runtests.sh checking sha256sums of test vectors, fixed namespacing fo…

…r keccak4x

.gitignore

@@ -1,6 +1,6 @@
-*.dSYM
-*.pyc
-.DS_Store
-*.req
-*.rsp
-*.o
+tvecs1024
+tvecs1024-90s
+tvecs512
+tvecs512-90s
+tvecs768
+tvecs768-90s

SHA256SUMS

@@ -0,0 +1,6 @@
+2d46d36da383c3ece6fc46512162dabb73bf73a92ee90f29fbf990873766c223  tvecs1024
+761e10eda09e1f02204ca5484c2a9f908e8866ade289c1deabe9d0769233df67  tvecs1024-90s
+005c74399ff14d2244905f089c6d408a6bb33828bc78d5b77a3a01b227affbca  tvecs512
+71099d65a79fc1881836dd7cf8802fcce5d81cd64acb35c80102090d800bf131  tvecs512-90s
+cca4a24c53f7e8baddddd63bff32ad1e0b1590c742b545e0f682cb4202ac2051  tvecs768
+0bf24733d4ee523b60e43a25e85b577eba7ccc6109ba57b798904dd8d08828c0  tvecs768-90s

avx2/.gitignore

@@ -0,0 +1 @@
+../ref/.gitignore

avx2/Makefile

@@ -1,9 +1,9 @@
 CC = /usr/bin/cc
 CFLAGS += -Wall -Wextra -Wpedantic -Wmissing-prototypes -Wredundant-decls \
   -Wshadow -Wpointer-arith -march=native -mtune=native -O3 \
-  -fomit-frame-pointer -fstrict-aliasing -flto
+  -fomit-frame-pointer -flto
 #CFLAGS += -DUSE_RDPMC
-NISTFLAGS = -Wno-unused-result -O3 -fomit-frame-pointer -march=native -mtune=native
+NISTFLAGS = -Wno-unused-result -O3 -march=native -mtune=native
 RM = /bin/rm
 
 SOURCES = kem.c indcpa.c polyvec.c poly.c fq.S shuffle.S ntt.S invntt.S \
@@ -16,6 +16,8 @@ HEADERS = params.h api.h indcpa.h polyvec.h poly.h reduce.h fq.inc shuffle.inc \
 HEADERSKECCAK   = $(HEADERS) fips202.h fips202x4.h
 HEADERSNINETIES = $(HEADERS) aes256ctr.h
 
+.PHONY: all shared clean
+
 all: \
   test_kyber512 \
   test_kyber768 \
@@ -43,17 +45,48 @@ all: \
   test_speed1024-90s \
   PQCgenKAT_kem
 
-kyber512.so: $(SOURCES) $(HEADERS) symmetric-shake.c
+shared: \
+  kyber512_avx2.so \
+  kyber768_avx2.so \
+  kyber1024_avx2.so \
+  kyber512-90s_avx2.so \
+  kyber768-90s_avx2.so \
+  kyber1024-90s_avx2.so \
+  pqcrystals_ref.so \
+  pqcrystals_avx2.so \
+  pqcrystals_avx2_aes256ctr.so
+
+keccak4x/KeccakP-1600-times4-SIMD256.o: \
+  keccak4x/KeccakP-1600-times4-SIMD256.c \
+  keccak4x/KeccakP-1600-times4-SnP.h \
+  keccak4x/KeccakP-1600-unrolling.macros \
+  keccak4x/SIMD256-config.h \
+  keccak4x/align.h \
+  keccak4x/brg_endian.h
+	$(CC) $(CFLAGS) -c $< -o $@
+
+pqcrystals_ref.so: fips202.c fips202.h
+	$(CC) -shared -fPIC $(CFLAGS) -o $@ $<
+
+pqcrystals_avx2.so: fips202x4.c fips202x4.h \
+  keccak4x/KeccakP-1600-times4-SIMD256.o
+	$(CC) -shared -fPIC $(CFLAGS) -o $@ $< \
+	  keccak4x/KeccakP-1600-times4-SIMD256.o
+
+pqcrystals_avx2_aes256ctr.so: aes256ctr.c aes256ctr.h
+	$(CC) -shared -fPIC $(CFLAGS) -o $@ $<
+
+kyber512_avx2.so: $(SOURCES) $(HEADERS) symmetric-shake.c
 	$(CC) -shared -fpic $(CFLAGS) -DKYBER_K=2 $(SOURCES) \
-	  symmetric-shake.c -o kyber512.so
+	  symmetric-shake.c -o kyber512_avx2.so
 
