From 3bee1f57cf9b5f7fe309692d1e13239d1b23b4e2 Mon Sep 17 00:00:00 2001
From: Jeffrey Walton <noloader@gmail.com>
Date: Wed, 10 May 2017 06:34:22 -0400
Subject: [PATCH] XOR block in specialized ProcessBlock Use Put and Get blocks
 consistently

---
 kalyna.cpp | 168 +++++++++++++++++++++--------------------------------
 kalyna.h   |  10 ++--
 2 files changed, 72 insertions(+), 106 deletions(-)

diff --git a/kalyna.cpp b/kalyna.cpp
index 864f139d..eba64846 100644
--- a/kalyna.cpp
+++ b/kalyna.cpp
@@ -929,20 +929,16 @@ void Kalyna::Base::SetKey_88(const word64 key[8])
 
 // *********************** ProcessAndXorBlock specializations *********************** //
 
-void Kalyna::Base::ProcessBlock_22(const word64 inBlock[2], word64 outBlock[2]) const
+void Kalyna::Base::ProcessBlock_22(const word64 inBlock[2], const word64 xorBlock[2], word64 outBlock[2]) const
 {
-#if defined(IS_BIG_ENDIAN)
     word64 *t1 = m_wspace+0, *t2 = m_wspace+2, *msg = m_wspace+4;
-    {
-        typedef GetBlock<word64, LittleEndian, false> Block;
-        Block block(inBlock);
-        block(msg[0])(msg[1]);
-        inBlock = msg;
-    }
-#else
-    word64 *t1 = outBlock, *t2 = m_wspace+2;
-#endif
 
+    // Reverse bytes on BigEndian; Align pointer on LittleEndian
+    typedef GetBlock<word64, LittleEndian, false> InBlock;
+    InBlock iblk(inBlock);
+    iblk(msg[0])(msg[1]);
+
+    inBlock = msg;
     if (IsForwardTransformation())
     {
         AddKey<2>(inBlock, t1, m_rkeys);
@@ -973,29 +969,22 @@ void Kalyna::Base::ProcessBlock_22(const word64 inBlock[2], word64 outBlock[2])
         IGL128(t2, t1, &m_rkeys[0]);
     }
 
-#if defined(IS_BIG_ENDIAN)
-    {
-        typedef GetBlock<word64, LittleEndian, false> Block;
-        Block block(t1);
-        block(outBlock[0])(outBlock[1]);
-    }
-#endif
+    // Reverse bytes on BigEndian; Align pointer on LittleEndian
+    typedef PutBlock<word64, LittleEndian, false> OutBlock;
+    OutBlock oblk(xorBlock, outBlock);
+    oblk(t1[0])(t1[1]);
 }
 
-void Kalyna::Base::ProcessBlock_24(const word64 inBlock[2], word64 outBlock[2]) const
+void Kalyna::Base::ProcessBlock_24(const word64 inBlock[2], const word64 xorBlock[2], word64 outBlock[2]) const
 {
-#if defined(IS_BIG_ENDIAN)
     word64 *t1 = m_wspace+0, *t2 = m_wspace+2, *msg = m_wspace+4;
-    {
-        typedef GetBlock<word64, LittleEndian, false> Block;
-        Block block(inBlock);
-        block(msg[0])(msg[1]);
-        inBlock = msg;
-    }
-#else
-    word64 *t1 = outBlock, *t2 = m_wspace+2;
-#endif
 
+    // Reverse bytes on BigEndian; Align pointer on LittleEndian
+    typedef GetBlock<word64, LittleEndian, false> InBlock;
+    InBlock iblk(inBlock);
+    iblk(msg[0])(msg[1]);
+
+    inBlock = msg;
     if (IsForwardTransformation())
     {
         AddKey<2>(inBlock, t1, m_rkeys);
@@ -1034,29 +1023,22 @@ void Kalyna::Base::ProcessBlock_24(const word64 inBlock[2], word64 outBlock[2])
         IGL128(t2, t1, &m_rkeys[0]);
     }
 
-#if defined(IS_BIG_ENDIAN)
-    {
-        typedef GetBlock<word64, LittleEndian, false> Block;
-        Block block(t1);
-        block(outBlock[0])(outBlock[1]);
-    }
-#endif
+    // Reverse bytes on BigEndian; Align pointer on LittleEndian
+    typedef PutBlock<word64, LittleEndian, false> OutBlock;
+    OutBlock oblk(xorBlock, outBlock);
+    oblk(t1[0])(t1[1]);
 }
 
-void Kalyna::Base::ProcessBlock_44(const word64 inBlock[4], word64 outBlock[4]) const
+void Kalyna::Base::ProcessBlock_44(const word64 inBlock[4], const word64 xorBlock[4], word64 outBlock[4]) const
 {
-#if defined(IS_BIG_ENDIAN)
     word64 *t1 = m_wspace+0, *t2 = m_wspace+4, *msg = m_wspace+8;
-    {
-        typedef GetBlock<word64, LittleEndian, false> Block;
-        Block block(inBlock);
-        block(msg[0])(msg[1])(msg[2])(msg[3]);
-        inBlock = msg;
-    }
-#else
-    word64 *t1 = outBlock, *t2 = m_wspace+4;
-#endif
 
