libftdi1学习笔记 8 - MPSSE SPI优化速度

之前用GPIO模拟SPI速度有点慢，加上直接发送字节的方式实现SPI。

1. IO的定义

如果不使用模拟的方式，则SCK，MOSI，MISO必须固定。通过这3个IO判断是哪种模式

if(spi[port].sck != 0 || spi[port].mosi_io0 != 1 || spi[port].miso_io1 != 2)
{
    //GPIO模拟的方式
}
else
{
    
}

2. 模式的限制

似乎只支持模式0，虽然文档上写的命令应该支持0或3。

3. 实现

之前8位数据是通过IO分8次发送出去，这里是采用0x31或0x39发送。

    else
    {
        uint8_t *pwrDat;
        if(wrBuf == NULL)
        {
            pwrDat = malloc(len);
            for(int i = 0; i < len; i++)
                pwrDat[i] = 0xff;
        }
        else
            pwrDat = wrBuf;
        switch(spi[port].mode)
        {
            case SPI_MODE0:
                if(spi[port].msb == true)
                    spi[port].pCommand[spi[port].iCommand++] = 0x31;
                else
                    spi[port].pCommand[spi[port].iCommand++] = 0x39;
                spi[port].pCommand[spi[port].iCommand++] = (uint8_t)((len - 1) & 0xff);
                spi[port].pCommand[spi[port].iCommand++] = (uint8_t)(((len - 1) >> 8) & 0xff);
                for(int i = 0; i < len; i++)
                {
                    spi[port].pCommand[spi[port].iCommand++] = pwrDat[i];
                }
                break;

            case SPI_MODE3:
                if(spi[port].msb == true)
                    spi[port].pCommand[spi[port].iCommand++] = 0x31;
                else
                    spi[port].pCommand[spi[port].iCommand++] = 0x39;
                spi[port].pCommand[spi[port].iCommand++] = (uint8_t)((len - 1) & 0xff);
                spi[port].pCommand[spi[port].iCommand++] = (uint8_t)(((len - 1) >> 8) & 0xff);
                for(int i = 0; i < len; i++)
                {
                    spi[port].pCommand[spi[port].iCommand++] = pwrDat[i];
                }
                break;
            default:
                printf("no support mode:%d\n", spi[port].mode);
                break;
        }
    }

读数据也不需要读入len * 8个字节了

    if(spi[port].sck != 0 || spi[port].mosi_io0 != 1 || spi[port].miso_io1 != 2)
        rdLen = len * 8;
    else
        rdLen = len;

读入的数据也不需要按位组合位字节。

        else
        {
            for(int i = 0; i < rdLen; i++)
            {
                rdBuf[i] = pReadBuf[i];
            }
        }

4. 验证

将IO的定义改成

    spiSetting.sck = 0;
    spiSetting.mosi_io0 = 1;
    spiSetting.miso_io1 = 2;

同样基于nor flash读写1MB的数据

sflash id is:ef4018
Erase Chip Finish
erase time:41027ms
write time:14672ms
read time:3537ms
sflash test finish

速度是1M / 0.3537 = 2.83MB/s