Re: [PATCH v2] soc/lapis: add machine driver

[Tomoya MORINAGA]

2011/12/6 Mark Brown <broonie@xxxxxxxxxxxxxxxxxxxxxxxxxxx>:
>> According to "machine.txt", I created this machine driver referring
>> corgi.c/spitz.c.
>> However, corgi.c/spitz.c. don't look obeying your saying.
>
> You really need to look at current drivers to make sure you're following
> current best practice - older drivers, particularly for obsolte hardware
> like the Zaurus machines, may well not reflect the current best practices.

I see.
I won't read this docs.

BTW, Let me know the best practice of machine driver?
I can't identify which driver is modern or not.


>> 2011/12/3 Mark Brown <broonie@xxxxxxxxxxxxxxxxxxxxxxxxxxx>:
>
>> >> +     {I2C_BOARD_INFO("ioh_i2c-5", CODEC_DEV_ADDR + 0)},
>
>> > This looks completely wrong.  I'd not expect to see any I2C_BOARD_INFO
>> > usage at all in a machine driver (that should be done by whatever
>> > enumerates the system as a whole)
>
>> Do you mean machine driver must not use I2C_BOARD_INFO ? Is this true ?
>> Grepping I2C_BOARD_INFO at sound/soc,  the following drivers use
>> I2C_BOARD_INFO.
>> pxa/raumfeld.c: I2C_BOARD_INFO("max9485", 0x63),
>> pxa/magician.c:         I2C_BOARD_INFO("uda1380", 0x18),
>> s6000/s6105-ipcam.c:    { I2C_BOARD_INFO("tlv320aic33", 0x18), }
>
> You'll note that this is *really* rare; there are vastly more machine
> drivers that don't do this.  If there's some specific reason then
> there's nothing technical stopping this happening but it's very unusual.
I see.
I want to know how to use i2c device.
Let me know the best practice driver uses i2c device.

>
>> >> +     ret = snd_soc_dai_set_fmt(codec_dai, SND_SOC_DAIFMT_I2S |
>> >> +                             SND_SOC_DAIFMT_NB_NF | SND_SOC_DAIFMT_CBS_CFS);
>> >> +     if (ret < 0)
>> >> +             return ret;
>
>> > Use the dai_fmt field in the dai_link to set this.
>
>> Sorry, I can't understand your saying.
>> Let me know in more detail.
>
> I'm not sure what more to say...  Have you looked at that field and how
> it is both implemented and used in other drivers, or at the commit logs
> for relevant changes?  What do you find unclear?

You said "Use the dai_fmt field in the dai_link to set this."
However, both dai_fmt and dai_link are already implemented like below.
static struct snd_soc_dai_link ioh_i2s_dai = {
...
};

static struct snd_soc_card ioh_i2s_card = {
...
	.dai_link	= &ioh_i2s_dai,
...
};
So, I can't understand your saying.

>
>> > Should have MODULE_ALIAS too.
>
>> Do you mean machine driver should have MODULE_ALIAS ?
>> Grepping  MODULE_ALIAS at sound/soc, it seems MODULE_ALIAS is used in
>> platform driver only like below.
>> pxa/pxa2xx-i2s.c:MODULE_ALIAS("platform:pxa2xx-i2s");
>
> I don't think you're looking at a current kernel...
>
>> Does machine driver need like "MODULE_ALIAS("machine:ml7213CRB")" ?
>
> No, it should match the name of the platform driver you created.  The
> platform: refers to the bus type.

I understood.

thanks,

tomoya
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/