Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1210160 > unrolled thread

Re: [PATCH 1/2] regulator: pbias: use untranslated address to program pbias regulator

Back to article view | Back to linux.kernel

This discussion starts older than the indexed window; earlier articles aren't shown. The article labeled Started by below is the oldest one visible, not the original post.

Contents

  Re: [PATCH 1/2] regulator: pbias: use untranslated address to program  pbias regulator Kishon Vijay Abraham I <kishon@ti.com> - 2015-08-20 08:00 +0200
    Re: [PATCH 1/2] regulator: pbias: use untranslated address to  program pbias regulator Mark Brown <broonie@kernel.org> - 2015-08-20 19:50 +0200

#1210160 — Re: [PATCH 1/2] regulator: pbias: use untranslated address to program pbias regulator

Hi Mark Brown,

On Wednesday 19 August 2015 11:41 PM, Mark Brown wrote:
> On Tue, Aug 18, 2015 at 11:23:54AM +0530, Kishon Vijay Abraham I wrote:
>> On Friday 14 August 2015 11:30 PM, Mark Brown wrote:
>>> On Mon, Jul 27, 2015 at 04:54:09PM +0530, Kishon Vijay Abraham I wrote:
> 
>>>> is moved as a child node of syscon, vsel_reg and enable_reg has the
>>>> absolute address because of the address translation that happens while
>>>> creating device from device tree node.
>>>> So avoid using platform_get_resource and use of_get_address in order to
>>>> get only the offset (untranslated address) and populate these in
>>>> vsel_reg and enable_reg.
> 
>>> This sounds like we're going in the wrong direction, we're moving from a
>>> more generic API to a firmware specific one.  Why is this a good fix?
> 
>> platform_get_resource can be used if we need the absolute address but here we
>> need only the offset.
> 
> So substract this address from the start of the resource to get the

That would mean from the offset (provided in dt) get the absolute address and
then again from the absolute address get the offset.
> offset?  Or provide a wrapper function in the resource code which does

Why not use 'of_get_address' which does the same thing? Moreover it's not a
resource we are dealing with here. It's a resource only for the syscon driver.
> that.  What you're saying above is pretty much "this happens to work"
> but my concern is that the solution that happens to work isn't really
> what we want to do.

Not just makes this work, this is also the most reasonable solution available IMHO.

The most ideal way would have been to use something like what Grygorii
mentioned to use syscon = <&scm_conf 0xe00> and then use the phandle to get the
offset. But then with this we'll be breaking older dtbs.

Thanks
Kishon


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

[toc] | [next] | [standalone]

#1210651 — Re: [PATCH 1/2] regulator: pbias: use untranslated address to program pbias regulator

[Multipart message — attachments visible in raw view] — view raw

On Thu, Aug 20, 2015 at 11:21:06AM +0530, Kishon Vijay Abraham I wrote:
> On Wednesday 19 August 2015 11:41 PM, Mark Brown wrote:
> > On Tue, Aug 18, 2015 at 11:23:54AM +0530, Kishon Vijay Abraham I wrote:

Please fix your mail client to word wrap within paragraphs.

> >> platform_get_resource can be used if we need the absolute address but here we
> >> need only the offset.

> > So substract this address from the start of the resource to get the

> That would mean from the offset (provided in dt) get the absolute address and
> then again from the absolute address get the offset.

Sure, that's how the Linux APIs work right now and why I'm suggesting
you might want a wrapper.

> > offset?  Or provide a wrapper function in the resource code which does

> Why not use 'of_get_address' which does the same thing? Moreover it's not a
> resource we are dealing with here. It's a resource only for the syscon driver.

This is how the OF description you've done is intended to be parsed and
hence is interpreted by the generic code, it's just a detail of the
syscon implementation that you need to translate it into an offset.

> > that.  What you're saying above is pretty much "this happens to work"
> > but my concern is that the solution that happens to work isn't really
> > what we want to do.

> Not just makes this work, this is also the most reasonable solution available IMHO.

In general moving to a lower level inteface is not usually a step
forward.  As far as I can tell the driver already has all the
information it needs from the more generic APIs, it's just not
interpreting it in the way that the APIs it's trying to use wants.  It
just needs to fix the way it interprets the data it's passing through.

> The most ideal way would have been to use something like what Grygorii
> mentioned to use syscon = <&scm_conf 0xe00> and then use the phandle to get the
> offset. But then with this we'll be breaking older dtbs.

There is no need to break older DTs, that would clearly be a bad idea.

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web