[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20150212023252.GS2593@finisterre.sirena.org.uk>
Date: Thu, 12 Feb 2015 10:32:52 +0800
From: Mark Brown <broonie@...nel.org>
To: Tim Bird <tim.bird@...ymobile.com>
Cc: "Andersson, Björn"
<Bjorn.Andersson@...ymobile.com>,
"lgirdwood@...il.com" <lgirdwood@...il.com>,
"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] regulator: Support different config and dev of_nodes in
regulator_register
On Wed, Feb 11, 2015 at 09:21:32AM -0800, Tim Bird wrote:
> Also, when registering a regulator, you can pass a different
> of_node in config (struct regulator_config) than the one in dev
> (struct device)
> However, this has problems in the current code, as the test in
> regulator_dev_lookup requires that the device_node found by of_get_regulator()
> match the dev.of_node in the regulator in the regulator list.
...
> It took me a while to figure this out, because a regulator defined
> with dev.of_node != config.of_node worked fine, when accessed
> directly by name (not using a supply-name). I only had problems
> when I accessed the regulator using the "-supply" indirection technique
> in DT.
I'm not quite sure exactly what you've done here but I'm pretty sure
there's some confusion somewhere since the -supply based lookups are of
course what everything in mainline is doing except for CPU supplies.
The fact that you are using of_get_regulator_init_data() is a bit of a
red flag here, you shouldn't have to use that unless your bindings are
in some way unusual.
Download attachment "signature.asc" of type "application/pgp-signature" (474 bytes)
Powered by blists - more mailing lists