> On Thu, Oct 25, 2007 at 11:20:27AM +0900, YAMAMOTO Takashi wrote:
> 
> > > Module Name:	src
> > > Committed By:	ad
> > > Date:		Wed Oct 24 14:50:42 UTC 2007
> > > 
> > > Modified Files:
> > > 	src/sys/arch/acorn26/acorn26: cpuswitch.c
> > > 	src/sys/arch/amd64/amd64: trap.c
> > > 	src/sys/arch/hp300/hp300: trap.c
> > > 	src/sys/arch/i386/i386: trap.c
> > > 	src/sys/arch/mips/mips: trap.c
> > > 	src/sys/arch/powerpc/powerpc: trap.c
> > > 	src/sys/arch/sh3/sh3: locore_c.c
> > > 	src/sys/arch/sparc64/sparc64: trap.c
> > > 	src/sys/arch/xen/i386: trap.c
> > > 	src/sys/kern: kern_exec.c kern_exit.c kern_fork.c kern_proc.c
> > > 	    kern_ras.c sys_process.c
> > > 	src/sys/sys: proc.h ras.h
> > > 
> > > Log Message:
> > > Make ras_lookup() lockless.
> > 
> > is NO_SOFTWARE_PATENTS for rcu?
> 
> Strictly speaking it's not RCU but a related technique. I'm still
> researching whether or not the patents apply.
> 
> > how is the kpause() version supposed to work?
> 
> - kpause + mi_switch will flush the write buffer on the calling cpu
> - ras_lookup + loads in advance will take no longer than 1/8s to complete
> 
> Andrew

can't kpause wake up spontaneously?
anyway, can you put a comment about the assumptions?

YAMAMOTO Takashi