-kyber768.so: $(SOURCES) $(HEADERS) symmetric-shake.c
+kyber768_avx2.so: $(SOURCES) $(HEADERS) symmetric-shake.c
 	$(CC) -shared -fpic $(CFLAGS) -DKYBER_K=3 $(SOURCES) \
-	  symmetric-shake.c -o kyber768.so
+	  symmetric-shake.c -o kyber768_avx2.so
 
-kyber1024.so: $(SOURCES) $(HEADERS) symmetric-shake.c
+kyber1024_avx2.so: $(SOURCES) $(HEADERS) symmetric-shake.c
 	$(CC) -shared -fpic $(CFLAGS) -DKYBER_K=4 $(SOURCES) \
-	  symmetric-shake.c -o kyber1024.so
+	  symmetric-shake.c -o kyber1024_avx2.so
 
 test_kyber512: $(SOURCESKECCAK) $(HEADERSKECCAK) test_kyber.c randombytes.c
 	$(CC) $(CFLAGS) -DKYBER_K=2 $(SOURCESKECCAK) randombytes.c test_kyber.c -o test_kyber512
@@ -82,7 +115,7 @@ test_vectors768: $(SOURCESKECCAK) $(HEADERSKECCAK) test_vectors.c
 test_vectors1024: $(SOURCESKECCAK) $(HEADERSKECCAK) test_vectors.c
 	$(CC) $(CFLAGS) -DKYBER_K=4 $(SOURCESKECCAK) test_vectors.c -o test_vectors1024
 
-test_speed512: $(SOURCESKECCAK) $(HEADERSKECCAK) cpucycles.h cpucycles.c test_speed.c speed_print.h speed_print.c randombytes.c kex.c kex.h
+test_speed512: $(SOURCESKECCAK) $(HEADERSKECCAK) cpucycles.h cpucycles.c speed_print.h speed_print.c test_speed.c randombytes.c kex.c kex.h
 	$(CC) $(CFLAGS) -DKYBER_K=2 $(SOURCESKECCAK) randombytes.c cpucycles.c speed_print.c kex.c test_speed.c -o test_speed512
 
 test_speed768: $(SOURCESKECCAK) $(HEADERSKECCAK) cpucycles.h cpucycles.c speed_print.h speed_print.c test_speed.c randombytes.c kex.c kex.h
@@ -91,17 +124,14 @@ test_speed768: $(SOURCESKECCAK) $(HEADERSKECCAK) cpucycles.h cpucycles.c speed_p
 test_speed1024: $(SOURCESKECCAK) $(HEADERSKECCAK) cpucycles.h cpucycles.c speed_print.h speed_print.c test_speed.c randombytes.c kex.c kex.h
 	$(CC) $(CFLAGS) -DKYBER_K=4 $(SOURCESKECCAK) randombytes.c cpucycles.c speed_print.c kex.c test_speed.c -o test_speed1024
 
-kyber512-90s.so: $(SOURCES) $(HEADERS)
-	$(CC) -shared -fpic $(CFLAGS) -DKYBER_K=2 -DKYBER_90S $(SOURCES) \
-	  -o kyber512-90s.so
+kyber512-90s_avx2.so: $(SOURCES) $(HEADERS)
+	$(CC) -shared -fpic $(CFLAGS) -DKYBER_K=2 -DKYBER_90S $(SOURCES) -o kyber512-90s_avx2.so
 
-kyber768-90s.so: $(SOURCES) $(HEADERS)
-	$(CC) -shared -fpic $(CFLAGS) -DKYBER_K=3 -DKYBER_90S $(SOURCES) \
-	  -o kyber768-90s.so
+kyber768-90s_avx2.so: $(SOURCES) $(HEADERS)
+	$(CC) -shared -fpic $(CFLAGS) -DKYBER_K=3 -DKYBER_90S $(SOURCES) -o kyber768-90s_avx2.so
 
-kyber1024-90s.so: $(SOURCES) $(HEADERS)
-	$(CC) -shared -fpic $(CFLAGS) -DKYBER_K=4 -DKYBER_90S $(SOURCES) \
-	  -o kyber1024-90s.so
+kyber1024-90s_avx2.so: $(SOURCES) $(HEADERS)
+	$(CC) -shared -fpic $(CFLAGS) -DKYBER_K=4 -DKYBER_90S $(SOURCES) -o kyber1024-90s_avx2.so
 
 test_kyber512-90s: $(SOURCESNINETIES) $(HEADERSNINETIES) test_kyber.c randombytes.c
 	$(CC) $(CFLAGS) -D KYBER_90S -DKYBER_K=2 $(SOURCESNINETIES) randombytes.c test_kyber.c -o test_kyber512-90s -lcrypto
@@ -142,16 +172,11 @@ test_speed1024-90s: $(SOURCESNINETIES) $(HEADERSNINETIES) cpucycles.h cpucycles.
 PQCgenKAT_kem: $(SOURCESKECCAK) $(HEADERSKECCAK) PQCgenKAT_kem.c rng.c rng.h
 	$(CC) $(NISTFLAGS) -o $@ $(SOURCESKECCAK) -I. rng.c PQCgenKAT_kem.c -lcrypto
 
