Re: [avr-chat] Avrdude : how to slow it down ?

[Bernard Fouché]

Joerg Wunsch wrote:

> > ?! :-/
> > The overhead of calling usleep() is apparently already larger than the
> > amount of delay caused by it.  Also, there's a system-dependent
> > granularity of which delays can be had with this kind of sleep
> > functions, often it's just 10 ms granularity (i.e. 100 Hz clock
> > frequency).  You can just write usleep(1) as well.
> >    
On kernel 2.6.x usleep(3) works well now. This was posted some time ago 
when kernel 2.4.x had some
granularity problems, I just ran it on kernel 2.6.12 and it was okay by 
1 or 2 ms:

#include <stdio.h>
#include <unistd.h>
#include <time.h>
#include <sys/time.h>

#define MILLISEC_DIFF(new, old, diff)   {                              \
       (diff) = (((new).tv_sec  - (old).tv_sec)  * 1000L ) +           \
                (((new).tv_usec - (old).tv_usec) / 1000L );             \
}

int main(int args, char ** argv) {

       int time;
       struct timeval time1,time2;
       long elapsed;

       if(args < 2) {
               printf("USAGE: sleep <time in milliseconds>\n");
               exit(0);
       }

       time = atoi(argv[1]);

       while(1) {

               gettimeofday(&time1, NULL);
               usleep(time*1000);
               gettimeofday(&time2, NULL);

               MILLISEC_DIFF(time2, time1, elapsed);

               printf("time 1 = <%ld:%ld>, time 2 = <%ld:%ld> ,elapsed 
=%ld\n\n",time1.tv_sec, time1.tv_usec, time2.tv_sec, time2.tv_usec, 
elapsed);
       }
}