LVGL Sample Does Not Work With STM32H747I-DISCO Board

[nluckblurpd]

I am attempting to run the LVGL sample application with the STM32H747i-Disco board. I follow the steps on the Zephyr board page [https://docs.zephyrproject.org/latest/boards/arm/stm32h747i_disco/doc/index.html]:
Build:
west build -b stm32h747i_disco_m7 samples/subsys/display/lvgl -- -DSHIELD=st_b_lcd40_dsi1_mb1166
Flash:
west flash
Both building and flashing report success.
During execution, the display backlight comes on, but the unit just sits there and reboots. No graphics are displayed on the screen.

I activated the console output to usuart1 which directs to the USB CDC on the dev kit. The console output indicates a boot failure:
[00:00:00.032,000] otm8009a: ID 0x0 (should 0x40)
[00:00:00.032,000] otm8009a: Panel ID check failed! (-22)
Therefore, it seems the sample cannot initialize the DSI controller on the display module.
I have tried two different dev kits, and get the same results on each, therefore I am fairly certain it is not a hardware issue.

Could the LVLG sample be made to work with the dev kit?
Am I more than willing to help investigate and fix the issue if anyone can point me in the right direction.

[github-actions[bot]]

Hi @nluckblurpd! We appreciate you submitting your first issue for our open-source project. 🌟

Even though I'm a bot, I can assure you that the whole community is genuinely grateful for your time and effort. 🤖💙

[FRASTM]

Testing on my stm32h747 disco kit (note that I removed CONFIG_LV_Z_SHELL=y and added CONFIG_LV_Z_POINTER_KSCAN=y to the prj.conf)

*** Booting Zephyr OS build zephyr-v3.4.0-1545-g67ec97aa9664 ***
uart:~$ help
Please press the <Tab> button to see all available commands.
You can also use the <Tab> button to prompt or auto-complete all commands or it.
You can try to call commands with <-h> or <--help> parameter for more informati.

Shell supports following meta-keys:
  Ctrl + (a key from: abcdefklnpuw)
  Alt  + (a key from: bf)
Please refer to shell documentation for more details.

Available commands:
  clear    :Clear screen.
  device   :Device commands
  devmem   :Read/write physical memory
            Usage:
            Read memory at address with optional width:
            devmem address [width]
            Write memory at address with mandatory width and value:
            devmem address <width> <value>
  help     :Prints the help message.
  history  :Command history.
  kernel   :Kernel commands
  log      :Commands for controlling logger
  resize   :Console gets terminal screen size or assumes default in case the
            readout fails. It must be executed after each terminal width change
            to ensure correct text display.
  retval   :Print return value of most recent command
  shell    :Useful, not Unix-like shell commands.
uart:~$ 

[nluckblurpd]

Thanks for checking on it. I made the changes to proj.conf you indicated above. Still, it doesn't run on my two dev kits. On each, I get the following console output:

[00:00:00.032,000] <err> otm8009a: ID 0x0 (should 0x40)
[00:00:00.032,000] <err> otm8009a: Panel ID check failed! (-22)
*** Booting Zephyr OS build zephyr-v3.4.0-1676-gd7c00ee7c25b ***

I wonder if you have an older dev kit with different hardware? I just ordered my two from Mouser last week. I can't think of anything else that might be different. Both my display modules indicate a revision of MB1166-A09, whereas yours indicates MB1166-A03 on the sticker.

Do you have any suggestions on how I can debug further?

[maarten1C96]

Replacing st_b_lcd40_dsi1_mb1166 by st_b_lcd40_dsi1_mb1166_a09 made it work with the newer STM32H747I-DISCO over here :

west build -p -b stm32h747i_disco_m7 ~/zephyrproject/zephyr/samples/subsys/display/lvgl -- -DSHIELD=st_b_lcd40_dsi1_mb1166_a09

[FRASTM]

yes, probably old LCD shield : mine is MB1166 -A03

[nluckblurpd]

I searched through my spare parts bin and found an A03 display like yours. It works! So, there is some hardware difference between A03 and A09. But, the base LVGL sample runs correctly with the A03 display.

