dri-devel.lists.freedesktop.org archive mirror
 help / color / mirror / Atom feed
* Re: [PATCH v1 01/11] dt-bindings: arm: qcom: add LG V60 ThinQ 5G (timelm)
       [not found] ` <20261117123946.668557-2-rofikcoga@gmail.com>
@ 2026-11-17 12:47   ` sashiko-bot
  0 siblings, 0 replies; 11+ messages in thread
From: sashiko-bot @ 2026-11-17 12:47 UTC (permalink / raw)
  To: rofikkerneldev
  Cc: devicetree, dri-devel, linux-input, robh, conor+dt,
	dmitry.torokhov

Thank you for your contribution! Sashiko AI review found 2 potential
issues to consider:
- [High] dt-bindings: arm: qcom: username in Signed-off-by instead of
  real name
- [Low] dt-bindings: arm: qcom: duplicate lge vendor prefix instead of
  lg
--

commit 3057431969983ea736dadadbf8f06be7752765b6
Author: rofikkerneldev <rofikcoga@gmail.com>

> dt-bindings: arm: qcom: add LG V60 ThinQ 5G (timelm)
>
> Document the compatible for the LG V60 ThinQ 5G (timelm), based on
> SM8250, and register the LG Electronics vendor prefix.
>
> Signed-off-by: rofikkerneldev <rofikcoga@gmail.com>

[Severity: High]
The Author header and Signed-off-by trailer use a handle (rofikkerneldev)
instead of a known real human name, which is required by the Developer's
Certificate of Origin in Documentation/process/submitting-patches.rst.

> diff --git a/Documentation/devicetree/bindings/arm/qcom.yaml b/Documentation/devicetree/bindings/arm/qcom.yaml
> index 121ca557ce1ef..f8b34109171d0 100644
> --- a/Documentation/devicetree/bindings/arm/qcom.yaml
> +++ b/Documentation/devicetree/bindings/arm/qcom.yaml
> @@ -1139,6 +1139,7 @@ properties:
>  
>        - items:
>            - enum:
> +              - lge,timelm
>                - qcom,qrb5165-rb5
>                - qcom,sm8250-hdk
>                - qcom,sm8250-mtp
> diff --git a/Documentation/devicetree/bindings/vendor-prefixes.yaml b/Documentation/devicetree/bindings/vendor-prefixes.yaml
> index 5471a06f93f10..fa50a81880f0b 100644
> --- a/Documentation/devicetree/bindings/vendor-prefixes.yaml
> +++ b/Documentation/devicetree/bindings/vendor-prefixes.yaml
> @@ -971,6 +971,8 @@ patternProperties:
>      description: Lenovo Group Ltd.
>    "^lg,.*":
>      description: LG Corporation
> +  "^lge,.*":
> +    description: LG Electronics
>    "^lgphilips,.*":
>      description: LG Display
>    "^libretech,.*":

[Severity: Low]
Should this use the existing "lg" vendor prefix (lg,timelm) instead of
adding a duplicate "lge" prefix?

Documentation/devicetree/bindings/vendor-prefixes.yaml already defines
"^lg,.*" for LG Corporation, and commit 5e4aac2caf12 ("dt-bindings: arm:
qcom: rename vendor of apq8026-lenok") replaced "lge" with "lg" in
Documentation/devicetree/bindings/arm/qcom.yaml to avoid separate vendor
prefixes for internal divisions of LG.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261117123946.668557-1-rofikcoga@gmail.com?part=1

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: [PATCH v1 06/11] dt-bindings: input: touchscreen: add Silicon Works SW42902
       [not found] ` <20261117123946.668557-7-rofikcoga@gmail.com>
@ 2026-11-17 12:51   ` sashiko-bot
  0 siblings, 0 replies; 11+ messages in thread
From: sashiko-bot @ 2026-11-17 12:51 UTC (permalink / raw)
  To: rofikkerneldev
  Cc: devicetree, dri-devel, linux-input, robh, conor+dt,
	dmitry.torokhov

Thank you for your contribution! Sashiko AI review found 2 potential
issues to consider:
- [High] dt-bindings: input: touchscreen: username handle used in
  sign-off and maintainer
- [Medium] dt-bindings: input: lg,sw42902: power rails modeled as GPIOs
  instead of supplies
--

commit c2a344c3130db163c68195828fdefbdfd36cd07a
Author: rofikkerneldev <rofikcoga@gmail.com>

> dt-bindings: input: touchscreen: add Silicon Works SW42902
>
> Document the I2C touchscreen and its interrupt, reset, power-enable,
> and panel-size properties.
>
> Signed-off-by: rofikkerneldev <rofikcoga@gmail.com>

[Severity: High]
The Developer's Certificate of Origin requires a real human name (known
identity) rather than an account handle or pseudonym in Signed-off-by
trailers, the commit Author header, the binding's maintainers list, and
MAINTAINERS.

> diff --git a/Documentation/devicetree/bindings/input/touchscreen/lg,sw42902.yaml b/Documentation/devicetree/bindings/input/touchscreen/lg,sw42902.yaml
> new file mode 100644
> index 0000000000000..c56cfb8530eeb
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/input/touchscreen/lg,sw42902.yaml
> @@ -0,0 +1,74 @@

[ ... ]

> +  vcl-gpios:
> +    maxItems: 1
> +    description: Active-high enable for the 3.3 V rail
> +
> +  vdd-gpios:
> +    maxItems: 1
> +    description: Active-high enable for the 1.8 V rail
> +
> +required:
> +  - compatible
> +  - reg
> +  - interrupts
> +  - reset-gpios
> +  - vcl-gpios
> +  - vdd-gpios

[Severity: Medium]
Should the 3.3 V VCL and 1.8 V VDD power inputs be described as
vcl-supply and vdd-supply regulator properties instead of vcl-gpios and
vdd-gpios?

VCL and VDD are power supply rails on the SW42902 controller rather than
GPIO pins, so requiring GPIO properties in the binding prevents it from
describing boards powered by PMIC regulators or always-on supplies.
GPIO-switched rails on a specific board can be modeled as regulator-fixed
nodes in the board DTS connected to vcl-supply and vdd-supply.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261117123946.668557-1-rofikcoga@gmail.com?part=6

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: [PATCH v1 04/11] dt-bindings: display: panel: add LG SW43103
       [not found] ` <20261117123946.668557-5-rofikcoga@gmail.com>
@ 2026-11-17 12:51   ` sashiko-bot
  0 siblings, 0 replies; 11+ messages in thread
From: sashiko-bot @ 2026-11-17 12:51 UTC (permalink / raw)
  To: rofikkerneldev
  Cc: devicetree, dri-devel, linux-input, robh, conor+dt,
	dmitry.torokhov

Thank you for your contribution! Sashiko AI review found 2 potential
issues to consider:
- [High] dt-bindings: display: panel: lg,sw43103: handle used instead of
  real name
- [Low] MAINTAINERS: LG SW43103 PANEL DRIVER added out of alphabetical
  order
--

commit 8f35306c3ad9479ca0f821358a8333dce583f152
Author: rofikkerneldev <rofikcoga@gmail.com>

> dt-bindings: display: panel: add LG SW43103
>
> Document the SW43103 FHD+ AMOLED panel used by the LG V60. The panel
> uses DSI command mode with DSC and DCS backlight control.
>
> Signed-off-by: rofikkerneldev <rofikcoga@gmail.com>
> ---
>  .../bindings/display/panel/lg,sw43103.yaml    | 70 +++++++++++++++++++
>  MAINTAINERS                                   |  5 ++
>  2 files changed, 75 insertions(+)
>  create mode 100644 Documentation/devicetree/bindings/display/panel/lg,sw43103.yaml
>
> diff --git a/Documentation/devicetree/bindings/display/panel/lg,sw43103.yaml b/Documentation/devicetree/bindings/display/panel/lg,sw43103.yaml
> new file mode 100644
> index 0000000000000..edd007e2c0d76
> --- /dev/null
> +++ b/Documentation/devicetree/bindings/display/panel/lg,sw43103.yaml
> @@ -0,0 +1,70 @@
> +# SPDX-License-Identifier: (GPL-2.0-only OR BSD-2-Clause)
> +%YAML 1.2
> +---
> +$id: http://devicetree.org/schemas/display/panel/lg,sw43103.yaml#
> +$schema: http://devicetree.org/meta-schemas/core.yaml#
> +
> +title: LG / BOE SW43103 1080x2460 DSI cmd mode panel
> +
> +maintainers:
> +  - rofikkerneldev <rofikcoga@gmail.com>

[Severity: High]
The commit Author, Signed-off-by trailer, DT binding schema maintainers
entry, and MAINTAINERS M: entry all use the handle "rofikkerneldev" instead
of a full real name.  Under the Developer's Certificate of Origin in
Documentation/process/submitting-patches.rst and the MAINTAINERS file
requirements, these must use a real name in the form
"FullName <address@domain>".