-test_ntt: $(SOURCESKECCAK) $(HEADERSKECCAK) test_ntt.c randombytes.c
-	$(CC) $(CFLAGS) -DKYBER_K=2 $(SOURCESKECCAK) randombytes.c test_ntt.c -o test_ntt
-
-.PHONY: clean test
-
 clean:
 	-$(RM) -rf *.o
-	-$(RM) -rf kyber512.so
-	-$(RM) -rf kyber768.so
-	-$(RM) -rf kyber1024.so
+	-$(RM) -rf kyber512_avx2.so
+	-$(RM) -rf kyber768_avx2.so
+	-$(RM) -rf kyber1024_avx2.so
 	-$(RM) -rf test_kyber512
 	-$(RM) -rf test_kyber768
 	-$(RM) -rf test_kyber1024
@@ -164,9 +189,9 @@ clean:
 	-$(RM) -rf test_speed512
 	-$(RM) -rf test_speed768
 	-$(RM) -rf test_speed1024
-	-$(RM) -rf kyber512-90s.so
-	-$(RM) -rf kyber768-90s.so
-	-$(RM) -rf kyber1024-90s.so
+	-$(RM) -rf kyber512-90s_avx2.so
+	-$(RM) -rf kyber768-90s_avx2.so
+	-$(RM) -rf kyber1024-90s_avx2.so
 	-$(RM) -rf test_kyber512-90s
 	-$(RM) -rf test_kyber768-90s
 	-$(RM) -rf test_kyber1024-90s
@@ -180,3 +205,7 @@ clean:
 	-$(RM) -rf test_speed768-90s
 	-$(RM) -rf test_speed1024-90s
 	-$(RM) -rf PQCgenKAT_kem
+	-$(RM) -rf pqcrystals_ref.so
+	-$(RM) -rf pqcrystals_avx2.so
+	-$(RM) -rf pqcrystals_avx2_aes256ctr.so
+	-$(RM) -rf keccak4x/KeccakP-1600-times4-SIMD256.o

avx2/fips202x4.c

@@ -6,8 +6,8 @@
 #include "fips202x4.h"
 
 /* Use implementation from the Keccak Code Package */
-extern void KeccakP1600times4_PermuteAll_24rounds(__m256i *s);
-#define KeccakF1600_StatePermute4x KeccakP1600times4_PermuteAll_24rounds
+extern void pqcrystals_avx2_KeccakP1600times4_PermuteAll_24rounds(__m256i *s);
+#define KeccakF1600_StatePermute4x pqcrystals_avx2_KeccakP1600times4_PermuteAll_24rounds
 
 static inline uint64_t load64(const uint8_t x[8]) {
   unsigned int i;

avx2/fips202x4.h

@@ -41,6 +41,7 @@ void shake256x4_squeezeblocks(uint8_t *out0,
                               size_t nblocks,
                               keccakx4_state *state);
 
+#define shake128x4 pqcrystals_avx2_shake128x4
 void shake128x4(uint8_t *out0,
                 uint8_t *out1,
                 uint8_t *out2,
@@ -52,6 +53,7 @@ void shake128x4(uint8_t *out0,
                 const uint8_t *in3,
                 size_t inlen);
 
+#define shake256x4 pqcrystals_avx2_shake256x4
 void shake256x4(uint8_t *out0,
                 uint8_t *out1,
                 uint8_t *out2,

avx2/keccak4x/KeccakP-1600-times4-SnP.h

@@ -30,21 +30,35 @@ and related or neighboring rights to the source code in this file.
 #include <stddef.h>
 
