Hi Guys, working on a new TIDA and playing with controlSTICK I ran into a strange problem. I am sending out 4B data over SPI (master) and for that reason I created following function which works....
/* PUSH data to SPI buffer - left adjusted */
#define SPIA_PUSH(TX) SpiaRegs.SPITXBUF = (TX << 8)
/* POP data from SPI buffer - right adjusted */
#define SPIA_POP SpiaRegs.SPIRXBUF
Uint32 hal_Xmit4BSPI(Uint32 frame)
{
Uint32 rxval = 0x00000000; /* incomming data */
/* empty RX and TX 4B FIFOs */
SpiaRegs.SPIFFTX.bit.TXFIFO = 0; /* reset FIFOs */
SpiaRegs.SPIFFRX.bit.RXFIFORESET = 0;
asm(" NOP"); /* probably not needed */
asm(" NOP"); /* probably not needed */
SpiaRegs.SPIFFTX.bit.TXFIFO = 1;
SpiaRegs.SPIFFRX.bit.RXFIFORESET = 1; /* release from reset */
/* quick & dirty ! */
SPIA_PUSH(frame>>24); /* first byte (from most significant) */
SPIA_PUSH(frame>>16);
SPIA_PUSH(frame>>8);
SPIA_PUSH(frame); /* last byte */
/* wait until the transmition is finished */
while(SpiaRegs.SPISTS.bit.INT_FLAG == 0){EMPTY_LOOP};
// asm(" NOP"); /* probably not needed */
// asm(" NOP"); /* probably not needed */
/* get data from RX FIFO */
rxval = SPIA_POP; /* first byte (from most significant) */
rxval <<= 8;
rxval |= SPIA_POP;
rxval <<= 8;
rxval |= SPIA_POP;
rxval <<= 8;
rxval |= SPIA_POP;
/* NCS must remain high at least 200ns */
HAL_DELAY_US(1);
return(rxval);
}
And I call my function temporary from my main loop which looks like this:
int main(void)
{
hal_InitSysCtrl(); /* inits clock domains, PLL */
hal_InitGPIO(); /* practically default config + LED */
hal_InitUART(); /* inits UART 115200 8N1 */
xdev_out(hal_PutcUART); /* inits xprintf function */
pga411_Init(); /* inits PGA411 driver */
/* endless loop */
while (1)
{
hal_ToggleLED();
HAL_DELAY_US(500000);
xprintf("Hello UART \r");
asm(" NOP"); /* probably not needed */
hal_Xmit4BSPI(0x12345678);
}
#pragma diag_suppress 112 /* to get rid of warning that code is unreachable */
return 0;
#pragma diag_default 112 /* restore error level */
}
Problem description:
I found out that I need to have purple or yellow assembly to have the code running properly. Otherwise the first hal_Xmit4BSPI call passes properly BUT later calls will end in program never leaving the while condition (INT_FLAG never set).
I could understand that the yellow ASM lines may help some internal synchronization but solving the problem with a NOP even before the function call is strange. xprintf is just an UART output.
CPU is running at 80 MHz from the internal A oscillator and PLL, LowSpeed clock is also 80 MHz. SPI clock is 8 MHz.
I am not asking for help but if you were interested in further data and whole CCS project we may get in touch via webex or internal mail.
Cheers,
Jiri, Freising.
