Re: Re: [PATCH] ARM: dts: sun4i: Add dts file for the pov protab2-ips9 tablet

[Maxime Ripard <maxime.ripard-wi1+55ScJUtKEb57/3fJTNBPR1lH4CV8@public.gmane.org>]

[-- Attachment #1: Type: text/plain, Size: 2080 bytes --]

On Tue, Sep 08, 2015 at 09:48:48AM +0200, Hans de Goede wrote:
> Hi,
> 
> On 07-09-15 22:52, Maxime Ripard wrote:
> >Hi,
> >
> >On Mon, Sep 07, 2015 at 11:30:03AM +0200, Hans de Goede wrote:
> >>>>>bma250 already has devicetree support. It is used in Gemei G9
> >>>>>tablet (sun4i-gemei-g9.dts).
> >>>>
> >>>>Yes I've seen that, but does it actually work? I've not tried but
> >>>>I do not see any compatible string in the actual bma250 code in
> >>>>the kernel, so I believe that this part of the sun4i-gemei-g9.dts
> >>>>file does not work ?
> >>>
> >>>It worked (even without IRQs) when I submitted the patch. Driver
> >>>itself is under iio/accel/bma180.c
> >>
> >>That is really weird, because:
> >>
> >>https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/drivers/iio/accel/bma180.c
> >>
> >>Does not have an of_match_table ... ??
> >
> >Not really, when using DT, i2c will set the i2c_client name to the
> >device part of the compatible [1] [2], and then if the of_device_id
> >lookup fails, will fallback to matching the i2c_client name to the
> >i2c_device_id [3]. Which in our case matches.
> >
> >Maxime
> >
> >[1] http://lxr.free-electrons.com/source/drivers/i2c/i2c-core.c#L1281
> >[2] http://lxr.free-electrons.com/source/drivers/i2c/i2c-core.c#L969
> >[3] http://lxr.free-electrons.com/source/drivers/i2c/i2c-core.c#L461
> 
> Hmm, not sure whether that is a useful feature or an ugly hack :)

Don't shoot the messenger :)

But I don't know either.

> It is sorta unexpected either way. But I can make good use of this to
> enable the accelerometer and maybe also some touchscreens on a bunch of
> boards I have.
> 
> Still should we rely on this? This means relying on Linux kernel behavior,
> rather then something documented in Documentation/devicetree/bindings/...

AFAIK, the behaviour isn't going away, there's been a few patches to
remove the need for i2c_device_id, and they've all been nacked, so I'd
say we can use it as is.

Maxime

-- 
Maxime Ripard, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com