gpio模拟以太网SMI接口，读寄存器失败

EDA3rd · 发表于 2017-10-26 15:56:54

因为F407不支持op code的设置，所以采用gpio模拟smi接口操作，但是读出来的数据有问题，全是0xffff，找不出问题，是delay_us(1)延时太长吗？

[mw_shl_code=c,true]//IO方向
#define MDIO_IN()  {GPIOA->MODER&=~(3<<(2*2));GPIOA->MODER|=0<<2*2;} //PA2
#define MDIO_OUT() {GPIOA->MODER&=~(3<<(2*2));GPIOA->MODER|=1<<2*2;} //PA2

#define MDIO PAout(2) //SCL
#define MDC    PCout(1) //SDA
#define READ_MDIO  PCin(1) //SDA

void SMI_Init(void)
{
//配置PA2 PC1为推完输出，GPIO模拟SMI
GPIO_InitStructure.GPIO_Pin = GPIO_Pin_2;
GPIO_InitStructure.GPIO_Speed = GPIO_Speed_100MHz;
GPIO_InitStructure.GPIO_Mode = GPIO_Mode_OUT;
GPIO_InitStructure.GPIO_OType = GPIO_OType_PP; //推完输出
GPIO_InitStructure.GPIO_PuPd = GPIO_PuPd_UP ;//GPIO_PuPd_UP GPIO_PuPd_NOPULL
GPIO_Init(GPIOA, &GPIO_InitStructure);//PA2 MDIO

GPIO_InitStructure.GPIO_Pin = GPIO_Pin_1;
GPIO_InitStructure.GPIO_PuPd = GPIO_PuPd_NOPULL ;//GPIO_PuPd_UP GPIO_PuPd_NOPULL
GPIO_Init(GPIOC, &GPIO_InitStructure);//PC1 MDC ，不需要上拉
MDIO = 1;
MDC = 0;
}

u8 SMI_Read_One_Byte(void)
{
u8 i, receive=0;
// MDIO_IN();
for(i = 0; i < 8; i++)
{
receive<<=1;
      MDC=0;//拉低时钟
delay_us(1);
MDC=1;  //上升沿传输数据
delay_us(1);
if(READ_MDIO) receive|= 0x01; //上升沿，高电平期间读数据
MDC=0;//时钟恢复低电平
}
return receive;
}

u8 SMI_Read_2Bit(void)
{
u8 i, receive=0;
// MDIO_IN();
for(i = 0; i < 2; i++)
{
      receive<<=1;
      MDC=0;//拉低时钟
delay_us(1);
MDC=1;  //上升沿传输数据
delay_us(1);
      if(READ_MDIO) receive|= 0x01; //上升沿，高电平期间读数据
      MDC=0;//恢复低电平
}
return receive;
}

void SMI_Write_One_Byte(u8 data)
{
u8 i;
// MDIO_OUT();
for(i = 0; i < 8; i++)
{
      MDC=0;
      MDIO=(data&0x80)>>7;//准备数据
      delay_us(1);
      MDC=1;  //上升沿传输数据
      delay_us(1); //保持一段时间
      MDC = 0;//恢复低电平
      data<<=1;
}
}

void SMI_Write_2Bit(u8 data)
{
u8 i;
MDIO_OUT();
for(i = 0; i < 2; i++)
{
      MDC=0;
      MDIO=(data&0x2)>>1;
      delay_us(1);
      MDC=1;    //上升沿传输数据
      delay_us(1); //保持一段时间
      MDC = 0;//恢复低电平
      data<<=1;
}
}

#define START_OF_FRAME_2bit 0x01
#define READ_OP_CODE_2bit 0x2
#define WRITE_OP_CODE_2bit 0x1
#define SMI_OP_CODE_2bit 0x00
#define SMI_TA 0x02
void SMI_Write_Frame(u16 PHYAddress, u16 PHYReg, u16 PHYValue)
{
u8 addr;

addr = (PHYAddress & 0x7) <<5 | (PHYReg & 0x1F);

MDIO_OUT();
////32 bit Preamble
SMI_Write_One_Byte(0xFF);
SMI_Write_One_Byte(0xFF);
SMI_Write_One_Byte(0xFF);
SMI_Write_One_Byte(0xFF);
SMI_Write_One_Byte(0xFF);
SMI_Write_One_Byte(0xFF);
SMI_Write_One_Byte(0xFF);
SMI_Write_One_Byte(0xFF);

SMI_Write_2Bit(START_OF_FRAME_2bit);
SMI_Write_2Bit(WRITE_OP_CODE_2bit);
SMI_Write_2Bit(0x00);    //dir: 0x0 is write
SMI_Write_One_Byte(addr);
SMI_Write_2Bit(SMI_TA);
SMI_Write_One_Byte(0xFF);
SMI_Write_One_Byte(PHYValue);
MDIO_IN();//恢复输入高阻态
}

u16 SMI_Read_Reg(u16 PHYAddress, u16 PHYReg)
{
u8 addr;
u16 data;

addr = (PHYAddress & 0x7)<<5 | (PHYReg & 0x1F);

////32 bit Preamble
MDIO_OUT();
SMI_Write_One_Byte(0xFF);
SMI_Write_One_Byte(0xFF);
SMI_Write_One_Byte(0xFF);
SMI_Write_One_Byte(0xFF);

SMI_Write_2Bit(START_OF_FRAME_2bit);
SMI_Write_2Bit(READ_OP_CODE_2bit);
SMI_Write_2Bit(0x0);    //PHYAD[4:3] decide dir: 0x2/0x3 is write
SMI_Write_One_Byte(addr);

MDIO_IN();
data = SMI_Read_2Bit();//SMI_TA should be Z0
data = SMI_Read_One_Byte();
data = data << 8;
data = data | SMI_Read_One_Byte();
return data;
}

u8 KSZ8863_Get_Reg_Value(u8 port, u8 reg_num)
{
u16 reg;
// reg = ETH_ReadPHYRegister(port, reg_num); //从port1的1号寄存器中读取网络速度和双工模式

reg = SMI_Read_Reg(port, reg_num);

printf("port %d, reg 0x%x = 0x%x \r\n", port, reg_num, reg);
return reg;
}

void KSZ8863_Print_Reg_Value(void)
{
u8 i, port;
for(port = 1; port < 4; port++){
for(i = 0; i < 6; i++){
KSZ8863_Get_Reg_Value(port, i);
delay_ms(100);
}
}
}
[/mw_shl_code]

EDA3rd · 发表于 2017-10-26 15:56:55

解决了，READ_MDIO 读引脚定义错了，应该定义为PA2：
#define READ_MDIO PAin(2) //SDA
可以读全局寄存器了

正点原子 · 发表于 2017-10-27 01:03:12

帮顶

EDA3rd · 发表于 2017-10-27 10:02:41

谢谢原子哥，网友博客的分析文章 http://blog.csdn.net/jasonchen_gbd/article/details/51628992
中提到“当PHY向MDIO写数据时，PHY要先发出一个MDC的上升沿，然后等待0-300ns” ，这句话意思是，mac读数据时，mdc时钟是由phy提供的？？？是这样子么？