On 26/08/2024 16:03, Andrei Stefanescu wrote:
> Hi Krzysztof,
>
> On 26/08/2024 12:10, Krzysztof Kozlowski wrote:
>> On 26/08/2024 10:42, Andrei Stefanescu wrote:
>
> Thank you for the quick review!
>
>>> + raw_spin_lock_init(&gpio_dev->lock);
>>
>> Why do you use raw spin? Are you sure you need it (some people just
>> replace it thinking this will help them in PREEMPT_RT without actually
>> thinking if it is needed). IOW, do you have here irqchip anywhere?
>
> I don't have an irqchip in this current patch series. There are, however,
> other patches which add support for interrupts and implementations for
> power management callbacks. I thought it would be easier for review if I
> sent those after the current series gets merged.
power management callbacks do not need raw spinlock.
>
>>> +
>>> +static const struct of_device_id siul2_gpio_dt_ids[] = {
>>> + { .compatible = "nxp,s32g2-siul2-gpio", .data = &s32g2_device_data },
>>
>> Why do you have match data? There are no other variants.
>
>
> We do have another match data in our downstream version. Could I keep it
> here or should I remove it?
If you already work on new version and you are going to send it soon,
then it is fine. Otherwise you will add matchdata once there is such need.
Best regards,
Krzysztof
