Hello,

Last week I've done total rework of handling userspace
window traps. Now we have two handlers for save and
restore traps: fast and slow. Trap handling code for
save and restore traps checks whether stack page
for corresponding %sp is mapped on the MMU. If it is,
fast handler is used, which just saves registers
to location pointed by %sp and returns. If not,
preemptive trap handler is launched, which inserts
mapping to the MMU and returns. save/restore
instruction is reexecuted, which cause next trap,
this time handled with the 'fast' handler.

Currently I'm working on interrupt traps and
timer-based scheduling support.

Regards,
Jakub

_______________________________________________
HelenOS-devel mailing list
[email protected]
http://lists.modry.cz/listinfo/helenos-devel

Reply via email to