Completed the implementation, including testing, and fixed bugs

.gitignore

@@ -1 +1,3 @@
-*~
+*~
+*.o
+test

Makefile

@@ -1,3 +1,5 @@
+CFLAGS = -Wall -Wextra -Wpedantic
+
 test: test.o sha-256.o
 
 .PHONY: clean

README.md

@@ -1,2 +1,27 @@
 # sha-2
-SHA-2 algorithm implementations
+
+## Contents
+
+SHA-2 algorithm implementations.
+
+At the moment, only a SHA-256 is implemented.
+
+## Design criteria
+
+- Easy to test, include in any project, compile and link.
+
+- ANSI C with as little specific C99 as possible (e.g. extended
+  integer types are used, but not bool).
+
+- Portable. Makes no assumptions on the target system's endianess or
+  word size.
+
+- The SHA-256 implementation is a straightforward implementation of
+  the algorithm specified on
+  [Wikipedia](https://en.wikipedia.org/wiki/SHA-2). At the moment,
+  no effort at all has been put in optimization.
+
+## License
+
+This repository is made available in the public domain. See [LICENSE
+FILE](LICENSE).

sha-256.c

@@ -71,6 +71,8 @@ static int calc_chunk(uint8_t chunk[CHUNK_SIZE], struct buffer_state * state)
 	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) {
@@ -83,21 +85,39 @@ static int calc_chunk(uint8_t chunk[CHUNK_SIZE], struct buffer_state * state)
 	 * 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 pass.
+	 * 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;
-		const size_t len = state->total_len;
+		const size_t len = state->total_len * 8;
 		memset(chunk, 0x00, left);
 		chunk += left;
 
-		chunk[0] = (uint8_t) (len >> 56);
+		if (sizeof len > 4) {
-		chunk[1] = (uint8_t) (len >> 48);
+			chunk[0] = (uint8_t) (len >> 56);
-		chunk[2] = (uint8_t) (len >> 40);
+			chunk[1] = (uint8_t) (len >> 48);
-		chunk[3] = (uint8_t) (len >> 32);
+			chunk[2] = (uint8_t) (len >> 40);
-		chunk[4] = (uint8_t) (len >> 24);
+			chunk[3] = (uint8_t) (len >> 32);
-		chunk[5] = (uint8_t) (len >> 16);
+		} else {
-		chunk[6] = (uint8_t) (len >> 8);
+			chunk[0] = 0;
+			chunk[1] = 0;
+			chunk[2] = 0;
+			chunk[3] = 0;
+		}
+		if (sizeof len > 2) {
+			chunk[4] = (uint8_t) (len >> 24);
+			chunk[5] = (uint8_t) (len >> 16);
+		} else {
+			chunk[4] = 0;
+			chunk[5] = 0;
+		}
+		
+		if (sizeof len > 1) {
+			chunk[6] = (uint8_t) (len >> 8);
+		} else {
+			chunk[6] = 0;
+		}
+		
 		chunk[7] = (uint8_t) len;
 
 		state->total_len_delivered = 1;
@@ -108,6 +128,13 @@ static int calc_chunk(uint8_t chunk[CHUNK_SIZE], struct buffer_state * state)
 	return 1;
 }
 
+/*
+ * Limitations:
+ * - len must be small enough for (8 * len) to fit in len. Otherwise, the results are unpredictable.
+ * - sizeof size_t is assumed to be either 8, 16, 32 or 64. Otherwise, the results are unpredictable.
+ * - Since input is a pointer in RAM, the data to hash should be in RAM, which could be a problem
+ *   for large data sizes.
+ */
 void calc_sha_256(uint8_t hash[32], const void * input, size_t len)
 {
 	/*
@@ -161,8 +188,8 @@ void calc_sha_256(uint8_t hash[32], const void * input, size_t len)
 
 		/* Extend the first 16 words into the remaining 48 words w[16..63] of the message schedule array: */
 		for (int i = 16; i < 64; i++) {
-			const uint32_t s0 = right_rot(w[i - 15], 7) ^ right_rot(w[i - 15], 18) ^ w[i - 15] >> 3;
+			const uint32_t s0 = right_rot(w[i - 15], 7) ^ right_rot(w[i - 15], 18) ^ (w[i - 15] >> 3);
-			const uint32_t s1 = right_rot(w[i - 2], 17) ^ right_rot(w[i - 2], 19) ^ w[i - 2] >> 10;
+			const uint32_t s1 = right_rot(w[i - 2], 17) ^ right_rot(w[i - 2], 19) ^ (w[i - 2] >> 10);
 			w[i] = w[i - 16] + s0 + w[i - 7] + s1;
 		}
 		
