ARM: udelay: prevent math rounding resulting in short udelays

We perform the microseconds to loops calculation using a number of
multiplies and shift rights.  Each shift right rounds down the
resulting value, which can result in delays shorter than requested.
Ensure that we always round up.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>

arch/arm/lib/delay.S

@@ -25,11 +25,15 @@ ENTRY(__udelay)
 		ldr	r2, .LC1
 		mul	r0, r2, r0
 ENTRY(__const_udelay)				@ 0 <= r0 <= 0x7fffff06
+		mov	r1, #-1
 		ldr	r2, .LC0
 		ldr	r2, [r2]		@ max = 0x01ffffff
+		add	r0, r0, r1, lsr #32-14
 		mov	r0, r0, lsr #14		@ max = 0x0001ffff
+		add	r2, r2, r1, lsr #32-10
 		mov	r2, r2, lsr #10		@ max = 0x00007fff
 		mul	r0, r2, r0		@ max = 2^32-1
+		add	r0, r0, r1, lsr #32-6
 		movs	r0, r0, lsr #6
 		moveq	pc, lr