[plug] Scheduling Problems

[Brad Campbell]

"Anthony J. Breeds-Taurima" wrote:
> 
> On Thu, 18 May 2000, Brad Campbell wrote:
> 
> > A:- I know Linux
> > B:- I don't know C, and my Pascal compiler is for Linux
> > C:- By increasing the HZ value, I can get the results I need, I just don't see
> >       why I should have to do it, if the kernel is doing the Right Thing(tm)
> > D:- I can't afford the time/money to invest into the development tools of a RTOS
> >       like QNX or similar.
> 
> I don't mean to sound rude BUT have  you looked at RTLinux
> (http://www.rtlinux.org/)

Yup, BUT
I don't need a RTOS.
I don't have a problem with the odd missed deadline, and I only need relative accuracy
to 2ms. Hardly a requirement for a RTOS.
If I needed accuracy to 50uS then I'd NEED to look at something else.
The problem I'm experiencing theoreticaly should not exist.

I had a look at the kernel serial code, and what happens is the interrupt handler
schedules the bottom half handler to run. What appears to be happening is my task 
is not being preempted on it's system call, thus the bottom half handler does not
run until my task gets preempted at the end of it's timeslice.