On Mon, Dec 02, 2013 at 11:38:59AM -0600, Kumar Gala wrote: > > On Dec 2, 2013, at 7:28 AM, Thierry Reding <thierry.reding@xxxxxxxxx> wrote: > > > Introduce device tree bindings for the MIPI pad calibration controller > > found on Tegra SoCs. The controller can be used to perform calibration > > of pads used for DSI and CSI peripherals. > > > > Signed-off-by: Thierry Reding <treding@xxxxxxxxxx> > > --- > > .../bindings/misc/nvidia,tegra114-mipi.txt | 37 ++++++++++++++++++++++ > > Should this just be nvidia,tegra-mipi.txt (as you have it supported > for tegra<chip>)? Tegra114 is the first chip to support it. Subsequent chips should be backwards-compatible. But it's probably safer to change the compatible property documentation to only include nvidia,tegra114-mipi for now. Newer SoC generations can always be added when support for them is added. > > 1 file changed, 37 insertions(+) > > create mode 100644 Documentation/devicetree/bindings/misc/nvidia,tegra114-mipi.txt > > Is this really pad control or pin? I ask as its rare to see pad control. It's not really control at all. Rather it is a hardware block that performs calibration to obtain the optimal parameters for a given set of pads. Since you mention it, is there a better location for this file? I couldn't find anything that matched, so I just put it here. I suppose I could add it to Documentation/devicetree/bindings/gpu where the rest of the bindings reside for the Tegra GPU, but this block is also used by CSI, which is a capture interface and not really related to the GPU at all. > > diff --git a/Documentation/devicetree/bindings/misc/nvidia,tegra114-mipi.txt b/Documentation/devicetree/bindings/misc/nvidia,tegra114-mipi.txt > > new file mode 100644 > > index 000000000000..beb75ec7f6fc > > --- /dev/null > > +++ b/Documentation/devicetree/bindings/misc/nvidia,tegra114-mipi.txt > > @@ -0,0 +1,37 @@ > > +NVIDIA Tegra MIPI pad calibration controller > > + > > +Required properties: > > +- compatible: "nvidia,tegra<chip>-mipi" > > +- reg: Physical base address and length of the controller's registers. > > +- clocks: The clock consumed by the controller. > > +- #nvidia,mipi-calibrate-cells: Should be 1. The cell is a bitmask of the pads > > + that need to be calibrated for a given device. > > Is there a case that this would not be set to 1? If so, why bother > with the property right now? One would assume for a non-1 case a new > compatible will be required anyways. There is no such case now, and I'm not aware of a future chip that has more than 32 pads and would therefore require this to be anything other than 1. However having this property allows the DT to be parsed without specific knowledge about the hardware block. If this was implied by the value of the compatible property, a generic parser wouldn't know how to interpret the data. A schema validator wouldn't be able to check that the content was correct either. Thierry
Attachment:
pgpAbfiUqThhr.pgp
Description: PGP signature
