[v2] arm64: dts: qcom: sc7280: eDP for herobrine boards

Add eDP support to herobrine boards, splitting up amongst the
different files as makes sense. Rationale for the current split of
things:
* The eDP connector itself is on qcard. However, not all devices with
  a qcard will use an eDP panel. Some might use MIPI and, presumably,
  someone could build a device with qcard that had no display at all.
* The qcard provides a PWM for backlight that goes to the eDP
  connector. This PWM is also provided to the board and it's expected
  that it would be used as the backlight PWM even for herobrine
  devices with MIPI displays.
* It's currently assumed that all herobrine boards will have some sort
  of display, either MIPI or eDP (but not both).
* We will assume herobrine-rev1 has eDP. The schematics allow for a
  MIPI panel to be hooked up but, aside from some testing, nobody is
  doing this and most boards don't have all the parts stuffed for
  it. The two panels would also share a PWM for backlight, which is
  weird.
* herobrine-villager and herobrine-hoglin (crd) also have eDP.
* herobrine-hoglin (crd) has slightly different regulator setup for
  the backlight. It's expected that this is unique to this board. See
  comments in the dts file.
* There are some regulators that are defined in the qcard schematic
  but provided by the board like "vreg_edp_bl" and
  "vreg_edp_3p3". While we could put references to these regulators
  straight in the qcard.dtsi file, this would force someone using
  qcard that didn't provide those regulators to provide a dummy or do
  an ugly /delete-node/. Instead, we'll add references in
  herobrine.dtsi.

Signed-off-by: Douglas Anderson <dianders@chromium.org>
---
This patch most directly depends on ("arm64: dts: qcom: pm8350c: Add
pwm support") [1] and it won't even compile without that. To be
functional, of course, we also need the driver support for the PWM to
land.

In order for this patch to function we also need Sankeerth's eDP
series [2] to land.

To be schema happy, this patch also relies on my patch to add
regulators into the schema for the PHY an the eDP controller [3].
Several other device tree files reference these supplies without being
in the schema, though, so we'll have to decide whether to block on
those bindings landing. It does sound as if the regulator names may
not be exactly correct right now, though. :(

In theory, I could break this patch up into separate patches adding
the basic support to the qcard, then to herobrine.dtsi, and then one
patch each for herobrine-r1, villager, and crd. Doing so didn't seem
to make sense to me--I think it's easier to make sense of the change
as one patch. However, if someone feels strongly that it should be
broken up I'm happy to do so.

I've managed to get the display on my herobrine-rev1 up and running on
today's linuxnext (next-20220422) with this series. For whatever
reason the eDP PHY wouldn't probe unless I hacked `fw_devlink=off` in
the config. I don't believe that problem is related to this patch,
though.

[1] https://lore.kernel.org/r/1645509309-16142-4-git-send-email-quic_c_skakit@quicinc.com
[2] https://lore.kernel.org/r/1650887072-16652-1-git-send-email-quic_sbillaka@quicinc.com/
[3] https://lore.kernel.org/r/20220425210643.2420919-1-dianders@chromium.org

Changes in v2:
- Commit message and comment cleanups from Stephen.

 .../boot/dts/qcom/sc7280-herobrine-crd.dts    | 40 ++++++++++++++
 .../qcom/sc7280-herobrine-herobrine-r1.dts    |  8 +++
 .../dts/qcom/sc7280-herobrine-villager-r0.dts |  8 +++
 .../arm64/boot/dts/qcom/sc7280-herobrine.dtsi | 24 +++++++++
 arch/arm64/boot/dts/qcom/sc7280-qcard.dtsi    | 54 +++++++++++++++++++
 5 files changed, 134 insertions(+)