 #define KeccakP1600times4_StaticInitialize()
+#define KeccakP1600times4_InitializeAll pqcrystals_avx2_KeccakP1600times4_InitializeAll
 void KeccakP1600times4_InitializeAll(void *states);
 #define KeccakP1600times4_AddByte(states, instanceIndex, byte, offset) \
     ((unsigned char*)(states))[(instanceIndex)*8 + ((offset)/8)*4*8 + (offset)%8] ^= (byte)
+#define KeccakP1600times4_AddBytes pqcrystals_avx2_KeccakP1600times4_AddBytes
 void KeccakP1600times4_AddBytes(void *states, unsigned int instanceIndex, const unsigned char *data, unsigned int offset, unsigned int length);
+#define KeccakP1600times4_AddLanesAll pqcrystals_avx2_KeccakP1600times4_AddLanesAll
 void KeccakP1600times4_AddLanesAll(void *states, const unsigned char *data, unsigned int laneCount, unsigned int laneOffset);
+#define KeccakP1600times4_OverwriteBytes pqcrystals_avx2_KeccakP1600times4_OverwriteBytes
 void KeccakP1600times4_OverwriteBytes(void *states, unsigned int instanceIndex, const unsigned char *data, unsigned int offset, unsigned int length);
+#define KeccakP1600times4_OverwriteLanesAll pqcrystals_avx2_KeccakP1600times4_OverwriteLanesAll
 void KeccakP1600times4_OverwriteLanesAll(void *states, const unsigned char *data, unsigned int laneCount, unsigned int laneOffset);
+#define KeccakP1600times4_OverwriteWithZeroes pqcrystals_avx2_KeccakP1600times4_OverwriteWithZeroes
 void KeccakP1600times4_OverwriteWithZeroes(void *states, unsigned int instanceIndex, unsigned int byteCount);
+#define KeccakP1600times4_PermuteAll_12rounds pqcrystals_avx2_KeccakP1600times4_PermuteAll_12rounds
 void KeccakP1600times4_PermuteAll_12rounds(void *states);
+#define KeccakP1600times4_PermuteAll_24rounds pqcrystals_avx2_KeccakP1600times4_PermuteAll_24rounds
 void KeccakP1600times4_PermuteAll_24rounds(void *states);
+#define KeccakP1600times4_ExtractBytes pqcrystals_avx2_KeccakP1600times4_ExtractBytes
 void KeccakP1600times4_ExtractBytes(const void *states, unsigned int instanceIndex, unsigned char *data, unsigned int offset, unsigned int length);
+#define KeccakP1600times4_ExtractLanesAll pqcrystals_avx2_KeccakP1600times4_ExtractLanesAll
 void KeccakP1600times4_ExtractLanesAll(const void *states, unsigned char *data, unsigned int laneCount, unsigned int laneOffset);
+#define KeccakP1600times4_ExtractAndAddBytes pqcrystals_avx2_KeccakP1600times4_ExtractAndAddBytes
 void KeccakP1600times4_ExtractAndAddBytes(const void *states, unsigned int instanceIndex,  const unsigned char *input, unsigned char *output, unsigned int offset, unsigned int length);
+#define KeccakP1600times4_ExtractAndAddLanesAll pqcrystals_avx2_KeccakP1600times4_ExtractAndAddLanesAll
 void KeccakP1600times4_ExtractAndAddLanesAll(const void *states, const unsigned char *input, unsigned char *output, unsigned int laneCount, unsigned int laneOffset);
+#define KeccakF1600times4_FastLoop_Absorb pqcrystals_avx2_KeccakF1600times4_FastLoop_Absorb
 size_t KeccakF1600times4_FastLoop_Absorb(void *states, unsigned int laneCount, unsigned int laneOffsetParallel, unsigned int laneOffsetSerial, const unsigned char *data, size_t dataByteLen);
+#define KeccakP1600times4_12rounds_FastLoop_Absorb pqcrystals_avx2_KeccakP1600times4_12rounds_FastLoop_Absorb
 size_t KeccakP1600times4_12rounds_FastLoop_Absorb(void *states, unsigned int laneCount, unsigned int laneOffsetParallel, unsigned int laneOffsetSerial, const unsigned char *data, size_t dataByteLen);
 
 #endif

ref/.gitignore

@@ -0,0 +1,40 @@
+kyber1024-90s_ref.so
+kyber1024_ref.so
+kyber512-90s_ref.so
+kyber512_ref.so
+kyber768-90s_ref.so
+kyber768_ref.so
+pqcrystals_ref.so
+test_kex1024
+test_kex1024-90s
+test_kex512
+test_kex512-90s
+test_kex768
+test_kex768-90s
+test_kyber1024
+test_kyber1024-90s
+test_kyber512
+test_kyber512-90s
+test_kyber768
+test_kyber768-90s
+test_speed1024
+test_speed1024-90s
+test_speed512
+test_speed512-90s
+test_speed768
+test_speed768-90s
+test_vectors1024
+test_vectors1024-90s
+test_vectors512
+test_vectors512-90s
+test_vectors768
+test_vectors768-90s
+PQCgenKAT_kem
+kyber1024-90s_avx2.so
+kyber1024_avx2.so
+kyber512-90s_avx2.so
+kyber512_avx2.so
+kyber768-90s_avx2.so
+kyber768_avx2.so
+pqcrystals_avx2.so
+pqcrystals_avx2_aes256ctr.so