From 2876371cea69139379dea4ec54f3dbd6f6d8699e Mon Sep 17 00:00:00 2001
From: Jeffrey Walton <noloader@gmail.com>
Date: Thu, 16 Aug 2018 18:24:22 -0400
Subject: [PATCH] Add GNUmakefile-cross flags SIMON and SPECK

---
 GNUmakefile-cross | 46 ++++++++++++++++++++++++++++++----------------
 blake2-simd.cpp   |  4 ++++
 2 files changed, 34 insertions(+), 16 deletions(-)

diff --git a/GNUmakefile-cross b/GNUmakefile-cross
index 7692d443..b688f192 100755
--- a/GNUmakefile-cross
+++ b/GNUmakefile-cross
@@ -264,8 +264,10 @@ ifeq ($(IS_NEON),1)
     BLAKE2_FLAG += -mfpu=neon
     LEA_FLAG += -mfpu=neon
     SIMECK_FLAG += -mfpu=neon
-    SIMON_FLAG += -mfpu=neon
-    SPECK_FLAG += -mfpu=neon
+    SIMON64_FLAG += -mfpu=neon
+    SIMON128_FLAG += -mfpu=neon
+    SPECK64_FLAG += -mfpu=neon
+    SPECK128_FLAG += -mfpu=neon
     ifeq ($(IS_ANDROID),1)
       ifeq ($(findstring -mfloat-abi=softfp,$(CXXFLAGS)),)
         NEON_FLAG += -mfloat-abi=softfp
@@ -274,8 +276,10 @@ ifeq ($(IS_NEON),1)
         BLAKE2_FLAG += -mfloat-abi=softfp
         LEA_FLAG += -mfloat-abi=softfp
         SIMECK_FLAG += -mfloat-abi=softfp
-        SIMON_FLAG += -mfloat-abi=softfp
-        SPECK_FLAG += -mfloat-abi=softfp
+        SIMON64_FLAG += -mfloat-abi=softfp
+        SIMON128_FLAG += -mfloat-abi=softfp
+        SPECK64_FLAG += -mfloat-abi=softfp
+        SPECK128_FLAG += -mfloat-abi=softfp
       endif
     endif
   endif
@@ -290,8 +294,10 @@ ifneq ($(IS_ARMv8),0)
     LEA_FLAG = -march=armv8-a
     NEON_FLAG = -march=armv8-a
     SIMECK_FLAG = -march=armv8-a
-    SIMON_FLAG = -march=armv8-a
-    SPECK_FLAG = -march=armv8-a
+    SIMON64_FLAG = -march=armv8-a
+    SIMON128_FLAG = -march=armv8-a
+    SPECK64_FLAG = -march=armv8-a
+    SPECK128_FLAG = -march=armv8-a
   endif
   HAVE_CRC := $(shell $(CXX) $(CXXFLAGS) -DADHOC_MAIN -march=armv8-a+crc -dM -E adhoc.cpp 2>&1 | $(EGREP) -i -c __ARM_FEATURE_CRC32)
   ifeq ($(HAVE_CRC),1)
@@ -318,13 +324,13 @@ ifneq ($(IS_i686)$(IS_x86_64),00)
     LEA_FLAG = -mssse3
     SSSE3_FLAG = -mssse3
     SIMECK_FLAG = -mssse3
-    SIMON_FLAG = -mssse3
-    SPECK_FLAG = -mssse3
+    SIMON128_FLAG = -mssse3
+    SPECK128_FLAG = -mssse3
   endif
   HAVE_SSE4 = $(shell $(CXX) $(CXXFLAGS) -DADHOC_MAIN -msse4.1 -dM -E adhoc.cpp 2>&1 | $(EGREP) -i -c __SSE4_1__)
   ifeq ($(HAVE_SSE4),1)
-    SIMON_FLAG = -msse4.1
-    SPECK_FLAG = -msse4.1
+    SIMON64_FLAG = -msse4.1
+    SPECK64_FLAG = -msse4.1
   endif
   HAVE_SSE4 = $(shell $(CXX) $(CXXFLAGS) -DADHOC_MAIN -msse4.2 -dM -E adhoc.cpp 2>&1 | $(EGREP) -i -c __SSE4_2__)
   ifeq ($(HAVE_SSE4),1)
@@ -604,13 +610,21 @@ shacal2-simd.o : shacal2-simd.cpp
 simeck-simd.o : simeck-simd.cpp
 	$(CXX) $(strip $(CXXFLAGS) $(SIMECK_FLAG) -c) $<
 
