Re: [RFC PATCH 0/3] gpio: ep93xx: convert to multi irqchips

[nikita . shubin]

27.12.2020, 16:55, "Linus Walleij" <linus.walleij@xxxxxxxxxx>:
> Paging Hartley who I think uses this chip.
>
> On Thu, Dec 24, 2020 at 12:22 PM Nikita Shubin
> <nikita.shubin@xxxxxxxxxxx> wrote:
>
>>  I was lucky enough to became an owner of some splendid piece's of
>>  antiques called ts7250 based on the top of Cirrus Logic EP9302.
>>
>>  I don't know what fate expects this hardware (it's not EOL it's just Not
>>  recommended for new designs) but i wanted to share fixes in ep93xx gpio area.
>>
>>  It seems ep93xx is deadly broken at current state usage of AB gpiochips
>>  interrupts leads to deadlocks coused by irq_unmask/irq_mask recursions.
>>
>>  Port F is not working at all:
>
> Yeah when I converted this gpio driver I did not have access to a hardware
> which made use of port F :(
>
> I used the SIM.One board but they all died on me, then I acquired another
> EP93xx board that I just didn't have time to test.
>
> The long term plan is to convert these boards to use device tree and
> multiplatform. Interested in the job? ;)
>

I had such a thought, but I was stopped by the delusion that no one is using this 
hardware anymore.

Not promising anything right now - but such a job sounds like a real fun.

> Thanks for fixing up the GPIO chip, let's get the patches in shape and
> merge as fixes. Bartosz is handling the fixes right now, I'll try to review too!

I am glad that this work is interesting not only for me. For this, I'm willing to put in 
a little effort to make the proposed changes look decent.

>
> Yours,
> Linus Walleij