[ ... ]

> diff --git a/MAINTAINERS b/MAINTAINERS
> index c88fc0e019296..e7e32c23ba2fd 100644
> --- a/MAINTAINERS
> +++ b/MAINTAINERS
> @@ -15231,6 +15231,11 @@ Q:	http://patchwork.linuxtv.org/project/linux-media/list/
>  T:	git git://linuxtv.org/mkrufky/tuners.git
>  F:	drivers/media/dvb-frontends/lgdt3305.*
>  
> +LG SW43103 PANEL DRIVER
> +M:	rofikkerneldev <rofikcoga@gmail.com>
> +S:	Maintained
> +F:	Documentation/devicetree/bindings/display/panel/lg,sw43103.yaml
> +

[Severity: Low]
This new section is out of alphabetical order in MAINTAINERS, which causes
scripts/parse-maintainers.pl --order to warn.  Since a space sorts before
digits and letters, "LG SW43103 PANEL DRIVER" should be placed after
"LG LAPTOP EXTRAS" and before "LG2160 MEDIA DRIVER".

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261117123946.668557-1-rofikcoga@gmail.com?part=4

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: [PATCH v1 09/11] arm64: dts: qcom: sm8250-lge-timelm: add SW42902 touchscreen
       [not found] ` <20261117123946.668557-10-rofikcoga@gmail.com>
@ 2026-11-17 12:56   ` sashiko-bot
  0 siblings, 0 replies; 11+ messages in thread
From: sashiko-bot @ 2026-11-17 12:56 UTC (permalink / raw)
  To: rofikkerneldev
  Cc: devicetree, dri-devel, linux-input, robh, conor+dt,
	dmitry.torokhov

Thank you for your contribution! Sashiko AI review found 2 potential
issues to consider:
- [High] arm64: dts: qcom: Signed-off-by and Author use handle instead
  of real name
- [Low] arm64: dts: qcom: sm8250-lge-timelm: i2c13 and tlmm nodes out of
  order
--

commit c8220e43d297448ae1467ca0fc69fcaa8385a92a
Author: rofikkerneldev <rofikcoga@gmail.com>

> arm64: dts: qcom: sm8250-lge-timelm: add SW42902 touchscreen
>
> Describe the SW42902 on GENI I2C13 with its interrupt, reset,
> power-enable GPIOs, pinctrl, and native 1080x2460 touch area.
>
> Signed-off-by: rofikkerneldev <rofikcoga@gmail.com>

[Severity: High]
Both the commit Author and the Signed-off-by trailer use the account
handle "rofikkerneldev" instead of a real human name.  Under the
Developer Certificate of Origin in
Documentation/process/submitting-patches.rst, patches must be authored
and signed off using a known real name rather than a pseudonym or
username.

> diff --git a/arch/arm64/boot/dts/qcom/sm8250-lge-timelm.dts b/arch/arm64/boot/dts/qcom/sm8250-lge-timelm.dts
> index fca48472f5dfe..a094aa81ad225 100644
> --- a/arch/arm64/boot/dts/qcom/sm8250-lge-timelm.dts
> +++ b/arch/arm64/boot/dts/qcom/sm8250-lge-timelm.dts
> @@ -596,3 +596,39 @@ &mdss_dsi0_phy {
>  	phy-type = <PHY_TYPE_DPHY>;
>  	status = "okay";
>  };
> +
> +&i2c13 {
> +	clock-frequency = <400000>;
> +	status = "okay";
> +
> +	touchscreen@28 {
> +		compatible = "lg,sw42902";
> +		reg = <0x28>;
> +
> +		pinctrl-names = "default";
> +		pinctrl-0 = <&touch_reset_default &touch_irq_default>;
> +		interrupt-parent = <&tlmm>;
> +		interrupts = <39 IRQ_TYPE_LEVEL_LOW>;
> +		reset-gpios = <&tlmm 38 GPIO_ACTIVE_LOW>;
> +		vcl-gpios = <&tlmm 67 GPIO_ACTIVE_HIGH>;
> +		vdd-gpios = <&tlmm 68 GPIO_ACTIVE_HIGH>;
> +		touchscreen-size-x = <1080>;
> +		touchscreen-size-y = <2460>;
> +	};
> +};
> +
> +&tlmm {
> +	touch_reset_default: touch-reset-default-state {
> +		pins = "gpio38";
> +		function = "gpio";
> +		drive-strength = <6>;
> +		bias-pull-up;
> +	};
> +
> +	touch_irq_default: touch-irq-default-state {
> +		pins = "gpio39";
> +		function = "gpio";
> +		drive-strength = <2>;
> +		bias-pull-up;
> +	};
> +};

[Severity: Low]
This isn't a bug, but &i2c13 and &tlmm are appended at the bottom of the
file after &mdss_dsi0_phy rather than in alphabetical order (&i2c13
after &gpu and &tlmm before &uart12) as expected by
Documentation/devicetree/bindings/dts-coding-style.rst and the rest of
this board file.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261117123946.668557-1-rofikcoga@gmail.com?part=9

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: [PATCH v1 03/11] arm64: dts: qcom: sm8250-lge-timelm: add reserved memory regions
       [not found] ` <20261117123946.668557-4-rofikcoga@gmail.com>
@ 2026-11-17 12:57   ` sashiko-bot
  0 siblings, 0 replies; 11+ messages in thread
From: sashiko-bot @ 2026-11-17 12:57 UTC (permalink / raw)
  To: rofikkerneldev
  Cc: devicetree, dri-devel, linux-input, robh, conor+dt,
	dmitry.torokhov

Thank you for your contribution! Sashiko AI review found 4 potential
issues to consider:
- [High] arm64: dts: qcom: Signed-off-by and Author use handle instead
  of real name
- [Medium] arm64: dts: qcom: sm8250-lge-timelm: ramoops sub-buffers
  leave 0 bytes for dmesg
- [Low] arm64: dts: qcom: sm8250-lge-timelm: changelog omits
  cont_splash_mem and ramoops
- [Low] arm64: dts: qcom: sm8250-lge-timelm: reserved-memory nodes not
  sorted by address
--

commit d0da8e8448ef005db2a8bdee2ccfef831c824926
Author: rofikkerneldev <rofikcoga@gmail.com>

> arm64: dts: qcom: sm8250-lge-timelm: add reserved memory regions
>
> Reserve the XBL/UEFI log and DFPS data regions used by the firmware.
> Keep these regions protected from kernel reuse.
>
> Signed-off-by: rofikkerneldev <rofikcoga@gmail.com>

[Severity: High]
The Author header, Signed-off-by trailer, and file copyright header in
sm8250-lge-timelm.dts use the handle "rofikkerneldev" instead of a real
human name, which does not meet the Developer's Certificate of Origin
requirements in Documentation/process/submitting-patches.rst.

