From mboxrd@z Thu Jan  1 00:00:00 1970
From: mperttunen@nvidia.com (Mikko Perttunen)
Date: Thu, 3 Jul 2014 17:15:36 +0300
Subject: [PATCH 1/6] thermal: of: Add support for hardware-tracked trip
 points
In-Reply-To: <53B2FADC.8060708@wwwdotorg.org>
References: <1403856699-2140-1-git-send-email-mperttunen@nvidia.com>
 <1403856699-2140-2-git-send-email-mperttunen@nvidia.com>
 <53B1D1C8.5010907@wwwdotorg.org> <53B262EA.9030806@nvidia.com>
 <53B2FADC.8060708@wwwdotorg.org>
Message-ID: <53B56588.6090703@nvidia.com>
To: linux-arm-kernel@lists.infradead.org
List-Id: linux-arm-kernel.lists.infradead.org

On 01/07/14 21:15, Stephen Warren wrote:
>> The thermal core only supports a fixed number of trip points for each
>> driver and the core informs the driver of any changes to those, so
>> drivers using the core framework can already have hardware trip points,
>> but just a fixed number of them.
>>
>> The way of-thermal works, is it reads all the trip points from the
>> device tree, registers a new thermal_zone_device with that number of
>> trip points and then handles the trip points completely independently.
>> Of course, if we're just polling, this is fine, since the thermal core
>> also knows about those trip points and will trigger cooling when polling
>> the each zone. However, the driver doesn't, so it cannot setup any
>> interrupts to call thermal_zone_device_update.
>
> Is there any possibility of cleaning that up? It's obviously horribly
> inconsistent if core driver functionality works completely differently
> simply because the list of trip-points comes from DT rather than a
> static table in the driver. of_thermal should be limited to DT parsing
> and related device instantiation/lookup, not introducing a completely
> different functionality model.

I guess the smallest possible change would be to add a 
#hardware-trip-cells property to the thermal driver node (this would 
need to designate both the thermal zone and the trip point) and a 
hardware-trip-point phandle node to trip points. Then trip points could 
point to a hardware trip point that would get programmed. Since this is 
just adding properties, it would be backwards-compatible as well.
This is starting to sound like a good idea. Will have to give think 
about it some more.