Re: [RFC PATCH] ACPI: bus: match of_device_id using acpi device

From: Nikolaus Voss <nikolaus.voss@loewensteinmedical.de>
To: Andy Shevchenko <andy.shevchenko@gmail.com>
Cc: Sudeep Holla <sudeep.holla@arm.com>,
	Srinath Mannam <srinath.mannam@broadcom.com>,
	"Voss, Dr. Nikolaus" <Nikolaus.Dr.Voss@loewensteinmedical.de>,
	"Rafael J. Wysocki" <rafael@kernel.org>,
	Len Brown <lenb@kernel.org>, Ray Jui <ray.jui@broadcom.com>,
	Vladimir Olovyannikov <vladimir.olovyannikov@broadcom.com>,
	Vikram Prakash <vikram.prakash@broadcom.com>,
	Scott Branden <scott.branden@broadcom.com>,
	ACPI Devel Maling List <linux-acpi@vger.kernel.org>,
	Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
	nv@vosn.de
Subject: Re: [RFC PATCH] ACPI: bus: match of_device_id using acpi device
Date: Wed, 4 Jul 2018 12:50:16 +0200 (CEST)	[thread overview]
Message-ID: <alpine.DEB.2.20.1807041239080.9273@fox.voss.local> (raw)
In-Reply-To: <CAHp75VeuoTM=RoUgk0qWyiRJYk2AncVhxKDvms_2Hv0yE=g8uQ@mail.gmail.com>

On Wed, 4 Jul 2018, Andy Shevchenko wrote:
> On Wed, Jul 4, 2018 at 1:17 PM, Nikolaus Voss
> <nikolaus.voss@loewensteinmedical.de> wrote:
>> On Wed, 4 Jul 2018, Sudeep Holla wrote:
>>> On 04/07/18 10:32, Andy Shevchenko wrote:
>>>> On Wed, Jul 4, 2018 at 6:37 AM, Srinath Mannam
>>>> <srinath.mannam@broadcom.com> wrote:
>
>>> +1 on NACK for this and anything else that abuse PRP0001 as a short cut
>>> approach.
>> This is no abuse but exactly what PRP0001 is meant for. The basic idea of
>> PRP0001 is to reuse DT "compatible" strings in ACPI namespace, see
>> Documentation/acpi/enumeration.txt. Reusing also means getting access to the
>> of_device_id.
>
> The idea was for almost DIY and / or manufacturer to develop a
> prototype without modifying match code and faking ACPI IDs.
> That's why the target of PRP0001 is almost sensors connected to I2C and SPI.
>
> That's why I agreed on your patch to help with this. But! The proper
> solution for the devices (device manufacturer) is to allocate an ACPI
> ID and provide a corresponding table to the driver.
>
> This is my understanding of that exercise. Rafael can correct me.

This is not meant as a short cut for device manufacturers. The patch is meant to 
make PRP0001 work when access to specific driver_data is needed. I see 
nothing bad with it.

>> Allocating an ACPI id for an already existing DT driver is redundant, isn't
>> it?
>
> It is not.

The driver won't work any better with it. The driver code gets another 
table as big as of_device_id table. Can you give me a hint what the added 
value is?

Niko