Merge branch 'mppa-cse2' of gricad-gitlab.univ-grenoble-alpes.fr:sixcy/CompCert into mppa-work

7 files changed, 869 insertions, 0 deletions

diff --git a/test/monniaux/sandbox/Makefile b/test/monniaux/sandbox/Makefile
new file mode 100644
index 00000000..0fa2a2ae
--- /dev/null
+++ b/test/monniaux/sandbox/Makefile
@@ -0,0 +1,148 @@
+# This Makefile does not depend on ../rules.mk
+SHELL=bash
+
+# You can modify ALL_CFILES to include the files that should be linked
+ALL_CFILES=$(wildcard *.c)
+
+# Name of the target
+TARGET=toto
+
+# Name of the clock object
+CLOCK=../clock.gcc.k1c.o
+
+# Maximum amount of time measures (see cycles.h)
+MAX_MEASURES=10
+
+# Flags common to both compilers, then to gcc, then to ccomp
+ALL_CFLAGS=-Wall -D__K1C_COS__ -DMAX_MEASURES=$(MAX_MEASURES)
+#ALL_CFLAGS+=-g
+ALL_GCCFLAGS=$(ALL_CFLAGS) -std=c99 -Wextra -Werror=implicit
+ALL_CCOMPFLAGS=$(ALL_CFLAGS)
+
+# The compilers
+K1C_CC=k1-cos-gcc
+K1C_CCOMP=ccomp
+
+# Command to execute
+EXECUTE_CYCLES=k1-cluster --syscall=libstd_scalls.so --cycle-based --
+
+# You can define up to GCC4FLAGS and CCOMP4FLAGS
+GCC0FLAGS=
+GCC1FLAGS=$(ALL_GCCFLAGS) -O1
+GCC2FLAGS=$(ALL_GCCFLAGS) -O2
+GCC3FLAGS=$(ALL_GCCFLAGS) -O3
+GCC4FLAGS=
+CCOMP0FLAGS=
+CCOMP1FLAGS=$(ALL_CCOMPFLAGS) -fno-postpass
+CCOMP2FLAGS=$(ALL_CCOMPFLAGS)
+CCOMP3FLAGS=
+CCOMP4FLAGS=
+
+# Prefix names
+GCC0PREFIX=
+GCC1PREFIX=.gcc.o1
+GCC2PREFIX=.gcc.o2
+GCC3PREFIX=.gcc.o3
+GCC4PREFIX=
+CCOMP0PREFIX=.ccomp.o0
+CCOMP1PREFIX=.ccomp.o1
+CCOMP2PREFIX=.ccomp.o2
+CCOMP3PREFIX=
+CCOMP4PREFIX=
+
+# List of outfiles, updated by gen_rules
+OUTFILES:=
+BINFILES:=
+
+# First line of the CSV file
+FIRSTLINE:=benches
+
+firstrule: all
+
+# $1: compiler
+# $2: compilation flags
+# $3: extension prefix
+define gen_rules
+
+.SECONDARY:
+asm/%$(3).s: %.c
+	@mkdir -p $$(@D)
+	$(1) $(2) -S $$< -o $$@
+
+.SECONDARY:
+bin/$(TARGET)$(3).bin: $(addprefix obj/,$(ALL_CFILES:.c=$(3).o)) $(CLOCK)
+	@mkdir -p $$(@D)
+	$(K1C_CC) $$+ -lm -o $$@
+
+BINFILES:=$(BINFILES) bin/$(TARGET)$(3).bin
+OUTFILES:=$(OUTFILES) out/$(TARGET)$(3).out
+FIRSTLINE:=$(FIRSTLINE), $(3)
+
+endef
+
+# Generic rules
+obj/%.o: asm/%.s
+	@mkdir -p $(@D)
+	$(K1C_CC) $< -c -o $@
+
+out/%.out: bin/%.bin
+	@mkdir -p $(@D)
+	$(EXECUTE_CYCLES) $< | tee $@
+
+##
+# Generating the rules for all the compiler/flags..
