Hi Ben,
On Wed, 09 Jul 2008 13:51:48 +0100, Ben Dooks wrote:
> Allow the platform data to specify the bus bumber that the
> new I2C bus will be given. This is to allow the use of the
> board registration mechanism to specify the new style of
> I2C device registration which allows boards to provide a
> list of attached devices.
>
> Note, as discussed on the mailing list, we have dropped
> backwards compatibility of adding an dynamic bus number
> as it should not affect most boards to have the bus pinned
> to 0 if they have either not specified platform data for
either?
> driver. Any board supplying platform data will automatically
> have the bus_num field set to 0, and anyone who needs the
> driver on a different bus number can supply platform data
> to set bus_num.
Sounds OK to me.
>
> Signed-off-by: Ben Dooks <[EMAIL PROTECTED]>
>
> Index: linux-2.6.26-rc4-quilt3/drivers/i2c/busses/i2c-s3c2410.c
> ===================================================================
> --- linux-2.6.26-rc4-quilt3.orig/drivers/i2c/busses/i2c-s3c2410.c
> 2008-06-02 22:55:20.000000000 +0100
> +++ linux-2.6.26-rc4-quilt3/drivers/i2c/busses/i2c-s3c2410.c 2008-06-02
> 22:55:29.000000000 +0100
> @@ -752,9 +752,12 @@ static int s3c24xx_i2c_init(struct s3c24
> static int s3c24xx_i2c_probe(struct platform_device *pdev)
> {
> struct s3c24xx_i2c *i2c = &s3c24xx_i2c;
> + struct s3c2410_platform_i2c *pdata;
> struct resource *res;
> int ret;
>
> + pdata = s3c24xx_i2c_get_platformdata(&pdev->dev);
> +
> /* find the clock and enable it */
>
> i2c->dev = &pdev->dev;
> @@ -832,7 +835,15 @@ static int s3c24xx_i2c_probe(struct plat
> dev_dbg(&pdev->dev, "irq resource %p (%lu)\n", res,
> (unsigned long)res->start);
>
> - ret = i2c_add_adapter(&i2c->adap);
> + /* Note, previous versions of the driver used i2c_add_adapter()
> + * to add an bus at any number. We now pass the bus number via
"an bus" doesn't look correct.
> + * the platform data, so if unset it will now default to always
> + * being bus 0.
> + */
> +
> + i2c->adap.nr = pdata->bus_num;
> + ret = i2c_add_numbered_adapter(&i2c->adap);
> +
> if (ret < 0) {
Best practice is to not have a blank line between a function call and
the test of its return value.
> dev_err(&pdev->dev, "failed to add bus to i2c core\n");
> goto err_irq;
> Index: linux-2.6.26-rc4-quilt3/include/asm-arm/plat-s3c/iic.h
> ===================================================================
> --- linux-2.6.26-rc4-quilt3.orig/include/asm-arm/plat-s3c/iic.h
> 2008-06-02 20:12:47.000000000 +0100
> +++ linux-2.6.26-rc4-quilt3/include/asm-arm/plat-s3c/iic.h 2008-06-02
> 22:55:29.000000000 +0100
> @@ -21,6 +21,7 @@
> */
>
> struct s3c2410_platform_i2c {
> + int bus_num; /* bus number to use */
> unsigned int flags;
> unsigned int slave_addr; /* slave address for controller */
> unsigned long bus_freq; /* standard bus frequency */
>
Other than these details, the patch looks OK to me.
--
Jean Delvare
_______________________________________________
i2c mailing list
[email protected]
http://lists.lm-sensors.org/mailman/listinfo/i2c
