events to be propagated to user space. Per our discussion at
> IPC, there is way to signal user-space through kcontrol. Is it through 
> calling snd_ctl_notify()?

Yes.

> What is the ALSA library API to intercept the event?

snd_ctl_open() and snd_ctl_subscribe(), then snd_ctl_read() in
a loop.  To integrate this into a poll()-based event loop, use
the snd_ctl_poll_descriptors*() functions.

Do you want to write a specific tool that does nothing but waiting
for these events, or will this be integrated into another program?

> Is there an sample user-space code which gets kcontrol event?

Not really.  There is alsamixer, but it uses a different API (the
snd_mixer* functions implement an abstracation on top of the snd_ctl*
stuff).  In the main loop (alsa-utils/alsamixer/mainloop.c), the
snd_mixer_poll_descriptors* functions work similar to the snd_ctl*
functions, but the snd_mixer_handle_events() function then reads the
event(s) from the ctl device and calls callback functions, in this case
elem_callback() and mixer_callback() in mixer_widget.c.


Regards,
Clemens

Re: Signaling user-space through kcontrol

[Sebastian H.]

> Is there an sample user-space code which gets kcontrol event?

QasMixer supports snd_hctl_* controls beside snd_mixer_selem_* controls.
(select the element mixer view)
AFAIK snd_hctl_* is only  thin wrapper over snd_ctl_*.

Sebastian