[PATCH] hdspm: native float

[PATCH] hdspm: native float

[Florian Faber]

Adds native float support to the driver. The interface has hardware
support for CoreAudio[tm], why shouldn't jack users benefit from it.

diff a/sound/pci/rme9652/hdspm.c b/sound/pci/rme9652/hdspm.c

13a14,16
>  *      Modified 2009-04-14 for native float support by Florian Faber
>  *                                               <faber@faberman.de>
>  *
39a43
> #include <sound/pcm_params.h>
198a203
> #define HDSPe_FLOAT_FORMAT         0x2000000
3726a3732,3747
> 
> 
> 	/* Switch to native float format if requested */
> 	if (SNDRV_PCM_FORMAT_FLOAT_LE == params_format(params)) {
> 	  if (!(hdspm->control_register & HDSPe_FLOAT_FORMAT)) {
> 	    snd_printk(KERN_INFO "hdspm: Switching to native 32bit LE float
format.\n");
> 	  }
> 	  hdspm->control_register |= HDSPe_FLOAT_FORMAT;
> 	} else if (SNDRV_PCM_FORMAT_S32_LE == params_format(params)) {
> 	  if (hdspm->control_register & HDSPe_FLOAT_FORMAT) {
> 	    snd_printk(KERN_INFO "hdspm: Switching to native 32bit LE integer
format.\n");
> 	  }
> 	  hdspm->control_register &= ~HDSPe_FLOAT_FORMAT;
> 	}
> 	hdspm_write(hdspm, HDSPM_controlRegister, hdspm->control_register);
> 
3870c3891
< 	.formats = SNDRV_PCM_FMTBIT_S32_LE,
---
> 	.formats = SNDRV_PCM_FMTBIT_S32_LE | SNDRV_PCM_FMTBIT_FLOAT_LE,
3895c3916
< 	.formats = SNDRV_PCM_FMTBIT_S32_LE,
---
> 	.formats = SNDRV_PCM_FMTBIT_S32_LE | SNDRV_PCM_FMTBIT_FLOAT_LE,

Flo
-- 
Machines can do the work, so people have time to think.
public key 6C002249          x-hkp://wwwkeys.eu.pgp.net

Re: [PATCH] hdspm: native float

[Clemens Ladisch]

Florian Faber wrote:
> diff a/sound/pci/rme9652/hdspm.c b/sound/pci/rme9652/hdspm.c

Please use unified diff format and add Signed-off-by; see
linux/Documentation/SubmittingPatches.

> >  *      Modified 2009-04-14 for native float support by Florian Faber
> >  *                                               <faber@faberman.de>
> >  *

Changelogs are now in git.

> > 	/* Switch to native float format if requested */
> > ...
> > 	hdspm_write(hdspm, HDSPM_controlRegister, hdspm->control_register);

This looks like a global change.  What happens when playback and
recording want to use different formats?


Best regards,
Clemens

Re: [PATCH] hdspm: native float

[Florian Faber]

Clemens,

> > > 	/* Switch to native float format if requested */
> > > ...
> > > 	hdspm_write(hdspm, HDSPM_controlRegister, hdspm->control_register);
> This looks like a global change.  What happens when playback and
> recording want to use different formats?

It is a global change. The only use case is for jack, and jack doesn't
use different formats for input/output.


-- 
Flo

Re: [PATCH] hdspm: native float

[Clemens Ladisch]

Florian Faber wrote:
> > > > 	/* Switch to native float format if requested */
> > > > ...
> > > > 	hdspm_write(hdspm, HDSPM_controlRegister, hdspm->control_register);
> > 
> > This looks like a global change.  What happens when playback and
> > recording want to use different formats?
> 
> It is a global change. The only use case is for jack, and jack doesn't
> use different formats for input/output.

"Using aplay + arecord" is a second use case.  :-)

Anyway, at least add a check that the formats are actually the same
(like the checks for sample rate and period size).


Best regards,
Clemens

Re: [PATCH] hdspm: native float

[Florian Faber]

Clemens,

> > It is a global change. The only use case is for jack, and jack doesn't
> > use different formats for input/output.
> "Using aplay + arecord" is a second use case.  :-)

And they would willingly choose different formats?

> Anyway, at least add a check that the formats are actually the same
> (like the checks for sample rate and period size).

Ok.

I just fixed a bug regarding the clock source anyway
(LAC-live-debugging :).

-- 
Flo

Re: [PATCH] hdspm: native float

[Takashi Iwai]

At Thu, 16 Apr 2009 14:12:21 +0200,
Florian Faber wrote:
> 
> > > It is a global change. The only use case is for jack, and jack doesn't
> > > use different formats for input/output.
> > "Using aplay + arecord" is a second use case.  :-)
> 
> And they would willingly choose different formats?

Possibly.  It's not always the same user who play and record.


thanks,

Takashi

Re: [PATCH] hdspm: native float

[Mark Brown]

On Thu, Apr 16, 2009 at 02:12:21PM +0200, Florian Faber wrote:

> > > It is a global change. The only use case is for jack, and jack doesn't
> > > use different formats for input/output.
> > "Using aplay + arecord" is a second use case.  :-)

> And they would willingly choose different formats?

A moderately common case is recording voice quality audio (at 8kHz or
so) and playing back higher quality audio.

Re: [PATCH] hdspm: native float

[John Rigg]

On Thu, Apr 16, 2009 at 01:41:59PM +0100, Mark Brown wrote:
> On Thu, Apr 16, 2009 at 02:12:21PM +0200, Florian Faber wrote:
> 
> > > > It is a global change. The only use case is for jack, and jack doesn't
> > > > use different formats for input/output.
> > > "Using aplay + arecord" is a second use case.  :-)
> 
> > And they would willingly choose different formats?
> 
> A moderately common case is recording voice quality audio (at 8kHz or
> so) and playing back higher quality audio.

Perhaps a little less common on a 64-channel MADI interface
like this :-)

John

Re: [PATCH] hdspm: native float

[John Rigg]

On Thu, Apr 16, 2009 at 01:41:59PM +0100, Mark Brown wrote:
> On Thu, Apr 16, 2009 at 02:12:21PM +0200, Florian Faber wrote:
> 
> > > > It is a global change. The only use case is for jack, and jack doesn't
> > > > use different formats for input/output.
> > > "Using aplay + arecord" is a second use case.  :-)
> 
> > And they would willingly choose different formats?
> 
> A moderately common case is recording voice quality audio (at 8kHz or
> so) and playing back higher quality audio.

Perhaps a little less common on a 64-channel MADI interface
like this :-)

John