On Fri, Dec 21, 2012 at 01:47:18AM -0800, Andrey Smirnov wrote:

This looks really good, the issues and questions I have below are pretty
detailed.

> -	int (*reg_read)(struct regmap *map, unsigned int reg, unsigned int *val);
> -	int (*reg_write)(struct regmap *map, unsigned int reg, unsigned int val);
> +	int (*reg_read)(void *context, unsigned int reg, unsigned int *val);
> +	int (*reg_write)(void *context, unsigned int reg, unsigned int val);

I'd be inclined to just do this in the initial refectoring patches
rather than rerefactoring here.

> +	if (!bus || !bus->fast_io) {
>  		mutex_init(&map->mutex);
>  		map->lock = regmap_lock_mutex;
>  		map->unlock = regmap_unlock_mutex;
> +	} else {
> +		spin_lock_init(&map->spinlock);
> +		map->lock = regmap_lock_spinlock;
> +		map->unlock = regmap_unlock_spinlock;

It's not immediately obvious to me that no-bus should be forced to use
mutexes - is there any great reason for tying the two together?  I'd add
a flag to allow no-bus devices to choose, possibly as part of a separate
"bus" configuration thing that gets configured with a separate init
function.

> +	if (!bus) {
> +		map->cache_registers = true;
> +		goto skip_format_initialization;
> +	} else {
> +		map->reg_read = _regmap_bus_read;
> +	}

Not sure I understand cache_registers here.  Why has this flag been
added?

> + * @reg_read: Optional callback that if filled will be used to perform
> + *            all the reads from the registers.
> + * @reg_write: Optional callback that if filled will be used to perform
> + *             all the writes to the registers.

I'd probably add some comment about not using this in conjunction with
SPI or I2C.