On Tue, May 5, 2015 at 4:35 PM, James Hogan <james.hogan@xxxxxxxxxx> wrote: > On Tue, May 05, 2015 at 04:09:31PM -0700, Andrew Bresticker wrote: >> On Tue, May 5, 2015 at 3:43 PM, James Hogan <james.hogan@xxxxxxxxxx> wrote: >> > On Tue, May 05, 2015 at 03:16:23PM -0700, Andrew Bresticker wrote: >> >> Hi James, >> >> >> >> On Tue, May 5, 2015 at 3:01 PM, James Hogan <james.hogan@xxxxxxxxxx> wrote: >> >> > Hi Andrew, >> >> > >> >> > On Tue, Apr 07, 2015 at 03:04:16PM -0700, Andrew Bresticker wrote: >> >> >> Add a binding document for the USB2.0 PHY found on the IMG Pistachio SoC. >> >> >> >> >> >> Signed-off-by: Andrew Bresticker <abrestic@xxxxxxxxxxxx> >> >> >> Cc: Rob Herring <robh+dt@xxxxxxxxxx> >> >> >> Cc: Pawel Moll <pawel.moll@xxxxxxx> >> >> >> Cc: Mark Rutland <mark.rutland@xxxxxxx> >> >> >> Cc: Ian Campbell <ijc+devicetree@xxxxxxxxxxxxxx> >> >> >> Cc: Kumar Gala <galak@xxxxxxxxxxxxxx> >> >> >> --- >> >> >> No changes from v1. >> >> >> --- >> >> >> .../devicetree/bindings/phy/pistachio-usb-phy.txt | 29 ++++++++++++++++++++++ >> >> >> include/dt-bindings/phy/phy-pistachio-usb.h | 16 ++++++++++++ >> >> >> 2 files changed, 45 insertions(+) >> >> >> create mode 100644 Documentation/devicetree/bindings/phy/pistachio-usb-phy.txt >> >> >> create mode 100644 include/dt-bindings/phy/phy-pistachio-usb.h >> >> >> >> >> >> diff --git a/Documentation/devicetree/bindings/phy/pistachio-usb-phy.txt b/Documentation/devicetree/bindings/phy/pistachio-usb-phy.txt >> >> >> new file mode 100644 >> >> >> index 0000000..afbc7e2 >> >> >> --- /dev/null >> >> >> +++ b/Documentation/devicetree/bindings/phy/pistachio-usb-phy.txt >> >> >> @@ -0,0 +1,29 @@ >> >> >> +IMG Pistachio USB PHY >> >> >> +===================== >> >> >> + >> >> >> +Required properties: >> >> >> +-------------------- >> >> >> + - compatible: Must be "img,pistachio-usb-phy". >> >> >> + - #phy-cells: Must be 0. See ./phy-bindings.txt for details. >> >> >> + - clocks: Must contain an entry for each entry in clock-names. >> >> >> + See ../clock/clock-bindings.txt for details. >> >> >> + - clock-names: Must include "usb_phy". >> >> >> + - img,cr-top: Must constain a phandle to the CR_TOP syscon node. >> >> >> + - img,refclk: Indicates the reference clock source for the USB PHY. >> >> >> + See <dt-bindings/phy/phy-pistachio-usb.h> for a list of valid values. >> >> > >> >> > Possibly dumb question: why isn't the reference clock source specified >> >> > in the normal ways like the "usb_phy" clock is? >> >> > >> >> > Does the value required here depend on what usb_phy clock gets muxed >> >> > from or something? >> >> >> >> Right, the value indicates what clock "usb_phy" is: whether it comes >> >> from the core clock controller, the XO crystal, or is some external >> >> clock. It's a mux internal to the PHY. >> > >> > Okay. If its a software controllable mux, is there a particular reason >> > the DT doesn't describe it as such, i.e. have all 3 clock inputs, and >> > the driver somehow work out which to use? >> >> Well, I'm not sure how the driver would determine which clock to use >> without a device-tree property like the one I've got here :). Also, > > Does it make sense to just look for the "best" usable source clock based > on the supported rates listed in fsel_rate_map[] (for some definition of > "best" such as "fastest" / "slowest" / "first usable"), or are things > just not that simple? > > I'm just wondering how the DT writer would decide, since it seems to > come down to a policy decision rather than a description of the > hardware, which should probably be avoided in DT bindings if possible. Ah, sorry if that was unclear - this *is* describing a hardware property. The DT author would pick a value by looking at which clock is connected to the PHY in the schematic. -Andrew
