Re: [PATCH 04/10 v3] ARM: mach-shmobile: add fixed voltage regulators to armadillo800eva

[Guennadi Liakhovetski <g.liakhovetski@gmx.de>]

On Thu, 28 Jun 2012, Rafael J. Wysocki wrote:

> On Wednesday, June 27, 2012, Guennadi Liakhovetski wrote:
> > On Wed, 27 Jun 2012, Guennadi Liakhovetski wrote:
> > 
> > > On armadillo800eva provide a 3.3V supply for its SD/MMC-card interfaces.
> > > 
> > > Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
> > > ---
> > >  arch/arm/mach-shmobile/board-armadillo800eva.c |   16 ++++++++++++++++
> > >  1 files changed, 16 insertions(+), 0 deletions(-)
> > > 
> > > diff --git a/arch/arm/mach-shmobile/board-armadillo800eva.c b/arch/arm/mach-shmobile/board-armadillo800eva.c
> > > index 9e37026..819cee5 100644
> > > --- a/arch/arm/mach-shmobile/board-armadillo800eva.c
> > > +++ b/arch/arm/mach-shmobile/board-armadillo800eva.c
> > > @@ -28,6 +28,8 @@
> > >  #include <linux/platform_device.h>
> > >  #include <linux/gpio.h>
> > >  #include <linux/gpio_keys.h>
> > > +#include <linux/regulator/fixed.h>
> > > +#include <linux/regulator/machine.h>
> > >  #include <linux/sh_eth.h>
> > >  #include <linux/videodev2.h>
> > >  #include <linux/usb/renesas_usbhs.h>
> > > @@ -407,6 +409,17 @@ static struct platform_device gpio_keys_device = {
> > >  	},
> > >  };
> > >  
> > > +/* Fixed 3.3V regulator to be used by SDHI0, SDHI1, MMCIF */
> > > +static struct regulator_consumer_supply fixed3v3_power_consumers[] > > > +{
> > > +	REGULATOR_SUPPLY("vmmc", "sh_mobile_sdhi.0"),
> > > +	REGULATOR_SUPPLY("vqmmc", "sh_mobile_sdhi.0"),
> > > +	REGULATOR_SUPPLY("vmmc", "sh_mobile_sdhi.1"),
> > > +	REGULATOR_SUPPLY("vqmmc", "sh_mobile_sdhi.1"),
> > > +	REGULATOR_SUPPLY("vmmc", "sh_mmcif.0"),
> > > +	REGULATOR_SUPPLY("vqmmc", "sh_mmcif.0"),
> > 
> > There is a slight problem with this patch: a1 (r8a7740) SoCs are currently 
> > the only ones registering their MMCIF device without an index, i.e., with 
> > .id = -1. This is reflected in board-armadillo800eva.c and in 
> > clock-r8a7740.c. This patch erroneously uses sh_mmcif.0 for device name, 
> > which is also what every single other ARM and sh platform in the mainline, 
> > using mmcif, AFAICS, currently does.
> > 
> > Therefore question: Magnus, Rafael: what would you prefer - fix the above 
> > regulator entries to not use a device ID, or unify a1 with other SoCs to 
> > use .id = 0?
> 
> I wonder what's going to be more straightforward?

The easier fix is, of course, to redo the above hunk with just "sh_mmcif" 
as a device name. This is also what Magnus wanted to have and what I've 
sent a few minutes ago. But I'm pretty certain - if we keep this one 
platform different from the rest, I've made this mistake once, I'll do it 
again :-) So, eventually, maybe trying to keep the error probability as 
low as possible is good _if_ we don't have to make any further sacrifices 
for that.

Thanks
Guennadi
---
Guennadi Liakhovetski, Ph.D.
Freelance Open-Source Software Developer
http://www.open-technology.de/