[U-Boot] [PATCH 0/3] Bring in new I2C framework

[Simon Glass]

Hi,

On Tue, Oct 30, 2012 at 9:50 AM, Stephen Warren <swarren at wwwdotorg.org> wrote:
> On 10/29/2012 11:57 PM, Heiko Schocher wrote:
>> Hello Stephen,
>>
>> On 29.10.2012 16:34, Stephen Warren wrote:
> ...
>>> If there are e.g. 4 I2C controllers in an SoC, the driver needs to know
>>> which one is in use. Passing that information directly to the driver
>>> functions is much simple than requiring the SoC I2C driver to go grovel
>>> in some I2C core global variables to find out the same information.
>>
>> Ah, do you mean we should change the i2c adapter struct from:
>>
>> struct i2c_adapter {
>>        void            (*init)(int speed, int slaveaddr);
>>        int             (*probe)(uint8_t chip);
>>        int             (*read)(uint8_t chip, uint addr, int alen,
>>                                uint8_t *buffer, int len);
>>        int             (*write)(uint8_t chip, uint addr, int alen,
>>                                uint8_t *buffer, int len);
>>        uint            (*set_bus_speed)(uint speed);
>> [...]
>>
>> to
>>
>> struct i2c_adapter {
>>        void            (*init)(struct i2c_adapter *adap, int speed, int
>> slaveaddr);
>>        int             (*probe)(struct i2c_adapter *adap, uint8_t chip);
>>        int             (*read)(struct i2c_adapter *adap, uint8_t chip,
>> uint addr, int alen,
>>                                uint8_t *buffer, int len);
>>        int             (*write)(struct i2c_adapter *adap, uint8_t chip,
>> uint addr, int alen,
>>                                uint8_t *buffer, int len);
>>        uint            (*set_bus_speed)(struct i2c_adapter *adap, uint
>> speed);
>> [...]
>> ?
>>
>> We can do this. Simon suggested this too ...
>
> Yes, exactly. (the functions will need some way to get information out
> of the struct i2c_adapter; I assume there's some kind of driver_private
> field in there too).

Yes.

I will post a few patches to move Tegra over to the new framework as
it is, so that people can see the impact. Given that every driver has
to change, it would be my preference to set the i2c API once. But I
suppose within a short while everything will move to the device model,
so perhaps the job of this series is just to move things in the right
direction?

Regards,
Simon