+##
+
+ifneq ($(GCC0FLAGS),)
+$(eval $(call gen_rules,$(K1C_CC),$(GCC0FLAGS),$(GCC0PREFIX)))
+endif
+ifneq ($(GCC1FLAGS),)
+$(eval $(call gen_rules,$(K1C_CC),$(GCC1FLAGS),$(GCC1PREFIX)))
+endif
+ifneq ($(GCC2FLAGS),)
+$(eval $(call gen_rules,$(K1C_CC),$(GCC2FLAGS),$(GCC2PREFIX)))
+endif
+ifneq ($(GCC3FLAGS),)
+$(eval $(call gen_rules,$(K1C_CC),$(GCC3FLAGS),$(GCC3PREFIX)))
+endif
+ifneq ($(GCC4FLAGS),)
+$(eval $(call gen_rules,$(K1C_CC),$(GCC4FLAGS),$(GCC4PREFIX)))
+endif
+
+ifneq ($(CCOMP0FLAGS),)
+$(eval $(call gen_rules,$(K1C_CCOMP),$(CCOMP0FLAGS),$(CCOMP0PREFIX)))
+endif
+ifneq ($(CCOMP1FLAGS),)
+$(eval $(call gen_rules,$(K1C_CCOMP),$(CCOMP1FLAGS),$(CCOMP1PREFIX)))
+endif
+ifneq ($(CCOMP2FLAGS),)
+$(eval $(call gen_rules,$(K1C_CCOMP),$(CCOMP2FLAGS),$(CCOMP2PREFIX)))
+endif
+ifneq ($(CCOMP3FLAGS),)
+$(eval $(call gen_rules,$(K1C_CCOMP),$(CCOMP3FLAGS),$(CCOMP3PREFIX)))
+endif
+ifneq ($(CCOMP4FLAGS),)
+$(eval $(call gen_rules,$(K1C_CCOMP),$(CCOMP4FLAGS),$(CCOMP4PREFIX)))
+endif
+
+measures.csv: $(OUTFILES)
+	@echo $(FIRSTLINE) > $@
+	@for ((i=0;i<=$(MAX_MEASURES);i++)) do\
+		first=$$(grep "($$i) cycles" $(firstword $(OUTFILES)));\
+		if test ! -z "$$first"; then\
+			line="$(TARGET) $$i";\
+			$(foreach outfile,$(OUTFILES),line="$$line, $$(grep "($$i) cycles" $(outfile) | cut -d':' -f2)"; ):;\
+			echo "$$line" >> $@;\
+		fi;\
+	done;\
+	echo "$@ created!"
+
+.PHONY: all clean run
+all: $(BINFILES)
+
+run: measures.csv
+
+clean:
+	rm -f *.o *.s *.bin *.out
+	rm -f asm/*.s bin/*.bin obj/*.o out/*.out
+
diff --git a/test/monniaux/sandbox/example.c b/test/monniaux/sandbox/example.c
new file mode 100644
index 00000000..2b8fc8c5
--- /dev/null
+++ b/test/monniaux/sandbox/example.c
@@ -0,0 +1,42 @@
+#include <stdio.h>
+#include "f.h"
+#include "../cycles.h"
+
+#if 0
+int main(void){
+  cycle_count_config();
+
+  int i;
+  int S = 0;
+
+  TIMEINIT
+  for (i = 0; i < 1000; i++){
+    S += f(i, i*2);
+  }
+  TIMESTOP(0)
+
+  printf("Final value: %d\n", S);
+  TIMESTOP(1)
+
+  TIMEPRINT(1)
+  return 0;
+}
+#endif
+
+#if 0
+int main(void){
+  cycle_count_config();
+
+  TIMEINIT
+  int a = 42;
+  int b = 21;
+  int c = 42+b;
+  TIMESTOP(0)
+  int d = a + b;
+  int e = a + b + c + d;
+  TIMESTOP(1)
+
+  TIMEPRINT(1)
+  return e;
+}
+#endif
diff --git a/test/monniaux/sandbox/f.c b/test/monniaux/sandbox/f.c
new file mode 100644
index 00000000..7705c4f7
--- /dev/null
+++ b/test/monniaux/sandbox/f.c
@@ -0,0 +1,3 @@
+int f(int x, int y){
+  return x+y;
+}
diff --git a/test/monniaux/sandbox/f.h b/test/monniaux/sandbox/f.h
new file mode 100644
index 00000000..6ecab534
--- /dev/null
+++ b/test/monniaux/sandbox/f.h
@@ -0,0 +1 @@
+int f(int, int);
diff --git a/test/monniaux/sandbox/sha-256.c b/test/monniaux/sandbox/sha-256.c
new file mode 100644
index 00000000..9a9e7802
--- /dev/null
+++ b/test/monniaux/sandbox/sha-256.c
@@ -0,0 +1,387 @@
+#include <stdint.h>
+#include <string.h>
+#if 0 /* __COMPCERT__ */
+#define my_memcpy(dst, src, size) __builtin_memcpy_aligned(dst, src, size, 1)
+#else
+#define my_memcpy(dst, src, size) memcpy(dst, src, size)
+#endif
+
+#include "../cycles.h"
+
+#include "sha-256.h"
+
+#define USE_ORIGINAL 1
+#define AUTOINCREMENT 1
+
+#define CHUNK_SIZE 64
+#define TOTAL_LEN_LEN 8
+
+/*
+ * ABOUT bool: this file does not use bool in order to be as pre-C99 compatible as possible.
