When I think a bit more of it, one probably should update the PSR after
the heir has been acquired, as the task could potentially be acquired
and released again with a new PSR by another core before the swap.
On 2015-11-12 11:27, Daniel Cederman wrote:
Hello,
I experienced a bug when using 4.11 SMP on a LEON4 system. One of the
tasks in my program suddenly lost the FPU enable bit in the PSR register
after a task switch. I tracked down the problem being due to the task
switching code not updating the PSR if the heir changes during the
switch (which happens if the original heir gets scheduled on another
CPU). I am submitting a patch that updates the PSR register to match the
value for the new heir. It can probably be improved by moving the PSR
update to after the heir has been acquired, but I wanted to change as
little as possible.
--
Daniel Cederman
Software Engineer
Cobham Gaisler
T : +46 (0) 31 775 8665
F : +46 (0) 31 421407
daniel.ceder...@gaisler.com
Cobham Gaisler AB, Kungsgatan 12, SE-411 19, GÖTEBORG, Sweden.
+46 (0) 31 775 8650, www.cobham.com/gaisler
_______________________________________________
devel mailing list
devel@rtems.org
http://lists.rtems.org/mailman/listinfo/devel