+    // Reverse bytes on BigEndian; Align pointer on LittleEndian
+    typedef GetBlock<word64, LittleEndian, false> InBlock;
+    InBlock iblk(inBlock);
+    iblk(msg[0])(msg[1])(msg[2])(msg[3]);
+
+    inBlock = msg;
     if (IsForwardTransformation())
     {
         AddKey<4>(inBlock, t1, m_rkeys);
@@ -1095,32 +1077,25 @@ void Kalyna::Base::ProcessBlock_44(const word64 inBlock[4], word64 outBlock[4])
         IGL256(t2, t1, &m_rkeys[0]);
     }
 
-#if defined(IS_BIG_ENDIAN)
-    {
-        typedef GetBlock<word64, LittleEndian, false> Block;
-        Block block(t1);
-        block(outBlock[0])(outBlock[1])(outBlock[2])(outBlock[3]);
-    }
-#endif
+    // Reverse bytes on BigEndian; Align pointer on LittleEndian
+    typedef PutBlock<word64, LittleEndian, false> OutBlock;
+    OutBlock oblk(xorBlock, outBlock);
+    oblk(t1[0])(t1[1])(t1[2])(t1[3]);
 }
 
-void Kalyna::Base::ProcessBlock_48(const word64 inBlock[4], word64 outBlock[4]) const
+void Kalyna::Base::ProcessBlock_48(const word64 inBlock[4], const word64 xorBlock[4], word64 outBlock[4]) const
 {
-#if defined(IS_BIG_ENDIAN)
     word64 *t1 = m_wspace+0, *t2 = m_wspace+4, *msg = m_wspace+8;
-    {
-        typedef GetBlock<word64, LittleEndian, false> Block;
-        Block block(inBlock);
-        block(msg[0])(msg[1])(msg[2])(msg[3]);
-        inBlock = msg;
-    }
-#else
-    word64 *t1 = outBlock, *t2 = m_wspace+4;
-#endif
 
+    // Reverse bytes on BigEndian; Align pointer on LittleEndian
+    typedef GetBlock<word64, LittleEndian, false> InBlock;
+    InBlock iblk(inBlock);
+    iblk(msg[0])(msg[1])(msg[2])(msg[3]);
+
+    inBlock = msg;
     if (IsForwardTransformation())
     {
-        AddKey<4>((const word64*)inBlock, t1, m_rkeys);
+        AddKey<4>(inBlock, t1, m_rkeys);
         G256(t1, t2, &m_rkeys[4]); // 1
         G256(t2, t1, &m_rkeys[8]); // 2
         G256(t1, t2, &m_rkeys[12]); // 3
@@ -1164,29 +1139,22 @@ void Kalyna::Base::ProcessBlock_48(const word64 inBlock[4], word64 outBlock[4])
         IGL256(t2, t1, &m_rkeys[0]);
     }
 
-#if defined(IS_BIG_ENDIAN)
-    {
-        typedef GetBlock<word64, LittleEndian, false> Block;
-        Block block(t1);
-        block(outBlock[0])(outBlock[1])(outBlock[2])(outBlock[3]);
-    }
-#endif
+    // Reverse bytes on BigEndian; Align pointer on LittleEndian
+    typedef PutBlock<word64, LittleEndian, false> OutBlock;
+    OutBlock oblk(xorBlock, outBlock);
+    oblk(t1[0])(t1[1])(t1[2])(t1[3]);
 }
 
-void Kalyna::Base::ProcessBlock_88(const word64 inBlock[8], word64 outBlock[8]) const
+void Kalyna::Base::ProcessBlock_88(const word64 inBlock[8], const word64 xorBlock[8], word64 outBlock[8]) const
 {
-#if defined(IS_BIG_ENDIAN)
     word64 *t1 = m_wspace+0, *t2 = m_wspace+8, *msg = m_wspace+16;
-    {
-        typedef GetBlock<word64, LittleEndian, false> Block;
-        Block block(inBlock);
-        block(msg[0])(msg[1])(msg[2])(msg[3])(msg[4])(msg[5])(msg[6])(msg[7]);
-        inBlock = msg;
-    }
-#else
-    word64 *t1 = outBlock, *t2 = m_wspace+8;
-#endif
 
+    // Reverse bytes on BigEndian; Align pointer on LittleEndian
+    typedef GetBlock<word64, LittleEndian, false> InBlock;
+    InBlock iblk(inBlock);
+    iblk(msg[0])(msg[1])(msg[2])(msg[3])(msg[4])(msg[5])(msg[6])(msg[7]);
+
+    inBlock = msg;
     if (IsForwardTransformation())
     {
         AddKey<8>(inBlock, t1, m_rkeys);
@@ -1233,14 +1201,10 @@ void Kalyna::Base::ProcessBlock_88(const word64 inBlock[8], word64 outBlock[8])
         IGL512(t2, t1, &m_rkeys[0]);
     }
 
