Does anyone really have success experience with SPI codecs with new name-based SOC API?

Does anyone really have success experience with SPI codecs with new name-based SOC API?

[Alexander]

Does anyone really have success experience with SPI codecs with new
name-based SOC API?
There are really few SPI codecs in ASoC subsystem, so I wonder if only I
cannot bind my SPI codec to machine driver.

The codec is registered with name spi0.0 it's the behavior of SPI
subsystem, I suppose. So the form is %s%d.%d and not %s.%d as supposed
in fmt_single_name() in soc_core.c

For me
dev_name(dev)==spi0.0
dev->driver->name==cs4271-codec

Is this really working code?
-----
	strncpy(name, dev_name(dev), NAME_SIZE);

	/* are we a "%s.%d" name (platform and SPI components) */
	found = strstr(name, dev->driver->name);
	if (found) {
		/* get ID */
		if (sscanf(&found[strlen(dev->driver->name)], ".%d", id) == 1) {

			/* discard ID from name if ID == -1 */
			if (*id == -1)
				found[strlen(dev->driver->name)] = '\0';
		}

	} else {
-----

Re: Does anyone really have success experience with SPI codecs with new name-based SOC API?

[Dimitris Papastamos]

On Thu, 2010-11-18 at 14:31 +0300, Alexander wrote:
> Does anyone really have success experience with SPI codecs with new
> name-based SOC API?
> There are really few SPI codecs in ASoC subsystem, so I wonder if only I
> cannot bind my SPI codec to machine driver.

Just use the registered codec name in the machine driver so soc-core can
match it correctly.  You should not worry about incompatibilities in the
naming convention between SPI registered devices and I2C ones.  The only
thing that we try to guarantee is to be able to uniquely identify the
devices on the system.

Thanks,
Dimitrios

Re: Does anyone really have success experience with SPI codecs with new name-based SOC API?

[Dimitris Papastamos]

On Thu, 2010-11-18 at 14:31 +0300, Alexander wrote:
> Does anyone really have success experience with SPI codecs with new
> name-based SOC API?
> There are really few SPI codecs in ASoC subsystem, so I wonder if only I
> cannot bind my SPI codec to machine driver.
> 
> The codec is registered with name spi0.0 it's the behavior of SPI
> subsystem, I suppose. So the form is %s%d.%d and not %s.%d as supposed
> in fmt_single_name() in soc_core.c

Just use the registered codec name in the machine driver so soc-core can
match it correctly.  You should not worry about incompatibilities in the
naming convention between SPI registered devices and I2C ones.  The only
thing that we try to guarantee is to be able to uniquely identify the
devices on the system.

Thanks,
Dimitrios

Re: Does anyone really have success experience with SPI codecs with new name-based SOC API?

[Fabio Estevam]

Hi Alexander,

On Thu, Nov 18, 2010 at 9:31 AM, Alexander <subaparts@yandex.ru> wrote:
> Does anyone really have success experience with SPI codecs with new
> name-based SOC API?
> There are really few SPI codecs in ASoC subsystem, so I wonder if only I
> cannot bind my SPI codec to machine driver.
>
> The codec is registered with name spi0.0 it's the behavior of SPI
> subsystem, I suppose. So the form is %s%d.%d and not %s.%d as supposed
> in fmt_single_name() in soc_core.c
>
> For me
> dev_name(dev)==spi0.0
> dev->driver->name==cs4271-codec
>
> Is this really working code?
> -----
>        strncpy(name, dev_name(dev), NAME_SIZE);
>
>        /* are we a "%s.%d" name (platform and SPI components) */
>        found = strstr(name, dev->driver->name);
>        if (found) {
>                /* get ID */
>                if (sscanf(&found[strlen(dev->driver->name)], ".%d", id) == 1) {
>
>                        /* discard ID from name if ID == -1 */
>                        if (*id == -1)
>                                found[strlen(dev->driver->name)] = '\0';
>                }
>
>        } else {
> -----

Have you managed to bind the SPI codec? I am trying the same thing
here. Please let me know.

Regards,

Fabio Estevam
_______________________________________________
Alsa-devel mailing list
Alsa-devel@alsa-project.org
http://mailman.alsa-project.org/mailman/listinfo/alsa-devel

Re: Does anyone really have success experience with SPI codecs with new name-based SOC API?

[Alexander]

On Thu, 2010-12-02 at 16:12 -0200, Fabio Estevam wrote:
> On Thu, Nov 18, 2010 at 9:31 AM, Alexander <subaparts@yandex.ru> wrote:
> > Does anyone really have success experience with SPI codecs with new
> > name-based SOC API?
> > There are really few SPI codecs in ASoC subsystem, so I wonder if only I
> > cannot bind my SPI codec to machine driver.
> >
> > The codec is registered with name spi0.0 it's the behavior of SPI
> > subsystem, I suppose. So the form is %s%d.%d and not %s.%d as supposed
> > in fmt_single_name() in soc_core.c
> >
> Have you managed to bind the SPI codec? I am trying the same thing
> here. Please let me know.

I've received a reply from Dimitris Papastamos, SoC subsystem developer,
and he written, that I should use it as is, ie spi0.0
So I'm finished with names and successfully binded it as
codecname==spi0.0

But for now, I have no sound anyway, it seems that something in SPI path
is broken (but I suppose, not in my code, as it works on 2.6.36-rc6), I
cannot see any communications going to codec with oscilloscope.
So I'm trying to figure it out now...

Best regards,
	Alexander A. Sverdlin.