On Fri 22 Feb 11:57 PST 2019, Rob Herring wrote:
> On Mon, Feb 18, 2019 at 10:04:02PM -0800, Bjorn Andersson wrote:
> > The Qualcomm PCIe2 PHY is a Synopsys based PCIe PHY found in a number of
> > Qualcomm platforms, add a binding to describe this.
> >
> > Signed-off-by: Bjorn Andersson <bjorn.andersson@xxxxxxxxxx>
> > ---
> > .../bindings/phy/qcom-pcie2-phy.txt | 40 +++++++++++++++++++
> > 1 file changed, 40 insertions(+)
> > create mode 100644 Documentation/devicetree/bindings/phy/qcom-pcie2-phy.txt
> >
> > diff --git a/Documentation/devicetree/bindings/phy/qcom-pcie2-phy.txt b/Documentation/devicetree/bindings/phy/qcom-pcie2-phy.txt
> > new file mode 100644
> > index 000000000000..7da02f9d78c7
> > --- /dev/null
> > +++ b/Documentation/devicetree/bindings/phy/qcom-pcie2-phy.txt
> > @@ -0,0 +1,40 @@
> > +Qualcomm PCIe2 PHY controller
> > +=============================
> > +
> > +The Qualcomm PCIe2 PHY is a Synopsys based phy found in a number of Qualcomm
> > +platforms.
> > +
> > +Required properties:
> > + - compatible: compatible list, should be:
> > + "qcom,qcs404-pcie2-phy", "qcom,pcie2-phy"
> > +
> > + - reg: offset and length of the PHY register set.
> > + - #phy-cells: must be 0.
> > +
> > + - clocks: a clock-specifier pair for the "pipe" clock
> > +
> > + - vdda-vp-supply: phandle to low voltage regulator
> > + - vdda-vph-supply: phandle to high voltage regulator
> > +
> > + - resets: reset-specifier pairs for the "phy" and "pipe" resets
> > + - reset-names: list of resets, should contain:
> > + "phy" and "pipe"
> > +
> > + - clock-output-names: name of the outgoing clock signal from the PHY PLL
>
> Not valid to have this without '#clock-cells'.
>
This happens to work today as we don't describe &gcc's dependency on
this, but just rely on the global clock namespace. But it should be
there, so I'll respin this.
> Though I'm confused how this and the input clock name seem to match.
>
The PHY generates the clock we call "pcie_0_pipe_clk" which is fed into
the global clock controller (gcc), where it can be gated (and divided)
and then fed back to both the PHY and the controller as
"gcc_pcie_0_pipe_clk".
Thanks for the review Rob.
Regards,
Bjorn
> > +
> > +Example:
> > + phy@7786000 {
> > + compatible = "qcom,qcs404-pcie2-phy", "qcom,pcie2-phy";
> > + reg = <0x07786000 0xb8>;
> > +
> > + clocks = <&gcc GCC_PCIE_0_PIPE_CLK>;
> > + resets = <&gcc GCC_PCIEPHY_0_PHY_BCR>,
> > + <&gcc GCC_PCIE_0_PIPE_ARES>;
> > + reset-names = "phy", "pipe";
> > +
> > + vdda-vp-supply = <&vreg_l3_1p05>;
> > + vdda-vph-supply = <&vreg_l5_1p8>;
> > +
> > + clock-output-names = "pcie_0_pipe_clk";
> > + #phy-cells = <0>;
> > + };
> > --
> > 2.18.0
> >
