Skip to content

tdo_panel: bcm27xx-gpu: Add dsi panel driver;#7318

Open
flash-fea wants to merge 3 commits intoraspberrypi:rpi-6.18.yfrom
flash-fea:rpi-6.18.y
Open

tdo_panel: bcm27xx-gpu: Add dsi panel driver;#7318
flash-fea wants to merge 3 commits intoraspberrypi:rpi-6.18.yfrom
flash-fea:rpi-6.18.y

Conversation

@flash-fea
Copy link
Copy Markdown

- Add vc4-kms-dsi-panel-overlay.dts for tdo panel compatiable.
- Simplify add panel-tdo-dsi-v1.c as module to driver tdo panel.
- __overrides__ parameter can be add in config.txt for customers.

@flash-fea flash-fea force-pushed the rpi-6.18.y branch 2 times, most recently from 7a2e2f8 to 840f6f9 Compare April 16, 2026 06:19
@flash-fea
Copy link
Copy Markdown
Author

whether the addition was successful?

@pelwell
Copy link
Copy Markdown
Contributor

pelwell commented Apr 16, 2026

The partitioning of the code looks good, and checkpatch is happy.

This isn't my field of expertise so there is limited input I can give regarding the driver itself - perhaps @6by9 has some comments - but there are a few things I have spotted:

  1. The authorship of the code (and therefore the permission to publish it) is a bit vague. The comment at the top of the driver file says Copyright (C) 2024 Waveshare International Limited, but the footer says MODULE_AUTHOR("LK <support@tdo.com>"); and MODULE_DESCRIPTION("TDO DSI panel driver");, none of which seems to match your GitHub user account.
  2. I see a number of dev_err() used for what look like debug output. I do the same myself, but it shouldn't still be around in a production driver. If the information is useful to users, demote it to INFO level or lower, otherwise go for dev_dbg or deletion.

@6by9
Copy link
Copy Markdown
Contributor

6by9 commented Apr 16, 2026

Sorry I had all my review comments pending on the 6.12 version. Now posted.
I suspect that they all apply here too.

Largely the same as pelwell has commented with regard pr_err debugging and authorship, but also a couple of other points.
I suspect that you've copied the panel-waveshare-dsi-v2 driver and amended for your panels, but it'd nice to have that confirmed. That may not have been the best base as it isn't a driver that has been reviewed by upstream, but that isn't critical for a Pi specific product.

@flash-fea
Copy link
Copy Markdown
Author

yes,some informations need to change,about two dev_err() only use in debug to find by text easyier,but forget to recover.

@@ -0,0 +1,69 @@
/*
* Device Tree overlay for Waveshare DSI Touchscreens
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This also needs updating

compatible = "tdo,4.0-dsi-tl040wvs17";
status = "okay";
reg = <0>;
reset-gpios = <&gpio 47 1>; // Dummy GPIO , Unused or change
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Delete, as 6by9 says

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

delete all? reset-gpios maybe use in future,i see these were added also in other dts.

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You can add it at a later stage via an override if it is needed.

As it stands the driver will claim a totally unrelated GPIO for no particular reason. GPIO 47 isn't necessarily unused or even present on all devices, so could actually cause issues if left in.

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

some my device's driver need it, as how i write to overlay in use

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

For those devices that do require it you can add an additional fragment to configure it which can be enabled by the relevant override.

A similar case is https://github.com/raspberrypi/linux/blob/rpi-6.12.y/arch/arm/boot/dts/overlays/edt-ft5406.dtsi#L11-L16 where we want to add the touchscreen-inverted-x; property on demand, triggered by the invx override

With the driver calling devm_gpiod_get_optional, the driver doesn't need it. It's permitted to call any of the gpiod_ functions with the returned (NULL) handle.

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

like this?
reset_pin = <&panel>,"reset-gpios:47";
and modify driver for devm_gpiod_get_optional.

Copy link
Copy Markdown
Contributor

@pelwell pelwell Apr 16, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Close:

reset_pin = <&panel>,"reset-gpios:0=",<&gpio>, 
            <&panel>,"reset-gpios:4=47",
            <&panel>,"reset-gpios:8=1";

will create the reset-gpios property if you use the reset_pin parameter, but 6by9 was suggesting that you would build that into a product declaration:

new_panel_needing_reset = <&panel>, "compatible=tdo,new-panel-needing-reset",
            <&panel>,"reset-gpios:0=",<&gpio>,
            <&panel>,"reset-gpios:4=47",
            <&panel>,"reset-gpios:8=1";

But it might be cleaner to have a dormant fragment that gets enabled as needed.

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The driver is already using

ctx->reset = devm_gpiod_get_optional(&dsi->dev, "reset", GPIOD_OUT_LOW);

That means that ctx->reset will not return an error if reset-gpios is missing.

Add an additional fragment along the lines of:

	fragment@10 {
		target = <&panel>;
		reset_gpio_frag: __dormant__ {
			reset-gpios = <&gpio 47 1>;
		};
	};
	__overrides__ {
		reset_pin = <0>, "+10", 
				<&reset_gpio_frag>, "reset-gpios:4";
	};

You can then have dtoverlay=vc4-kms-dsi-tdo-panel,reset_pin=47 to enable the reset gpio and set it to 47 (or whatever other GPIO number you wish to use).

