bcm64: fix kernels cmpswap() function

spectacular bug. cmpswap() had a sign extension bug
using sign extending MOV to load the old compare
value and LDXRW using zero extension while the CMP
instruction compared 64 bit registers.

this caused cmpswap with negative old value always
to fail.

interestingly, libc's version of this function was
fine.

sys/src/9/bcm64/l.s

@@ -229,8 +229,8 @@ TEXT touser(SB), 1, $-4
 
 TEXT cas(SB), 1, $-4
 TEXT cmpswap(SB), 1, $-4
-	MOVW	ov+8(FP), R1
-	MOVW	nv+16(FP), R2
+	MOVWU	ov+8(FP), R1
+	MOVWU	nv+16(FP), R2
 _cas1:
 	LDXRW	(R0), R3
 	CMP	R3, R1