+ */
+
+/*
+ * Comments from pseudo-code at https://en.wikipedia.org/wiki/SHA-2 are reproduced here.
+ * When useful for clarification, portions of the pseudo-code are reproduced here too.
+ */
+
+/*
+ * Initialize array of round constants:
+ * (first 32 bits of the fractional parts of the cube roots of the first 64 primes 2..311):
+ */
+static const uint32_t k[] = {
+	0x428a2f98, 0x71374491, 0xb5c0fbcf, 0xe9b5dba5, 0x3956c25b, 0x59f111f1, 0x923f82a4, 0xab1c5ed5,
+	0xd807aa98, 0x12835b01, 0x243185be, 0x550c7dc3, 0x72be5d74, 0x80deb1fe, 0x9bdc06a7, 0xc19bf174,
+	0xe49b69c1, 0xefbe4786, 0x0fc19dc6, 0x240ca1cc, 0x2de92c6f, 0x4a7484aa, 0x5cb0a9dc, 0x76f988da,
+	0x983e5152, 0xa831c66d, 0xb00327c8, 0xbf597fc7, 0xc6e00bf3, 0xd5a79147, 0x06ca6351, 0x14292967,
+	0x27b70a85, 0x2e1b2138, 0x4d2c6dfc, 0x53380d13, 0x650a7354, 0x766a0abb, 0x81c2c92e, 0x92722c85,
+	0xa2bfe8a1, 0xa81a664b, 0xc24b8b70, 0xc76c51a3, 0xd192e819, 0xd6990624, 0xf40e3585, 0x106aa070,
+	0x19a4c116, 0x1e376c08, 0x2748774c, 0x34b0bcb5, 0x391c0cb3, 0x4ed8aa4a, 0x5b9cca4f, 0x682e6ff3,
+	0x748f82ee, 0x78a5636f, 0x84c87814, 0x8cc70208, 0x90befffa, 0xa4506ceb, 0xbef9a3f7, 0xc67178f2
+};
+
+struct buffer_state {
+	const uint8_t * p;
+	size_t len;
+	size_t total_len;
+	int single_one_delivered; /* bool */
+	int total_len_delivered; /* bool */
+};
+
+static inline uint32_t right_rot(uint32_t value, unsigned int count)
+{
+	/*
+	 * Defined behaviour in standard C for all count where 0 < count < 32,
+	 * which is what we need here.
+	 */
+	return value >> count | value << (32 - count);
+}
+
+/* BEGIN DM */
+#define DEF_ROT(n) \
+static inline uint32_t right_rot##n(uint32_t value) \
+{ \
+	return value >> n | value << (32 - n); \
+}
+DEF_ROT(2)
+DEF_ROT(6)
+DEF_ROT(7)
+DEF_ROT(11)
+DEF_ROT(13)
+DEF_ROT(17)
+DEF_ROT(18)
+DEF_ROT(19)
+DEF_ROT(22)
+DEF_ROT(25)
+/* END DM */
+
+static void init_buf_state(struct buffer_state * state, const void * input, size_t len)
+{
+	state->p = input;
+	state->len = len;
+	state->total_len = len;
+	state->single_one_delivered = 0;
+	state->total_len_delivered = 0;
+}
+
+/* Return value: bool */
+static int calc_chunk(uint8_t chunk[CHUNK_SIZE], struct buffer_state * state)
+{
+	size_t space_in_chunk;
+
+	if (state->total_len_delivered) {
+		return 0;
+	}
+
+	if (state->len >= CHUNK_SIZE) {
+		my_memcpy(chunk, state->p, CHUNK_SIZE);
+		state->p += CHUNK_SIZE;
+		state->len -= CHUNK_SIZE;
+		return 1;
+	}
+
+	memcpy(chunk, state->p, state->len);
+	chunk += state->len;
+	space_in_chunk = CHUNK_SIZE - state->len;
+	state->p += state->len;
+	state->len = 0;
+
+	/* If we are here, space_in_chunk is one at minimum. */
+	if (!state->single_one_delivered) {
+		*chunk++ = 0x80;
+		space_in_chunk -= 1;
+		state->single_one_delivered = 1;
+	}
+
+	/*
+	 * Now:
+	 * - either there is enough space left for the total length, and we can conclude,
+	 * - or there is too little space left, and we have to pad the rest of this chunk with zeroes.