@@ -179,10 +206,10 @@ void calc_sha_256(uint8_t hash[32], const void * input, size_t len)
 		/* Compression function main loop: */
 		for (int i = 0; i < 64; i++) {
 			const uint32_t s1 = right_rot(e, 6) ^ right_rot(e, 11) ^ right_rot(e, 25);
-			const uint32_t ch = e & f ^ ~e & g;
+			const uint32_t ch = (e & f) ^ (~e & g);
 			const uint32_t temp1 = h + s1 + ch + k[i] + w[i];
-			const uint32_t s0 = right_rot(a, 2) ^right_rot(a, 13) ^ right_rot(a, 22);
+			const uint32_t s0 = right_rot(a, 2) ^ right_rot(a, 13) ^ right_rot(a, 22);
-			const uint32_t maj = a & b ^ a & c ^ b & c;
+			const uint32_t maj = (a & b) ^ (a & c) ^ (b & c);
 			const uint32_t temp2 = s0 + maj;
 
 			h = g;

test.c

@@ -1,14 +1,75 @@
 #include <stdint.h>
 #include <stdio.h>
+#include <string.h>
 
 #include "sha-256.h"
 
-int main()
+struct vector {
+	const char *input;
+	const char *output;
+};
+
+static const struct vector VECTORS[] = {
+	{
+		"",
+		"e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
+	},
+	{
+		"abc",
+		"ba7816bf8f01cfea414140de5dae2223b00361a396177a9cb410ff61f20015ad"
+	},
+	{
+		"0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef",
+		"a8ae6e6ee929abea3afcfc5258c8ccd6f85273e0d4626d26c7279f3250f77c8e"
+	},
+	{
+		"0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcde",
+		"057ee79ece0b9a849552ab8d3c335fe9a5f1c46ef5f1d9b190c295728628299c"
+	},
+	{
+		"0123456789abcdef0123456789abcdef0123456789abcdef0123456789abcdef0",
+		"2a6ad82f3620d3ebe9d678c812ae12312699d673240d5be8fac0910a70000d93"
+	},
+	{
+		"abcdbcdecdefdefgefghfghighijhijkijkljklmklmnlmnomnopnopq",
+		"248d6a61d20638b8e5c026930c3e6039a33ce45964ff2167f6ecedd419db06c1"
+	},
+	{
+		"abcdefghbcdefghicdefghijdefghijkefghijklfghijklmghijklmnhijklmnoi"
+		"jklmnopjklmnopqklmnopqrlmnopqrsmnopqrstnopqrstu",
+		"cf5b16a778af8380036ce59e7b0492370b249b11e8f07a51afac45037afee9d1"
+	}
+};
+
+static void hash_to_string(char string[65], const uint8_t hash[32])
+{
+	for (size_t i = 0; i < 32; i++) {
+		string += sprintf(string, "%02x", hash[i]);
+	}
+}	
+	
+static void test(const char input[], const char output[])
 {
 	uint8_t hash[32];
-	calc_sha_256(hash, "hello\n", 6);
+	char hash_string[65];
-	for (size_t i = 0; i < sizeof hash; i++) {
+	calc_sha_256(hash, input, strlen(input));
-		printf("%02x", hash[i]);
+	hash_to_string(hash_string, hash);
-	}
+	printf("input: %s\n", input);
-	printf("\n");
+	printf("hash : %s\n", hash_string);
+	if (strcmp(output, hash_string)) {
+		printf("FAILURE!\n\n");
+	} else {
+		printf("SUCCESS!\n\n");
+	}		
+}
+	
+
+int main(void)
+{
+	printf("%zu\n", sizeof (size_t));
+	for (size_t i = 0; i < (sizeof VECTORS / sizeof (struct vector)); i++) {
+		const struct vector *vector = &VECTORS[i];
+		test(vector->input, vector->output);
+	}
+	return 0;
 }

tmp.txt

@@ -1 +0,0 @@
-hello