Note

This is the documentation for the latest development branch and may refer to features that are not available in released versions. If you are looking for a specific release, use the drop-down menu on the left.

LVGL Examples and VG-Lite Acceleration

Overview

K230 RT-Smart integrates LVGL 9.4 and provides adaptations for display, touch/HID, Sensor images, and K230 image format conversion. After enabling the VG-Lite renderer, LVGL submits supported fill, border, shadow, text, image, arc, line, layer, triangle, and mask tasks to the K230 GPU, while unsupported tasks are still handled by the software renderer.

Source code locations:

src/rtsmart/libs/3rd-party/lvgl/
src/rtsmart/examples/3rd-party/lvgl/

Current examples include:

Example	Purpose
lvgl_basic	Displays common controls and verifies connector, OSD layer, rotation, and color format
lvgl_hid	Verifies USB HID input devices
lvgl_sensor	Displays Sensor images and overlays the LVGL UI
lvgl_image_convert_test	Verifies RVV pixel conversion and MPP CSC image conversion

Configuration

Enable the following in make menuconfig:

RTSMART_3RD_PARTY_ENABLE_LVGL=y
RTSMART_ENABLE_3RD_PARTY_EXAMPLES=y
RTSMART_3RD_PARTY_ENABLE_LVGL_SAMPLES=y

The LVGL color depth is selected by the choice corresponding to RTSMART_3RD_PARTY_LVGL_COLOR_DEPTH，supporting 8、16、24 and 32 bit。The VG-Lite acceleration configuration is as follows：

RTSMART_3RD_PARTY_LVGL_USE_VGLITE=y
RTSMART_HAL_ENABLE_VG_LITE=y
RT_USING_VG_LITE=y

Enabling RTSMART_3RD_PARTY_LVGL_USE_VGLITE automatically selects the user-space VG-Lite HAL；the kernel still needs to include RT_USING_VG_LITE。The default tessellation size is 1920 x 1080，which can be adjusted through RTSMART_3RD_PARTY_LVGL_VGLITE_TESS_WIDTH and RTSMART_3RD_PARTY_LVGL_VGLITE_TESS_HEIGHT。

Running lvgl_basic

First use list_connector to query the connector types supported by the current firmware, then run:

lvgl_basic.elf -c <connector_type>

Optional parameters:

-c, --connector <type>  connector type
-l, --layer <id>        OSD layer, default 1
-f, --format <format>   l8, rgb565, rgb888, or argb8888
-r, --rotate <deg>      0, 90, 180, or 270
-H, --help              Display help

The example creates an LVGL display that matches the connector size and displays a label, button, bar, slider, and switch. Pressing Ctrl+C to exit releases the LVGL, display, and VB resources.

For custom interfaces, continue using the standard LVGL API:

lv_obj_t *label = lv_label_create(lv_screen_active());
lv_label_set_text(label, "Hello LVGL");
lv_obj_center(label);

while (running) {
    lv_timer_handler();
    usleep(5000);
}

VG-Lite renderer

The K230 port creates a VG-Lite draw unit during LVGL initialization and retains the software draw unit as a fallback. Applications do not need to rewrite every LVGL object to use the VG-Lite API, but the following should be noted:

• The draw buffer must be accessible by the GPU; the K230 port uses an appropriate allocator for LVGL’s own buffers;

• Contiguous buffers from MPP/VB must first have their virtual address, physical address, size, cache attributes, and pixel format registered;

• When the CPU and GPU alternately access an external cacheable buffer, access must go through the port’s synchronization path;

• Complex scenes should avoid frequently forcing vg_lite_finish(); instead, allow multiple draw tasks to form larger batches.

The low-level driver, native VG-Lite API, and Tiger performance examples are documented in VG-Lite GPU and examples.

Image Conversion

The K230 port provides a layered conversion path for lv_draw_buf_copy():

1. RVV is preferred between common packed LVGL formats;

2. When converting an MPP RGB/YUV buffer to an LVGL buffer, MPP CSC is used, combined with SDMA or memcpy as appropriate;

3. When the acceleration conditions are not met, it falls back to the scalar converter.

lvgl_image_convert_test.elf covers the following ranges:

• L8, A8, AL88, RGB565, RGB565 swapped, RGB888, XRGB8888;

• ARGB8888, premultiplied ARGB8888, ARGB1555, ARGB4444, ARGB2222, ARGB8565;

• MPP RGB/BGR packed formats, and NV12, NV21, I420.

Run the complete test:

lvgl_image_convert_test.elf

To run only one backend set or output each case:

lvgl_image_convert_test.elf --rvv-only
lvgl_image_convert_test.elf --csc-only --verbose

The program ultimately outputs Total: passed/total, where a return value of 0 indicates that all tests passed. CSC tests require successful MPP VB initialization; if initialization fails, first check the VB configuration and memory rather than attributing all failures to LVGL.

Sensor and Input Examples

lvgl_sensor.elf demonstrates the combined path of Sensor frame, LVGL draw buffer, GPU/CSC conversion, and display; lvgl_hid.elf demonstrates USB HID input. Before running, ensure that the target sensor, display, touch, or HID driver is included in the current board-level firmware.

FAQ

1. connector initialization failure: use list_connector to verify the enumeration values and the board-level panel configuration.

2. The UI is displayed but GPU acceleration is not enabled: check the three VG-Lite configuration items and the startup log. The software fallback itself does not report an error.

3. Garbled external image: verify the physical address, stride, pixel format, cache attributes, and buffer lifecycle.

4. The conversion test only fails in the CSC case: first confirm that the MPP VB pool is available and that the YUV plane address and stride meet the format requirements.