CH592 memcpy

memcpy提速方法：

#define ALIGN_MASK       0x3
#define COPY32 *d32 = *s32; d32++; s32++;
#define COPY8 *d8 = *s8; d8++; s8++;
#define SET32(x) *d32 = x; d32++;
#define SET8(x) *d8 = x; d8++;
#define REPEAT8(expr) expr expr expr expr expr expr expr expr
typedef uintptr_t lv_uintptr_t;

#define LV_ATTRIBUTE_FAST_MEM   __attribute__((section(".highcode")))

//__attribute__((section(".highcode")))
LV_ATTRIBUTE_FAST_MEM void * lv_memcpy(void * dst, const void * src, size_t len)
{
    uint8_t * d8 = dst;
    const uint8_t * s8 = src;

    lv_uintptr_t d_align = (lv_uintptr_t)d8 & ALIGN_MASK;
    lv_uintptr_t s_align = (lv_uintptr_t)s8 & ALIGN_MASK;

    /*Byte copy for unaligned memories*/
    if(s_align != d_align) {
        while(len > 32) {
            REPEAT8(COPY8);
            REPEAT8(COPY8);
            REPEAT8(COPY8);
            REPEAT8(COPY8);
            len -= 32;
        }
        while(len) {
            COPY8
            len--;
        }
        return dst;
    }

    /*Make the memories aligned*/
    if(d_align) {
        d_align = ALIGN_MASK + 1 - d_align;
        while(d_align && len) {
            COPY8;
            d_align--;
            len--;
        }
    }

    uint32_t * d32 = (uint32_t *)d8;
    const uint32_t * s32 = (uint32_t *)s8;
    while(len > 32) {
        REPEAT8(COPY32)
        len -= 32;
    }

    while(len > 4) {
        COPY32;
        len -= 4;
    }

    d8 = (uint8_t *)d32;
    s8 = (const uint8_t *)s32;
    while(len) {
        COPY8
        len--;
    }

    return dst;
}

验证效果，4K字节传输花费116us。