Quoting Douglas Anderson (2022-04-26 12:41:03)
> Add eDP support to herobrine boards, splitting up amongst the
> different files as makes sense. Rationale for the current split of
> things:
> * The eDP connector itself is on qcard. However, not all devices with
>   a qcard will use an eDP panel. Some might use MIPI and, presumably,
>   someone could build a device with qcard that had no display at all.
> * The qcard provides a PWM for backlight that goes to the eDP
>   connector. This PWM is also provided to the board and it's expected
>   that it would be used as the backlight PWM even for herobrine
>   devices with MIPI displays.
> * It's currently assumed that all herobrine boards will have some sort
>   of display, either MIPI or eDP (but not both).
> * We will assume herobrine-rev1 has eDP. The schematics allow for a
>   MIPI panel to be hooked up but, aside from some testing, nobody is
>   doing this and most boards don't have all the parts stuffed for
>   it. The two panels would also share a PWM for backlight, which is
>   weird.
> * herobrine-villager and herobrine-hoglin (crd) also have eDP.
> * herobrine-hoglin (crd) has slightly different regulator setup for
>   the backlight. It's expected that this is unique to this board. See
>   comments in the dts file.
> * There are some regulators that are defined in the qcard schematic
>   but provided by the board like "vreg_edp_bl" and
>   "vreg_edp_3p3". While we could put references to these regulators
>   straight in the qcard.dtsi file, this would force someone using
>   qcard that didn't provide those regulators to provide a dummy or do
>   an ugly /delete-node/. Instead, we'll add references in
>   herobrine.dtsi.
>
> Signed-off-by: Douglas Anderson <dianders@chromium.org>
> ---

Reviewed-by: Stephen Boyd <swboyd@chromium.org>

I have reviewed all the display bits in this and they are in-line with

https://patchwork.kernel.org/project/linux-arm-msm/patch/1647452154-16361-3-git-send-email-quic_sbillaka@quicinc.com/

which is a working reference for this.

