Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
Message-ID: <5da64b37.1c69fb81.fbe6e.51ad@mx.google.com>
Content-Type: text/plain; charset="utf-8"
MIME-Version: 1.0
Content-Transfer-Encoding: quoted-printable
In-Reply-To: <CAHp75Vcht2S7j64vUGw+DQdSEAyxuyBJ9oVwHyXxnL9bM1-wFA@mail.gmail.com>
References: <20191004214334.149976-1-swboyd@chromium.org> <20191004214334.149976-2-swboyd@chromium.org> <CAHp75Vebn48hbzqKWzV3aj4NEBCta_Fn7zOQHzsznW4=6cXLsQ@mail.gmail.com> <5da4e084.1c69fb81.567f9.4b9c@mx.google.com> <CAHp75Vcht2S7j64vUGw+DQdSEAyxuyBJ9oVwHyXxnL9bM1-wFA@mail.gmail.com>
From:   Stephen Boyd <swboyd@chromium.org>
To:     Andy Shevchenko <andy.shevchenko@gmail.com>
Cc:     Wolfram Sang <wsa@the-dreams.de>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        Arnd Bergmann <arnd@arndb.de>,
        Geert Uytterhoeven <geert@linux-m68k.org>,
        Riku Voipio <riku.voipio@iki.fi>,
        Rob Herring <robh+dt@kernel.org>,
        Frank Rowand <frowand.list@gmail.com>,
        Jacek Anaszewski <jacek.anaszewski@gmail.com>,
        Pavel Machek <pavel@ucw.cz>, Dan Murphy <dmurphy@ti.com>,
        Linux LED Subsystem <linux-leds@vger.kernel.org>
Subject: Re: [PATCH 01/10] leds: pca953x: Use of_device_get_match_data()
User-Agent: alot/0.8.1
Date:   Tue, 15 Oct 2019 15:41:58 -0700
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

Quoting Andy Shevchenko (2019-10-15 02:02:01)
> On Mon, Oct 14, 2019 at 11:54 PM Stephen Boyd <swboyd@chromium.org> wrote:
> > Quoting Andy Shevchenko (2019-10-14 10:50:06)
> > > On Sat, Oct 5, 2019 at 12:47 AM Stephen Boyd <swboyd@chromium.org> wr=
ote:
> > > >
> > > > This driver can use the of_device_get_match_data() API to simplify =
the
> > > > code. Replace calls to of_match_device() with this newer API under =
the
> > > > assumption that where it is called will be when we know the device =
is
> > > > backed by a DT node. This nicely avoids referencing the match table=
 when
> > > > it is undefined with configurations where CONFIG_OF=3Dn.
> > >
> > > > +       devid =3D (int)(uintptr_t)of_device_get_match_data(dev);
> > >
> > > > +               devid =3D (int)(uintptr_t)of_device_get_match_data(=
&client->dev);
> > >
> > > This still leaves it OF-centric.
> > > Better to use device_get_match_data().
> > >
> > > Also, I'm thinking that following may help to clean a lot of the i2c
> > > client drivers
> > >
> > > static inline // perhaps no
> > > const void *i2c_device_get_match_data(struct i2c_client *client, const
> > > struct i2c_device_id *id)
> > > {
> > >   if (id)
> > >     return (const void *)id->driver_data;
> > >   return device_get_match_data(&client->dev);
> > > }
> > >
> >
> > Looks alright to me. Maybe device_get_match_data() can look at the bus
> > and call some bus op if the firmware match isn't present? Then we can
> > replace a bunch of these calls with device_get_match_data() and it will
> > "do the right thing" regardless of what bus or firmware the device is
> > running on.
>=20
> It will be something ugly like
>=20
> buses {
> #ifdef I2C
> &i2c_bus_type,
> #endif
> ...
> }
>=20
> in the code. I won't do this.
>=20
> See generic_match_buses[] for example.

Why is it like generic_match_buses[]? I thought it would look at
struct device::of_node or struct device::fw_node and try to extract
device data out that and if that fails it would fallback to some new
function like struct bus_type::get_match_data() that does the right
thing for the bus. In the case of i2c it would extract the i2c_client's
i2c_device_id pointer and return it onwards.