improve implementation of ainc and adec for 386 and amd64

Reference: /n/atom/patch/applied/ainc386
Date: Fri May 23 20:21:08 CES 2014
Signed-off-by: quanstro@quanstro.net

--- /sys/src/libc/amd64/atom.s	Fri May 23 20:21:08 2014
+++ /sys/src/libc/amd64/atom.s	Fri May 23 20:21:08 2014
@@ -1,21 +1,14 @@
 TEXT ainc(SB), 1, $0	/* int ainc(int *); */
-ainclp:
-	MOVL	(RARG), AX	/* exp */
-	MOVL	AX, BX
-	INCL	BX		/* new */
-	LOCK; CMPXCHGL BX, (RARG)
-	JNZ	ainclp
-	MOVL	BX, AX
+TEXT ainc32(SB), 1, $0	/* u32int ainc(u32int*); */
+	MOVL	$1, AX
+	LOCK; XADDL AX, (RARG)
+	ADDL	$1, AX
 	RET
 
 TEXT adec(SB), 1, $0	/* int adec(int*); */
-adeclp:
-	MOVL	(RARG), AX
-	MOVL	AX, BX
-	DECL	BX
-	LOCK; CMPXCHGL BX, (RARG)
-	JNZ	adeclp
-	MOVL	BX, AX
+	MOVL	$-1, AX
+	LOCK; XADDL AX, (RARG)
+	SUBL	$1, AX
 	RET
 
 /*
--- /sys/src/libc/386/atom.s	Fri May 23 20:21:08 2014
+++ /sys/src/libc/386/atom.s	Fri May 23 20:21:08 2014
@@ -1,25 +1,16 @@
 TEXT ainc(SB), $0	/* int ainc(int *); */
+TEXT ainc32(SB), 1, $0	/* u32int ainc(u32int*); */
 	MOVL	addr+0(FP), BX
-ainclp:
-	MOVL	(BX), AX
-	MOVL	AX, CX
-	INCL	CX
-	LOCK
-	BYTE	$0x0F; BYTE $0xB1; BYTE $0x0B	/* CMPXCHGL CX, (BX) */
-	JNZ	ainclp
-	MOVL	CX, AX
+	MOVL	$1, AX
+	LOCK; XADDL AX, (BX)
+	ADDL	$1, AX
 	RET
 
 TEXT adec(SB), $0	/* int adec(int*); */
 	MOVL	addr+0(FP), BX
-adeclp:
-	MOVL	(BX), AX
-	MOVL	AX, CX
-	DECL	CX
-	LOCK
-	BYTE	$0x0F; BYTE $0xB1; BYTE $0x0B	/* CMPXCHGL CX, (BX) */
-	JNZ	adeclp
-	MOVL	CX, AX
+	MOVL	$-1, AX
+	LOCK; XADDL AX, (BX)
+	SUBL	$1, AX
 	RET
 
 /*