Fix ppc4xx clear_bss() code

ppc4xx clear_bss() fails if BSS segment size is not
divisible by 4 without remainder. This patch provides
fix for this problem.

Signed-off-by: Anatolij Gustschin <agust@denx.de>
diff --git a/cpu/ppc4xx/start.S b/cpu/ppc4xx/start.S
index 645ac0b..52601ed 100644
--- a/cpu/ppc4xx/start.S
+++ b/cpu/ppc4xx/start.S
@@ -1494,16 +1494,25 @@
 	lwz	r4,GOT(_end)
 
 	cmplw	0, r3, r4
-	beq	6f
+	beq	7f
 
 	li	r0, 0
-5:
+
+	andi.	r5, r4, 3
+	beq	6f
+	sub	r4, r4, r5
+	mtctr	r5
+	mr	r5, r4
+5:	stb	r0, 0(r5)
+	addi	r5, r5, 1
+	bdnz	5b
+6:
 	stw	r0, 0(r3)
 	addi	r3, r3, 4
 	cmplw	0, r3, r4
-	bne	5b
-6:
+	bne	6b
 
+7:
 	mr	r3, r9		/* Init Data pointer		*/
 	mr	r4, r10		/* Destination Address		*/
 	bl	board_init_r