From 37c3a9dcf1a34cbe4241e3e056dc7e4c7d00bd9d Mon Sep 17 00:00:00 2001
From: orignal <romakoshelkin@yandex.ru>
Date: Fri, 9 May 2014 12:05:04 -0400
Subject: [PATCH] ChiperBlock XOR

---
 aes.cpp | 6 ++----
 aes.h   | 5 +++--
 2 files changed, 5 insertions(+), 6 deletions(-)

diff --git a/aes.cpp b/aes.cpp
index 2036ac35..48dc8fc2 100644
--- a/aes.cpp
+++ b/aes.cpp
@@ -165,8 +165,7 @@ namespace crypto
 	{
 		for (int i = 0; i < numBlocks; i++)
 		{
-			m_LastBlock.ll[0] ^= in[i].ll[0];
-			m_LastBlock.ll[1] ^= in[i].ll[1];
+			m_LastBlock ^= in[i];
 			m_ECBEncryption.Encrypt (&m_LastBlock, &m_LastBlock);
 			out[i] = m_LastBlock;
 		}
@@ -186,8 +185,7 @@ namespace crypto
 		{
 			ChipherBlock tmp = in[i];
 			m_ECBDecryption.Decrypt (in + i, out + i);
-			out[i].ll[0] ^= m_IV.ll[0];
-			out[i].ll[1] ^= m_IV.ll[1];
+			out[i] ^= m_IV;
 			m_IV = tmp;
 		}
 	}
diff --git a/aes.h b/aes.h
index 89a3a62d..dac7d378 100644
--- a/aes.h
+++ b/aes.h
@@ -14,13 +14,14 @@ namespace crypto
 		uint8_t buf[16];
 		uint64_t ll[2];
 
-		void operator^(const ChipherBlock& other) // XOR
+		void operator^=(const ChipherBlock& other) // XOR
 		{
 #ifdef __x86_64__
 			__asm__
 			(
 				"movups	(%[b1]), %%xmm0 \n"
-				"pxor (%[b2]), %%xmm0 \n"
+				"movups	(%[b2]), %%xmm1 \n" // b2 might not be 16-bytes aligned
+				"pxor %%xmm1, %%xmm0 \n" 
 				"movups	%%xmm0, (%[b1]) \n"	
 				: : [b1]"r"(buf), [b2]"r"(other.buf): "memory", "%xmm0"
 			);