Re: [RFC PATCH V2 1/8] irqdomain: Ensure type settings match for an existing mapping

[Jon Hunter <jonathanh-DDmLM1+adcrQT0dZR+AlfA@public.gmane.org>]

On 22/12/15 10:00, Linus Walleij wrote:
> On Tue, Dec 22, 2015 at 10:58 AM, Linus Walleij
> <linus.walleij-QSEj5FYQhm4dnm+yROfE0A@public.gmane.org> wrote:
>> On Fri, Dec 18, 2015 at 11:10 AM, Jon Hunter <jonathanh-DDmLM1+adcrQT0dZR+AlfA@public.gmane.org> wrote:
>>> On 17/12/15 13:16, Linus Walleij wrote:
>>>> On Thu, Dec 17, 2015 at 11:48 AM, Jon Hunter <jonathanh-DDmLM1+adcrQT0dZR+AlfA@public.gmane.org> wrote:
>>>>> +               } else {
>>>>> +                       virq = irq_domain_alloc_descs(-1, 1, hwirq,
>>>>> +                                                     of_node_to_nid(of_node));
>>>>
>>>> What is this all of a sudden? Not even mentioned in the
>>>> commit. Plus I bet ACPI need something else than OF nid
>>>> passed here.
>>>
>>> Do you mean the else part of all of the above?
>>
>> Yes
>>
>>> So in the current code, the else part calls irq_create_mapping() (...)
>>
>> No, not that... The fact that you all of a sudden have started
>> calling irq_domain_alloc_descs() which the function didn't do
>> before, totally changing the calling semantics for everyone in
>> the kernel, leading to the problem I then describe with this
>> potentially being called before the irqdomain for the irqchip
>> is initialized and descs getting "random" numbers.
> 
> I see I didn't go into those details in my first answer. Hm, I
> guess I got uncertain and deleted it because I remember
> writing it...
> 
> I am simply worries that starting to call irq_domain_alloc_descs()
> has unintended side effects, especially on platforms using
> legacy or simple irqdomains.

Ok, no problem I will keep the existing irq_create_mapping() instead then.

Cheers
Jon