RE: nRF52840 + SSD1680 (mipi-dbi-spi) + LVGL

Øyvind — Fri, 06 Jun 2025 13:07:47 GMT

Hello, happy to hear that you got past one hurdle.

Unfortunately we do not provide support for the LVGL library, this is something specific from the Zephyr community. You can reach out to them through Discord.

Good luck in getting a cool display up and running

Kind regards,
Øyvind

RE: nRF52840 + SSD1680 (mipi-dbi-spi) + LVGL

alex2103 — Fri, 06 Jun 2025 12:58:21 GMT

I want to share my progress and ask for advice about LVGL + Zephyr on a custom board with an SSD1680 E-Paper display (using mipi-dbi-spi), nRF52840 SoC, and NCS v3.0.2.

Previously, I was getting the error:

<err> lvgl: Display device not ready.

This happened during startup because LVGL was trying to initialize before the display had power.

I fixed this by adding to my prj.conf

CONFIG_LV_Z_AUTO_INIT=n

This stopped LVGL from initializing automatically at boot.

Now I initialize everything manually in my code:

void view_lvgl()
{
    LOG_DBG("Begin");
    gpio_pin_configure_dt(&out, GPIO_OUTPUT_ACTIVE);
    gpio_pin_set_dt(&out, 1);
    LOG_DBG("PowerON Disp");
    k_msleep(250);

    display_dev = DEVICE_DT_GET(DT_CHOSEN(zephyr_display));
    LOG_DBG("Init Disp");
    device_init(display_dev);

    while (!device_is_ready(display_dev)) {
        LOG_WRN("Device not ready...\n");
        k_msleep(100);
    }
    LOG_DBG("Start lvgl_init");

    lvgl_init();

    LOG_DBG("LVGL started");
    hello_world_label = lv_label_create(lv_scr_act());
    lv_label_set_text(hello_world_label, "Hello world!");
    lv_obj_align(hello_world_label, LV_ALIGN_CENTER, 0, 0);
    LOG_DBG("Label created");

    LOG_DBG("lv_task_handler starting");
    lv_task_handler();
    display_blanking_off(display_dev);
    LOG_DBG("lv_task_handler finish");

    while (1) {
        printk("view_lvgl loop\n");
        k_msleep(1000);
    }
}

Now, the "Display device not ready" error is gone.

However, I now have a new issue:
As soon as lv_task_handler() is called for the first time, the system crashes with a fatal error.

Here is the stack trace from the fault handler:

k_sys_fatal_error_handler(unsigned int reason, const struct arch_esf * esf) (e:\krisrianna\git\nrf\blinky\build\zephyr\include\generated\zephyr\syscalls\log_ctrl.h:34)
z_fatal_error(unsigned int reason, const struct arch_esf * esf) (e:\Segger\ncs\v3.0.2\zephyr\kernel\fatal.c:119)
z_arm_fatal_error(unsigned int reason, const struct arch_esf * esf) (e:\Segger\ncs\v3.0.2\zephyr\arch\arm\core\fatal.c:86)
z_arm_fault(uint32_t msp, uint32_t psp, uint32_t exc_return, _callee_saved_t * callee_regs) (e:\Segger\ncs\v3.0.2\zephyr\arch\arm\core\cortex_m\fault.c:1107)
z_arm_usage_fault() (e:\Segger\ncs\v3.0.2\zephyr\arch\arm\core\cortex_m\fault_s.S:102)
<signal handler called> (Unknown Source:0)
[Unknown code] (Unknown Source:0)
display_write(const void * buf, const struct display_buffer_descriptor * desc, const uint16_t y, const uint16_t x, const struct device * dev) (e:\Segger\ncs\v3.0.2\zephyr\include\zephyr\drivers\display.h:254)
lvgl_flush_cb_mono(lv_display_t * display, const lv_area_t * area, uint8_t * px_map) (e:\Segger\ncs\v3.0.2\zephyr\modules\lvgl\lvgl_display_mono.c:103)
call_flush_cb(uint8_t * px_map, const lv_area_t * area, lv_display_t * disp) (e:\Segger\ncs\v3.0.2\modules\lib\gui\lvgl\src\core\lv_refr.c:1303)
draw_buf_flush(lv_display_t * disp) (e:\Segger\ncs\v3.0.2\modules\lib\gui\lvgl\src\core\lv_refr.c:1270)
refr_invalid_areas() (e:\Segger\ncs\v3.0.2\modules\lib\gui\lvgl\src\core\lv_refr.c:611)
lv_display_refr_timer(lv_timer_t * tmr) (e:\Segger\ncs\v3.0.2\modules\lib\gui\lvgl\src\core\lv_refr.c:405)
lv_timer_exec(lv_timer_t * timer) (e:\Segger\ncs\v3.0.2\modules\lib\gui\lvgl\src\misc\lv_timer.c:327)
lv_timer_handler() (e:\Segger\ncs\v3.0.2\modules\lib\gui\lvgl\src\misc\lv_timer.c:107)
lv_timer_handler() (e:\Segger\ncs\v3.0.2\modules\lib\gui\lvgl\src\misc\lv_timer.c:63)
lv_task_handler() (e:\Segger\ncs\v3.0.2\modules\lib\gui\lvgl\src\lv_api_map_v8.h:77)
view_lvgl() (e:\krisrianna\git\nrf\blinky\src\main.c:136)
z_thread_entry(k_thread_entry_t entry, void * p1, void * p2, void * p3) (e:\Segger\ncs\v3.0.2\zephyr\lib\os\thread_entry.c:48)

It seems the crash occurs inside the display driver when LVGL tries to flush the buffer for the first time.

I did more debugging and found that the crash is caused by a NULL pointer passed to the display driver.

Just before the crash, in the function display_write() (from include/zephyr/drivers/display.h), the dev pointer is NULL:

As you can see, when display_write() is called, dev == 0x0 (NULL). This leads to a crash when the API tries to dereference the device pointer.

My questions:

Is it correct to call device_init(display_dev) directly?
In most Zephyr examples, I don't see this function being called.
Is my sequence of powering up, checking readiness, and then calling lvgl_init() correct?
Do I need to wait longer after powering up the display, or do anything special for SSD1680?
What is the minimal recommended LVGL event/task handling loop?
Right now I just call lv_task_handler() once before entering the main loop (which only prints a message). Should I be calling lv_task_handler() in the main loop with some interval, or do I need to process events differently for E-Paper?
What could be causing this crash in display_write()?
Could it be a problem with my devicetree, buffer sizes, or driver configuration? What is the best way to debug or verify the display initialization and flush pipeline for this type of E-Paper?

What could cause LVGL or the display pipeline to pass a NULL device pointer to display_write()?

Am I missing some required initialization or assignment step for the display device when doing manual LVGL init?
Anything else I should be aware of when using deferred LVGL init on Zephyr with E-Paper displays?

Thanks for any tips or corrections!