RE: [PATCH v1] ARM: tegra: Fix sd4 regulator in Jetson TK1 device tree

[Vidya Sagar <vidyas@nvidia.com>]

> -----Original Message-----
> From: Stephen Warren [mailto:swarren@wwwdotorg.org]
> Sent: Tuesday, September 23, 2014 12:36 AM
> To: Vidya Sagar
> Cc: thierry.reding@gmail.com; Laxman Dewangan; Krishna Thota; linux-
> tegra@vger.kernel.org; linux@arm.linux.org.uk; linux-
> kernel@vger.kernel.org
> Subject: Re: [PATCH v1] ARM: tegra: Fix sd4 regulator in Jetson TK1 device
> tree
> 
> On 09/22/2014 11:57 AM, Vidya Sagar wrote:
> > sd4 is an always on regulator which is turned on at boot time.
> > It is externally controller through gpio. This change reflects the
> > same in Jetson TK1 device tree
> 
> In the schematics, the "Power Sequencing" timing diagram says "S/W
> controlled" for SD4/+1.05V_RUN. I also don't see an "ENABLE1" pin on the
> AS3722, which would be required for ...
> 
> > +					ams,ext-control = <1>;
> 
> ... to be valid.
> 
> What's the source of information behind this change?
> 
> What symptoms does this patch correct?

I'm seeing one issue when I add support for PCIe suspend/resume functionality.
The issue is that, when regulator_bulk_diable() is called, disabling one of the power rails (which is deriving its voltage from SD4) of PCIe is failing.
The reason being, I2C controller is getting power gated before power rail disable is called.
Hence SD4 is made dependent on ENABLE1, which is nothing but the deep sleep signal coming from Tegra,
So eventually, SD4 will be powered off when system enters into deep-sleep state.
Source of information is from downstream kernel