Hi Mark and Marek,
My intention was to have a single commit, instead of two commits with
the second one making a part of the previous one obsolete.
Nevertheless I understand your points which are valid, so I will review
Marek's patch first before submitting another patch.
BR
Fabien
On 22/04/2022 12:08, Marc Zyngier wrote:
On Fri, 22 Apr 2022 10:20:36 +0100,
Fabien DESSENNE <fabien.dessenne@xxxxxxxxxxx> wrote:
Hi Marek,
I agree there is something wrong with the clock management in IRQ
context here and your patch goes in the right way.
There are also some other problems regarding performance (enabling /
disabling clock each time we want to change the IO value, ...).
For these both issues I have a patch, which basically keeps the GPIO
clocks enabled from probe.
I did not have time to submit it, but, considering your concerns, I
will do it in the coming days.
For the time being I suggest that we do not apply your patch.
Why? This fixes a glaring issue, and there are no alternatives at the
moment. So if there is something to improve on, please base your patch
on top of Marek's.
Thanks,
M.
