kcalloc: remove runtime division

While in all cases in the kernel we know the size of the elements to be
created, we don't always know the count of elements.  By commuting the size
and count in the overflow check, the compiler can reduce the runtime division
of size_t with a compare to a (unique) constant in these cases.

Signed-off-by: Milton Miller <miltonm@bga.com>
Cc: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>