Re: [alsa-devel] [PATCH] ASoC: da7210: Add support for spi regmap

Re: [alsa-devel] [PATCH] ASoC: da7210: Add support for spi regmap

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 1622 bytes --]

On Wed, Mar 21, 2012 at 09:22:50PM +0530, Ashish Chavan wrote:

> +#if defined(CONFIG_SPI_MASTER)
> +	/* Dummy read to give two pulses over nCS */
> +	regmap_read(da7210->regmap, DA7210_STATUS, &val);
> +	regmap_read(da7210->regmap, DA7210_STATUS, &val);
> +	regmap_read(da7210->regmap, DA7210_STATUS, &val);
> +#endif

This ifdef stuff should all be runtime configured based on the bus type,
unless the cost of the reads is considered immaterial in which case it
should be unconditional.

> +#if defined(CONFIG_SPI_MASTER)
> +	snd_soc_write(codec, 0x00, 0x80);
> +#endif

Hrm?

>  	/* unlock */
>  	regmap_write(da7210->regmap, DA7210_A_HID_UNLOCK,	0x8B);
> @@ -1035,6 +1056,10 @@ static int da7210_probe(struct snd_soc_codec *codec)
>  	regmap_write(da7210->regmap, DA7210_A_HID_UNLOCK,	0x00);
>  	regmap_write(da7210->regmap, DA7210_A_TEST_UNLOCK,	0x00);

I also note that you've not yet updated this to use a regmap patch as
was previously requested.

> +#if defined(CONFIG_SPI_MASTER)
> +static int __devinit da7210_spi_probe(struct spi_device *spi)
> +{
> +	struct da7210_priv *da7210;
> +	int ret;
> +
> +	da7210 = kzalloc(sizeof(struct da7210_priv), GFP_KERNEL);
> +	if (!da7210)
> +		return -ENOMEM;

devm_kzalloc().

> +	da7210_regmap.read_flag_mask = 0x01;
> +	da7210_regmap.write_flag_mask = 0x00;

Just have a second, static, regmap variable.  The regmap should be
declared const.

> +	da7210->regmap = regmap_init_spi(spi, &da7210_regmap);

devm_regmap_init_spi() (will come in in the merge window).

> +static struct spi_driver da7210_spi_driver = {
> +	.driver = {
> +		.name = "da7210-codec",

No -codec.

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [alsa-devel] [PATCH] ASoC: da7210: Add support for spi regmap

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 1293 bytes --]

On Wed, Mar 21, 2012 at 10:11:20PM +0530, Ashish Chavan wrote:
> On Wed, 2012-03-21 at 15:56 +0000, Mark Brown wrote:

> This is writing to page register. SPI register space is divided in to
> two pages. Registers from 0x01 to 0x80 fall in to first page. If we want
> to write to any register above 0x80, first we need to set page register
> with PAGE1. May be I should put comments to make it obvious.

You also need to make sure that the register cache doesn't get confused.

> > >  	/* unlock */
> > >  	regmap_write(da7210->regmap, DA7210_A_HID_UNLOCK,	0x8B);
> > > @@ -1035,6 +1056,10 @@ static int da7210_probe(struct snd_soc_codec *codec)
> > >  	regmap_write(da7210->regmap, DA7210_A_HID_UNLOCK,	0x00);
> > >  	regmap_write(da7210->regmap, DA7210_A_TEST_UNLOCK,	0x00);

> > I also note that you've not yet updated this to use a regmap patch as
> > was previously requested.

> Actually I don't have enough details about the problem that you
> mentioned with this. I am waiting for inputs from some body else. Can

...and didn't bother asking any questions so it's unlikely anyone will
say anything...

> you please elaborate the problem if you have details? so that I can
> correct it.

What is unclear in the previous feedback?  You should be converting this
to use a regmap patch...

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [alsa-devel] [PATCH] ASoC: da7210: Add support for spi regmap

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 959 bytes --]

On Wed, Mar 21, 2012 at 10:35:53PM +0530, Ashish Chavan wrote:

> > > Actually I don't have enough details about the problem that you
> > > mentioned with this. I am waiting for inputs from some body else. Can

> > ...and didn't bother asking any questions so it's unlikely anyone will
> > say anything...

> No, actually it's not like that. You mentioned that the sequence should
> be changed, which will fix an issue that the driver has with suspend and
> resume. I asked other users of this driver about the issue and waiting
> for their feedback. I just wanted to have clear picture about the issue
> before I commit any change there. Just to make sure that only changing
> of sequence is enough or some thing extra need to be done.

No, I didn't say that the sequence should be changed.  I said that it
should be implemented using the relevant framework features rather than
being open coded.  The sequence itself is fine from a framework point of
view.

[-- Attachment #2: Digital signature --]
[-- Type: application/pgp-signature, Size: 836 bytes --]