-#if defined(IS_BIG_ENDIAN)
-    {
-        typedef GetBlock<word64, LittleEndian, false> Block;
-        Block block(t1);
-        block(outBlock[0])(outBlock[1])(outBlock[2])(outBlock[3])
-             (outBlock[4])(outBlock[5])(outBlock[6])(outBlock[7]);
-    }
-#endif
+    // Reverse bytes on BigEndian; Align pointer on LittleEndian
+    typedef PutBlock<word64, LittleEndian, false> OutBlock;
+    OutBlock oblk(xorBlock, outBlock);
+    oblk(t1[0])(t1[1])(t1[2])(t1[3])(t1[4])(t1[5])(t1[6])(t1[7]);
 }
 
 // *********************** Library routines *********************** //
@@ -1324,32 +1288,34 @@ void Kalyna::Base::ProcessAndXorBlock(const byte *inBlock, const byte *xorBlock,
 
     const byte* p = reinterpret_cast<const byte*>(KalynaTab::S);
     for (unsigned int i=0; i<256; i+=cacheLineSize)
-        u &= *reinterpret_cast<const word64*>(p+i);
+        u ^= *reinterpret_cast<const word64*>(p+i);
     m_wspace[0] = u;
 
     switch ((m_nb << 8) | m_nk)
     {
     case (2 << 8) | 2:
-        ProcessBlock_22(reinterpret_cast<const word64*>(inBlock), reinterpret_cast<word64*>(outBlock));
+        ProcessBlock_22(reinterpret_cast<const word64*>(inBlock),
+            reinterpret_cast<const word64*>(xorBlock), reinterpret_cast<word64*>(outBlock));
         break;
     case (2 << 8) | 4:
-        ProcessBlock_24(reinterpret_cast<const word64*>(inBlock), reinterpret_cast<word64*>(outBlock));
+        ProcessBlock_24(reinterpret_cast<const word64*>(inBlock),
+            reinterpret_cast<const word64*>(xorBlock), reinterpret_cast<word64*>(outBlock));
         break;
     case (4 << 8) | 4:
-        ProcessBlock_44(reinterpret_cast<const word64*>(inBlock), reinterpret_cast<word64*>(outBlock));
+        ProcessBlock_44(reinterpret_cast<const word64*>(inBlock),
+            reinterpret_cast<const word64*>(xorBlock), reinterpret_cast<word64*>(outBlock));
         break;
     case (4 << 8) | 8:
-        ProcessBlock_48(reinterpret_cast<const word64*>(inBlock), reinterpret_cast<word64*>(outBlock));
+        ProcessBlock_48(reinterpret_cast<const word64*>(inBlock),
+            reinterpret_cast<const word64*>(xorBlock), reinterpret_cast<word64*>(outBlock));
         break;
     case (8 << 8) | 8:
-        ProcessBlock_88(reinterpret_cast<const word64*>(inBlock), reinterpret_cast<word64*>(outBlock));
+        ProcessBlock_88(reinterpret_cast<const word64*>(inBlock),
+            reinterpret_cast<const word64*>(xorBlock), reinterpret_cast<word64*>(outBlock));
         break;
     default:
         CRYPTOPP_ASSERT(0);
     }
-
-    for (size_t i=0; xorBlock && i<m_nb; ++i)
-        outBlock[i] ^= xorBlock[i];
 }
 
 NAMESPACE_END
diff --git a/kalyna.h b/kalyna.h
index 1440a4bb..047685ef 100644
--- a/kalyna.h
+++ b/kalyna.h
@@ -134,11 +134,11 @@ public:
         // Visual Studio and C2910: 'Kalyna::Base::ProcessBlock_Template': cannot be explicitly specialized
         //template <unsigned int NB, unsigned int NK>
         //  void ProcessBlock_Template(const word64 inBlock[NB], const word64 outBlock[NB]) const;
-        void ProcessBlock_22(const word64 inBlock[2], word64 outBlock[2]) const;
-        void ProcessBlock_24(const word64 inBlock[2], word64 outBlock[2]) const;
-        void ProcessBlock_44(const word64 inBlock[4], word64 outBlock[4]) const;
-        void ProcessBlock_48(const word64 inBlock[4], word64 outBlock[4]) const;
-        void ProcessBlock_88(const word64 inBlock[8], word64 outBlock[8]) const;
+        void ProcessBlock_22(const word64 inBlock[2], const word64 xorBlock[2], word64 outBlock[2]) const;
+        void ProcessBlock_24(const word64 inBlock[2], const word64 xorBlock[2] ,word64 outBlock[2]) const;
+        void ProcessBlock_44(const word64 inBlock[4], const word64 xorBlock[4], word64 outBlock[4]) const;
+        void ProcessBlock_48(const word64 inBlock[4], const word64 xorBlock[4], word64 outBlock[4]) const;
+        void ProcessBlock_88(const word64 inBlock[8], const word64 xorBlock[8], word64 outBlock[8]) const;
 
     private:
         typedef SecBlock<word64, AllocatorWithCleanup<word64, true> > AlignedSecBlock64;