A secondary issue: I would like to turn on ChromeArt acceleration. I activated the proper (I think) define in lv_conf.h:

/*Use STM32's DMA2D (aka Chrom Art) GPU*/
#define LV_USE_GPU_STM32_DMA2D 1
#if LV_USE_GPU_STM32_DMA2D
    /*Must be defined to include path of CMSIS header of target processor
    e.g. "stm32f7xx.h" or "stm32f4xx.h"*/
    #define LV_GPU_DMA2D_CMSIS_INCLUDE "stm32h7xx.h"
#endif

It builds just fine. But the video output is incorrect:

Could you give me any pointers on how to fix that issue? I assume it's some ChromeArt setup issue, but I don't know where to look in Zephyr.

[FRASTM]

I am not really aware of this dma2d of the stm32H747
Even if the application is building, I do not see where this dma2d node is defined and if it should be neither. There is a register bank @ 0x52001000, clock enable bit (4) on RCC_AHB3ENR, IRQ 90. That could be a first step to declare the peripheral in the dtsi.

[erwango]

Moved to a discussion as now issue for now.
Regarding 2D accelaration, you could be interested in #60840.
That won't help directly as it's just he beginning but that will give you the current status of 2D acceleration support in Zephyr right now.

[maxhbr]

I just received my devkit with an A09 shield and I have the same issue that it is not working. Was that solved somehow?

[maxhbr]

maybe a helpful link: STMicroelectronics/STM32CubeF7@6eed5bb#diff-7f4e9ae42bf83eb3d8371345a2edafd6345f79f2d658eb863382f01c8a8f8ed4 . It seems to come with NT35510 instead of OTM8009A

[gizmo1904]

I experienced the same issue. Any chance of fixing this?

[maxhbr]

I think to fix this, someone would need to contribute the drivers for the chips used on the newer display. Not sure if that specific driver is on the roadmap.

[ajarmouni-st]

I experienced the same issue. Any chance of fixing this?

Hi @maxhbr, as you can see here STM32CubeH7 & here community.st.com the newer stm32h747i_disco kits use a different revision of MB1166 display shield that has the panel controller NT35510, which is not currently supported in zephyr. Only the revisions of MB1166 prior to A09 (that use OTM8009A) are currently supported.

[erian747]

Hi, i have encountered the same issue with the A09 version and have written a driver for it based on the driver from ST found at
https://github.com/STMicroelectronics/STM32CubeH7/tree/master/Drivers/BSP/Components/nt35510
and the OTM8900A driver in zephyr
I can make a PR if it would be worth adding it to main repo

[erwango]

I can make a PR if it would be worth adding it to main repo

That would be very much welcome indeed.

[erian747]

Added new PR #70350 :)

[erian747]

@nluckblurpd

I also want to use DMA2D, almost a must to get reasonable performance on demos
The i realized CONFIG_STM32_LTDC_RGB888=y was set in the shield config
To use DMA2D, CONFIG_STM32_LTDC_ARGB8888 must be configured, then it worked fine
I guess the shield config should not force RGB888, istead it shall select ARGB8888 if
CONFIG_LV_USE_GPU_STM32_DMA2D=y

[erian747]

On my hardware i also noted that i got inconsistent touch coordinates, see log below
Seems like bit errors
So i tested lowering I2C frequency in shields board overlay to I2C_BITRATE_STANDARD instead of FAST
That made the problem dissapear, haven't investigated what could be cause, software or hardware
Just writing here if anyone else having the same issue

[00:13:51.856,000] ft5336: ft5336_process: points: 1, touch_id: 1, row: 0, col: 290
[00:13:51.956,000] ft5336: ft5336_process: points: 1, touch_id: 1, row: 0, col: 290
[00:13:52.056,000] ft5336: ft5336_process: points: 1, touch_id: 1, row: 0, col: 290
[00:13:52.156,000] ft5336: ft5336_process: points: 1, touch_id: 1, row: 0, col: 290
[00:13:52.256,000] ft5336: ft5336_process: points: 1, touch_id: 1, row: 128, col: 257
[00:13:52.356,000] ft5336: ft5336_process: points: 1, touch_id: 1, row: 0, col: 290
[00:13:52.456,000] ft5336: ft5336_process: points: 1, touch_id: 1, row: 0, col: 290
[00:13:52.556,000] ft5336: ft5336_process: points: 1, touch_id: 1, row: 0, col: 290
[00:13:52.656,000] ft5336: ft5336_process: points: 1, touch_id: 1, row: 0, col: 290

