On 07/29/2024, Liu Ying wrote:
> Hi Dmitry,
>
> On 07/27/2024, Dmitry Baryshkov wrote:
>> On Fri, Jul 26, 2024 at 02:50:12PM GMT, Liu Ying wrote:
>>> MCIMX-LVDS1[1] display module integrates a HannStar HSD100PXN1 LVDS
>>> display panel and a touch IC. Add an overlay to support the LVDS
>>> panel on i.MX53 QSB / QSRB platforms.
>>>
>>> [1] https://www.nxp.com/part/MCIMX-LVDS1
>>>
>>> Signed-off-by: Liu Ying <victor....@nxp.com>
>>> ---
>>> I mark RFC in patch subject prefix because if the DT overlay is used, both
>>> ldb
>>> and panel devices end up as devices deferred. However, if the DT overlay is
>>> not used and the devices are defined in imx53-qsb-common.dtsi, then they
>>> can be
>>> probed ok.
>>>
>>> With a dev_err_probe() added to imx_ldb_probe() in imx-ldb.c,
>>> devices_deferred
>>> indicates 53fa8008.ldb and panel-lvds kind of depend on each other.
>>>
>>> root@imx53qsb:~# cat /sys/kernel/debug/devices_deferred
>>> 53fa8008.ldb imx-ldb: failed to find panel or bridge for channel0
>>> panel-lvds platform: wait for supplier
>>> /soc/bus@50000000/ldb@53fa8008/lvds-channel@0
>>>
>>> It looks like the issue is related to fw_devlink, because if
>>> "fw_devlink=off"
>>> is added to kernel bootup command line, then the issue doesn't happen.
>>
>> Could you please fdtdump /sys/firmware/fdt (or just generated DTB files)
>> in both cases and compare the dumps for sensible differences?
>
> I fdtdump imx53-qsrb-mcimx-lvds1.dtb and imx53-qsrb.dtb.
>
> I see three sensible differences.
> 1) panel-lvds node position.
> For imx53-qsrb-mcimx-lvds1.dtb, it comes very early and is next to
> 'compatible = "fsl,imx53-qsrb", "fsl,imx53";'.
> For imx53-qsrb.dtb, it comes later and is next to panel node in '/' node.
It turns out only 1) panel-lvds node position matters.
I can reproduce the issue with imx53-qsrb.dtb(no DT overlay) if I put
the panel-lvds node before the soc node. If the panel-lvds node is
after the soc node, then the issue doesn't happen with imx53-qsrb.dtb.
The ldb node(LVDS display bridge) and IPU(display controller) node are
in the soc node. Maybe, the order of the ldb node and the panel-lvds
node in DT blob matters(be my guess).
>
> 2) properties order in panel-lvds node.
> For imx53-qsrb-mcimx-lvds1.dtb, it shows
> panel-lvds {
>
> power-supply = <0x0000001c>;
>
> backlight = <0x00000030>;
>
> compatible = "hannstar,hsd100pxn1";
>
> port {
>
> endpoint {
>
> phandle = <0x0000007d>;
>
> remote-endpoint = <0x0000007c>;
>
> };
>
> };
>
> };
> For imx53-qsrb.dtb, it shows
> panel-lvds {
>
> compatible = "hannstar,hsd100pxn1";
>
> backlight = <0x00000031>;
>
> power-supply = <0x0000001d>;
>
> port {
>
> endpoint {
>
> remote-endpoint = <0x00000033>;
>
> phandle = <0x00000017>;
>
> };
>
> };
>
> };
>
> 3) No 'lvds0_out' and 'panel_lvds_in' in __symbols__ node for
> imx53-qsrb-mcimx-lvds1.dtb, but for imx53-qsrb.dtb they are in it.
> lvds0_out = "/soc/bus@50000000/ldb@53fa8008/lvds-channel@0/port@2/endpoint";
> panel_lvds_in = "/panel-lvds/port/endpoint";
>
> BTW, reverting Saravana's commits
> 7cb50f6c9fba ("of: property: fw_devlink: Fix stupid bug in remote-endpoint
> parsing")
> and/or
> 7fddac12c382 ("driver core: Fix device_link_flag_is_sync_state_only()")
> avoids the issue from happening.
>
>>
>>>
>>> Saravana, DT folks, any ideas?
>>>
>>> Thanks.
