>Note the 'code' as passed to systrace_enter() is already masked,
>and thus guaranteed to be lower than nsysent for given emulation.
>Yes, i've been looking on the problem reported on openbsd-bugs;
>we are not vulnerable.

	i see, i see that on the code.  i still think it safer to have
	e_nsysent outside, to avoid mistakes that happen when syscall table
	size is not 2^n.  thanks anyways.

itojun