Add M55 native implementations.

Author: bremoran

Makefile

@@ -2,14 +2,14 @@
 # Copyright (c) The mldsa-native project authors
 # SPDX-License-Identifier: Apache-2.0 OR ISC OR MIT
 
-.PHONY: func kat acvp stack \
+.PHONY: func kat acvp stack keccak \
 	func_44 kat_44 acvp_44 stack_44 \
-	func_65 kat_65 acvp_65 stack_65 \
-	func_87 kat_87 acvp_87 stack_87 \
-	run_func run_kat run_acvp run_stack \
-	run_func_44 run_kat_44 run_stack_44 \
-	run_func_65 run_kat_65 run_stack_65 \
-	run_func_87 run_kat_87 run_stack_87 \
+	func_65 kat_65 acvp_65 stack_65 keccak_65 \
+	func_87 kat_87 acvp_87 stack_87 keccak_87 \
+	run_func run_kat run_acvp run_stack run_keccak \
+	run_func_44 run_kat_44 run_stack_44 run_keccak_44 \
+	run_func_65 run_kat_65 run_stack_65 run_keccak_65 \
+	run_func_87 run_kat_87 run_stack_87 run_keccak_87 \
 	bench_44 bench_65 bench_87 bench \
 	run_bench_44 run_bench_65 run_bench_87 run_bench \
 	bench_components_44 bench_components_65 bench_components_87 bench_components \
@@ -108,6 +108,23 @@ run_stack_87: stack_87
 	$(Q)python3 scripts/stack $(MLDSA87_DIR)/bin/test_stack87 --build-dir $(MLDSA87_DIR) $(STACK_ANALYSIS_FLAGS)
 run_stack: run_stack_44 run_stack_65 run_stack_87
 
+# Keccak tests
+keccak_44: $(MLDSA44_DIR)/bin/test_keccak44
+	$(Q)echo "  KECCAK    ML-DSA-44:   $^"
+keccak_65: $(MLDSA65_DIR)/bin/test_keccak65
+	$(Q)echo "  KECCAK    ML-DSA-65:   $^"
+keccak_87: $(MLDSA87_DIR)/bin/test_keccak87
+	$(Q)echo "  KECCAK    ML-DSA-87:  $^"
+keccak: keccak_44 keccak_65 keccak_87
+
+run_keccak_44: keccak_44
+	$(W) $(MLDSA44_DIR)/bin/test_keccak44
+run_keccak_65: keccak_65
+	$(W) $(MLDSA65_DIR)/bin/test_keccak65
+run_keccak_87: keccak_87
+	$(W) $(MLDSA87_DIR)/bin/test_keccak87
+run_keccak: run_keccak_44 run_keccak_65 run_keccak_87
+
 lib: $(BUILD_DIR)/libmldsa.a $(BUILD_DIR)/libmldsa44.a $(BUILD_DIR)/libmldsa65.a $(BUILD_DIR)/libmldsa87.a
 
 # Enforce setting CYCLES make variable when

mldsa/fips202/keccakf1600.c

@@ -40,7 +40,10 @@ void mld_keccakf1600_extract_bytes(uint64_t *state, unsigned char *data,
                                    unsigned offset, unsigned length)
 {
   unsigned i;
-#if defined(MLD_SYS_LITTLE_ENDIAN)
+#if defined(MLD_USE_FIPS202_X1_NATIVE)
+  (void) i;
+  mld_keccakf1600_extract_bytes_native(state, data, offset, length);
+#elif defined(MLD_SYS_LITTLE_ENDIAN)
   uint8_t *state_ptr = (uint8_t *)state + offset;
   for (i = 0; i < length; i++)
   __loop__(invariant(i <= length))
@@ -61,7 +64,10 @@ void mld_keccakf1600_xor_bytes(uint64_t *state, const unsigned char *data,
                                unsigned offset, unsigned length)
 {
   unsigned i;
-#if defined(MLD_SYS_LITTLE_ENDIAN)
+#if defined(MLD_USE_FIPS202_X1_NATIVE)
+  (void) i;
+  mld_keccakf1600_xor_bytes_native(state, data, offset, length);
+#elif defined(MLD_SYS_LITTLE_ENDIAN)
   uint8_t *state_ptr = (uint8_t *)state + offset;
   for (i = 0; i < length; i++)
   __loop__(invariant(i <= length))
@@ -84,6 +90,12 @@ void mld_keccakf1600x4_extract_bytes(uint64_t *state, unsigned char *data0,
                                      unsigned char *data3, unsigned offset,
                                      unsigned length)
 {
+#if defined(MLD_USE_FIPS202_X4_XOR_NATIVE)
+  mld_keccakf1600_extract_bytes_x4_native(state, data0,
+                                          data1, data2,
+                                          data3, offset,
+                                          length);
+#else /* MLD_USE_FIPS202_X4_XOR_NATIVE */
   mld_keccakf1600_extract_bytes(state + MLD_KECCAK_LANES * 0, data0, offset,
                                 length);
   mld_keccakf1600_extract_bytes(state + MLD_KECCAK_LANES * 1, data1, offset,
@@ -92,6 +104,7 @@ void mld_keccakf1600x4_extract_bytes(uint64_t *state, unsigned char *data0,
                                 length);
   mld_keccakf1600_extract_bytes(state + MLD_KECCAK_LANES * 3, data3, offset,
                                 length);
+#endif /* !MLD_USE_FIPS202_X4_NATIVE && !MLD_USE_FIPS202_X2_NATIVE */
 }
 
 void mld_keccakf1600x4_xor_bytes(uint64_t *state, const unsigned char *data0,
@@ -100,6 +113,13 @@ void mld_keccakf1600x4_xor_bytes(uint64_t *state, const unsigned char *data0,
                                  const unsigned char *data3, unsigned offset,
                                  unsigned length)
 {
+#if defined(MLD_USE_FIPS202_X4_XOR_NATIVE)
+  mld_keccakf1600_xor_bytes_x4_native(state, data0,
+                                      data1,
+                                      data2,
+                                      data3, offset,
+                                      length);
+#else /* MLD_USE_FIPS202_X4_XOR_NATIVE */
   mld_keccakf1600_xor_bytes(state + MLD_KECCAK_LANES * 0, data0, offset,
                             length);
   mld_keccakf1600_xor_bytes(state + MLD_KECCAK_LANES * 1, data1, offset,
@@ -108,6 +128,7 @@ void mld_keccakf1600x4_xor_bytes(uint64_t *state, const unsigned char *data0,
                             length);
   mld_keccakf1600_xor_bytes(state + MLD_KECCAK_LANES * 3, data3, offset,
                             length);
+#endif /* MLD_USE_FIPS202_X4_XOR_NATIVE */
 }
 
 void mld_keccakf1600x4_permute(uint64_t *state)

mldsa/fips202/native/armv8.1-m/auto.h

@@ -0,0 +1,13 @@
+/*
+ * Copyright (c) The mlkem-native project authors
+ * Copyright (c) The mldsa-native project authors
+ * SPDX-License-Identifier: Apache-2.0 OR ISC OR MIT
+ */
+
+#ifndef MLD_FIPS202_NATIVE_ARMV81M_AUTO_H
+#define MLD_FIPS202_NATIVE_ARMV81M_AUTO_H
+/* Default FIPS202 assembly profile for ARMV81M systems */
+
+#include "x4_v8m_mve.h"
+
+#endif /* !MLD_FIPS202_NATIVE_ARMV81M_AUTO_H */