> Like David, I do not like the idea of pre-empting kernel tasks. It implies a
> more complex locking setup than I'd really like.
I think any realtime task should be able to preempt the kernel,
while the realtime task executes code in the USER-SEGMENT.
If the task does a System call, the control has to be given
back to the task, which holds the kernel lock.
This would give a very fast response which is required by
realtime tasks, if the realtime-task uses locked memory.
I think this change should cause no problems. As far as
I know its already possible to execute code in a tasks User
Segment while a second processor executes Kernel-Code
in SMP-Linux, so why not doing something similar with
a single processor ?
> HOWEVER, I'm all for having "user processes" in kernel mode. That's not too
> hard - if only implies that we have to change the way we test for "user vs
> kernel" a bit.
This in combination with my above suggestion would make possible
very nice realtime tasks.
What do You think about this ?
Jochen
---Jochen Karrer http://cip.physik.uni-wuerzburg.de/~cip307
---