Just looking through the various Pi variants, gpio 47 is used for:

  • Pi1A/B: SD card detect
  • Pi1A+, B+, 2B, 3B, Zero, ZeroW: Activity LED
  • Pi 3B+/A+, CM3, CM4, 02W, CM0 : I2C to SMPS
  • CM1: EMMC_ENABLE_N

This is why I say it can't be claimed generically by the overlay.
(Pi5 appears to use it for 2712_WAKE)

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

this reset pin is not decided yet,can you give me a suggest? now dts add this->
+fragment@4 {

  •   target = <&panel>;
    
  •   reset_gpio_frag: __dormant__ {
    
  •   	reset-gpios = <&gpio 47 1>;
    
  •   };
    
  • };

dsi0 = <&dsi_frag>, "target:0=",<&dsi0>,
<&i2c_frag>, "target:0=",<&i2c_csi_dsi0>;

  • reset_pin = <0>, "+10",
  •          <&reset_gpio_frag>, "reset-gpios:4";
    

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

ctx->reset = devm_gpiod_get_optional(&dsi->dev, "reset", GPIOD_OUT_LOW);
if (IS_ERR(ctx->reset))
dev_warn(&dsi->dev, "Couldn't get the reset GPIO\n");

struct drm_display_mode *mode;

mode = drm_mode_duplicate(connector->dev, ctx->desc->mode);
dev_info(&ctx->dsi->dev, "%ux,%ux,%ux\n", ctx->desc->mode->hdisplay,
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

No user needs to see this - dev_dbg or delete.

struct tdo_panel *ctx;
int ret;

dev_info(&dsi->dev, "dsi panel: %s\n",
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It is preferred for successful probing to generate minimal output. Perhaps defer this line to after the number of lanes is available and combine the two?

struct tdo_panel *ctx = mipi_dsi_get_drvdata(dsi);

if (ctx->reset) {
dev_info(&dsi->dev, "shutdown\n");
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You really don't need this.

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

struct tdo_panel *ctx = mipi_dsi_get_drvdata(dsi);

gpiod_set_value_cansleep(ctx->reset, 0);

like this,is it ok?

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I was referring to the dev_info()

panel-tdo-dsi-v1.c as module to driver tdo panel.

Signed-off-by: kun_liu <kun_liu@shtdo.com>
@6by9
Copy link
Copy Markdown
Contributor

6by9 commented Apr 16, 2026

I've just noticed that someone is upstreaming the Waveshare DSI display drivers to mainline at present

V2 displays https://lore.kernel.org/dri-devel/20260413-waveshare-dsi-touch-v3-21-3aeb53022c32@oss.qualcomm.com/

https://github.com/torvalds/linux/blob/master/drivers/gpu/drm/bridge/waveshare-dsi.c for the V1 displays is already merged.
https://lore.kernel.org/dri-devel/20260412-ws-lcd-v3-0-db22c2631828@oss.qualcomm.com/T/ is extending that.

At some point I'll update the overlays and delete the old panel-waveshare-dsi and panel-waveshare-dsi-v2 drivers.

There are a couple of tdo panel drivers already in mainline, so if any of them match these new panels then the configuration really should be added to those drivers rather than this new one.
https://github.com/torvalds/linux/blob/master/drivers/gpu/drm/panel/panel-tdo-tl070wsh30.c
https://github.com/torvalds/linux/blob/master/drivers/gpu/drm/panel/panel-tpo-td028ttec1.c
https://github.com/torvalds/linux/blob/master/drivers/gpu/drm/panel/panel-tpo-td043mtea1.c

@flash-fea
Copy link
Copy Markdown
Author

my company need only one driver to use all our mipi panel,it seems to add some after this request finish.

Comment thread arch/arm/boot/dts/overlays/README Outdated
rotation Display rotation {0,90,180,270}
dsi0 Use DSI0 and i2c_csi_dsi0 (rather than
the default DSI1 and i2c_csi_dsi).
reset_pin GPIO pin for RESET (default 25)
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is not true - there is no default value, because you can't enable the reset pin without assigning a value.

rotation = <&panel>, "rotation:0";
dsi0 = <&dsi_frag>, "target:0=",<&dsi0>,
<&i2c_frag>, "target:0=",<&i2c_csi_dsi0>;
reset_pin = <0>, "+10",
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The number after the "+" is the number of the fragment to enable, which in this case should be 4.


/* And reset it */
if (ctx->reset) {
gpiod_set_value_cansleep(ctx->reset, 0);
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is this correct? You declare the reset pin to be active low (that's what the 1 in <&gpio 25 1> means), which suggests that the device is held in reset as long as the GPIO is low. When using the gpiod_ API, the value that you set is a logical value, not a physical line level - 1 means assert, 0 means deassert. This sequence will end up driving the GPIO high (deasserted) then low (asserted), which will leave the device in reset.

Have a look at the uses of the reset GPIO, decide in each case whether reset should be asserted or deasserted, then set the values appropriately. The flags value in the overlay then lets you specify the sense of the reset line on a per-board basis.

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

yes,you say is right,i review it carefulless.the pin not use in our first panel.so just add but not check.

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Unless you really feel the need to abuse the gpiod API and confuse the reader, any reset sequence should be assert then deassert. All you've done is invert the sense of the GPIO.

kun_liu added 2 commits April 17, 2026 17:37
__overrides__ parameter can be add in config.txt for customers.
README describe the dts config info.

Signed-off-by: kun_liu <kun_liu@shtdo.com>
Share panel drivers to bcmxxx_defconfig as module to use in all.

Signed-off-by: kun_liu <kun_liu@shtdo.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants