All of lore.kernel.org
 help / color / mirror / Atom feed
From: sashiko-bot@kernel.org
To: "Gopikishan Thawre" <gopikishan.thawre@oss.qualcomm.com>
Cc: devicetree@vger.kernel.org, robh@kernel.org, conor+dt@kernel.org
Subject: Re: [PATCH v3 3/3] arm64: dts: qcom: lemans: Add LeMans Draco EVK board
Date: Fri, 20 Nov 2026 07:11:09 +0000	[thread overview]
Message-ID: <sashiko-outbox-224600@kernel.org> (raw)
In-Reply-To: <20261119-lemans-draco-final-changes-bringup-v3-3-db08f12f0cd5@oss.qualcomm.com>

Thank you for your contribution! Sashiko AI review found 8 potential
issues to consider:
- [High] arm64: dts: qcom: wrong deprecated phy-supply on serdes0 and
  serdes1
- [High] arm64: dts: qcom: missing enable-active-high on DP power
  regulators
- [High] arm64: dts: qcom: misplaced PCIe PERST#/WAKE# GPIOs and wrong
  WAKE# polarity
- [High] arm64: dts: qcom: missing pciclass,0604 compatible on TC9563
  pcie@1,0 port
- [High] arm64: dts: qcom: swapped vdda-phy and vdda-pll on USB QMP PHYs
  in Draco SoM
- [Medium] arm64: dts: qcom: TC9563 endpoint functions under pcie@3,0
  modeled as bridges
- [Low] arm64: dts: qcom: undocumented vdda-refgen-supply on pcie0_phy
  and pcie1_phy
- [Low] arm64: dts: qcom: Makefile entries for lemans-draco-evk out of
  order
--

commit cde350979c0cc65bfe5158a7e63d40d7f1a15e33
Author: Gopikishan Thawre <gopikishan.thawre@oss.qualcomm.com>

arm64: dts: qcom: lemans: Add LeMans Draco EVK board

This commit adds the device tree and Makefile entries for the Qualcomm
LeMans Draco EVK board, building on top of the LeMans Draco SoM DTSI.

> diff --git a/arch/arm64/boot/dts/qcom/Makefile b/arch/arm64/boot/dts/qcom/Makefile
> index 9553e3aaeccb0..11448d8c70876 100644
> --- a/arch/arm64/boot/dts/qcom/Makefile
> +++ b/arch/arm64/boot/dts/qcom/Makefile
> @@ -62,6 +62,11 @@ dtb-$(CONFIG_ARCH_QCOM)	+= kaanapali-mtp.dtb
>  dtb-$(CONFIG_ARCH_QCOM)	+= kaanapali-qrd.dtb
>  dtb-$(CONFIG_ARCH_QCOM)	+= kalambo-crd.dtb
>  dtb-$(CONFIG_ARCH_QCOM)	+= lemans-evk.dtb
> +dtb-$(CONFIG_ARCH_QCOM)	+= lemans-draco-evk.dtb
> +
> +lemans-draco-evk-el2-dtbs := lemans-draco-evk.dtb lemans-el2.dtbo
> +
> +dtb-$(CONFIG_ARCH_QCOM)	+= lemans-draco-evk-el2.dtb

[Severity: Low]
This isn't a bug, but lemans-draco-evk.dtb and lemans-draco-evk-el2.dtb
sort alphabetically before lemans-evk.dtb.

> diff --git a/arch/arm64/boot/dts/qcom/lemans-draco-evk.dts b/arch/arm64/boot/dts/qcom/lemans-draco-evk.dts
> new file mode 100644
> index 0000000000000..7ac3a867d29a6
> --- /dev/null
> +++ b/arch/arm64/boot/dts/qcom/lemans-draco-evk.dts
> @@ -0,0 +1,778 @@
> +// SPDX-License-Identifier: BSD-3-Clause
> +/*
> + * Copyright (c) Qualcomm Technologies, Inc. and/or its subsidiaries.
> + */
> +/dts-v1/;
> +
> +#include <dt-bindings/pwm/pwm.h>
> +#include "lemans-draco-som.dtsi"

[Severity: High]
Are vdda-phy-supply and vdda-pll-supply swapped on usb_0_qmpphy and
usb_1_qmpphy in lemans-draco-som.dtsi (added in commit "arm64: dts: qcom:
lemans: Add LeMans Draco EVK SoM platform")?

