Re: Fwd: BUG#200905 pxa_gpio conflicing with others arches

[Robert Jarzmik <robert.jarzmik@xxxxxxx>]

Nicolas Chauvet <kwizart@xxxxxxxxx> writes:

> 2018-08-25 10:06 GMT+02:00 Robert Jarzmik <robert.jarzmik@xxxxxxx>:
>> Nicolas Chauvet <kwizart@xxxxxxxxx> writes:
>>
>>> I will test the patch when possible. But I wonder if hw availability
>>> can be verified at probe also ?
>> Mmh unfortunately no.
>>
>> The hardware IP is simply memory mapped (with an interrupt source as well), and
>> doesn't have any discoverable ID so rely upon AFAIK.
>
> Hi Robert,
>
> I've just tested today.
> Unfortunately, the patch doesn't work for me. The error looks the same.
Mmmm that's strange ...
Could you provide me (or attach them to the bugzilla) :
 - your .config
 - your .dtb
 - your vmlinux
 - the new stackstrace

> I wondered about a check if using a device-tree, then if there is no
> pxa_gpio node, no registration is done.
> As I understand, the current driver mix device-tree and platform code,
> so if there is no device-tree node, then it falls back to platform
> code.
That's not exactly what happens. The driver mixes both platform and device-tree
yes, but the suspend hooks registration happens regardless of the mode, and even
when no registration happened, or if a registration failed.
The goal of the patch was to address all the problems together.

Cheers.

-- 
Robert