The more I read about the use of floating points the more I get confused
about it.
I am about to control a tricept-robot with rtlinux-2.3-pre2. Since I
started trying and working with your kernel, I read about the use of
fp's in the mailing list every night. Now I have set up the
communication with the powerboard with the rt_com-module and I'm ready
to implement the whole computation of the control-values into the
"thread_code" of Barabanov's testcom.c, which seems very suitable for my
project. - But I still don't know how to make it work.
My computation is quite large with plenty of pow's, sqrt's, sin's and
cos's in it. But it has to be done in realtime to be able to control the
machine (it worked with DOS at a period of 20ms).
Now I did the following test:
... all unchanged untill here...
void *thread_code(void *t) {
pthread_setfp_np(thread, 1);
pthread_make_periodic_np(thread,...)
do {
int n;
char buf[210]
double a, b;
a = 2.0;
b = sqrt(a);
...then no change anymore...
} while(1);
return 0;
}
...and added the following lines in the Makefile:
testfp: testcom_fp.o
$(CC) $(CFLAGS) -c testcom_fp.c
I had no errors with make testfp, but - I guess the cracks between you
are already laughing - the insmod testcom_fp.o brought down my machine
immediatly and completly. Is it may not allowed to use more than one
pthread_function or is the math-function itself the problem? I couldn't
get any messages after the insmod, also after the poweroff, the kernel
seems really to stop working there.
Thanks for helping
Cheers
Martin
-- [rtl] ---
To unsubscribe:
echo "unsubscribe rtl" | mail [EMAIL PROTECTED] OR
echo "unsubscribe rtl <Your_email>" | mail [EMAIL PROTECTED]
---
For more information on Real-Time Linux see:
http://www.rtlinux.org/rtlinux/