-# SSSE3 or NEON available
-simon-simd.o : simon-simd.cpp
-	$(CXX) $(strip $(CXXFLAGS) $(SIMON_FLAG) -c) $<
+# SSE4.1, NEON or POWER7 available
+simon64-simd.o : simon64-simd.cpp
+	$(CXX) $(strip $(CXXFLAGS) $(SIMON64_FLAG) -c) $<
 
-# SSSE3 or NEON available
-speck-simd.o : speck-simd.cpp
-	$(CXX) $(strip $(CXXFLAGS) $(SPECK_FLAG) -c) $<
+# SSSE3, NEON or POWER8 available
+simon128-simd.o : simon128-simd.cpp
+	$(CXX) $(strip $(CXXFLAGS) $(SIMON128_FLAG) -c) $<
+
+# SSE4.1, NEON or POWER7 available
+speck64-simd.o : speck64-simd.cpp
+	$(CXX) $(strip $(CXXFLAGS) $(SPECK64_FLAG) -c) $<
+
+# SSSE3, NEON or POWER8 available
+speck128-simd.o : speck128-simd.cpp
+	$(CXX) $(strip $(CXXFLAGS) $(SPECK128_FLAG) -c) $<
 
 # AESNI available
 sm4-simd.o : sm4-simd.cpp
diff --git a/blake2-simd.cpp b/blake2-simd.cpp
index 1aea7bdc..6646b166 100644
--- a/blake2-simd.cpp
+++ b/blake2-simd.cpp
@@ -342,6 +342,7 @@ void BLAKE2_Compress32_SSE4(const byte* input, BLAKE2_State<word32, false>& stat
     row2 = ff1 = LOADU( &state.h[4] );
     row3 = LOADU( &BLAKE2S_IV[0] );
     row4 = _mm_xor_si128( LOADU( &BLAKE2S_IV[4] ), LOADU( &state.t[0] ) );
+
     BLAKE2S_ROUND( 0 );
     BLAKE2S_ROUND( 1 );
     BLAKE2S_ROUND( 2 );
@@ -352,6 +353,7 @@ void BLAKE2_Compress32_SSE4(const byte* input, BLAKE2_State<word32, false>& stat
     BLAKE2S_ROUND( 7 );
     BLAKE2S_ROUND( 8 );
     BLAKE2S_ROUND( 9 );
+
     STOREU( &state.h[0], _mm_xor_si128( ff0, _mm_xor_si128( row1, row3 ) ) );
     STOREU( &state.h[4], _mm_xor_si128( ff1, _mm_xor_si128( row2, row4 ) ) );
 }
@@ -752,6 +754,7 @@ void BLAKE2_Compress64_SSE4(const byte* input, BLAKE2_State<word64, true>& state
     row3h = LOADU( &BLAKE2B_IV[2] );
     row4l = _mm_xor_si128( LOADU( &BLAKE2B_IV[4] ), LOADU( &state.t[0] ) );
     row4h = _mm_xor_si128( LOADU( &BLAKE2B_IV[6] ), LOADU( &state.f[0] ) );
+
     BLAKE2B_ROUND( 0 );
     BLAKE2B_ROUND( 1 );
     BLAKE2B_ROUND( 2 );
@@ -764,6 +767,7 @@ void BLAKE2_Compress64_SSE4(const byte* input, BLAKE2_State<word64, true>& state
     BLAKE2B_ROUND( 9 );
     BLAKE2B_ROUND( 10 );
     BLAKE2B_ROUND( 11 );
+
     row1l = _mm_xor_si128( row3l, row1l );
     row1h = _mm_xor_si128( row3h, row1h );
     STOREU( &state.h[0], _mm_xor_si128( LOADU( &state.h[0] ), row1l ) );