More SPI issues

[ewan]

I have the SPI where I can read and write to a periperal without error when I do a couple of reads and writes then have a breakpoint to check the results. When I let the program run, it gets stuck in the while loop (see below).

Code: [Select]

// Write one byte to the reader IC address space
/*!
* -o  address  (IN) reader ic register address
* -o  value    (IN) 8 bit value
* return: none
*
* Function for writting one char to the reader module
*
*/

void WriteRawRC(unsigned char addr, unsigned char value)
{
  addr <<= 1;
  addr &= ~0x81;                              // fix up addr

  _CLEARBITS(AS, PORT_AS_BIT3);               // lower cs
 
  QIR = 0xd00f;                               // reset errors
  QAR = QSPI_COMMAND_RAM_0;                   // set address to first command location
  QDR = 0x0f00;                               // command to use no cs
  QDR = 0x0f00;                               // command to use no cs
  QAR = QSPI_TRANSMIT_RAM_0;                  // set address to first transmit location
  QDR = addr;                                 // save it
  QDR = value;                                // save it
  QWR = QSPI_CSIV | 0x100;                    // two sends
  QDLYR = QSPI_SPE;                           // QSPI enable (start)

  while (!(QIR & QSPI_SPIF)) {}               // wait for transfer to complete [stuck here]
  QIR |= QSPI_SPIF;                           // reset SPIF et al.

  _SETBITS(AS, PORT_AS_BIT3);                 // raise cs
}
The funny thing is that when I add a variable to it {test = QIR;} to test the loop, QIR equals 1 but the while loop continues anyway. Also the CS line has already been released. I cannot fathom how this is possible from several points of view. Does anyone have any ideas about this? When I disable the part of the program which deals with this stuff the ethernet/HTTP etc. works fine.

[ewan]

Well, perhaps just writing it down gets the brain thinking. Anyway -- problem solved. The routine above was being called by both the main program and an interrupt routine. (And reentrant is not sufficient with an I/O device.) I saved the state of the interrupt enable and dissabled that interrupt at the start of the routine. At the end I restored it.

Code: [Select]

void WriteRawRC(unsigned char addr, unsigned char value)
{
  unsigned char eport;
 
  eport = EPIER0;                             // save EPORT
 
  RC531_IRQ_Disable();                        // disable interrupt from RC531

  addr <<= 1;
  addr &= ~0x81;                              // fix up addr

  _CLEARBITS(AS, PORT_AS_BIT3);               // lower cs
 
  QIR = 0xd00f;                               // reset errors
  QAR = QSPI_COMMAND_RAM_0;                   // set address to first command location
  QDR = 0x0f00;                               // command to use no cs
  QDR = 0x0f00;                               // command to use no cs
  QAR = QSPI_TRANSMIT_RAM_0;                  // set address to first transmit location
  QDR = addr;                                 // save it
  QDR = value;                                // save it
  QWR = QSPI_CSIV | 0x100;                    // two sends
  QDLYR = QSPI_SPE;                           // QSPI enable (start)

  while (!(QIR & QSPI_SPIF)) {}               // wait for transfer to complete
  QIR |= QSPI_SPIF;                           // reset SPIF

  _SETBITS(AS, PORT_AS_BIT3);                 // raise cs
 
  EPIER0 = eport;                             // restore EPORT & reenable ints from RC531
}
I don't understand how all the weirdness happened in the first place, but it has gone away.
Ewan.