feat(lvgl_port_simd): RBG565 image blend to RGB565

    - RGB565 blend to RGB565 (optimized memcpy for RGB565 type)
    - esp32s3 assembly implementation using SIMD instructions
    - esp32 assembly fallback

components/esp_lvgl_port/CMakeLists.txt

@@ -85,6 +85,10 @@ if((lvgl_ver VERSION_GREATER_EQUAL "9.1.0") AND (lvgl_ver VERSION_LESS "9.2.0"))
         else()
             file(GLOB_RECURSE ASM_SRCS ${PORT_PATH}/simd/*_esp32.S)      # Select only esp32 related files
         endif()
+
+        # Explicitly add all assembly macro files
+        file(GLOB_RECURSE ASM_MACROS ${PORT_PATH}/simd/lv_macro_*.S)
+        list(APPEND ADD_SRCS ${ASM_MACROS})
         list(APPEND ADD_SRCS ${ASM_SRCS})
 
         # Include component libraries, so lvgl component would see lvgl_port includes
@@ -94,6 +98,7 @@ if((lvgl_ver VERSION_GREATER_EQUAL "9.1.0") AND (lvgl_ver VERSION_LESS "9.2.0"))
         # Force link .S files
         set_property(TARGET ${COMPONENT_LIB} APPEND PROPERTY INTERFACE_LINK_LIBRARIES "-u lv_color_blend_to_argb8888_esp")
         set_property(TARGET ${COMPONENT_LIB} APPEND PROPERTY INTERFACE_LINK_LIBRARIES "-u lv_color_blend_to_rgb565_esp")
+        set_property(TARGET ${COMPONENT_LIB} APPEND PROPERTY INTERFACE_LINK_LIBRARIES "-u lv_rgb565_blend_normal_to_rgb565_esp")
     endif()
 endif()
 

components/esp_lvgl_port/include/esp_lvgl_port_lv_blend.h

@@ -32,6 +32,10 @@ extern "C" {
     _lv_color_blend_to_rgb565_esp(dsc)
 #endif
 
+#ifndef LV_DRAW_SW_RGB565_BLEND_NORMAL_TO_RGB565
+#define LV_DRAW_SW_RGB565_BLEND_NORMAL_TO_RGB565(dsc)  \
+    _lv_rgb565_blend_normal_to_rgb565_esp(dsc)
+#endif
 
 /**********************
  *      TYPEDEFS
@@ -83,6 +87,22 @@ static inline lv_result_t _lv_color_blend_to_rgb565_esp(_lv_draw_sw_blend_fill_d
     return lv_color_blend_to_rgb565_esp(&asm_dsc);
 }
 
+extern int lv_rgb565_blend_normal_to_rgb565_esp(asm_dsc_t *asm_dsc);
+
+static inline lv_result_t _lv_rgb565_blend_normal_to_rgb565_esp(_lv_draw_sw_blend_image_dsc_t *dsc)
+{
+    asm_dsc_t asm_dsc = {
+        .dst_buf = dsc->dest_buf,
+        .dst_w = dsc->dest_w,
+        .dst_h = dsc->dest_h,
+        .dst_stride = dsc->dest_stride,
+        .src_buf = dsc->src_buf,
+        .src_stride = dsc->src_stride
+    };
+
+    return lv_rgb565_blend_normal_to_rgb565_esp(&asm_dsc);
+}
+
 #endif // CONFIG_LV_DRAW_SW_ASM_CUSTOM
 