>>>
>>> arch/arm/boot/dts/nxp/imx/Makefile | 4 ++
>>> .../boot/dts/nxp/imx/imx53-qsb-common.dtsi | 4 +-
>>> .../dts/nxp/imx/imx53-qsb-mcimx-lvds1.dtso | 43 +++++++++++++++++++
>>> 3 files changed, 49 insertions(+), 2 deletions(-)
>>> create mode 100644 arch/arm/boot/dts/nxp/imx/imx53-qsb-mcimx-lvds1.dtso
>>>
>>> diff --git a/arch/arm/boot/dts/nxp/imx/Makefile
>>> b/arch/arm/boot/dts/nxp/imx/Makefile
>>> index 92e291603ea1..7116889e1515 100644
>>> --- a/arch/arm/boot/dts/nxp/imx/Makefile
>>> +++ b/arch/arm/boot/dts/nxp/imx/Makefile
>>> @@ -46,8 +46,10 @@ dtb-$(CONFIG_SOC_IMX53) += \
>>> imx53-ppd.dtb \
>>> imx53-qsb.dtb \
>>> imx53-qsb-hdmi.dtb \
>>> + imx53-qsb-mcimx-lvds1.dtb \
>>> imx53-qsrb.dtb \
>>> imx53-qsrb-hdmi.dtb \
>>> + imx53-qsrb-mcimx-lvds1.dtb \
>>> imx53-sk-imx53.dtb \
>>> imx53-sk-imx53-atm0700d4-lvds.dtb \
>>> imx53-sk-imx53-atm0700d4-rgb.dtb \
>>> @@ -57,7 +59,9 @@ dtb-$(CONFIG_SOC_IMX53) += \
>>> imx53-usbarmory.dtb \
>>> imx53-voipac-bsb.dtb
>>> imx53-qsb-hdmi-dtbs := imx53-qsb.dtb imx53-qsb-hdmi.dtbo
>>> +imx53-qsb-mcimx-lvds1-dtbs := imx53-qsb.dtb imx53-qsb-mcimx-lvds1.dtbo
>>> imx53-qsrb-hdmi-dtbs := imx53-qsrb.dtb imx53-qsb-hdmi.dtbo
>>> +imx53-qsrb-mcimx-lvds1-dtbs := imx53-qsrb.dtb imx53-qsb-mcimx-lvds1.dtbo
>>> dtb-$(CONFIG_SOC_IMX6Q) += \
>>> imx6dl-alti6p.dtb \
>>> imx6dl-apf6dev.dtb \
>>> diff --git a/arch/arm/boot/dts/nxp/imx/imx53-qsb-common.dtsi
>>> b/arch/arm/boot/dts/nxp/imx/imx53-qsb-common.dtsi
>>> index 05d7a462ea25..430792a91ccf 100644
>>> --- a/arch/arm/boot/dts/nxp/imx/imx53-qsb-common.dtsi
>>> +++ b/arch/arm/boot/dts/nxp/imx/imx53-qsb-common.dtsi
>>> @@ -16,7 +16,7 @@ memory@70000000 {
>>> <0xb0000000 0x20000000>;
>>> };
>>>
>>> - backlight_parallel: backlight-parallel {
>>> + backlight: backlight {
>>
>> Nit: this seems unrelated to the LVDS support
>
> Do you suggest to do this in a separate patch?
> If yes, is it worth adding a Fixes tag?
>
>>
>>> compatible = "pwm-backlight";
>>> pwms = <&pwm2 0 5000000 0>;
>>> brightness-levels = <0 4 8 16 32 64 128 255>;
>>> @@ -89,7 +89,7 @@ panel_dpi: panel {
>>> compatible = "sii,43wvf1g";
>>> pinctrl-names = "default";
>>> pinctrl-0 = <&pinctrl_display_power>;
>>> - backlight = <&backlight_parallel>;
>>> + backlight = <&backlight>;
>>> enable-gpios = <&gpio3 24 GPIO_ACTIVE_HIGH>;
>>>
>>> port {
>>> diff --git a/arch/arm/boot/dts/nxp/imx/imx53-qsb-mcimx-lvds1.dtso
>>> b/arch/arm/boot/dts/nxp/imx/imx53-qsb-mcimx-lvds1.dtso
>>> new file mode 100644
>>> index 000000000000..27f6bedf3d39
>>> --- /dev/null
>>> +++ b/arch/arm/boot/dts/nxp/imx/imx53-qsb-mcimx-lvds1.dtso
>>> @@ -0,0 +1,43 @@
>>> +// SPDX-License-Identifier: (GPL-2.0+ OR MIT)
>>> +/*
>>> + * Copyright 2024 NXP
>>> + */
>>> +
>>> +/dts-v1/;
>>> +/plugin/;
>>> +
>>> +&{/} {
>>> + panel-lvds {
>>
>> Nit: Just 'panel' should be enough.
>
> Nope.
>
> 'panel-lvds' is needed to differentiate it from 'panel' in
> imx53-qsb-common.dtsi which is a DPI panel.
>
> Using 'panel-lvds', procfs lists exactly the properties needed.
> root@imx53qsb:~# ls /proc/device-tree/panel-lvds/
> backlight compatible name port power-supply
>
> Using 'panel', more are listed.
> root@imx53qsb:~# ls /proc/device-tree/panel/
> backlight compatible enable-gpios name phandle
> pinctrl-0 pinctrl-names port power-supply
>
>>
>>> + compatible = "hannstar,hsd100pxn1";
>>> + backlight = <&backlight>;
>>> + power-supply = <®_3p2v>;
>>> +
>>> + port {
>>> + panel_lvds_in: endpoint {
>>> + remote-endpoint = <&lvds0_out>;
>>> + };
>>> + };
>>> + };
>>> +};
>>> +
>>> +&ldb {
>>> + #address-cells = <1>;
>>> + #size-cells = <0>;
>>> + status = "okay";
>>> +
>>> + lvds-channel@0 {
>>> + #address-cells = <1>;
>>> + #size-cells = <0>;
>>> + fsl,data-mapping = "spwg";
>>> + fsl,data-width = <18>;
>>> + status = "okay";
>>> +
>>> + port@2 {
>>> + reg = <2>;
>>> +
>>> + lvds0_out: endpoint {
>>> + remote-endpoint = <&panel_lvds_in>;
>>> + };
>>> + };
>>> + };
>>> +};
>>> --
>>> 2.34.1
>>>
>>
>
--
Regards,
Liu Ying
