[ARM] 5339/1: fix __fls() on ARM

Commit 0c65f459

 intended to fix truncation issues with fls() on
ARMv5+ by renaming it to __fls() and wrapping it into a C function.
However that didn't take into account the fact that __fls() already
already had different semantics in the kernel.

Let's move the __fls() code into fls() function directly, and redefine
__fls() with the appropriate semantics.  While at it, bring a generic
__fls() definition for pre ARMv5 too.

Signed-off-by: Nicolas Pitre <nico@marvell.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>