#include <stdio.h> 
#include <string.h>
#include <stdbool.h>
#include <math.h>
#include <stdint.h>

#include <common_func.h>
#include <confreg_time.h>
#include <dma.h>
#include <fft.h>

// BSP板级支持包所需全局变量
unsigned long UART_BASE = 0xbf000000;
unsigned long CONFREG_TIMER_BASE = 0xbf20f100;
unsigned long CONFREG_CLOCKS_PER_SEC = 50000000L;
unsigned long CORE_CLOCKS_PER_SEC = 33000000L;

const float PI = 3.14159265358979323846;

// 软件FFT实现 (基2 DIT-FFT 算法)
void sw_fft(float re[], float im[], int n) {
    int i, j, k, l;
    float tr, ti, ur, ui, wr, wi;

    j = 0;
    for (i = 0; i < n - 1; i++) {
        if (i < j) {
            tr = re[i]; ti = im[i];
            re[i] = re[j]; im[i] = im[j];
            re[j] = tr; im[j] = ti;
        }
        k = n / 2;
        while (k <= j) {
            j -= k;
            k /= 2;
        }
        j += k;
    }

    for (l = 1; l < n; l *= 2) {
        ur = 1.0; 
        ui = 0.0;
        wr = cos(PI / l);
        wi = -sin(PI / l);
        
        for (i = 0; i < n; i += 2 * l) {
            ur = 1.0; 
            ui = 0.0;
            for (j = 0; j < l; j++) {
                int p = i + j;
                int q = p + l;
                
                tr = re[q] * ur - im[q] * ui;
                ti = re[q] * ui + im[q] * ur;
                
                re[q] = re[p] - tr;
                im[q] = im[p] - ti;
                re[p] += tr;
                im[p] += ti;
                
                float next_ur = ur * wr - ui * wi;
                ui = ur * wi + ui * wr;
                ur = next_ur;
            }
        }
    }
}

// 全局对齐数组：作为 DMA 的源和目的内存
int32_t hw_in_re_arr[FFT_POINT_NUM]  __attribute__((aligned(64)));
int32_t hw_in_im_arr[FFT_POINT_NUM]  __attribute__((aligned(64)));
int32_t hw_out_re_arr[FFT_POINT_NUM] __attribute__((aligned(64)));
int32_t hw_out_im_arr[FFT_POINT_NUM] __attribute__((aligned(64)));

int main(int argc, char** argv)
{
    unsigned int fft_csr = RegRead(FFT_CSR_REG);
    printf("fft_csr init = %x\n", fft_csr);

    // 获取外设的纯物理地址 (屏蔽高3位 0x1FFFFFFF)
    uint32_t phys_fft_in_re  = FFT_IN_RE_BASE  & 0x1FFFFFFF;
    uint32_t phys_fft_in_im  = FFT_IN_IM_BASE  & 0x1FFFFFFF;
    uint32_t phys_fft_out_re = FFT_OUT_RE_BASE & 0x1FFFFFFF;
    uint32_t phys_fft_out_im = FFT_OUT_IM_BASE & 0x1FFFFFFF;

    // 获取内存数组的纯物理地址，并包装为 CPU 使用的无缓存(Uncached)指针
    volatile int32_t *uncached_in_re  = (volatile int32_t *)(((uint32_t)hw_in_re_arr  & 0x1FFFFFFF) | 0xA0000000);
    volatile int32_t *uncached_in_im  = (volatile int32_t *)(((uint32_t)hw_in_im_arr  & 0x1FFFFFFF) | 0xA0000000);
    volatile int32_t *uncached_out_re = (volatile int32_t *)(((uint32_t)hw_out_re_arr & 0x1FFFFFFF) | 0xA0000000);
    volatile int32_t *uncached_out_im = (volatile int32_t *)(((uint32_t)hw_out_im_arr & 0x1FFFFFFF) | 0xA0000000);

    // 准备软件 FFT 测试数据
    float sw_in_re[FFT_POINT_NUM];
    float sw_in_im[FFT_POINT_NUM];