> diff --git a/arch/arm64/boot/dts/qcom/sm8250-lge-timelm.dts b/arch/arm64/boot/dts/qcom/sm8250-lge-timelm.dts
> index 74b673ebbeb9d..ecbff27976fc2 100644
> --- a/arch/arm64/boot/dts/qcom/sm8250-lge-timelm.dts
> +++ b/arch/arm64/boot/dts/qcom/sm8250-lge-timelm.dts
> @@ -103,6 +103,12 @@ ramoops: ramoops@b0000000 {
>  			console-size = <0x40000>;
>  			ftrace-size = <0x40000>;
>  			pmsg-size = <0x200000>;
> +			no-map;
> +		};

[Severity: Medium]
Does ramoops@b0000000 leave any space for record-size = <0x1000>?

The node sets reg = <0x0 0xb0000000 0x0 0x280000>, and console-size
(0x40000), ftrace-size (0x40000), and pmsg-size (0x200000) already sum to
the full 0x280000 bytes. In ramoops_probe(), dump_mem_sz is computed as
cxt->size - cxt->console_size - cxt->ftrace_size - cxt->pmsg_size, which
evaluates to 0 and causes ramoops_init_przs() to skip registering dmesg
crash dump zones (PSTORE_FLAGS_DMESG).

> +
> +		res_xbl_uefi_log_region: memory@80880000 {
> +			reg = <0x0 0x80880000 0x0 0x14000>;
> +			no-map;
>  		};
>  
>  		adsp_mem: memory@8a100000 {

[Severity: Low]
This isn't a bug, but ramoops@b0000000 is placed ahead of memory@80880000
and the other lower-addressed child nodes under /reserved-memory instead
of being sorted in ascending unit-address order.

> @@ -119,6 +125,16 @@ cdsp_secure_heap: memory@8c700000 {
>  			reg = <0x0 0x8c700000 0x0 0x4600000>;
>  			no-map;
>  		};
> +
> +		cont_splash_mem: memory@9c000000 {
> +			reg = <0x0 0x9c000000 0x0 0x1200000>;
> +			no-map;
> +		};
> +
> +		dfps_data_mem: memory@9e300000 {
> +			reg = <0x0 0x9e300000 0x0 0x100000>;
> +			no-map;
> +		};
>  	};
>  
>  	vph_pwr: vph-pwr-regulator {

[Severity: Low]
The commit message mentions only the XBL/UEFI log and DFPS data regions,
omitting the addition of cont_splash_mem: memory@9c000000 and the no-map
property on ramoops@b0000000. In addition, both /chosen/framebuffer@9c000000
and ramoops@b0000000 were introduced in the preceding commit "arm64: dts:
qcom: add LG V60 ThinQ 5G (timelm) support", so folding these reservations
(and referencing &cont_splash_mem from framebuffer@9c000000) into the
initial board commit would avoid leaving the framebuffer unreserved across
commits in the series.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261117123946.668557-1-rofikcoga@gmail.com?part=3

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: [PATCH v1 07/11] Input: touchscreen: add Silicon Works SW42902 support
       [not found] ` <20261117123946.668557-8-rofikcoga@gmail.com>
@ 2026-11-17 13:01   ` sashiko-bot
  0 siblings, 0 replies; 11+ messages in thread
From: sashiko-bot @ 2026-11-17 13:01 UTC (permalink / raw)
  To: rofikkerneldev
  Cc: devicetree, dri-devel, linux-input, robh, conor+dt,
	dmitry.torokhov

Thank you for your contribution! Sashiko AI review found 6 potential
issues to consider:
- [High] Input: sw42902: username handle in Signed-off-by trailer and
  MAINTAINERS
- [Medium] Input: sw42902: IRQ disable imbalance on sw42902_resume()
  error path
- [Low] Input: sw42902: use named initializer in struct i2c_device_id
  table
- [Low] Input: sw42902: follow Input error variable naming and return
  style
- [Low] Input: sw42902: partial MT events queued before palm cancel in
  sw42902_irq()
- [Low] Input: sw42902: redundant dev_err_probe() after
  devm_request_threaded_irq()
--

commit 7d934418330ef369d907f6e1ac22a2b5372028a3
Author: rofikkerneldev <rofikcoga@gmail.com>

> Input: touchscreen: add Silicon Works SW42902 support
>
> Add basic multitouch support for the SW42902 controller. Implement the
> GPIO power and reset sequence and use the firmware already present in
> controller flash; no firmware update is performed.
>
> Signed-off-by: rofikkerneldev <rofikcoga@gmail.com>

[Severity: High]
The Signed-off-by trailer, Author header, and MAINTAINERS entry for this
driver use an account handle (rofikkerneldev) instead of a real human
name, which does not meet the Developer's Certificate of Origin
requirements in Documentation/process/submitting-patches.rst.

> diff --git a/drivers/input/touchscreen/sw42902.c b/drivers/input/touchscreen/sw42902.c
> new file mode 100644
> index 0000000000000..7b39a97053c63
> --- /dev/null
> +++ b/drivers/input/touchscreen/sw42902.c

[ ... ]

> +	for (i = 0; i < count; i++) {
> +		const u8 *contact = ts->report + 36 + 12 * i;
> +		u32 a = get_unaligned_le32(contact);
> +		u32 b = get_unaligned_le32(contact + 4);
> +		u32 c = get_unaligned_le32(contact + 8);
> +		unsigned int id = a & 0x1f;
> +		unsigned int event = (a >> 16) & 3;
> +
> +		if (id == 15) {
> +			sw42902_release_contacts(ts);
> +			return IRQ_HANDLED;
> +		}
> +		if (id >= SW42902_MAX_CONTACTS || (event != 1 && event != 2))
> +			continue;
> +
> +		input_mt_slot(input, id);
> +		input_mt_report_slot_state(input, MT_TOOL_FINGER, true);
> +		touchscreen_report_pos(input, &ts->prop, a >> 18, b & 0x3fff, true);
> +		input_report_abs(input, ABS_MT_PRESSURE, (b >> 14) & 0xff);
> +		input_report_abs(input, ABS_MT_TOUCH_MAJOR, (c >> 4) & 0x3fff);
> +		input_report_abs(input, ABS_MT_TOUCH_MINOR, c >> 18);
> +	}

[Severity: Low]
Can sw42902_irq() queue partial multitouch events before cancelling the
frame on a palm report?

If a report has active finger contacts at earlier indices followed by a
palm entry (id == 15), the loop already queues slot activation and
position events into input->vals[] for those earlier contacts before
calling sw42902_release_contacts(), flushing transient touch updates and
slot releases in the same SYN_REPORT frame.

[ ... ]

> +static int sw42902_start(struct sw42902 *ts)
> +{
> +	u8 version[4];
> +	int ret;
> +
> +	ret = sw42902_read(ts, SW42902_VERSION, version, sizeof(version));
> +	if (ret)
> +		return ret;
> +	if (version[2] != 10 || (version[3] & 0xf) != 4) {
> +		dev_err(&ts->client->dev, "unexpected chip/protocol: %u/%u\n",
> +			version[2], version[3] & 0xf);
> +		return -ENODEV;
> +	}
> +
> +	ret = sw42902_write(ts, SW42902_SERIAL_SPI_EN, 0);
> +	if (ret)
> +		return ret;
> +	ret = sw42902_write(ts, SW42902_DEVICE_CTL, 1);
> +	if (ret)
> +		return ret;
> +	ret = sw42902_write(ts, SW42902_INTERRUPT_CTL, 1);
> +	if (ret)
> +		return ret;
> +	ret = sw42902_write(ts, SW42902_DRIVING_CTL, SW42902_DRIVING_STOP);
> +	if (ret)
> +		return ret;
> +	msleep(30);
> +	return sw42902_write(ts, SW42902_DRIVING_CTL, SW42902_DRIVING_U3);
> +}

[Severity: Low]
This isn't a bug, but in the Input subsystem variables that only hold 0 or
a negative error code should be named error or err rather than ret (here
and in sw42902_irq(), sw42902_probe(), and sw42902_resume()).  Also,
since sw42902_start() has multiple failure points, the final
sw42902_write() call should check error and explicitly return 0 on
success.

