Date: Sun, 06 Nov 2005 23:04:36 -0200 From: "Giovanni P. Tirloni" <gpt@tirloni.org> To: Stephen Montgomery-Smith <stephen@math.missouri.edu> Cc: freebsd-stable@freebsd.org Subject: Re: kernel: calcru: runtime went backwards Message-ID: <436EA824.5030106@tirloni.org> In-Reply-To: <436E67F0.9050409@math.missouri.edu> References: <436E67F0.9050409@math.missouri.edu>
next in thread | previous in thread | raw e-mail | index | archive | help
Stephen Montgomery-Smith wrote:
> This is on a recent FreeBSD-6.0 stable SMP machine.
>
> As root, run "top -s0". Then, at the same time run this program:
>
> #include <pthread.h>
> #define D (1<<10)
> void *thread(void *n) {
> int i;
> double array[D];
> for (i=0;i<D;i++) array[i] = i;
> }
> int main() {
> void *i;
> pthread_t tid;
> while (1) {
> pthread_create(&tid,NULL,thread,i);
> pthread_join(tid,&i);
> }
> }
>
> Let this program run for about a minute. Then kill this program.
>
> At this point, I sometimes get some calcru messages from the kernel.
>
> (This is not a contrived situation - this happens much more frequently
> than this sample program when I run programs with threaded fftw3 calls -
> I suspect that it throws off the cpu times shown by top.)
Try this,
http://www.freebsd.org/doc/en/books/faq/book.html#CALCRU-NEGATIVE
--
Giovanni P. Tirloni
http://tirloni.blogspot.com
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?436EA824.5030106>