sw parameter boundary

sw parameter boundary

[Norbert van Bolhuis]

is the boundary sw parameter still used ?
where is it for ?

in the PCM kernel core (e.g. snd_pcm_playback_avail) it clearly serves
as a pointers wrap point, but which one ?
The one at the end of the DMA buffer or the one at the end of a process
(virtual) address range (4 GB) ?




-- 
This message has been scanned for viruses and is believed to be clean

Re: sw parameter boundary

[Takashi Iwai]

At Wed, 25 Jun 2008 10:44:19 +0200,
Norbert van Bolhuis wrote:
> 
> 
> is the boundary sw parameter still used ?

Yes.

> where is it for ?

It defines the wrap point to the position 0.

> in the PCM kernel core (e.g. snd_pcm_playback_avail) it clearly serves
> as a pointers wrap point, but which one ?

It's for the absolute position handled internally.  This isn't
exported outside, so you don't care much about it, normally.

> The one at the end of the DMA buffer or the one at the end of a process
> (virtual) address range (4 GB) ?

No.


Takashi

Re: sw parameter boundary

[Norbert van Bolhuis]

> It's for the absolute position handled internally.  This isn't
> exported outside, so you don't care much about it, normally.
> 

OK, but what does an ALSA driver need to do with boundary, nothing ?

-- 
This message has been scanned for viruses and is believed to be clean

Re: sw parameter boundary

[Takashi Iwai]

At Thu, 26 Jun 2008 10:43:08 +0200,
Norbert van Bolhuis wrote:
> 
> 
> > It's for the absolute position handled internally.  This isn't
> > exported outside, so you don't care much about it, normally.
> > 
> 
> OK, but what does an ALSA driver need to do with boundary, nothing ?

The boundary is aligned to buffer_size so that the offset can be
calculated properly even after the overwrap.


Takashi

Re: sw parameter boundary

[Norbert van Bolhuis]

 >> OK, but what does an ALSA driver need to do with boundary, nothing ?
 >
 > The boundary is aligned to buffer_size so that the offset can be
 > calculated properly even after the overwrap.
 >

OK, so I assume that's a yes.
I don't see any driver maniplating boundary though.


-- 
This message has been scanned for viruses and is believed to be clean