Hence,

Reviewed-by: Abhinav Kumar <quic_abhinavk@quicinc.com>

On 4/26/2022 12:41 PM, Douglas Anderson wrote:
> Add eDP support to herobrine boards, splitting up amongst the
> different files as makes sense. Rationale for the current split of
> things:
> * The eDP connector itself is on qcard. However, not all devices with
>    a qcard will use an eDP panel. Some might use MIPI and, presumably,
>    someone could build a device with qcard that had no display at all.
> * The qcard provides a PWM for backlight that goes to the eDP
>    connector. This PWM is also provided to the board and it's expected
>    that it would be used as the backlight PWM even for herobrine
>    devices with MIPI displays.
> * It's currently assumed that all herobrine boards will have some sort
>    of display, either MIPI or eDP (but not both).
> * We will assume herobrine-rev1 has eDP. The schematics allow for a
>    MIPI panel to be hooked up but, aside from some testing, nobody is
>    doing this and most boards don't have all the parts stuffed for
>    it. The two panels would also share a PWM for backlight, which is
>    weird.
> * herobrine-villager and herobrine-hoglin (crd) also have eDP.
> * herobrine-hoglin (crd) has slightly different regulator setup for
>    the backlight. It's expected that this is unique to this board. See
>    comments in the dts file.
> * There are some regulators that are defined in the qcard schematic
>    but provided by the board like "vreg_edp_bl" and
>    "vreg_edp_3p3". While we could put references to these regulators
>    straight in the qcard.dtsi file, this would force someone using
>    qcard that didn't provide those regulators to provide a dummy or do
>    an ugly /delete-node/. Instead, we'll add references in
>    herobrine.dtsi.
> 
> Signed-off-by: Douglas Anderson <dianders@chromium.org>
> ---
> This patch most directly depends on ("arm64: dts: qcom: pm8350c: Add
> pwm support") [1] and it won't even compile without that. To be
> functional, of course, we also need the driver support for the PWM to
> land.
> 
> In order for this patch to function we also need Sankeerth's eDP
> series [2] to land.
> 
> To be schema happy, this patch also relies on my patch to add
> regulators into the schema for the PHY an the eDP controller [3].
> Several other device tree files reference these supplies without being
> in the schema, though, so we'll have to decide whether to block on
> those bindings landing. It does sound as if the regulator names may
> not be exactly correct right now, though. :(
> 
> In theory, I could break this patch up into separate patches adding
> the basic support to the qcard, then to herobrine.dtsi, and then one
> patch each for herobrine-r1, villager, and crd. Doing so didn't seem
> to make sense to me--I think it's easier to make sense of the change
> as one patch. However, if someone feels strongly that it should be
> broken up I'm happy to do so.
> 
> I've managed to get the display on my herobrine-rev1 up and running on
> today's linuxnext (next-20220422) with this series. For whatever
> reason the eDP PHY wouldn't probe unless I hacked `fw_devlink=off` in
> the config. I don't believe that problem is related to this patch,
> though.
> 
> [1] https://lore.kernel.org/r/1645509309-16142-4-git-send-email-quic_c_skakit@quicinc.com
> [2] https://lore.kernel.org/r/1650887072-16652-1-git-send-email-quic_sbillaka@quicinc.com/
> [3] https://lore.kernel.org/r/20220425210643.2420919-1-dianders@chromium.org
> 
> Changes in v2:
> - Commit message and comment cleanups from Stephen.
> 
>   .../boot/dts/qcom/sc7280-herobrine-crd.dts    | 40 ++++++++++++++
>   .../qcom/sc7280-herobrine-herobrine-r1.dts    |  8 +++
>   .../dts/qcom/sc7280-herobrine-villager-r0.dts |  8 +++
>   .../arm64/boot/dts/qcom/sc7280-herobrine.dtsi | 24 +++++++++
>   arch/arm64/boot/dts/qcom/sc7280-qcard.dtsi    | 54 +++++++++++++++++++
>   5 files changed, 134 insertions(+)
> 
> diff --git a/arch/arm64/boot/dts/qcom/sc7280-herobrine-crd.dts b/arch/arm64/boot/dts/qcom/sc7280-herobrine-crd.dts
> index b06f61e9b90b..a4ac33c4fd59 100644
> --- a/arch/arm64/boot/dts/qcom/sc7280-herobrine-crd.dts
> +++ b/arch/arm64/boot/dts/qcom/sc7280-herobrine-crd.dts
> @@ -12,6 +12,27 @@
>   / {
>   	model = "Qualcomm Technologies, Inc. sc7280 CRD platform (rev5+)";
>   	compatible = "google,hoglin", "qcom,sc7280";
> +
> +	/* FIXED REGULATORS */
> +
> +	/*
> +	 * On most herobrine boards PPVAR_SYS directly provides VREG_EDP_BL.
> +	 * However, on CRD there's an extra regulator in the way. Since this
> +	 * is expected to be uncommon, we'll leave the "vreg_edp_bl" label
> +	 * in the baseboard herobrine.dtsi point at "ppvar_sys" and then
> +	 * make a "_crd" specific version here.
> +	 */
> +	vreg_edp_bl_crd: vreg-edp-bl-crd-regulator {
> +		compatible = "regulator-fixed";
> +		regulator-name = "vreg_edp_bl_crd";
> +
> +		gpio = <&pm8350c_gpios 6 GPIO_ACTIVE_HIGH>;
> +		enable-active-high;
> +		pinctrl-names = "default";
> +		pinctrl-0 = <&edp_bl_reg_en>;