> +static int sw42902_probe(struct i2c_client *client)
> +{

[ ... ]

> +	ret = devm_request_threaded_irq(dev, client->irq, NULL, sw42902_irq,
> +					IRQF_ONESHOT | IRQF_NO_AUTOEN,
> +					dev_name(dev), ts);
> +	if (ret)
> +		return dev_err_probe(dev, ret, "interrupt request failed\n");

[Severity: Low]
This isn't a bug, but devm_request_threaded_irq() already calls
dev_err_probe() via devm_request_result() on failure, so wrapping its
return value in dev_err_probe() logs a duplicate error message and
overwrites the deferred probe reason.

[ ... ]

> +static int sw42902_suspend(struct device *dev)
> +{
> +	struct sw42902 *ts = i2c_get_clientdata(to_i2c_client(dev));
> +
> +	disable_irq(ts->client->irq);
> +	sw42902_release_contacts(ts);
> +	sw42902_write(ts, SW42902_DRIVING_CTL, SW42902_DRIVING_STOP);
> +	sw42902_power_off(ts);
> +	return 0;
> +}
> +
> +static int sw42902_resume(struct device *dev)
> +{
> +	struct sw42902 *ts = i2c_get_clientdata(to_i2c_client(dev));
> +	int ret;
> +
> +	sw42902_power_on(ts);
> +	ret = sw42902_start(ts);
> +	if (ret) {
> +		sw42902_power_off(ts);
> +		return ret;
> +	}
> +	enable_irq(ts->client->irq);
> +	return 0;
> +}

[Severity: Medium]
What happens on the next suspend if sw42902_start() fails during
sw42902_resume()?

When sw42902_start() fails, sw42902_resume() powers off the controller and
returns without calling enable_irq(ts->client->irq).  Because the PM core
clears dev->power.is_suspended before invoking the resume callback, the
next system suspend still calls sw42902_suspend(), which writes to the
powered-off chip over I2C and calls disable_irq(ts->client->irq) a second
time, leaving the IRQ disabled even if a later resume succeeds.

[ ... ]

> +static const struct i2c_device_id sw42902_ids[] = {
> +	{ "sw42902" },
> +	{ }
> +};

[Severity: Low]
This isn't a bug, but per I2C subsystem conventions new struct
i2c_device_id entries should use named initializers ({ .name = "sw42902" }).

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261117123946.668557-1-rofikcoga@gmail.com?part=7

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: [PATCH v1 05/11] drm/panel: add LG SW43103 panel driver
       [not found] ` <20261117123946.668557-6-rofikcoga@gmail.com>
@ 2026-11-17 13:04   ` sashiko-bot
  0 siblings, 0 replies; 11+ messages in thread
From: sashiko-bot @ 2026-11-17 13:04 UTC (permalink / raw)
  To: rofikkerneldev
  Cc: devicetree, dri-devel, linux-input, robh, conor+dt,
	dmitry.torokhov

Thank you for your contribution! Sashiko AI review found 4 potential
issues to consider:
- [High] drm/panel: username handle used instead of real name in DCO and
  authorship
- [Medium] drm/panel: sw43103_remove() calls sw43103_unprepare()
  unconditionally
- [Medium] drm/panel: sw43103: unsynchronized DSI transfers in
  sw43103_backlight_update_status()
- [Low] drm/panel: sw43103_enable() hardcodes full brightness before
  backlight_enable()
--

commit c8fd387b550027d7b2fde8a05702145f363e56ba
Author: rofikkerneldev <rofikcoga@gmail.com>

> drm/panel: add LG SW43103 panel driver
>
> Add support for the 1080x2460 SW43103 AMOLED panel. Handle power and
> reset sequencing, DCS backlight control, and the DSC configuration
> required by the panel.
>
> Signed-off-by: rofikkerneldev <rofikcoga@gmail.com>

[Severity: High]
Under Documentation/process/submitting-patches.rst and the Developer
Certificate of Origin, patches must use a real human name rather than a
username handle (rofikkerneldev) in the commit author, Signed-off-by tag,
maintainer entry, and module author macro.

> diff --git a/drivers/gpu/drm/panel/panel-lg-sw43103.c b/drivers/gpu/drm/panel/panel-lg-sw43103.c
> new file mode 100644
> index 0000000000000..caab54e0d1153
> --- /dev/null
> +++ b/drivers/gpu/drm/panel/panel-lg-sw43103.c
[ ... ]
> +static int sw43103_enable(struct drm_panel *panel)
> +{
> +	struct sw43103_panel *sw43103 = to_panel_info(panel);
> +	struct mipi_dsi_multi_context ctx = { .dsi = sw43103->link };
> +
> +	/* Display On */
> +	mipi_dsi_dcs_set_display_on_multi(&ctx);
> +	mipi_dsi_msleep(&ctx, 20);
> +
> +	/* Enable display backlight control and full brightness */
> +	mipi_dsi_dcs_write_seq_multi(&ctx, MIPI_DCS_WRITE_CONTROL_DISPLAY, 0x2c);
> +	mipi_dsi_dcs_write_seq_multi(&ctx, MIPI_DCS_SET_DISPLAY_BRIGHTNESS, 0x07, 0xff);
> +
> +	return ctx.accum_err;
> +}

[Severity: Low]
Should sw43103_enable() leave brightness programming to the backlight
device?

drm_panel_enable() calls backlight_enable(panel->backlight) right after
panel->funcs->enable(panel), which invokes sw43103_backlight_update_status()
to write the control display register and the configured brightness
(defaulting to 1024).  Hardcoding 0x07ff (2047) here duplicates those
transfers and briefly flashes the panel at maximum brightness whenever the
display is enabled at a lower brightness level.

[ ... ]

> +static int sw43103_backlight_update_status(struct backlight_device *bl)
> +{
> +	struct mipi_dsi_device *dsi = bl_get_data(bl);
> +	u16 brightness = backlight_get_brightness(bl);
> +	u8 ctrl = 0x2c;
> +	int ret;
> +
> +	dsi->mode_flags &= ~MIPI_DSI_MODE_LPM;
> +
> +	ret = mipi_dsi_dcs_write(dsi, MIPI_DCS_WRITE_CONTROL_DISPLAY, &ctrl, sizeof(ctrl));
> +	if (ret < 0)
> +		goto out;
> +
> +	ret = mipi_dsi_dcs_set_display_brightness_large(dsi, brightness);
> +
> +out:
> +	dsi->mode_flags |= MIPI_DSI_MODE_LPM;
> +
> +	return ret < 0 ? ret : 0;
> +}

[Severity: Medium]
Can a sysfs brightness write race with panel preparation or issue transfers
while the panel is powered off?

sw43103_backlight_init() registers the backlight device during probe, and
backlight_device_set_brightness() calls sw43103_backlight_update_status()
under only bd->ops_lock and bd->update_lock.  If userspace writes to sysfs
while the panel is unprepared, mipi_dsi_dcs_write() runs while the panel is
in reset or powered down.  If it writes during sw43103_prepare(), it also
races on dsi->mode_flags:

CPU0                              CPU1
sw43103_prepare()
  sw43103_program()
    sw43103_init_sequence()
                                  backlight_device_set_brightness()
                                    sw43103_backlight_update_status()
                                      dsi->mode_flags &= ~...
    mipi_dsi_dcs_write_seq_multi()
      /* sent in high-speed mode instead of low-power mode */

Conversely, restoring the low-power mode flag at the end of
sw43103_backlight_update_status() while sw43103_program() is sending the
picture parameter set forces mipi_dsi_picture_parameter_set_multi() into
low-power mode.