+	 * In the latter case, we will conclude at the next invokation of this function.
+	 */
+	if (space_in_chunk >= TOTAL_LEN_LEN) {
+		const size_t left = space_in_chunk - TOTAL_LEN_LEN;
+		size_t len = state->total_len;
+		int i;
+		memset(chunk, 0x00, left);
+		chunk += left;
+
+		/* Storing of len * 8 as a big endian 64-bit without overflow. */
+		chunk[7] = (uint8_t) (len << 3);
+		len >>= 5;
+		for (i = 6; i >= 0; i--) {
+			chunk[i] = (uint8_t) len;
+			len >>= 8;
+		}
+		state->total_len_delivered = 1;
+	} else {
+		memset(chunk, 0x00, space_in_chunk);
+	}
+
+	return 1;
+}
+
+/*
+ * Limitations:
+ * - Since input is a pointer in RAM, the data to hash should be in RAM, which could be a problem
+ *   for large data sizes.
+ * - SHA algorithms theoretically operate on bit strings. However, this implementation has no support
+ *   for bit string lengths that are not multiples of eight, and it really operates on arrays of bytes.
+ *   In particular, the len parameter is a number of bytes.
+ */
+
+#if USE_ORIGINAL
+void calc_sha_256(uint8_t hash[32], const void * input, size_t len)
+{
+  TIMEINIT(3)
+	/*
+	 * Note 1: All integers (expect indexes) are 32-bit unsigned integers and addition is calculated modulo 2^32.
+	 * Note 2: For each round, there is one round constant k[i] and one entry in the message schedule array w[i], 0 = i = 63
+	 * Note 3: The compression function uses 8 working variables, a through h
+	 * Note 4: Big-endian convention is used when expressing the constants in this pseudocode,
+	 *     and when parsing message block data from bytes to words, for example,
+	 *     the first word of the input message "abc" after padding is 0x61626380
+	 */
+
+	/*
+	 * Initialize hash values:
+	 * (first 32 bits of the fractional parts of the square roots of the first 8 primes 2..19):
+	 */
+	uint32_t h[] = { 0x6a09e667, 0xbb67ae85, 0x3c6ef372, 0xa54ff53a, 0x510e527f, 0x9b05688c, 0x1f83d9ab, 0x5be0cd19 };
+	int i, j;
+
+	/* 512-bit chunks is what we will operate on. */
+	uint8_t chunk[64];
+
+	struct buffer_state state;
+
+	init_buf_state(&state, input, len);
+  TIMEINIT(0)
+	while (calc_chunk(chunk, &state)) {
+		uint32_t ah[8];
+		
+		/*
+		 * create a 64-entry message schedule array w[0..63] of 32-bit words
+		 * (The initial values in w[0..63] don't matter, so many implementations zero them here)
+		 * copy chunk into first 16 words w[0..15] of the message schedule array
+		 */
+		uint32_t w[64];
+		const uint8_t *p = chunk;
+
+		memset(w, 0x00, sizeof w);
+		for (i = 0; i < 16; i++) {
+			w[i] = (uint32_t) p[0] << 24 | (uint32_t) p[1] << 16 |
+				(uint32_t) p[2] << 8 | (uint32_t) p[3];
+			p += 4;
+		}
+
+		/* Extend the first 16 words into the remaining 48 words w[16..63] of the message schedule array: */
+		for (i = 16; i < 64; i++) {
+			const uint32_t s0 = right_rot7(w[i - 15]) ^ right_rot18(w[i - 15]) ^ (w[i - 15] >> 3);
+			const uint32_t s1 = right_rot17(w[i - 2]) ^ right_rot19(w[i - 2]) ^ (w[i - 2] >> 10);
+			w[i] = w[i - 16] + s0 + w[i - 7] + s1;
+		}
+		
+		/* Initialize working variables to current hash value: */
+		for (i = 0; i < 8; i++)
+			ah[i] = h[i];