> +
> +		vin-supply = <&ppvar_sys>;
> +	};
>   };
>   
>   /* ADDITIONS TO NODES DEFINED IN PARENT DEVICE TREE FILES */
> @@ -81,6 +102,14 @@ ap_ts: touchscreen@5c {
>   	};
>   };
>   
> +&mdss_edp {
> +	status = "okay";
> +};
> +
> +&mdss_edp_phy {
> +	status = "okay";
> +};
> +
>   /* For nvme */
>   &pcie1 {
>   	status = "okay";
> @@ -91,6 +120,10 @@ &pcie1_phy {
>   	status = "okay";
>   };
>   
> +&pm8350c_pwm_backlight {
> +	power-supply = <&vreg_edp_bl_crd>;
> +};
> +
>   /* For eMMC */
>   &sdhc_1 {
>   	status = "okay";
> @@ -121,6 +154,13 @@ &pm8350c_gpios {
>   			  "PMIC_EDP_BL_EN",
>   			  "PMIC_EDP_BL_PWM",
>   			  "";
> +
> +	edp_bl_reg_en: edp-bl-reg-en {
> +		pins = "gpio6";
> +		function = "normal";
> +		bias-disable;
> +		qcom,drive-strength = <PMIC_GPIO_STRENGTH_LOW>;
> +	};
>   };
>   
>   &tlmm {
> diff --git a/arch/arm64/boot/dts/qcom/sc7280-herobrine-herobrine-r1.dts b/arch/arm64/boot/dts/qcom/sc7280-herobrine-herobrine-r1.dts
> index 29c4ca095294..b69ca09d9bfb 100644
> --- a/arch/arm64/boot/dts/qcom/sc7280-herobrine-herobrine-r1.dts
> +++ b/arch/arm64/boot/dts/qcom/sc7280-herobrine-herobrine-r1.dts
> @@ -100,6 +100,14 @@ ap_ts: touchscreen@5c {
>   	};
>   };
>   
> +&mdss_edp {
> +	status = "okay";
> +};
> +
> +&mdss_edp_phy {
> +	status = "okay";
> +};
> +
>   /* For nvme */
>   &pcie1 {
>   	status = "okay";
> diff --git a/arch/arm64/boot/dts/qcom/sc7280-herobrine-villager-r0.dts b/arch/arm64/boot/dts/qcom/sc7280-herobrine-villager-r0.dts
> index 6c2b9a14535a..d3d6ffad4eff 100644
> --- a/arch/arm64/boot/dts/qcom/sc7280-herobrine-villager-r0.dts
> +++ b/arch/arm64/boot/dts/qcom/sc7280-herobrine-villager-r0.dts
> @@ -58,6 +58,14 @@ &ap_sar_sensor1 {
>   	status = "okay";
>   };
>   
> +&mdss_edp {
> +	status = "okay";
> +};
> +
> +&mdss_edp_phy {
> +	status = "okay";
> +};
> +
>   /* For nvme */
>   &pcie1 {
>   	status = "okay";
> diff --git a/arch/arm64/boot/dts/qcom/sc7280-herobrine.dtsi b/arch/arm64/boot/dts/qcom/sc7280-herobrine.dtsi
> index d58045dd7334..9cb1bc8ed6b5 100644
> --- a/arch/arm64/boot/dts/qcom/sc7280-herobrine.dtsi
> +++ b/arch/arm64/boot/dts/qcom/sc7280-herobrine.dtsi
> @@ -367,6 +367,11 @@ &vreg_l2c_1p8 {
>   
>   /* ADDITIONS TO NODES DEFINED IN PARENT DEVICE TREE FILES */
>   
> +&edp_panel {
> +	/* Our board provides power to the qcard for the eDP panel. */
> +	power-supply = <&vreg_edp_3p3>;
> +};
> +
>   ap_sar_sensor_i2c: &i2c1 {
>   	clock-frequency = <400000>;
>   	status = "disabled";
> @@ -420,6 +425,14 @@ tpm@50 {
>   	};
>   };
>   
> +&mdss {
> +	status = "okay";
> +};
> +
> +&mdss_mdp {
> +	status = "okay";
> +};
> +
>   /* NVMe drive, enabled on a per-board basis */
>   &pcie1 {
>   	pinctrl-names = "default";
> @@ -429,6 +442,17 @@ &pcie1 {
>   	vddpe-3v3-supply = <&pp3300_ssd>;
>   };
>   
> +&pm8350c_pwm {
> +	status = "okay";
> +};
> +
> +&pm8350c_pwm_backlight {
> +	status = "okay";
> +
> +	/* Our board provides power to the qcard for the backlight */
> +	power-supply = <&vreg_edp_bl>;
> +};
> +
>   &pmk8350_rtc {
>   	status = "disabled";
>   };
> diff --git a/arch/arm64/boot/dts/qcom/sc7280-qcard.dtsi b/arch/arm64/boot/dts/qcom/sc7280-qcard.dtsi
> index b833ba1e8f4a..fbd9f9fa9d3a 100644
> --- a/arch/arm64/boot/dts/qcom/sc7280-qcard.dtsi
> +++ b/arch/arm64/boot/dts/qcom/sc7280-qcard.dtsi
> @@ -29,6 +29,16 @@ aliases {
>   		serial0 = &uart5;
>   		serial1 = &uart7;
>   	};
> +
> +	pm8350c_pwm_backlight: backlight {
> +		compatible = "pwm-backlight";
> +		status = "disabled";
> +
> +		enable-gpios = <&pm8350c_gpios 7 GPIO_ACTIVE_HIGH>;
> +		pinctrl-names = "default";
> +		pinctrl-0 = <&pmic_edp_bl_en>;
> +		pwms = <&pm8350c_pwm 3 65535>;
> +	};
>   };
>   
>   &apps_rsc {
> @@ -293,11 +303,50 @@ &ipa {
>   	modem-init;
>   };
>   
> +/* NOTE: Not all Qcards have eDP connector stuffed */
> +&mdss_edp {
> +	vdda-0p9-supply = <&vdd_a_edp_0_0p9>;
> +	vdda-1p2-supply = <&vdd_a_edp_0_1p2>;
> +
> +	aux-bus {
> +		edp_panel: panel {
> +			compatible = "edp-panel";
> +
> +			backlight = <&pm8350c_pwm_backlight>;
> +
> +			ports {
> +				#address-cells = <1>;
> +				#size-cells = <0>;
> +				port@0 {
> +					reg = <0>;
> +					edp_panel_in: endpoint {
> +						remote-endpoint = <&mdss_edp_out>;
> +					};
> +				};
> +			};
> +		};
> +	};
> +};
> +
> +&mdss_edp_out {
> +	remote-endpoint = <&edp_panel_in>;
> +};
> +
> +&mdss_edp_phy {
> +	vdda-pll-supply = <&vdd_a_edp_0_0p9>;
> +	vdda-phy-supply = <&vdd_a_edp_0_1p2>;
> +};
> +
>   &pcie1_phy {
>   	vdda-phy-supply = <&vreg_l10c_0p88>;
>   	vdda-pll-supply = <&vreg_l6b_1p2>;
>   };
>   
> +&pm8350c_pwm {
> +	pinctrl-names = "default";
> +	pinctrl-0 = <&pmic_edp_bl_pwm>;
> +};
> +
>   &pmk8350_vadc {
>   	pmk8350-die-temp@3 {
>   		reg = <PMK8350_ADC7_DIE_TEMP>;
> @@ -383,6 +432,11 @@ &usb_2_hsphy {
>    * baseboard or board device tree, not here.
>    */
>   
> +/* No external pull for eDP HPD, so set the internal one. */
> +&edp_hot_plug_det {
> +	bias-pull-down;
> +};
> +
>   /*
>    * For ts_i2c
>    *

