From 81ee80ab88941658d7d427c76e1a6c94debfb4a3 Mon Sep 17 00:00:00 2001
From: Andy Polyakov <appro@openssl.org>
Date: Wed, 6 Apr 2005 09:45:42 +0000
Subject: [PATCH] +45% RC4 performance boost on Intel EM64T core. Unrolled loop
 providing further +35% will follow...

Submitted by: Zou Nanhai
---
 crypto/rc4/asm/rc4-amd64.pl | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/crypto/rc4/asm/rc4-amd64.pl b/crypto/rc4/asm/rc4-amd64.pl
index 9e0da8af99..2d3dedde03 100755
--- a/crypto/rc4/asm/rc4-amd64.pl
+++ b/crypto/rc4/asm/rc4-amd64.pl
@@ -181,7 +181,7 @@ $code.=<<___;
 
 .align	16
 .LRC4_CHAR:
-	inc	$XX#b
+	add	\$1,$XX#b
 	movzb	`&PTR("BYTE:[$dat+$XX]")`,$TX#d
 	add	$TX#b,$YY#b
 	movzb	`&PTR("BYTE:[$dat+$YY]")`,$TY#d
@@ -191,9 +191,9 @@ $code.=<<___;
 	movzb	`&PTR("BYTE:[$dat+$TY]")`,$TY#d
 	xorb	`&PTR("BYTE:[$inp]")`,$TY#b
 	movb	$TY#b,`&PTR("BYTE:[$out]")`
-	inc	$inp
-	inc	$out
-	dec	$len
+	lea	1($inp),$inp
+	lea	1($out),$out
+	sub	\$1,$len
 	jnz	.LRC4_CHAR
 	jmp	.Lexit
 ___
-- 
2.40.0