[i2c] [patch 2.6.21-rc3-git +i2c] switch at91 to new-style i2c drivers

[David Brownell]

On Saturday 10 March 2007 12:33 pm, Jean Delvare wrote:

> >  
> > +/* work with "modprobe at91_i2c" from hotplugging or coldplugging */
> > +MODULE_ALIAS("at91_i2c");
> 
> What is it useful for?

Mostly coldplugging.  The general rule has some startup agent
scanning /sys/devices and then doing a "modprobe $(cat modalias)"
for any device that doesn't have a driver link.  Since the
name of that module is "i2c-at91", which doesn't match the
driver name (which does match the device name) such a modprobe
would fail without the MODULE_ALIAS directive.


> > --- at91.orig/arch/arm/mach-at91/board-dk.c	2007-03-04 15:40:41.000000000 -0800
> > +++ at91/arch/arm/mach-at91/board-dk.c	2007-03-08 09:21:56.000000000 -0800
> > @@ -99,6 +99,14 @@ static struct at91_mmc_data __initdata d
> >  	.wire4		= 1,
> >  };
> >  
> > +static struct i2c_board_info __initdata dk_i2c_devices[] = { {
> > +	I2C_BOARD_INFO("ics1523", 0x26, ""),
> 
> Why not simply NULL?

Does that even work when initializing an array?  I know that "" does.

Better would be to not include the type name in that macro.  Judging
by that small sample, it's not needed often enough.

> 
> > +}, {
> > +	I2C_BOARD_INFO("x9429", 0x28, ""),
> > +}, {
> > +	I2C_BOARD_INFO("at24c", 0x50, "24c1024"),
> > +} };
> 
> Also, I guess it's a matter of personal taste, but:
> 
> static struct i2c_board_info __initdata dk_i2c_devices[] = {
> 	{ I2C_BOARD_INFO("ics1523", 0x26, "") },
> 	{ I2C_BOARD_INFO("x9429", 0x28, "") },
> 	{ I2C_BOARD_INFO("at24c", 0x50, "24c1024") },

Except that the at24c lines are going to need to pass
some platform_data, so it won't fit on one line; there
will be the basic info, then another line.

- Dave


> };
> 
> looks much better IMVHO.
> 
> -- 
> Jean Delvare
>