On Tue, 26 Apr 2022 12:41:03 -0700, Douglas Anderson wrote:
> Add eDP support to herobrine boards, splitting up amongst the
> different files as makes sense. Rationale for the current split of
> things:
> * The eDP connector itself is on qcard. However, not all devices with
>   a qcard will use an eDP panel. Some might use MIPI and, presumably,
>   someone could build a device with qcard that had no display at all.
> * The qcard provides a PWM for backlight that goes to the eDP
>   connector. This PWM is also provided to the board and it's expected
>   that it would be used as the backlight PWM even for herobrine
>   devices with MIPI displays.
> * It's currently assumed that all herobrine boards will have some sort
>   of display, either MIPI or eDP (but not both).
> * We will assume herobrine-rev1 has eDP. The schematics allow for a
>   MIPI panel to be hooked up but, aside from some testing, nobody is
>   doing this and most boards don't have all the parts stuffed for
>   it. The two panels would also share a PWM for backlight, which is
>   weird.
> * herobrine-villager and herobrine-hoglin (crd) also have eDP.
> * herobrine-hoglin (crd) has slightly different regulator setup for
>   the backlight. It's expected that this is unique to this board. See
>   comments in the dts file.
> * There are some regulators that are defined in the qcard schematic
>   but provided by the board like "vreg_edp_bl" and
>   "vreg_edp_3p3". While we could put references to these regulators
>   straight in the qcard.dtsi file, this would force someone using
>   qcard that didn't provide those regulators to provide a dummy or do
>   an ugly /delete-node/. Instead, we'll add references in
>   herobrine.dtsi.
> 
> [...]

Applied, thanks!

[1/1] arm64: dts: qcom: sc7280: eDP for herobrine boards
      commit: 366a0a194b3b3e3e52bc7b7b1ac35b40a1187902

Best regards,