On Tue, Jun 17, 2014 at 02:20:25PM +0300, Mikko Perttunen wrote:

This v3 breaks SATA; see below

On 16/06/14 17:57, Thierry Reding wrote:
...

quoted

+                       regval = padctl_readl(padctl, lane->offset);
+
+                       if (!value)
+                               regval &= ~lane->iddq;
+                       else
+                               regval |= lane->iddq;
+
+                       padctl_writel(padctl, regval, lane->offset);

...

The bits at at lane->iddq are actually IDDQ_DISABLE bits, so this is the
wrong way around: having iddq = <0> in device tree leads to the device
turning off. So either the DT property should be renamed or the branches of
the if here should be switched around.

Indeed, s/!value/value/ in the above fixes this. I've sent a v4 with
that fix and changes to address Stephen's comments regarding the
#address-cells and #size-cells properties.

Thanks,
Thierry