Subj : Re: pthread_cond_timedwait problems
To   : comp.programming.threads
From : David Schwartz
Date : Sun Aug 14 2005 05:52 pm


<markh@compro.net> wrote in message 
news:1124009185.007178.164900@f14g2000cwb.googlegroups.com...
> This makes sense.
>
> Attempting  delay of 10000000 nsec (10000 usecs)
> gettimeofday reported delay  =  11019 usecs
> Actual usec delay from timer =  10952 usecs
>
> Attempting  delay of 20000000 nsec (20000 usecs)
> gettimeofday reported delay  =  21033 usecs
> Actual usec delay from timer =  20963 usecs
>
> Attempting  delay of 40000000 nsec (40000 usecs)
> gettimeofday reported delay  =  41298 usecs
> Actual usec delay from timer =  41234 usecs
>
> nanosleep does not relinquish the cpu. Using it in my env causes
> serious problems.

    The job of 'nanosleep' is to wait a particular amount of time. It may or 
may not relinquish the CPU.

> I guess the nsec member of the timeval struct was misleading me. I
> should have known better.

    Huh?

    DS

.