[ ... ]

> +static void sw43103_remove(struct mipi_dsi_device *dsi)
> +{
> +	struct sw43103_panel *ctx = mipi_dsi_get_drvdata(dsi);
> +	int ret;
> +
> +	ret = sw43103_unprepare(&ctx->base);
> +	if (ret < 0)
> +		dev_err(&dsi->dev, "failed to unprepare panel: %d\n", ret);
> +
> +	ret = mipi_dsi_detach(dsi);
> +	if (ret < 0)
> +		dev_err(&dsi->dev, "failed to detach from DSI host: %d\n", ret);
> +
> +	drm_panel_remove(&ctx->base);
> +}

[Severity: Medium]
Does calling sw43103_unprepare() directly in sw43103_remove() underflow the
regulator enable counts?

sw43103_unprepare() unconditionally calls regulator_bulk_disable() without
checking or updating ctx->base.prepared.  If the panel is not prepared when
the driver is unbound, regulator_bulk_disable() triggers an unbalanced
regulator disable warning and returns -EIO; if the panel is prepared,
powering it off before mipi_dsi_detach() and drm_panel_remove() leaves
ctx->base.prepared set, causing modeset teardown to access an unpowered
panel and disable the supplies a second time.  Removing the manual
sw43103_unprepare() call lets the modeset driver tear the panel down
cleanly.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261117123946.668557-1-rofikcoga@gmail.com?part=5

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: [PATCH v1 02/11] arm64: dts: qcom: add LG V60 ThinQ 5G (timelm) support
       [not found] ` <20261117123946.668557-3-rofikcoga@gmail.com>
@ 2026-11-17 13:06   ` sashiko-bot
  0 siblings, 0 replies; 11+ messages in thread
From: sashiko-bot @ 2026-11-17 13:06 UTC (permalink / raw)
  To: rofikkerneldev
  Cc: devicetree, dri-devel, linux-input, robh, conor+dt,
	dmitry.torokhov

Thank you for your contribution! Sashiko AI review found 8 potential
issues to consider:
- [High] arm64: dts: qcom: timelm: enabling gpu without gmu fails
  a6xx_gmu_init()
- [High] arm64: dts: qcom: timelm: invalid 1.05V constraint on PM8009
  ldo4
- [High] arm64: dts: qcom: handle used instead of real name in DCO and
  MAINTAINERS
- [Medium] arm64: dts: qcom: timelm: swapped vdda-phy and vdda-pll on
  usb_1_qmpphy
- [Medium] arm64: dts: qcom: timelm: missing pinctrl states on sdhc_2
  node
- [Medium] arm64: dts: qcom: ramoops sub-buffers exhaust region and
  disable dmesg dumps
- [Low] arm64: dts: qcom: missing DRMS properties on UFS supply
  regulators
- [Low] arm64: dts: qcom: unused dt-bindings headers in
  sm8250-lge-timelm.dts
--

commit e0c6ddf610453315a993fcf369aad59d6f783910
Author: rofikkerneldev <rofikcoga@gmail.com>

arm64: dts: qcom: add LG V60 ThinQ 5G (timelm) support

This commit adds initial device tree support and a MAINTAINERS entry for
the Qualcomm SM8250-based LG V60 ThinQ 5G (timelm) smartphone.

> diff --git a/MAINTAINERS b/MAINTAINERS
> index 222b7fc2c3aab..c88fc0e019296 100644
> --- a/MAINTAINERS
> +++ b/MAINTAINERS
> @@ -15231,6 +15231,12 @@ Q:	http://patchwork.linuxtv.org/project/linux-media/list/
>  T:	git git://linuxtv.org/mkrufky/tuners.git
>  F:	drivers/media/dvb-frontends/lgdt3305.*
>  
> +LG V60 THINQ 5G (TIMELM) SUPPORT
> +M:	rofikkerneldev <rofikcoga@gmail.com>
> +L:	linux-arm-msm@vger.kernel.org
> +S:	Maintained
> +F:	arch/arm64/boot/dts/qcom/sm8250-lge-timelm*
> +
>  LIBATA PATA ARASAN COMPACT FLASH CONTROLLER
>  M:	Viresh Kumar <vireshk@kernel.org>
>  L:	linux-ide@vger.kernel.org

[Severity: High]
The MAINTAINERS entry, commit author, Signed-off-by trailer, and
sm8250-lge-timelm.dts copyright header use the account handle
rofikkerneldev instead of a real name, which does not satisfy the
Developer's Certificate of Origin in
Documentation/process/submitting-patches.rst.

> diff --git a/arch/arm64/boot/dts/qcom/sm8250-lge-timelm.dts b/arch/arm64/boot/dts/qcom/sm8250-lge-timelm.dts
> new file mode 100644
> index 0000000000000..74b673ebbeb9d
> --- /dev/null
> +++ b/arch/arm64/boot/dts/qcom/sm8250-lge-timelm.dts
> @@ -0,0 +1,543 @@
> +// SPDX-License-Identifier: BSD-3-Clause
> +/*
> + * Copyright (c) 2026 rofikkerneldev <rofikcoga@gmail.com>
> + * Device Tree for LG V60 ThinQ 5G (timelm / LM-V600)
> + */
> +
> +/dts-v1/;
> +
> +#include <dt-bindings/gpio/gpio.h>
> +#include <dt-bindings/input/input.h>
> +#include <dt-bindings/leds/common.h>
> +#include <dt-bindings/phy/phy.h>
> +#include <dt-bindings/regulator/qcom,rpmh-regulator.h>
> +#include <dt-bindings/sound/qcom,q6afe.h>
> +#include <dt-bindings/sound/qcom,q6asm.h>
> +#include <dt-bindings/usb/pd.h>

[Severity: Low]
This isn't a bug, but dt-bindings/leds/common.h,
dt-bindings/sound/qcom,q6afe.h, and dt-bindings/sound/qcom,q6asm.h are
unused throughout the series, while dt-bindings/phy/phy.h and
dt-bindings/usb/pd.h are only needed starting in commit "arm64: dts: qcom:
sm8250-lge-timelm: enable SW43103 panel" and commit "arm64: dts: qcom:
sm8250-lge-timelm: enable USB Type-C".

[ ... ]

> +	reserved-memory {
> +		ramoops: ramoops@b0000000 {
> +			compatible = "ramoops";
> +			reg = <0x0 0xb0000000 0x0 0x280000>;
> +			record-size = <0x1000>;
> +			console-size = <0x40000>;
> +			ftrace-size = <0x40000>;
> +			pmsg-size = <0x200000>;
> +		};

[Severity: Medium]
Does ramoops@b0000000 leave any space for record-size?

console-size (0x40000), ftrace-size (0x40000), and pmsg-size (0x200000)
consume the entire 0x280000 region, so ramoops_probe() computes
dump_mem_sz as 0 and disables dmesg oops and panic dumps.  While commit
"arm64: dts: qcom: sm8250-lge-timelm: add reserved memory regions" adds
no-map to this node, the sizes remain unchanged.

