RE: [avrdude-dev] Win 32 port

[Alex Shepherd]

> > Here are a list of errors that I saw.  Note that I got multiples of 
> > these errors and most of the time the mismatch was at 0x0200. Funny 
> > that it is always on a page boundary?

I had this exact same issue months ago using the parallel port interface
on Win2k.

The issues tuned out to me that the timers were not delaying long
enough. The solution I finally found was to poll the Pentium 64 bit CPU
performance counter for the shorter duration delays.

Anyway all this code is already in CVS a while ago in the ppiwin.c file.

I suggest that if you append these changes to you test Win32 API
version, the problem will go away. The code changes replace the library
version of usleep() - see code fragment below.

Hopefully this will make this problem go away. :)

Cheers

Alex Shepherd
================================================================


// WARNING WARNING This function replaces the standard usleep() library
function
// because it doesn't appear to delay for the correct time

#ifndef MIN_SLEEP_USEC
#define MIN_SLEEP_USEC 20000
#endif

unsigned usleep( unsigned int uSeconds )
{
  struct timeval t1, t2;
  struct timespec nanoDelay ;

  if (usecPerfDelay(uSeconds))
    return;

  gettimeofday(&t1, NULL);

  if( uSeconds > MIN_SLEEP_USEC )
  {
    nanoDelay.tv_sec = uSeconds / 1000000UL;
    nanoDelay.tv_nsec = (uSeconds / 1000000UL) * 1000 ;
    nanosleep( &nanoDelay, NULL ) ;
  }

  /* loop for the remaining time */
  t2.tv_sec = uSeconds / 1000000UL;
  t2.tv_usec = uSeconds % 1000000UL;

  timeradd(&t1, &t2, &t1);

  do {
    gettimeofday(&t2, NULL);
  } while (timercmp(&t2, &t1, <));
}


#endif