Re: [PATCH] mfd: cros: Update EC protocol to match current EC code

From: Guenter Roeck <groeck@google.com>
To: Enric Balletbo Serra <eballetbo@gmail.com>
Cc: Gwendal Grignou <gwendal@chromium.org>,
	Guenter Roeck <groeck@chromium.org>,
	Enric Balletbo i Serra <enric.balletbo@collabora.com>,
	Benson Leung <bleung@chromium.org>,
	Lee Jones <lee.jones@linaro.org>,
	linux-kernel <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH] mfd: cros: Update EC protocol to match current EC code
Date: Wed, 6 Mar 2019 09:20:33 -0800	[thread overview]
Message-ID: <CABXOdTfqPKq34OVDmJ9qjTr0nqarhpN=f37m3xH9iuHxex+xrw@mail.gmail.com> (raw)
In-Reply-To: <CAFqH_52tWw8pGwY5fdAPQ6npbfWA7o4OhtPxvbUbCh5TdnbSew@mail.gmail.com>

[resending in plain text mode ]

On Wed, Mar 6, 2019 at 8:57 AM Enric Balletbo Serra <eballetbo@gmail.com> wrote:
>
> Hi Gwendal,
>
> Many thanks to send this upstream.
>
> Missatge de Gwendal Grignou <gwendal@chromium.org> del dia dj., 28 de
> febr. 2019 a les 1:31:
> >
> > Chromebook Embedded Controller protocol is defined in the kernel at
> > cros_ec_commands.h.
> > The source of trust for the EC protocol is at
> > https://chromium.googlesource.com/chromiumos/platform/ec/+/master/include/ec_commands.h
> >
> > Only needed changes have been picked up from this file to the kernel
> > include file leading to gaps between the upstream version and what the
> > latest ECs can do.
> >
> > Fill the gaps to ease future integrations. Changes from the original
> > files is header/footer for license and include files for alignment.
> >
> > Check this include file works on ChomeOS kernel 4.14 and 4.19 on eve.
> >
> > Signed-off-by: Gwendal Grignou <gwendal@chromium.org>
> > ---
> >  include/linux/mfd/cros_ec_commands.h | 3627 +++++++++++++++++++++-----
>
> I'm wondering if we should move this file to include/uapi at some
> point as this file is also used as user-space API for some userspace
> applications.
>
> While we are here I'd suggest if we can also fix the few errors (3)
> and warnings (5) spotted by checkpatch. With that it's an ack from my
> side.
>
> Being strict, though, on most cases the variables are going to be used
> in code that can be seen by user-space programs so maybe we should
> really need to switch to __u8/__u16/etc exportable data types instead
> of the uint8_t/uint16_t/etc types (those are not aimed to be used
> within the kernel). For those types that are internal we should use
> in-kernel type (u8/u16/etc)
>
> There is also the use of the BIT macro instead of the (1 << x), I know
> that this is a maintainer preference.
>
Is all that even possible ? After all, this is an imported file, and
we don't usually expect that imported files meet the Linux kernel
coding style.

Thanks,
Guenter

> [snip]
>
> Thanks,
>  Enric