[ ... ]

> +	regulators-2 {
> +		compatible = "qcom,pm8009-rpmh-regulators";
> +		qcom,pmic-id = "f";
> +
> +		vdd-s1-supply = <&vph_pwr>;
> +		vdd-s2-supply = <&vreg_bob>;
> +		vdd-l2-supply = <&vreg_s8c_1p2>;
> +		vdd-l5-l6-supply = <&vreg_bob>;
> +		vdd-l7-supply = <&vreg_s4a_1p8>;
> +
> +		vreg_s1f_1p2: smps1 {
> +			regulator-name = "vreg_s1f_1p2";
> +			regulator-min-microvolt = <1200000>;
> +			regulator-max-microvolt = <1200000>;
> +			regulator-initial-mode = <RPMH_REGULATOR_MODE_HPM>;
> +		};
> +
> +		vreg_s2f_0p5: smps2 {
> +			regulator-name = "vreg_s2f_0p5";
> +			regulator-min-microvolt = <512000>;
> +			regulator-max-microvolt = <1100000>;
> +			regulator-initial-mode = <RPMH_REGULATOR_MODE_HPM>;
> +		};
> +
> +		vreg_l2f_1p3: ldo2 {
> +			regulator-name = "vreg_l2f_1p3";
> +			regulator-min-microvolt = <1304000>;
> +			regulator-max-microvolt = <1304000>;
> +			regulator-initial-mode = <RPMH_REGULATOR_MODE_HPM>;
> +		};
> +
> +		vreg_l4f_1p05: ldo4 {
> +			regulator-name = "vreg_l4f_1p05";
> +			regulator-min-microvolt = <1050000>;
> +			regulator-max-microvolt = <1050000>;
> +			regulator-initial-mode = <RPMH_REGULATOR_MODE_HPM>;
> +		};

[Severity: High]
Does vreg_l4f_1p05 cause rpmh_regulator_probe() to fail for the entire
regulators-2 node?

In pm8009_vreg_data, ldo4 uses pmic5_nldo, which has a 320000 uV base and
an 8000 uV step.  Because 1050000 uV lies between the 1048000 uV and
1056000 uV steps, machine_constraints_voltage() returns -EINVAL and aborts
registration of all PM8009 regulators in regulators-2.  In addition,
regulators-2 omits vdd-l4-supply.

[ ... ]

> +&gpu {
> +	status = "okay";
> +
> +	zap-shader {
> +		memory-region = <&gpu_mem>;
> +		firmware-name = "qcom/sm8250/a650_zap.mbn";
> +	};
> +};

[Severity: High]
Does enabling &gpu without enabling &gmu cause GPU initialization to fail?

In sm8250.dtsi, gmu has status = "disabled" by default, so no platform
device is created for it and a6xx_gpu_init() -> a6xx_gmu_init() returns
-ENODEV.  Also, memory-region = <&gpu_mem> is already set on
gpu_zap_shader in sm8250.dtsi.

[ ... ]

> +&ufs_mem_hc {
> +	status = "okay";
> +
> +	vcc-supply = <&vreg_l17a_3p0>;
> +	vcc-max-microamp = <800000>;
> +	vccq-supply = <&vreg_l6a_1p2>;
> +	vccq-max-microamp = <800000>;
> +	vccq2-supply = <&vreg_s4a_1p8>;
> +	vccq2-max-microamp = <800000>;
> +};

[Severity: Low]
Should vreg_l17a_3p0, vreg_l6a_1p2, and vreg_s4a_1p8 set
regulator-allow-set-load and regulator-allowed-modes?

Without those properties on the regulator nodes, the max-microamp limits
here have no effect when ufshcd calls regulator_set_load(), leaving the
three UFS rails in RPMH_REGULATOR_MODE_HPM during UFS low-power states.

[ ... ]

> +&sdhc_2 {
> +	status = "okay";
> +
> +	cd-gpios = <&tlmm 77 GPIO_ACTIVE_LOW>;
> +	vmmc-supply = <&vreg_l9c_2p9>;
> +	vqmmc-supply = <&vreg_l6c_2p9>;
> +	bus-width = <4>;
> +	no-sdio;
> +	no-mmc;
> +};

[Severity: Medium]
Should &sdhc_2 configure default and sleep pinctrl states?

sdhc_2 in sm8250.dtsi does not define pinctrl properties, so the SDCC2
pads (sdc2_clk, sdc2_cmd, sdc2_data) and card-detect pin (gpio77) are left
unconfigured by the kernel and will not transition to sdc2_sleep_state
across suspend and resume.

[ ... ]

> +&usb_1_qmpphy {
> +	status = "okay";
> +
> +	vdda-phy-supply = <&vreg_l9a_1p2>;
> +	vdda-pll-supply = <&vreg_l18a_0p9>;
> +};

[Severity: Medium]
Are vdda-phy-supply and vdda-pll-supply swapped on &usb_1_qmpphy?

On SM8250, the QMP PHY core rail (vdda-phy-supply) is powered by the 0.9V
vreg_l18a_0p9 regulator and the PLL rail (vdda-pll-supply) is powered by
the 1.2V vreg_l9a_1p2 regulator; swapping them applies the driver's
21800 uA and 36000 uA load votes to the wrong LDOs.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261117123946.668557-1-rofikcoga@gmail.com?part=2

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: [PATCH v1 11/11] arm64: dts: qcom: sm8250-lge-timelm: add QCA6390 Bluetooth
       [not found] ` <20261117123946.668557-12-rofikcoga@gmail.com>
@ 2026-11-17 13:09   ` sashiko-bot
  0 siblings, 0 replies; 11+ messages in thread
From: sashiko-bot @ 2026-11-17 13:09 UTC (permalink / raw)
  To: rofikkerneldev
  Cc: devicetree, dri-devel, linux-input, robh, conor+dt,
	dmitry.torokhov

Thank you for your contribution! Sashiko AI review found 2 potential
issues to consider:
- [High] arm64: dts: qcom: non-real-name handle in Signed-off-by trailer
- [High] arm64: dts: qcom: sm8250-lge-timelm: under-volted QCA6390 PMU
  supplies
--

commit c689e882729fccea2deb0c720da66bf756537580
Author: rofikkerneldev <rofikcoga@gmail.com>

> arm64: dts: qcom: sm8250-lge-timelm: add QCA6390 Bluetooth
>
> Describe the QCA6390 power sequencer and its PMU regulators. Enable the
> Bluetooth HCI controller on GENI UART6 with the supplies provided by the
> PMU.
>
> Signed-off-by: rofikkerneldev <rofikcoga@gmail.com>

[Severity: High]
The Author header and Signed-off-by trailer use an account handle
(rofikkerneldev) instead of a known real human name, which does not meet
the Developer's Certificate of Origin requirements in
Documentation/process/submitting-patches.rst.

> diff --git a/arch/arm64/boot/dts/qcom/sm8250-lge-timelm.dts b/arch/arm64/boot/dts/qcom/sm8250-lge-timelm.dts
> index 6b57c658cf918..38816268fcf53 100644
> --- a/arch/arm64/boot/dts/qcom/sm8250-lge-timelm.dts
> +++ b/arch/arm64/boot/dts/qcom/sm8250-lge-timelm.dts
> @@ -95,6 +95,67 @@ key-assistant {
>  		};
>  	};
>  
> +	qca6390-pmu {
> +		compatible = "qcom,qca6390-pmu";
> +
> +		pinctrl-names = "default";
> +		pinctrl-0 = <&bt_en_state>, <&wlan_en_state>;
> +
> +		vddaon-supply = <&vreg_s6a_0p6>;
> +		vddpmu-supply = <&vreg_s6a_0p6>;
> +		vddrfa0p95-supply = <&vreg_s6a_0p6>;
> +		vddrfa1p3-supply = <&vreg_s8c_1p2>;
> +		vddrfa1p9-supply = <&vreg_s5a_1p9>;
> +		vddpcie1p3-supply = <&vreg_s8c_1p2>;
> +		vddpcie1p9-supply = <&vreg_s5a_1p9>;
> +		vddio-supply = <&vreg_s4a_1p8>;

[Severity: High]
Does wiring vddaon-supply, vddpmu-supply and vddrfa0p95-supply to
vreg_s6a_0p6, and vddrfa1p3-supply and vddpcie1p3-supply to vreg_s8c_1p2,
under-volt the QCA6390 PMU?

