[PATCH v2 3/7] pwm: rockchip: Remove the dumplicate rockchip_pwm_ops ops

David.Wu david.wu at rock-chips.com
Tue Aug 8 08:41:46 PDT 2017


Hi Boris,

在 2017/8/4 15:09, Boris Brezillon 写道:
> On Fri, 4 Aug 2017 10:38:26 +0800
> "David.Wu" <david.wu at rock-chips.com> wrote:
> 
>> Hi Boris,
>>
>> 在 2017/8/2 19:40, Boris Brezillon 写道:
>>> Yep, just define 3 different pwm_ops (one for each IP), each of them
>>> implementing ->apply() and ->get_state() and that's all.
>>>
>>> Something like:
>>>
>>> static const struct pwm_ops rockchip_pwm_ops_v1 = {
>>> 	.get_state = rockchip_pwm_v1_get_state,
>>> 	.apply = rockchip_pwm_v1_apply,
>>> 	.owner = THIS_MODULE,
>>> };
>>>
>>> static const struct pwm_ops rockchip_pwm_ops_v2 = {
>>> 	.get_state = rockchip_pwm_v2_get_state,
>>> 	.apply = rockchip_pwm_v2_apply,
>>> 	.owner = THIS_MODULE,
>>> };
>>>
>>> static const struct pwm_ops rockchip_pwm_ops_vop = {
>>> 	.get_state = rockchip_pwm_vop_get_state,
>>> 	.apply = rockchip_pwm_vop_apply,
>>> 	.owner = THIS_MODULE,
>>> };
>>>
>>> static const struct of_device_id rockchip_pwm_dt_ids[] = {
>>> 	{ .compatible = "rockchip,rk2928-pwm", .data = &rockchip_pwm_ops_v1 },
>>> 	{ .compatible = "rockchip,rk3288-pwm", .data = &rockchip_pwm_ops_v2 },
>>> 	{ .compatible = "rockchip,vop-pwm", .data = &rockchip_pwm_ops_vop },
>>> 	{ /* sentinel */ }
>>> };
>>> MODULE_DEVICE_TABLE(of, rockchip_pwm_dt_ids);
>>
>> I think we should keep the data members in the rockchip_pwm_data,like
>> supports_polarity and regs...
>>
>> The supports_polarity is needed for of_pwm_n_cells when pwm registered.
>> And the other data members is helpful for us to use common code.
>>
>> It's okay for just define 3 different pwm_ops (one for each IP), but
>> they are with other data members in the struct of rockchip_pwm_data.
>>
> 
> I think we could even get rid of the other fields in rockchip_pwm_data,
> but ok, let's do that.

I use the same pwm ops for each IP at V3's patch, but defined 3 
different rockchip_pwm_data for use. I think this might look more clean.

> 
> 
> 




More information about the linux-arm-kernel mailing list