+
+		/* Compression function main loop: */
+		for (i = 0; i < 64; i++) {
+      //TIMEINIT(4)
+			const uint32_t s1 = right_rot6(ah[4]) ^ right_rot11(ah[4]) ^ right_rot25(ah[4]);
+			const uint32_t ch = (ah[4] & ah[5]) ^ (~ah[4] & ah[6]);
+			const uint32_t temp1 = ah[7] + s1 + ch + k[i] + w[i];
+			const uint32_t s0 = right_rot2(ah[0]) ^ right_rot13(ah[0]) ^ right_rot22(ah[0]);
+			const uint32_t maj = (ah[0] & ah[1]) ^ (ah[0] & ah[2]) ^ (ah[1] & ah[2]);
+			const uint32_t temp2 = s0 + maj;
+      //TIMESTOP(4) TIMEINIT(5)
+			ah[7] = ah[6];
+			ah[6] = ah[5];
+			ah[5] = ah[4];
+			ah[4] = ah[3] + temp1;
+			ah[3] = ah[2];
+			ah[2] = ah[1];
+			ah[1] = ah[0];
+			ah[0] = temp1 + temp2;
+      //TIMESTOP(5)
+		}
+
+		/* Add the compressed chunk to the current hash value: */
+		for (i = 0; i < 8; i++)
+			h[i] += ah[i];
+    TIMESTOP(0)
+	}
+
+  TIMEINIT(2)
+	/* Produce the final hash value (big-endian): */
+	for (i = 0, j = 0; i < 8; i++)
+	{
+		hash[j++] = (uint8_t) (h[i] >> 24);
+		hash[j++] = (uint8_t) (h[i] >> 16);
+		hash[j++] = (uint8_t) (h[i] >> 8);
+		hash[j++] = (uint8_t) h[i];
+    TIMESTOP(2)
+	}
+  TIMESTOP(3)
+}
+#else
+/* Modified by D. Monniaux */
+void calc_sha_256(uint8_t hash[32], const void * input, size_t len)
+{
+	/*
+	 * Note 1: All integers (expect indexes) are 32-bit unsigned integers and addition is calculated modulo 2^32.
+	 * Note 2: For each round, there is one round constant k[i] and one entry in the message schedule array w[i], 0 = i = 63
+	 * Note 3: The compression function uses 8 working variables, a through h
+	 * Note 4: Big-endian convention is used when expressing the constants in this pseudocode,
+	 *     and when parsing message block data from bytes to words, for example,
+	 *     the first word of the input message "abc" after padding is 0x61626380
+	 */
+
+	/*
+	 * Initialize hash values:
+	 * (first 32 bits of the fractional parts of the square roots of the first 8 primes 2..19):
+	 */
+	uint32_t h[] = { 0x6a09e667, 0xbb67ae85, 0x3c6ef372, 0xa54ff53a, 0x510e527f, 0x9b05688c, 0x1f83d9ab, 0x5be0cd19 };
+	uint32_t h0 = h[0];
+	uint32_t h1 = h[1];
+	uint32_t h2 = h[2];
+	uint32_t h3 = h[3];
+	uint32_t h4 = h[4];
+	uint32_t h5 = h[5];
+	uint32_t h6 = h[6];
+	uint32_t h7 = h[7];
+	int i, j;
+
+	/* 512-bit chunks is what we will operate on. */
+	uint8_t chunk[64];
+
+	struct buffer_state state;
+
+	init_buf_state(&state, input, len);
+
+	while (calc_chunk(chunk, &state)) {
+	        uint32_t ah0, ah1, ah2, ah3, ah4, ah5, ah6, ah7;
+		
+		/*
+		 * create a 64-entry message schedule array w[0..63] of 32-bit words
+		 * (The initial values in w[0..63] don't matter, so many implementations zero them here)
+		 * copy chunk into first 16 words w[0..15] of the message schedule array
+		 */
+		uint32_t w[64];
+		const uint8_t *p = chunk;
+
+		memset(w, 0x00, sizeof w);
+#ifndef SKIP_SLOW_PARTS
+		for (i = 0; i < 16; i++) {
+			w[i] = (uint32_t) p[0] << 24 | (uint32_t) p[1] << 16 |
+				(uint32_t) p[2] << 8 | (uint32_t) p[3];
+			p += 4;
+		}
+
+		/* Extend the first 16 words into the remaining 48 words w[16..63] of the message schedule array: */