pwrseq-qcom-wcn only enables these regulators without calling
regulator_set_voltage(), so the RPMh regulator driver initializes them to
their regulator-min-microvolt constraints when enabled.  In this board DT,
vreg_s6a_0p6 has regulator-min-microvolt = <600000> and vreg_s8c_1p2 has
regulator-min-microvolt = <1200000>, with no other consumer voting a higher
voltage.  That drives the 0.95V PMU inputs at 0.6V (below the PMU's own
0.8V and 0.85V LDO outputs) and the 1.35V inputs at 1.2V.  Should the
minimum voltages on smps6 and smps8 be raised to ~950000 uV and ~1350000 uV
like on other SM8250 QCA6390 boards?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261117123946.668557-1-rofikcoga@gmail.com?part=11

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: [PATCH v1 10/11] arm64: dts: qcom: sm8250-lge-timelm: enable USB Type-C
       [not found] ` <20261117123946.668557-11-rofikcoga@gmail.com>
@ 2026-11-17 13:12   ` sashiko-bot
  0 siblings, 0 replies; 11+ messages in thread
From: sashiko-bot @ 2026-11-17 13:12 UTC (permalink / raw)
  To: rofikkerneldev
  Cc: devicetree, dri-devel, linux-input, robh, conor+dt,
	dmitry.torokhov

Thank you for your contribution! Sashiko AI review found 4 potential
issues to consider:
- [High] arm64: dts: qcom: username handle used in Signed-off-by trailer
- [High] arm64: dts: qcom: sm8250-lge-timelm: missing OF graph ports in
  usb-c-connector
- [High] arm64: dts: qcom: sm8250-lge-timelm: PDO_BATT in sink-pdos
  breaks 9V PD current
- [Medium] arm64: dts: qcom: sm8250-lge-timelm: swapped supplies in
  usb_1_qmpphy
--

commit 81ec31057ff34acf5ea3063aab5a376e5f194531
Author: rofikkerneldev <rofikcoga@gmail.com>

> arm64: dts: qcom: sm8250-lge-timelm: enable USB Type-C
>
> Enable the PM8150B Type-C controller as a sink-only port. Keep the
> primary DWC3 controller in peripheral mode for now; host mode and VBUS
> sourcing can be enabled after OTG power delivery is validated.
>
> Signed-off-by: rofikkerneldev <rofikcoga@gmail.com>

[Severity: High]
Under Documentation/process/submitting-patches.rst, the Developer's
Certificate of Origin requires a real human name rather than an account
handle in the Author header and Signed-off-by trailer.

> diff --git a/arch/arm64/boot/dts/qcom/sm8250-lge-timelm.dts b/arch/arm64/boot/dts/qcom/sm8250-lge-timelm.dts
> index a094aa81ad225..6b57c658cf918 100644
> --- a/arch/arm64/boot/dts/qcom/sm8250-lge-timelm.dts
> +++ b/arch/arm64/boot/dts/qcom/sm8250-lge-timelm.dts
> @@ -535,6 +535,24 @@ &sdhc_2 {
>  	no-mmc;
>  };
>  
> +&pm8150b_typec {
> +	status = "okay";
> +
> +	vdd-pdphy-supply = <&vreg_l2a_3p1>;
> +
> +	connector {
> +		compatible = "usb-c-connector";
> +
> +		power-role = "sink";
> +		data-role = "device";
> +		self-powered;
> +
> +		op-sink-microwatt = <15000000>;
> +		sink-pdos = <PDO_FIXED(5000, 3000, PDO_FIXED_USB_COMM)
> +			     PDO_BATT(5000, 9000, 15000)>;

[Severity: High]
Should the second entry in sink-pdos use PDO_VAR() or PDO_FIXED() instead
of PDO_BATT()?

PDO_BATT(5000, 9000, 15000) encodes 15000 mW in bits [9:0] in 250 mW units
(value 60).  When connected to a standard USB-PD adapter advertising a 9V
PDO_TYPE_FIXED source PDO, tcpm_pd_select_pdo() matches that 9V source PDO
against this sink PDO by voltage range, and tcpm_pd_build_request() then
checks the source PDO type:

drivers/usb/typec/tcpm/tcpm.c:tcpm_pd_build_request() {
	...
	type = pdo_type(pdo);
	...
	if (type == PDO_TYPE_BATT) {
		mw = min_power(pdo, matching_snk_pdo);
		ma = 1000 * mw / mv;
	} else {
		ma = min_current(pdo, matching_snk_pdo);
		mw = ma * mv / 1000;
	}
}

Because the source PDO is PDO_TYPE_FIXED, min_current() calls
pdo_max_current(matching_snk_pdo), which interprets bits [9:0] in 10 mA
units.  That clamps the requested current at 9V to 600 mA (5.4 W instead
of 15 W) and sets RDO_CAP_MISMATCH.

> +	};
> +};

[Severity: High]
Is the connector node missing its ports subnode linking port@0 to
usb_1_dwc3_hs_out and port@1 to usb_1_qmpphy_out (along with the matching
remote-endpoint properties on those nodes)?

In qcom_pmic_typec_port_set_polarity(), the PM8150B TCPC driver does not
program SuperSpeed lane polarity itself and leaves orientation switching to
usb_1_qmpphy, which registers a Type-C orientation switch and mux in
qmp_combo_probe().  Without the OF graph endpoints under connector,
fwnode_typec_switch_get(), fwnode_typec_mux_get() and
fwnode_usb_role_switch_get() find no matches and return NULL.  When a
Type-C cable is plugged in with reversed (CC2) orientation,
typec_set_orientation() is a no-op and usb_1_qmpphy stays stuck in
TYPEC_ORIENTATION_NORMAL, breaking SuperSpeed USB.  Omitting port@0 also
violates Documentation/devicetree/bindings/connector/usb-connector.yaml.

> +
>  &usb_1 {
>  	status = "okay";
>  };

[Severity: Medium]
Just below &usb_1, are vdda-phy-supply and vdda-pll-supply swapped in
&usb_1_qmpphy (introduced earlier in this series in commit "arm64: dts:
qcom: add LG V60 ThinQ 5G (timelm) support")?

&usb_1_qmpphy currently sets:

	vdda-phy-supply = <&vreg_l9a_1p2>;
	vdda-pll-supply = <&vreg_l18a_0p9>;

On SM8250 QMP PHYs, vdda-phy-supply is the 0.9V core rail (vreg_l18a_0p9)
and vdda-pll-supply is the 1.2V PLL rail (vreg_l9a_1p2).  Because
qmp_phy_vreg_l in drivers/phy/qualcomm/phy-qcom-qmp-combo.c requests
21800 uA on vdda-phy and 36000 uA on vdda-pll, swapping the supplies
programs inverted load votes on the two LDOs when &usb_1 brings the PHY up.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261117123946.668557-1-rofikcoga@gmail.com?part=10

^ permalink raw reply	[flat|nested] 11+ messages in thread

* Re: [PATCH v1 08/11] arm64: dts: qcom: sm8250-lge-timelm: enable SW43103 panel
       [not found] ` <20261117123946.668557-9-rofikcoga@gmail.com>
@ 2026-11-17 13:12   ` sashiko-bot
  0 siblings, 0 replies; 11+ messages in thread
From: sashiko-bot @ 2026-11-17 13:12 UTC (permalink / raw)
  To: rofikkerneldev
  Cc: devicetree, dri-devel, linux-input, robh, conor+dt,
	dmitry.torokhov

Thank you for your contribution! Sashiko AI review found 5 potential
issues to consider:
- [High] arm64: dts: qcom: Signed-off-by and Author use handle
  rofikkerneldev
- [High] arm64: dts: qcom: enabling mdss with gmu disabled breaks
  a6xx_gmu_init()
- [Medium] arm64: dts: qcom: missing enable-gpios in sm8250-lge-timelm
  panel@0 node
- [Medium] drm/panel: sw43103_remove() calls sw43103_unprepare()
  unconditionally
- [Low] arm64: dts: qcom: redundant dispcc status and unsorted mdss
  nodes
--

commit 3038d85d6373aad3b6104c371b0b0d8abde1a320
Author: rofikkerneldev <rofikcoga@gmail.com>

> arm64: dts: qcom: sm8250-lge-timelm: enable SW43103 panel
> 
> Enable DISPCC, MDSS, DSI0, and its PHY. Attach the SW43103 panel with
> its power rails, reset GPIO, and four-lane DSI link.
> 
> Signed-off-by: rofikkerneldev <rofikcoga@gmail.com>

[Severity: High]
The Author header and Signed-off-by trailer use an account handle
(rofikkerneldev) instead of a real human name, which does not meet the
Developer Certificate of Origin requirements in
Documentation/process/submitting-patches.rst.