Both USB QMP PHY nodes assign the 1.2V rail (vreg_l1c) to vdda-phy-supply
and the ~0.88V rail (vreg_l7a) to vdda-pll-supply:

arch/arm64/boot/dts/qcom/lemans-draco-som.dtsi:
&usb_0_qmpphy {
	vdda-phy-supply = <&vreg_l1c>;
	vdda-pll-supply = <&vreg_l7a>;

	status = "okay";
};

As fixed for other LeMans boards in 932113c3c778 ("arm64: dts: qcom:
lemans: Fix swapped USB QMP PHY vdda-phy/vdda-pll supplies") and matching
ufs_mem_phy in the same SoM file, the SA8775P QMP PHY expects the ~0.88V
rail (vreg_l7a) on vdda-phy-supply and the 1.2V rail (vreg_l1c) on
vdda-pll-supply.

[ ... ]

> +	vreg_dp0_pwr: regulator-dp0-pwr {
> +		compatible = "regulator-fixed";
> +		regulator-name = "VREG_DP0_PWR_3P3";
> +
> +		regulator-min-microvolt = <3300000>;
> +		regulator-max-microvolt = <3300000>;
> +
> +		gpio = <&expander1 7 GPIO_ACTIVE_HIGH>;
> +	};
> +
> +	vreg_dp1_pwr: regulator-dp1-pwr {
> +		compatible = "regulator-fixed";
> +		regulator-name = "VREG_DP1_PWR_3P3";
> +
> +		regulator-min-microvolt = <3300000>;
> +		regulator-max-microvolt = <3300000>;
> +
> +		gpio = <&expander1 8 GPIO_ACTIVE_HIGH>;
> +	};
> +
> +	vreg_dp2_pwr: regulator-dp2-pwr {
> +		compatible = "regulator-fixed";
> +		regulator-name = "VREG_DP2_PWR_3P3";
> +
> +		regulator-min-microvolt = <3300000>;
> +		regulator-max-microvolt = <3300000>;
> +
> +		gpio = <&expander1 9 GPIO_ACTIVE_HIGH>;
> +	};
> +
> +	vreg_dp3_pwr: regulator-dp3-pwr {
> +		compatible = "regulator-fixed";
> +		regulator-name = "VREG_DP3_PWR_3P3";
> +
> +		regulator-min-microvolt = <3300000>;
> +		regulator-max-microvolt = <3300000>;
> +
> +		gpio = <&expander1 10 GPIO_ACTIVE_HIGH>;
> +	};

[Severity: High]
Are vreg_dp0_pwr through vreg_dp3_pwr missing enable-active-high?

For regulator-fixed nodes, of_gpio_set_polarity_by_property() in
drivers/gpio/gpiolib-of.c ignores the GPIO_ACTIVE_HIGH flag in the gpio
phandle specifier and forces OF_GPIO_ACTIVE_LOW unless enable-active-high
is present, so enabling these regulators will drive expander1 pins 7..10
low and keep 3.3V DP_PWR disabled on all four DisplayPort connectors.

[ ... ]

> +&pcie0 {
> +	perst-gpios = <&tlmm 2 GPIO_ACTIVE_LOW>;
> +	wake-gpios = <&tlmm 0 GPIO_ACTIVE_HIGH>;
> +
> +	pinctrl-0 = <&pcie0_default_state>;
> +	pinctrl-names = "default";
> +
> +	iommu-map = <0x0 &pcie_smmu 0x0 0x0 0x1>,
> +		    <0x100 &pcie_smmu 0x1 0x0 0x1>,
> +		    <0x208 &pcie_smmu 0x2 0x0 0x1>,
> +		    <0x210 &pcie_smmu 0x3 0x0 0x1>,
> +		    <0x218 &pcie_smmu 0x4 0x0 0x1>,
> +		    <0x300 &pcie_smmu 0x5 0x0 0x1>,
> +		    <0x400 &pcie_smmu 0x6 0x0 0x1>,
> +		    <0x500 &pcie_smmu 0x7 0x0 0x1>,
> +		    <0x501 &pcie_smmu 0x8 0x0 0x1>;
> +
> +	status = "okay";
> +};
> +
> +&pcie0_phy {
> +	vdda-phy-supply = <&vreg_l5a>;
> +	vdda-pll-supply = <&vreg_l1c>;
> +	vdda-qref-supply = <&vreg_l4a>;
> +	vdda-refgen-supply = <&vreg_l7a>;
> +
> +	status = "okay";
> +};
> +
> +&pcie1 {
> +	perst-gpios = <&tlmm 4 GPIO_ACTIVE_LOW>;
> +	wake-gpios = <&tlmm 5 GPIO_ACTIVE_HIGH>;
> +
> +	pinctrl-0 = <&pcie1_default_state>;
> +	pinctrl-names = "default";
> +
> +	status = "okay";
> +};