+		/* DM this is a SLOW part with ccomp; awkward address computations. */
+		for (i = 16; i < 64; i++) {
+			const uint32_t s0 = right_rot7(w[i - 15]) ^ right_rot18(w[i - 15]) ^ (w[i - 15] >> 3);
+			const uint32_t s1 = right_rot17(w[i - 2]) ^ right_rot19(w[i - 2]) ^ (w[i - 2] >> 10);
+			w[i] = w[i - 16] + s0 + w[i - 7] + s1;
+		}
+#endif
+		/* Initialize working variables to current hash value: */
+		ah0 = h0;
+		ah1 = h1;
+		ah2 = h2;
+		ah3 = h3;
+		ah4 = h4;
+		ah5 = h5;
+		ah6 = h6;
+		ah7 = h7;
+
+		/* Compression function main loop: */
+#if AUTOINCREMENT
+		const uint32_t *ki=k, *wi=w;
+#define KI *ki
+#define WI *wi
+#define STEP i++; ki++; wi++;
+#else
+#define KI k[i]
+#define WI w[i]
+#define STEP i++;
+#endif
+		for (i = 0; i < 64; ) {
+#define CHUNK								\
+		  {							\
+			const uint32_t s1 = right_rot6(ah4) ^ right_rot11(ah4) ^ right_rot25(ah4); \
+			const uint32_t ch = (ah4 & ah5) ^ (~ah4 & ah6);	\
+			const uint32_t temp1 = ah7 + s1 + ch + KI + WI; \
+			const uint32_t s0 = right_rot2(ah0) ^ right_rot13(ah0) ^ right_rot22(ah0); \
+			const uint32_t maj = (ah0 & ah1) ^ (ah0 & ah2) ^ (ah1 & ah2); \
+			const uint32_t temp2 = s0 + maj;		\
+									\
+			ah7 = ah6;					\
+			ah6 = ah5;					\
+			ah5 = ah4;					\
+			ah4 = ah3 + temp1;				\
+			ah3 = ah2;					\
+			ah2 = ah1;					\
+			ah1 = ah0;					\
+			ah0 = temp1 + temp2;				\
+			STEP						\
+		  }
+		  CHUNK
+		  CHUNK
+		}
+
+		/* Add the compressed chunk to the current hash value: */
+		h0 += ah0;
+		h1 += ah1;
+		h2 += ah2;
+		h3 += ah3;
+		h4 += ah4;
+		h5 += ah5;
+		h6 += ah6;
+		h7 += ah7;
+	}
+	h[0]=h0;
+	h[1]=h1;
+	h[2]=h2;
+	h[3]=h3;
+	h[4]=h4;
+	h[5]=h5;
+	h[6]=h6;
+	h[7]=h7;
+
+	/* Produce the final hash value (big-endian): */
+	for (i = 0, j = 0; i < 8; i++)
+	{
+		hash[j++] = (uint8_t) (h[i] >> 24);
+		hash[j++] = (uint8_t) (h[i] >> 16);
+		hash[j++] = (uint8_t) (h[i] >> 8);
+		hash[j++] = (uint8_t) h[i];
+	}
+}
+#endif
+
+void print_all(void){
+  TIMEPRINT(5)
+}
diff --git a/test/monniaux/sandbox/sha-256.h b/test/monniaux/sandbox/sha-256.h
new file mode 100644
index 00000000..0753ea9e
--- /dev/null
+++ b/test/monniaux/sandbox/sha-256.h
@@ -0,0 +1,2 @@
+void calc_sha_256(uint8_t hash[32], const void *input, size_t len);
+void print_all(void);
diff --git a/test/monniaux/sandbox/sha-256_run.c b/test/monniaux/sandbox/sha-256_run.c
new file mode 100644
index 00000000..a1631bc6
--- /dev/null
+++ b/test/monniaux/sandbox/sha-256_run.c
@@ -0,0 +1,286 @@
+#include <stdint.h>
+#include <stdio.h>
+#include <string.h>
+#include <stdlib.h>
+#include <inttypes.h>
+#include "../cycles.h"
+#include "sha-256.h"
+
+struct string_vector {
+	const char *input;
+	const char *output;
+};
+
+static const struct string_vector STRING_VECTORS[] = {
+	{
+		"",
+		"e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
+	},
+	{
+		"abc",
+		"ba7816bf8f01cfea414140de5dae2223b00361a396177a9cb410ff61f20015ad"
+	},
+	{
+		"0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef",
+		"a8ae6e6ee929abea3afcfc5258c8ccd6f85273e0d4626d26c7279f3250f77c8e"
+	},
+	{
+		"0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcde",
