Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Fri, 1 Jul 2022 17:31:53 +0100
From:   Jonathan Cameron <jic23@kernel.org>
To:     Andy Shevchenko <andy.shevchenko@gmail.com>
Cc:     Nishanth Menon <nm@ti.com>,
        Javier Martinez Canillas <javier@osg.samsung.com>,
        Nuno =?UTF-8?B?U8Oh?= <nuno.sa@analog.com>,
        Christophe JAILLET <christophe.jaillet@wanadoo.fr>,
        Alexandru Ardelean <ardeleanalex@gmail.com>,
        Lars-Peter Clausen <lars@metafoo.de>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        linux-iio <linux-iio@vger.kernel.org>,
        Stable <stable@vger.kernel.org>
Subject: Re: [PATCH] iio: adc: ti-adc128s052: Fix number of channels when
 device tree is used
Message-ID: <20220701173153.04711237@jic23-huawei>
In-Reply-To: <CAHp75Vfm+NDzZEB1Qp-3+mbj=NOko=5jjcHr_A4J6-jMpTykhg@mail.gmail.com>
References: <20220630230107.13438-1-nm@ti.com>
        <CAHp75Vfm+NDzZEB1Qp-3+mbj=NOko=5jjcHr_A4J6-jMpTykhg@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Precedence: bulk

On Fri, 1 Jul 2022 12:13:24 +0200
Andy Shevchenko <andy.shevchenko@gmail.com> wrote:

> On Fri, Jul 1, 2022 at 1:02 AM Nishanth Menon <nm@ti.com> wrote:
> >
> > When device_match_data is called - with device tree, of_match list is  
> 
> device_get_match_data() ?
> 
> > looked up to find the data, which by default is 0. So, no matter which
> > kind of device compatible we use, we match with config 0 which implies
> > we enable 8 channels even on devices that do not have 8 channels.
> >
> > Solve it by providing the match data similar to what we do with the ACPI
> > lookup information.
> >
> > Fixes: 9e611c9e5a20 ("iio: adc128s052: Add OF match table")
> > Cc: <stable@vger.kernel.org> # 5.0+
> > Signed-off-by: Nishanth Menon <nm@ti.com>  
> 
> ...
> 
> > +       { .compatible = "ti,adc128s052", .data = 0},  
> 
> No assignment, 0 _is_ the default here.
> 
> > +       { .compatible = "ti,adc122s021", .data = 1},
> > +       { .compatible = "ti,adc122s051", .data = 1},
> > +       { .compatible = "ti,adc122s101", .data = 1},
> > +       { .compatible = "ti,adc124s021", .data = 2},
> > +       { .compatible = "ti,adc124s051", .data = 2},
> > +       { .compatible = "ti,adc124s101", .data = 2},  
> 
> What you need _ideally_ is rather use pointers to data structure where
> each of that chip is defined, then it will be as simple as
> 
> 
> const struct my_custom_drvdata *data;
> 
> data = device_get_match_data(dev);
> 
> Where my_custom_drvdata::num_of_channels will be already assigned to
> whatever you want on a per chip basis.

Agreed. That's much nicer and a not a lot larger change so still suitable as a fix.

> 
> If the number of channels is the only data you have, then yes, cast it
> to void * in the OF ID table and

It's not just the number of channels.  This is an index into an array of chip
type specific structures. Hence the driver is half way to what you suggested.
Using a pointer for the ACPI and DT paths is the right way to do this.
For the spi_device_id table, you could stick with an index, or move to casting
a pointer to an integer, I don't really mind.

Thanks,

Jonathan

> 
> num = (uintptr_t)device_get_match_data(dev);
> 
> will suffice.
>