[Severity: High]
Should perst-gpios and wake-gpios on pcie0 and pcie1 be moved to the root
port nodes (pcieport0 and pcie1_port0) as reset-gpios and wake-gpios, with
wake-gpios changed to GPIO_ACTIVE_LOW?

pci_configure_of_wake_gpio() looks up wake-gpios on the root port pci_dev
OF node rather than the host bridge node, so wake-gpios on pcie0 and pcie1
is ignored; furthermore, WAKE# is an active-low line pulled up in
pcie0_default_state and pcie1_default_state, so GPIO_ACTIVE_HIGH would
configure a level-high wake IRQ (see e3a8a50e2a04 ("arm64: dts: qcom:
lemans: Move PCIe phy and GPIOs to root port node") and cd8d4803bba6
("arm64: dts: qcom: lemans: Fix PCIe WAKE# GPIO polarity")).

Keeping perst-gpios on the Root Complex node also triggers a runtime
warning in qcom_pcie_parse_perst() ("Reusing PERST# from Root Complex
node. DT needs to be fixed!") and adds duplicate reset descriptors to
port->perst for every descendant PCI bridge node.

> +
> +&pcie1_phy {
> +	vdda-phy-supply = <&vreg_l5a>;
> +	vdda-pll-supply = <&vreg_l1c>;
> +	vdda-qref-supply = <&vreg_l4a>;
> +	vdda-refgen-supply = <&vreg_l7a>;
> +
> +	status = "okay";
> +};

[Severity: Low]
Does vdda-refgen-supply on pcie0_phy and pcie1_phy violate
qcom,sc8280xp-qmp-pcie-phy.yaml?  The binding only permits
vdda-phy-supply, vdda-pll-supply, and vdda-qref-supply (and the SA8775P
QMP PCIe PHY driver only requests vdda-phy and vdda-pll).

> +
> +&pcieport0 {
> +	#address-cells = <3>;
> +	#size-cells = <2>;
> +
> +	pcie@0,0 {
> +		compatible = "pci1179,0623";
> +		reg = <0x10000 0x0 0x0 0x0 0x0>;
> +		#address-cells = <3>;
> +		#size-cells = <2>;
> +		device_type = "pci";
> +		ranges;
> +		bus-range = <0x2 0xff>;
> +
> +		vddc-supply = <&vreg_0p9>;
> +		vdd18-supply = <&vreg_1p8>;
> +		vdd09-supply = <&vreg_0p9>;
> +		vddio1-supply = <&vreg_1p8>;
> +		vddio2-supply = <&vreg_1p8>;
> +		vddio18-supply = <&vreg_1p8>;
> +		i2c-parent = <&i2c18 0x77>;
> +		resx-gpios = <&tlmm 10 GPIO_ACTIVE_LOW>;
> +		pinctrl-0 = <&tc9563_resx_n>;
> +		pinctrl-names = "default";
> +
> +		pcie@1,0 {
> +			reg = <0x20800 0x0 0x0 0x0 0x0>;
> +			#address-cells = <3>;
> +			#size-cells = <2>;
> +			device_type = "pci";
> +			ranges;
> +			bus-range = <0x3 0xff>;
> +
> +			port {
> +				tc9563_pcie1_ep: endpoint {};
> +			};
> +		};

[Severity: High]
Is pcie@1,0 missing compatible = "pciclass,0604"?

pci_pwrctrl_is_required() returns false when a PCI bridge node has no
compatible property starting with "pci", so pci-pwrctrl-slot will not bind
to this downstream port and the "pcie" power sequencer target on
wifi-bt-connector (pwrseq-pcie-m2) will never be enabled for the M.2 E-key
slot.

