[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID:
<PH7PR12MB728290AB15E7F027186306EBD7D62@PH7PR12MB7282.namprd12.prod.outlook.com>
Date: Wed, 26 Jun 2024 21:45:23 +0000
From: Asmaa Mnebhi <asmaa@...dia.com>
To: Jonas Gorski <jonas.gorski@...il.com>
CC: Shiji Yang <yangshiji66@...look.com>, "linux-gpio@...r.kernel.org"
<linux-gpio@...r.kernel.org>, Linus Walleij <linus.walleij@...aro.org>,
Bartosz Golaszewski <brgl@...ev.pl>, Andy Shevchenko
<andy.shevchenko@...il.com>, "linux-kernel@...r.kernel.org"
<linux-kernel@...r.kernel.org>, Mark Mentovai <mark@...tovai.com>,
Lóránd Horváth <lorand.horvath82@...il.com>
Subject: RE: [PATCH] gpio: mmio: do not calculate bgpio_bits via "ngpios"
> > >I believe we should use only a single source for what we need. If we
> > >rely on ngpios, the bgpio_bits should be recalculated based on it.
> > >The expression doing this is very simple. Something like
> > >round_up(ngpios, 8);
> >
>
> Right, mistakes happen. Also I don't want to blame anyone, because it
> happens to work fine on little endian systems, and big endian systems are so
> rare it's easy to forget that they exist ;-)
>
> > Now, if we want to not modify bgpio_bits, we could go back to my v3 patch.
> > ngpio is the number of gpio pins supported while bgpio_bits is the register
> access bit type. These are 2 different entities.
>
> Exactly. And AFAICT this patch does exactly this, restoring the code to the
> state of v3.
>
LGTM then 😊. Thanks!
Powered by blists - more mailing lists