> > i meant, nothing except kernel_lock prevents the state changed before
> > you examine the return value of the function.
> > of course, whether it can actually be a problem or not depends
> > on its usage.  hence my question above.
> 
> I thought it might be useful. since I have no intention on discussing
> it, I will just revert it if you care that much.

you said we would use it soon, didn't you?

i'm not requesting you back it out.
if you have a good use of it, for which the race is not a problem, it's fine.

YAMAMOTO Takashi