Calin Culianu wrote:
> Let's say your real-time thread were (erroneously) making use of some
> kernel services that could potentially block. (Like, oh, writing to a
> Linux fifo that might not have its buffers in memory, for instance).
>
> Could such cases cause the whole system to lock? Basically our real-time
> thread is non-preemptible (not even by hardware-interrupts) so we would
> never get out of the linux kernel routines that are blocking (since no
> interrupts would ever make their way in to any sort of interrupt handler
> both in RTL and in the regular kernel).
Way are you using RTLinux for this, if it should bee non-preemptible????
Don't you have a pthread_wait_np call or other suspend funktion???
(every RTLinux suspend funktion will preemp the CPU (and Linux) when the
task is ready, if it is needed.)
>
> Is such a situation something I should be worried about? Am I
> misunderstanding something?
Do not know. I do not know if you will preempt the CPU.
> Should I NEVER call a regular kernel routine, EVER EVER EVER? :)
(I assume you meen Linux kernel routine.)
I would never use such a program in a critical part. Maybee for test but
thats it.
I tried once to use a standard linux task queue and it left linux in a undefined
state. (it crased).
Anders Gnistrup
-- [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/