+		"057ee79ece0b9a849552ab8d3c335fe9a5f1c46ef5f1d9b190c295728628299c"
+	},
+	{
+		"0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef0",
+		"2a6ad82f3620d3ebe9d678c812ae12312699d673240d5be8fac0910a70000d93"
+	},
+	{
+		"abcdbcdecdefdefgefghfghighijhijkijkljklmklmnlmnomnopnopq",
+		"248d6a61d20638b8e5c026930c3e6039a33ce45964ff2167f6ecedd419db06c1"
+	},
+	{
+		"abcdefghbcdefghicdefghijdefghijkefghijklfghijklmghijklmnhijklmno"
+		"ijklmnopjklmnopqklmnopqrlmnopqrsmnopqrstnopqrstu",
+		"cf5b16a778af8380036ce59e7b0492370b249b11e8f07a51afac45037afee9d1"
+	}
+};
+
+#define LARGE_MESSAGES 1
+#define LARGER_MESSAGES 0
+
+static uint8_t data1[] = { 0xbd };
+static uint8_t data2[] = { 0xc9, 0x8c, 0x8e, 0x55 };
+static uint8_t data7[1000];
+static uint8_t data8[1000];
+static uint8_t data9[1005];
+#if LARGE_MESSAGES
+#define SIZEOF_DATA11 536870912
+#define SIZEOF_DATA12 1090519040
+#define SIZEOF_DATA13 1610612798
+static uint8_t * data11;
+static uint8_t * data12;
+static uint8_t * data13;
+#endif
+
+struct vector {
+	const uint8_t *input;
+	size_t input_len;
+	const char *output;
+};
+
+static struct vector vectors[] = {
+	{
+		data1,
+		sizeof data1,
+		"68325720aabd7c82f30f554b313d0570c95accbb7dc4b5aae11204c08ffe732b"
+	},
+	{
+		data2,
+		sizeof data2,
+		"7abc22c0ae5af26ce93dbb94433a0e0b2e119d014f8e7f65bd56c61ccccd9504"
+	},
+	{
+		data7,
+		55,
+		"02779466cdec163811d078815c633f21901413081449002f24aa3e80f0b88ef7"
+	},
+	{
+		data7,
+		56,
+		"d4817aa5497628e7c77e6b606107042bbba3130888c5f47a375e6179be789fbb"
+	},
+	{
+		data7,
+		57,
+		"65a16cb7861335d5ace3c60718b5052e44660726da4cd13bb745381b235a1785"
+	},
+	{
+		data7,
+		64,
+		"f5a5fd42d16a20302798ef6ed309979b43003d2320d9f0e8ea9831a92759fb4b"
+	},
+	{
+		data7,
+		sizeof data7,
+		"541b3e9daa09b20bf85fa273e5cbd3e80185aa4ec298e765db87742b70138a53"
+	},
+	{
+		data8,
+		sizeof data8,
+		"c2e686823489ced2017f6059b8b239318b6364f6dcd835d0a519105a1eadd6e4"
+	},
+	{
+		data9,
+		sizeof data9,
+		"f4d62ddec0f3dd90ea1380fa16a5ff8dc4c54b21740650f24afc4120903552b0"
+	},
+#if LARGE_MESSAGES
+	{
+		NULL,
+		/* too big
+		1000000,
+		"d29751f2649b32ff572b5e0a9f541ea660a50f94ff0beedfb0b692b924cc8025"
+		*/
+		50000,
+		"5b4b67b5d68e02c992760de07640472efe53a7f7553865f83262d0a74efc3e5d"
+	},
+#if LARGER_MESSAGES
+	{
+		NULL,
+		SIZEOF_DATA11,
+		"15a1868c12cc53951e182344277447cd0979536badcc512ad24c67e9b2d4f3dd"
+	},
+	{
+		NULL,
+		SIZEOF_DATA12,
+		"461c19a93bd4344f9215f5ec64357090342bc66b15a148317d276e31cbc20b53"
+	},
+	{
+		NULL,
+		SIZEOF_DATA13,
+		"c23ce8a7895f4b21ec0daf37920ac0a262a220045a03eb2dfed48ef9b05aabea"
+	}
+#endif
+#endif
+};
+
+#if LARGE_MESSAGES
+static void *my_malloc(size_t size) {
+  void *p=malloc(size);
+  if (p==0) {
+    fprintf(stderr, "malloc(%zu) failed\n", size);
+    abort();
+  }
+  return p;
+}
+#endif
+
+static void construct_binary_messages(void)
+{
+	memset(data7, 0x00, sizeof data7);
+	memset(data8, 0x41, sizeof data8);
+	memset(data9, 0x55, sizeof data9);
+#if LARGE_MESSAGES
+#if LARGER_MESSAGES
+	/*
+	 * Heap allocation as a workaround for some linkers not liking
+	 * large BSS segments.