> diff --git a/arch/arm64/boot/dts/qcom/sm8250-lge-timelm.dts b/arch/arm64/boot/dts/qcom/sm8250-lge-timelm.dts
> index ecbff27976fc2..fca48472f5dfe 100644
> --- a/arch/arm64/boot/dts/qcom/sm8250-lge-timelm.dts
> +++ b/arch/arm64/boot/dts/qcom/sm8250-lge-timelm.dts
> @@ -557,3 +557,42 @@ &usb_1_qmpphy {
>  	vdda-phy-supply = <&vreg_l9a_1p2>;
>  	vdda-pll-supply = <&vreg_l18a_0p9>;
>  };
> +
> +&dispcc {
> +	status = "okay";
> +};

[Severity: Low]
This isn't a bug, but dispcc in arch/arm64/boot/dts/qcom/sm8250.dtsi is
already enabled by default, so this status override is redundant.  Also,
board-level label overrides in Qualcomm device trees are sorted
alphabetically, so the &mdss* nodes below belong between &gpu and
&pm8150_gpios rather than after &usb_1_qmpphy.

> +
> +&mdss {
> +	status = "okay";
> +};

[Severity: High]
Does enabling &mdss fail to bind the DRM device because &gmu is still
disabled?

In sm8250-lge-timelm.dts, &gpu is enabled, while gmu in sm8250.dtsi
defaults to status = "disabled" and is never enabled by the board DTS.
When dpu_dev_probe() calls msm_drv_probe(), add_gpu_components() includes
the GPU in the DRM component master match list alongside mdss_dsi0.
During component_bind_all() in msm_drm_init(), adreno_bind() ->
a6xx_gpu_init() -> a6xx_gmu_init() looks up the GMU platform device via
of_find_device_by_node(), which returns NULL because the gmu node is
disabled.  a6xx_gmu_init() returns -ENODEV, aborting component_bind_all()
and preventing both the display controller and the SW43103 panel from
initializing unless &gmu is also enabled.

> +
> +&mdss_dsi0 {
> +	vdda-supply = <&vreg_l9a_1p2>;
> +	status = "okay";
> +
> +	panel@0 {
> +		compatible = "lg,sw43103";

[Severity: Medium]
Does instantiating "lg,sw43103" expose an unbalanced regulator disable in
sw43103_remove() on driver unbind or module unload?

drivers/gpu/drm/panel/panel-lg-sw43103.c:sw43103_remove() {
	ret = sw43103_unprepare(&ctx->base);
	if (ret < 0)
		dev_err(&dsi->dev, "failed to unprepare panel: %d\n", ret);

	ret = mipi_dsi_detach(dsi);
	if (ret < 0)
		dev_err(&dsi->dev, "failed to detach from DSI host: %d\n", ret);

	drm_panel_remove(&ctx->base);
}

Calling the raw sw43103_unprepare() callback directly bypasses
drm_panel_unprepare() and ctx->base.prepared tracking, unconditionally
calling regulator_bulk_disable() on vddi and vpnl.  If the panel is
already unprepared, this underflows the regulator enable count and
triggers a WARN in _regulator_disable().  If the panel is prepared and
enabled, sw43103_unprepare() asserts reset and cuts power while
ctx->base.prepared remains true, before mipi_dsi_detach() triggers DRM
atomic shutdown to send MIPI DSI commands and call drm_panel_unprepare()
a second time.

> +		reg = <0>;
> +
> +		vddi-supply = <&vreg_l14a_1p8>;
> +		vpnl-supply = <&vreg_l11c_3p0>;
> +		reset-gpios = <&tlmm 75 GPIO_ACTIVE_LOW>;

[Severity: Medium]
Should this panel node also specify enable-gpios = <&pm8150l_gpios 4
GPIO_ACTIVE_HIGH>?

The binding example for the timelm panel in
Documentation/devicetree/bindings/display/panel/lg,sw43103.yaml specifies
that GPIO alongside the same vddi-supply, vpnl-supply, and reset-gpios.
Without enable-gpios here, devm_gpiod_get_optional(dev, "enable",
GPIOD_OUT_LOW) in panel-lg-sw43103 returns NULL, and sw43103_prepare() and
sw43103_unprepare() never toggle the panel's PM8150L GPIO 4 enable line
across display power cycles.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261117123946.668557-1-rofikcoga@gmail.com?part=8

^ permalink raw reply	[flat|nested] 11+ messages in thread

end of thread, other threads:[~2026-11-17 13:12 UTC | newest]

Thread overview: 11+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
     [not found] <20261117123946.668557-1-rofikcoga@gmail.com>
     [not found] ` <20261117123946.668557-2-rofikcoga@gmail.com>
2026-11-17 12:47   ` [PATCH v1 01/11] dt-bindings: arm: qcom: add LG V60 ThinQ 5G (timelm) sashiko-bot
     [not found] ` <20261117123946.668557-7-rofikcoga@gmail.com>
2026-11-17 12:51   ` [PATCH v1 06/11] dt-bindings: input: touchscreen: add Silicon Works SW42902 sashiko-bot
     [not found] ` <20261117123946.668557-5-rofikcoga@gmail.com>
2026-11-17 12:51   ` [PATCH v1 04/11] dt-bindings: display: panel: add LG SW43103 sashiko-bot
     [not found] ` <20261117123946.668557-10-rofikcoga@gmail.com>
2026-11-17 12:56   ` [PATCH v1 09/11] arm64: dts: qcom: sm8250-lge-timelm: add SW42902 touchscreen sashiko-bot
     [not found] ` <20261117123946.668557-4-rofikcoga@gmail.com>
2026-11-17 12:57   ` [PATCH v1 03/11] arm64: dts: qcom: sm8250-lge-timelm: add reserved memory regions sashiko-bot
     [not found] ` <20261117123946.668557-8-rofikcoga@gmail.com>
2026-11-17 13:01   ` [PATCH v1 07/11] Input: touchscreen: add Silicon Works SW42902 support sashiko-bot
     [not found] ` <20261117123946.668557-6-rofikcoga@gmail.com>
2026-11-17 13:04   ` [PATCH v1 05/11] drm/panel: add LG SW43103 panel driver sashiko-bot
     [not found] ` <20261117123946.668557-3-rofikcoga@gmail.com>
2026-11-17 13:06   ` [PATCH v1 02/11] arm64: dts: qcom: add LG V60 ThinQ 5G (timelm) support sashiko-bot
     [not found] ` <20261117123946.668557-12-rofikcoga@gmail.com>
2026-11-17 13:09   ` [PATCH v1 11/11] arm64: dts: qcom: sm8250-lge-timelm: add QCA6390 Bluetooth sashiko-bot
     [not found] ` <20261117123946.668557-11-rofikcoga@gmail.com>
2026-11-17 13:12   ` [PATCH v1 10/11] arm64: dts: qcom: sm8250-lge-timelm: enable USB Type-C sashiko-bot
     [not found] ` <20261117123946.668557-9-rofikcoga@gmail.com>
2026-11-17 13:12   ` [PATCH v1 08/11] arm64: dts: qcom: sm8250-lge-timelm: enable SW43103 panel sashiko-bot

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).