[erian747]

Have you tried running with CONFIG_STM32_LTDC_ARGB8888 configured for the shield?

[KoenVda88]

Is this a comment for my issue?

[KoenVda88]

Hi,

The patch is working great but I get these unalignment issues:

I tried to run the LVGL demo with and without the CONFIG_LV_USE_GPU_STM32_DMA2D=y setup in 16bit color mode. I see the FPS change but the align issue stays the same. So it seems to be something in the driver?

Thanks for any leads!

[KoenVda88]

I tried to run the benchmark tool also with CONFIG_STM32_LTDC_ARGB8888=y inside the shield config. But with or without DMA I get the same problem.
i also tried set the pixel-format = <PANEL_PIXEL_FORMAT_ARGB_8888>; in &ltdc which has no visible effect

For the DMA I do always get this warning during compilation: LVGL can't enable the clock of DMA2D. To solve it I just placed the clock inits outside the #elif defined(STM32H7). If I don't do this the DMA is not working at all.

These are my changes:
align_issue_stm32.patch
stm32_dma_clk.patch

[erian747]

You could also try updating LVGL to latest version, i did that when testing the shield I dont remember why i had to do that, maybe it was because the DMA clock problem you mentioned I have no access to my computer right now but can take o look tomorrow to see which LVGL version i use Den lör 4 maj 2024 14:43KoenVda88 ***@***.***> skrev:

…

I tried to run the benchmark tool also with CONFIG_STM32_LTDC_ARGB8888=y inside the shield config. But with or without DMA I get the same problem. i also tried set the pixel-format = <PANEL_PIXEL_FORMAT_ARGB_8888>; in &ltdc which has no visible effect For the DMA I do always get this warning during compilation: *LVGL can't enable the clock of DMA2D*. To solve it I just placed the clock inits outside the #elif defined(STM32H7). If I don't do this the DMA is not working at all. These are my changes: align_issue_stm32.patch <https://github.com/zephyrproject-rtos/zephyr/files/15209746/align_issue_stm32.patch> stm32_dma_clk.patch <https://github.com/zephyrproject-rtos/zephyr/files/15209747/stm32_dma_clk.patch> — Reply to this email directly, view it on GitHub <#60888 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AC4NPBTFNPI23L7B6FCP5QTZATJV7AVCNFSM6AAAAABBEXJUSWVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TGMJTGYZDO> . You are receiving this because you commented.Message ID: ***@***.*** com>

[KoenVda88]

I already have the last version of LVGL I'm using zephyrproject-rtos/lvgl@2b498e6

[erian747]

Ok, same version as i use then

This is my settings in proj.conf, note the CONFIG_LV_GPU_DMA2D_CMSIS_INCLUDE="stm32h7xx.h"

CONFIG_STM32_LTDC_ARGB8888=y
CONFIG_NT35510=y
CONFIG_DISPLAY_INIT_PRIORITY=87
CONFIG_MAIN_STACK_SIZE=4096
CONFIG_LV_Z_MEM_POOL_SIZE=49152
CONFIG_LV_USE_GPU_STM32_DMA2D=y
CONFIG_LV_GPU_DMA2D_CMSIS_INCLUDE="stm32h7xx.h"

[KoenVda88]

No luck on those. I was missing CONFIG_NT35510=y but it does not seem to matter. The rest was already the same

So when you run the CONFIG_LV_USE_DEMO_BENCHMARK you don't get these strange align problems?

Thanks!

[erian747]

Actually i haven't tried the benchmark demo, just tried the other demos, think it was called widgets or music