Clean up smp_idle's stack jumping
authorBarret Rhoden <brho@cs.berkeley.edu>
Fri, 22 Jul 2016 18:41:55 +0000 (14:41 -0400)
committerBarret Rhoden <brho@cs.berkeley.edu>
Wed, 27 Jul 2016 16:52:43 +0000 (12:52 -0400)
commit8c99aa704a1cdcb6257131a16ac93bfbe002d982
treec89bbaca269f684e579ee0019869eb33bff2b4ae
parent4a5d86c829b2b80be62bf95d073af9c963d5a988
Clean up smp_idle's stack jumping

First off, RESET_STACKS is no longer an option.  There are weird cases
where you want to backtrace beyond smp_idle.  You can comment that out
manually if you want, but there's no need to make it a regular thing.
If you don't RESET_STACKS, you'll potentially run off the stack.

Other things:
- that disable_irq in smp_idle() wasn't really protecting anything.  No
  sense in doing it there.  We might need it to protect parts of the
KMSG subsystem, so I moved it to __smp_idle() for now.
- that cmb() did nothing.  The compiler won't reorder those asm
  operations.

Signed-off-by: Barret Rhoden <brho@cs.berkeley.edu>
Kconfig
config-default
kern/src/smp.c