> +
> +		pcie@2,0 {
> +			reg = <0x21000 0x0 0x0 0x0 0x0>;
> +			#address-cells = <3>;
> +			#size-cells = <2>;
> +			device_type = "pci";
> +			ranges;
> +			bus-range = <0x4 0xff>;
> +		};
> +
> +		pcie@3,0 {
> +			reg = <0x21800 0x0 0x0 0x0 0x0>;
> +			#address-cells = <3>;
> +			#size-cells = <2>;
> +			device_type = "pci";
> +			ranges;
> +			bus-range = <0x5 0xff>;
> +
> +			pcie@0,0 {
> +				reg = <0x50000 0x0 0x0 0x0 0x0>;
> +				#address-cells = <3>;
> +				#size-cells = <2>;
> +				device_type = "pci";
> +				ranges;
> +			};
> +
> +			pcie@0,1 {
> +				reg = <0x50100 0x0 0x0 0x0 0x0>;
> +				#address-cells = <3>;
> +				#size-cells = <2>;
> +				device_type = "pci";
> +				ranges;
> +			};
> +		};

[Severity: Medium]
Should the two integrated TC9563 Ethernet endpoint functions under
pcie@3,0 omit device_type = "pci", #address-cells, #size-cells, and
ranges (and be named ethernet@0,0 and ethernet@0,1 per
toshiba,tc9563.yaml)?

These functions are PCIe endpoints rather than PCI-to-PCI bridges, and
setting device_type = "pci" causes kernel OF PCI helpers that check
of_node_is_type(child, "pci") to treat them as bridges (see ca3027cfe8c7
("arm64: dts: qcom: lemans-evk-ifp-mezzanine: Clean up TC9563 switch
nodes")).

[ ... ]

> +&serdes0 {
> +	phy-supply = <&vreg_l5a>;
> +
> +	status = "okay";
> +};
> +
> +&serdes1 {
> +	phy-supply = <&vreg_l5a>;
> +
> +	status = "okay";
> +};

[Severity: High]
Should serdes0 and serdes1 specify vdda-0p9-supply = <&vreg_l4a> and
vdda-1p2-supply = <&vreg_l1c> instead of phy-supply = <&vreg_l5a>?

In drivers/phy/qualcomm/phy-qcom-sgmii-eth.c, qcom_dwmac_sgmii_phy_probe()
only requests "vdda-0p9" and "vdda-1p2" and ignores the deprecated
phy-supply property:

drivers/phy/qualcomm/phy-qcom-sgmii-eth.c:
static const struct regulator_bulk_data qcom_dwmac_sgmii_phy_vregs[] = {
	{ .supply = "vdda-0p9", .init_load_uA = 46000 },
	{ .supply = "vdda-1p2", .init_load_uA = 15000 },
};

Without vdda-0p9-supply and vdda-1p2-supply (as fixed for lemans-evk in
457cf65a7997 ("arm64: dts: qcom: lemans-evk: fix SerDes PHY regulator
supplies")), dummy regulators are used and the 0.9V and 1.2V SerDes rails
are not voted for when bringing up ethernet0 and ethernet1.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261119-lemans-draco-final-changes-bringup-v3-0-db08f12f0cd5@oss.qualcomm.com?part=3

      reply	other threads:[~2026-11-20  7:11 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-19  7:10 [PATCH v3 0/3] arm64: dts: qcom: lemans: Add LeMans Draco EVK board support Gopikishan Thawre
2026-11-19  7:10 ` [PATCH v3 1/3] dt-bindings: arm: qcom: Add LeMans Draco EVK board Gopikishan Thawre
2026-11-19  7:10 ` [PATCH v3 2/3] arm64: dts: qcom: lemans: Add LeMans Draco EVK SoM platform Gopikishan Thawre
2026-11-20  7:11   ` sashiko-bot
2026-11-19  7:10 ` [PATCH v3 3/3] arm64: dts: qcom: lemans: Add LeMans Draco EVK board Gopikishan Thawre
2026-11-20  7:11   ` sashiko-bot [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=sashiko-outbox-224600@kernel.org \
    --to=sashiko-bot@kernel.org \
    --cc=conor+dt@kernel.org \
    --cc=devicetree@vger.kernel.org \
    --cc=gopikishan.thawre@oss.qualcomm.com \
    --cc=robh@kernel.org \
    --cc=sashiko-reviews@lists.linux.dev \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.