On Thu, Feb 15, 2024 at 09:19:39AM +0100, Krzysztof Kozlowski wrote:
> On 15/02/2024 03:41, Bjorn Andersson wrote:
> > On Tue, Feb 13, 2024 at 09:39:51AM +0100, Krzysztof Kozlowski wrote:
> >> On 13/02/2024 09:27, Krishna Kurapati wrote:
> >>> Multiport USB controller (host-only) of SA8295 ADP has 4 Type-A ports
> >>> exposed for connecting peripherals. The VBUS to these peripherals is
> >>> provided by TPS2559QWDRCTQ1 regulators connected to these ports. Each
> >>> regulator has an enable pin controlled by PMM8540. Since these regulators
> >>> are GPIO controlled regulators, model them as fixed regulators and keep
> >>> them Always-On at boot since we are wakeup capable and we don't need to
> >>> turn them off on suspend. Also since we don't enter device mode, these
> >>> regulators can be kept on.
> >>>
> >>> Signed-off-by: Krishna Kurapati <quic_kriskura@xxxxxxxxxxx>
> >>> ---
> >>> arch/arm64/boot/dts/qcom/sa8295p-adp.dts | 83 ++++++++++++++++++++++++
> >>> 1 file changed, 83 insertions(+)
> >>>
> >>> diff --git a/arch/arm64/boot/dts/qcom/sa8295p-adp.dts b/arch/arm64/boot/dts/qcom/sa8295p-adp.dts
> >>> index fd253942e5e5..49418843c214 100644
> >>> --- a/arch/arm64/boot/dts/qcom/sa8295p-adp.dts
> >>> +++ b/arch/arm64/boot/dts/qcom/sa8295p-adp.dts
> >>> @@ -9,6 +9,7 @@
> >>> #include <dt-bindings/gpio/gpio.h>
> >>> #include <dt-bindings/regulator/qcom,rpmh-regulator.h>
> >>> #include <dt-bindings/spmi/spmi.h>
> >>> +#include <dt-bindings/pinctrl/qcom,pmic-gpio.h>
> >>>
> >>> #include "sa8540p.dtsi"
> >>> #include "sa8540p-pmics.dtsi"
> >>> @@ -108,6 +109,46 @@ edp3_connector_in: endpoint {
> >>> };
> >>> };
> >>> };
> >>> +
> >>> + regulator-usb2-vbus {
> >>> + compatible = "regulator-fixed";
> >>> + regulator-name = "USB2_VBUS";
> >>> + gpio = <&pmm8540c_gpios 9 GPIO_ACTIVE_HIGH>;
> >>> + pinctrl-0 = <&usb2_en>;
> >>> + pinctrl-names = "default";
> >>> + enable-active-high;
> >>> + regulator-always-on;
> >>> + };
> >>> +
> >>> + regulator-usb3-vbus {
> >>> + compatible = "regulator-fixed";
> >>> + regulator-name = "USB3_VBUS";
> >>> + gpio = <&pmm8540e_gpios 5 GPIO_ACTIVE_HIGH>;
> >>> + pinctrl-0 = <&usb3_en>;
> >>> + pinctrl-names = "default";
> >>> + enable-active-high;
> >>> + regulator-always-on;
> >>> + };
> >>> +
> >>> + regulator-usb4-vbus {
> >>> + compatible = "regulator-fixed";
> >>> + regulator-name = "USB4_VBUS";
> >>> + gpio = <&pmm8540g_gpios 5 GPIO_ACTIVE_HIGH>;
> >>> + pinctrl-0 = <&usb4_en>;
> >>> + pinctrl-names = "default";
> >>> + enable-active-high;
> >>> + regulator-always-on;
> >>> + };
> >>> +
> >>> + regulator-usb5-vbus {
> >>> + compatible = "regulator-fixed";
> >>> + regulator-name = "USB5_VBUS";
> >>> + gpio = <&pmm8540g_gpios 9 GPIO_ACTIVE_HIGH>;
> >>> + pinctrl-0 = <&usb5_en>;
> >>> + pinctrl-names = "default";
> >>> + enable-active-high;
> >>> + regulator-always-on;
> >>
> >> Why all these regulators are always on? If USB controller does not probe
> >> for any reason, why keeping them enabled? These must not be always-on,
> >> but instead used by connector as VBUS supply (or by whatever you have
> >> there for USB).
> >>
> >
> > I'm not too concerned about keeping the lights on in this scenario, but
> > if we can describe this properly let's do so (and let's do so on other
> > boards with connectors as well).
> >
> > We'd have a set of usb-a-connector nodes, that we can tie to the nodes
> > in the USB/phy, and the supply. But so far we've associated a connector
> > with a port manager, here we don't have one of those, so where would the
> > node reside and who should acquire and drive the vbus-supply?
>
> usb-connector binding has vbus-supply and its node could be top-level.
Introducing usb-connector nodes toplevel, with vbus-supply sounds
reasonable. But to my knowledge there's today no way to acquire a
handle to that regulator, unless you have a struct device for the
connector node.
> However don't some USB phys also take that regulator?
>
I don't think it's appropriate to add the supply to any of the phys,
some of the connectors has 2 phys others has 1 phy.
Representing the vbus in the connector but driving it from the phy
(we will need to figure out which one) sounds reasonable. We just need
to make sure this doesn't conflict with the fact that some TCPM
implementations also seems to drive vbus.
I would like this to be properly modelled, but it seems reasonable to
punt that to the todo list for now.
Regards,
Bjorn
> Best regards,
> Krzysztof
>
