From mboxrd@z Thu Jan  1 00:00:00 1970
From: jmkrzyszt@gmail.com (Janusz Krzysztofik)
Date: Tue, 07 Aug 2018 18:50:22 +0200
Subject: [RFC PATCH v2 09/12] gpiolib: Identify GPIO descriptor arrays
 with direct mapping
In-Reply-To: <CACRpkdZtuU6zXBz6UqprE7-31vW9-1ZabFjyzHBgWaTgmqkbbQ@mail.gmail.com>
References: <20180718235710.18242-1-jmkrzyszt@gmail.com>
 <20180806222918.12644-10-jmkrzyszt@gmail.com>
 <CACRpkdZtuU6zXBz6UqprE7-31vW9-1ZabFjyzHBgWaTgmqkbbQ@mail.gmail.com>
Message-ID: <1793769.2WP40jSPqy@z50>
To: linux-arm-kernel@lists.infradead.org
List-Id: linux-arm-kernel.lists.infradead.org

Hi Linus,

On Tuesday, August 7, 2018 1:29:43 AM CEST Linus Walleij wrote:
> Hi Janusz!
> 
> On Tue, Aug 7, 2018 at 12:29 AM Janusz Krzysztofik <jmkrzyszt@gmail.com> 
wrote:
> 
> > Certain GPIO array lookup results may map directly to GPIO pins of a
> > single GPIO chip in hardware order.  If that condition is recognized
> > and handled efficiently, significant performance gain of get/set array
> > functions may be possible.
> >
> > While processing a request for an array of GPIO descriptors, verify if
> > the descriptors just collected represent consecutive pins of a single
> > GPIO chip.  Pass that information with the array to the caller so it
> > can benefit from enhanced performance as soon as bitmap based get/set
> > array functions which can make efficient use of that are available.
> >
> > Signed-off-by: Janusz Krzysztofik <jmkrzyszt@gmail.com>
> (...)
> >  This function returns a struct gpio_descs which contains an array of
> > -descriptors::
> > +descriptors.  It may also contain a valid descriptor of a single GPIO 
chip in
> > +case the array strictly matches pin hardware layout of the chip::
> >
> >         struct gpio_descs {
> >                 unsigned int ndescs;
> >                 struct gpio_desc *desc[];
> > +               struct gpio_chip *chip;
> 
> This must be motivated: if the only purpose is to indicate to the consumer 
that
> all GPIOs are on the same chip, why not just have a
> 
> bool all_on_same_chip;
> 
> That you set to true if these are all on the same chip?

My approach would probably save one or two instructions per get/set call, but 
I'm not stuck to it and will be happy to find a better solution.

How about folding the chip descriptor inside an additional structure, private 
to drivers, with internals not revealed to consumers?

Thanks,
Janusz