 #ifdef __cplusplus

components/esp_lvgl_port/src/lvgl9/simd/lv_macro_memcpy.S

@@ -0,0 +1,60 @@
+/*
+ * SPDX-FileCopyrightText: 2025 Espressif Systems (Shanghai) CO LTD
+ *
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+// Memcpy macros for modulo checking
+// After running the main loop, there is need to check remaining bytes to be copied out of the main loop
+// Macros work with both, aligned and unaligned (4-byte boundary) memories
+// but performance is significantly lower when using unaligned memory, because of the unaligned memory access exception
+
+
+// Macro for checking modulo 8
+ .macro macro_memcpy_mod_8 src_buf, dest_buf, condition, x1, x2, JUMP_TAG
+    // Check modulo 8 of the \condition, if - then copy 8 bytes
+    bbci \condition, 3, ._mod_8_check_\JUMP_TAG      // Branch if 3-rd bit of \condition is clear
+        l32i.n      \x1,        \src_buf,   0        // Load 32 bits from \src_buff to \x1, offset 0
+        l32i.n      \x2,        \src_buf,   4        // Load 32 bits from \src_buff to \x2, offset 4
+        s32i.n      \x1,        \dest_buf,  0        // Save 32 bits from \x1 to \dest_buff, offset 0
+        s32i.n      \x2,        \dest_buf,  4        // Save 32 bits from \x2 to \dest_buff, offset 4
+        addi.n      \src_buf,   \src_buf,   8        // Increment \src_buff pointer by 8
+        addi.n      \dest_buf,  \dest_buf,  8        // Increment \dest_buff pointer 8
+    ._mod_8_check_\JUMP_TAG:
+.endm // macro_memcpy_mod_8
+
+
+// Macro for checking modulo 4
+ .macro macro_memcpy_mod_4 src_buf, dest_buf, condition, x1, JUMP_TAG
+    // Check modulo 4 of the \condition, if - then copy 4 bytes
+    bbci \condition, 2, ._mod_4_check_\JUMP_TAG      // Branch if 2-nd bit of \condition is clear
+        l32i.n      \x1,        \src_buf,   0        // Load 32 bits from \src_buff to \x1, offset 0
+        addi.n      \src_buf,   \src_buf,   4        // Increment \src_buff pointer by 4
+        s32i.n      \x1,        \dest_buf,  0        // Save 32 bits from \x1 to \dest_buff, offset 0
+        addi.n      \dest_buf,  \dest_buf,  4        // Increment \dest_buff pointer 4
+    ._mod_4_check_\JUMP_TAG:
+.endm // macro_memcpy_mod_4
+
+
+// Macro for checking modulo 2
+ .macro macro_memcpy_mod_2 src_buf, dest_buf, condition, x1, JUMP_TAG
+    // Check modulo 2 of the \condition, if - then copy 2 bytes
+    bbci \condition, 1, ._mod_2_check_\JUMP_TAG      // Branch if 1-st bit of \condition is clear
+        l16ui       \x1,        \src_buf,   0        // Load 16 bits from \src_buff to \x1, offset 0
+        addi.n      \src_buf,   \src_buf,   2        // Increment \src_buff pointer by 2
+        s16i        \x1,        \dest_buf,  0        // Save 16 bits from \x1 to \dest_buff, offset 0
+        addi.n      \dest_buf,  \dest_buf,  2        // Increment \dest_buff pointer 2
+    ._mod_2_check_\JUMP_TAG:
+.endm // macro_memcpy_mod_2
+
+
+// Macro for checking modulo 1
+ .macro macro_memcpy_mod_1 src_buf, dest_buf, condition, x1, JUMP_TAG
+    // Check modulo 1 of the \condition, if - then copy 1 byte
+    bbci \condition, 0, ._mod_1_check_\JUMP_TAG      // Branch if 0-th bit of \condition is clear
+        l8ui        \x1,        \src_buf,   0        // Load 8 bits from \src_buff to \x1, offset 0
+        addi.n      \src_buf,   \src_buf,   1        // Increment \src_buff pointer by 1
+        s8i         \x1,        \dest_buf,  0        // Save 8 bits from \x1 to \dest_buff, offset 0
+        addi.n      \dest_buf,  \dest_buf,  1        // Increment \dest_buff pointer 1
+    ._mod_1_check_\JUMP_TAG:
+.endm // macro_memcpy_mod_1