[linux-sunxi] Re: [PATCH v4 2/5] soc: sunxi: Add driver for Allwinner Reduced Serial Bus

[Chen-Yu Tsai]

On Thu, Sep 24, 2015 at 12:45 AM, Mark Brown <broonie at kernel.org> wrote:
> On Thu, Sep 24, 2015 at 12:05:18AM +0800, Chen-Yu Tsai wrote:
>
>> +static int sunxi_rsb_device_probe(struct device *dev)
>> +{
>> +     const struct sunxi_rsb_driver *drv = to_sunxi_rsb_driver(dev->driver);
>> +     struct sunxi_rsb_device *rdev = to_sunxi_rsb_device(dev);
>> +     int ret;
>> +
>> +     if (dev->bus != &sunxi_rsb_bus)
>> +             return 0;
>
> That's a bit worrying...  why might this get called for something other
> than a rsb device?

You're right. sunxi_rsb_bus is self contained (not exported) in sunxi_rsb.c,
so it's safe to assume that the devices going through sunxi_rsb_device_probe()
and sunxi_rsb_device_remove() are indeed sunxi_rsb slave devices.

I'll remove the checks from both functions.

>> +struct regmap *devm_regmap_init_sunxi_rsb(struct sunxi_rsb_device *rdev,
>> +                                       const struct regmap_config *config);
>
> This should work most of the time but you might want to take a look at
> the changes introduced in 3cfe7a74d42b (regmap: Use different lockdep
> class for each regmap init call) and follow a similar pattern for RSB to
> avoid spurious lockdep warnings in some configurations - see the commit
> log for that change for a full explanation.

This seems easy enough. I'll follow the patterns found in linux/regmap.h.

> Otherwise the regmap bit of this looks good, I'll try to have another
> look through at the rest of the code later.

Thanks!


Regards
ChenYu