    // 初始化测试波形
    for (int i = 0; i < FFT_POINT_NUM; i++) {
        float dc_part   = 4000.0f;
        float f10_part  = 8000.0f * cos(2 * PI * 10.0 * i / FFT_POINT_NUM);
        float f200_part = 6000.0f * sin(2 * PI * 200.0 * i / FFT_POINT_NUM);
        float f400_part = 3000.0f * sin(2 * PI * 400.0 * i / FFT_POINT_NUM);
        
        sw_in_re[i] = dc_part + f10_part + f200_part + f400_part;
        sw_in_im[i] = 0.0f;

        // 硬件输入需要转为整数存入无缓存内存
        uncached_in_re[i]  = (int32_t)sw_in_re[i];
        uncached_in_im[i]  = 0;
        
        // 清理输出内存以防干扰
        uncached_out_re[i] = 0xDEADBEEF;
        uncached_out_im[i] = 0xDEADBEEF;
    }

    unsigned int tick_start, tick_end;
    unsigned int hw_time, sw_time;

    // 硬件加速 FFT 测试 (纯 DMA 搬运)
    printf("\n--- Starting Hardware FFT with DMA ---\n");
    tick_start = get_ns(); // 开始计时

    uint32_t transfer_bytes = FFT_POINT_NUM * 4; // 1024个点 * 4字节

    // MA 将数据从内存搬运到 FFT 输入外设
    dma_start_transfer(0, ((uint32_t)hw_in_re_arr & 0x1FFFFFFF), phys_fft_in_re, transfer_bytes, 100);
    dma_start_transfer(1, ((uint32_t)hw_in_im_arr & 0x1FFFFFFF), phys_fft_in_im, transfer_bytes, 200);
    dma_wait_polling(0);
    dma_wait_polling(1);

    
    // 启动 FFT 并等待计算完成
    fft_start();
    fft_wait();
    
    // DMA 将结果从 FFT 输出外设搬回内存
    dma_start_transfer(0, phys_fft_out_re, ((uint32_t)hw_out_re_arr & 0x1FFFFFFF), transfer_bytes, 10);
    dma_start_transfer(1, phys_fft_out_im, ((uint32_t)hw_out_im_arr & 0x1FFFFFFF), transfer_bytes, 20);
    dma_wait_polling(0);
    dma_wait_polling(1);


    tick_end = get_ns(); // 结束计时
    hw_time = tick_end - tick_start;

    // 纯软件 FFT 测试
    printf("--- Starting Software FFT ---\n");
    tick_start = get_ns();
    
    sw_fft(sw_in_re, sw_in_im, FFT_POINT_NUM);
    
    tick_end = get_ns();
    sw_time = tick_end - tick_start;

    // 打印对比结果
    printf("\n--- Performance Comparison ---\n");
    printf("Timer Clock Freq : %lu Hz\n", CONFREG_CLOCKS_PER_SEC);
    printf("Hardware FFT Time: %u ns (%.3f ms)\n", hw_time, (float)hw_time / 1000000.0);
    printf("Software FFT Time: %u ns (%.3f ms)\n", sw_time, (float)sw_time / 1000000.0);
    
    if (hw_time > 0) {
        printf("Speedup Ratio    : %.2fx\n", (float)sw_time / hw_time);
    }

    printf("\n--- Verification (Only showing Bins with energy > 10) ---\n");
    for (int i = 0; i < FFT_POINT_NUM; i++) {
        // CPU 通过无缓存指针读取 DMA 搬运回来的结果
        int32_t hw_re = uncached_out_re[i];
        int32_t hw_im = uncached_out_im[i];

        if (fabs((float)hw_re) > 10 || fabs((float)hw_im) > 10) {
            printf("Bin [%4d] Hz: HW(Re:%6d, Im:%6d) | SW(Re:%6d, Im:%6d)\n", 
                    i, 
                    hw_re, hw_im, 
                    (int)sw_in_re[i] / FFT_POINT_NUM, (int)sw_in_im[i] / FFT_POINT_NUM);
        }
    }

    return 0;
}