+	 */
+	data11 = my_malloc(SIZEOF_DATA11);
+	data12 = my_malloc(SIZEOF_DATA12);
+	data13 = my_malloc(SIZEOF_DATA13);
+	memset(data11, 0x5a, SIZEOF_DATA11);
+	memset(data12, 0x00, SIZEOF_DATA12);
+	memset(data13, 0x42, SIZEOF_DATA13);
+	vectors[9].input = data12;
+	vectors[10].input = data11;
+	vectors[11].input = data12;
+	vectors[12].input = data13;
+#else
+	vectors[9].input = data12 = my_malloc(vectors[9].input_len);
+	memset(data12, 0x00, vectors[9].input_len);
+#endif
+#endif
+}
+
+static void destruct_binary_messages(void)
+{
+#if LARGE_MESSAGES
+#if LARGER_MESSAGES
+	free(data11);
+	free(data12);
+	free(data13);
+#else
+	free(data12);
+#endif
+#endif
+}
+
+static void hash_to_string(char string[65], const uint8_t hash[32])
+{
+	size_t i;
+	for (i = 0; i < 32; i++) {
+		string += sprintf(string, "%02x", hash[i]);
+	}
+}	
+
+static cycle_t cycle_total, cycle_start_time;
+
+static void cycle_count_start(void) {
+  cycle_start_time=get_cycle();
+}
+
+static void cycle_count_end(void) {
+  cycle_total += get_cycle()-cycle_start_time;
+}
+
+static int string_test(const char input[], const char output[])
+{
+	uint8_t hash[32];
+	char hash_string[65];
+
+	cycle_count_start();
+	calc_sha_256(hash, input, strlen(input));
+	cycle_count_end();
+	
+	hash_to_string(hash_string, hash);
+	printf("input: %s\n", input);
+	printf("hash : %s\n", hash_string);
+	if (strcmp(output, hash_string)) {
+		printf("FAILURE!\n\n");
+		return 1;
+	} else {
+		printf("SUCCESS!\n\n");
+		return 0;
+	}		
+}
+
+/*
+ * Limitation:
+ * - The variable input_len will be truncated to its LONG_BIT least
+ * significant bits in the print output. This will never be a problem
+ * for values that in practice are less than 2^32 - 1. Rationale: ANSI
+ * C-compatibility and keeping it simple.
+ */
+static int test(const uint8_t * input, size_t input_len, const char output[])
+{
+	uint8_t hash[32];
+	char hash_string[65];
+
+	cycle_count_start();
+	calc_sha_256(hash, input, input_len);
+	cycle_count_end();
+
+	hash_to_string(hash_string, hash);
+	printf("input starts with 0x%02x, length %lu\n", *input, (unsigned long) input_len);
+	printf("hash : %s\n", hash_string);
+	if (strcmp(output, hash_string)) {
+		printf("FAILURE!\n\n");
+		return 1;
+	} else {
+		printf("SUCCESS!\n\n");
+		return 0;
+	}
+}
+
+int main(void)
+{
+        cycle_count_config();
+	size_t i;
+	for (i = 0; i < (sizeof STRING_VECTORS / sizeof (struct string_vector)); i++) {
+		const struct string_vector *vector = &STRING_VECTORS[i];
+		if (string_test(vector->input, vector->output))
+		  {} /* DM return 1; */
+	}
+	construct_binary_messages();
+	for (i = 0; i < (sizeof vectors / sizeof (struct vector)); i++) {
+		const struct vector *vector = &vectors[i];
+		if (test(vector->input, vector->input_len, vector->output))
+		{ /* DM
+			destruct_binary_messages();
+			return 1; */
+		}
+	}
+	destruct_binary_messages();
+	printf("total cycles : %" PRIu64 "\n", cycle_total);
+  print_all();
+	return 0;
+}