Convert bcrypt to use OpenBSD code (#68)

* swap to using openbsd bcrypt

* we should probably call this 3.0

* update tests to handle slight change in behavior, test better

* strip out code we're not using

* define this for linux

* py3 fix

* add a changelog to the readme

* maybe work with windows

* portable endian header, replace swaps, other windows fixes

* handle older windows compilers properly, handle glibc < 2.9, retab

* remove a todo, that's definitely the limit

* make these definitions conditional since some BSDs may already have them

* add $2a$ tests from crypt_blowfish-1.3

* update readme to note supported prefixes

* almost pointless commit

* add support for $2y$

test vectors from openwall crypt-blowfish1.3

MANIFEST.in

@@ -1,3 +1,3 @@
 include LICENSE README.rst
 include src/build_bcrypt.py
-recursive-include src/crypt_blowfish-1.3 *
+recursive-include src/_csrc *

README.rst

@@ -34,6 +34,19 @@ For Fedora and RHEL-derivatives, the following command will ensure that the requ
 
     $ sudo yum install gcc libffi-devel python-devel
 
+Changelog
+=========
+
+3.0.0
+-----
+* Switched the C backend to code obtained from the OpenBSD project rather than
+  openwall.
+
+2.0.0
+-----
+* Added support for an adjustible prefix when calling `gensalt`.
+* Switched to CFFI 1.0+
+
 Usage
 -----
 
@@ -84,6 +97,8 @@ Another one of bcrypt's features is an adjustable prefix to let you define what
 libraries you'll remain compatible with. To adjust this, pass either ``2a`` or
 ``2b`` (the default) to ``bcrypt.gensalt(prefix=b"2b")`` as a bytes object.
 
+As of 3.0.0 the `$2y$` prefix is still supported in `hashpw` but deprecated.
+
 Maxmimum Password Length
 ~~~~~~~~~~~~~~~~~~~~~~~~
 
@@ -107,6 +122,11 @@ Compatibility
 This library should be compatible with py-bcrypt and it will run on Python
 2.6+, 3.3+, and PyPy 2.6+.
 
+C Code
+------
+
+This library uses code from OpenBSD.
+
 Security
 --------
 

src/_csrc/bcrypt.c

@@ -0,0 +1,274 @@
+/*	$OpenBSD: bcrypt.c,v 1.55 2015/09/13 15:33:48 guenther Exp $	*/
+
+/*
+ * Copyright (c) 2014 Ted Unangst <tedu@openbsd.org>
+ * Copyright (c) 1997 Niels Provos <provos@umich.edu>
+ *
+ * Permission to use, copy, modify, and distribute this software for any
+ * purpose with or without fee is hereby granted, provided that the above
+ * copyright notice and this permission notice appear in all copies.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
+ * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
+ * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
+ * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
+ * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
+ * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
+ */
+/* This password hashing algorithm was designed by David Mazieres
+ * <dm@lcs.mit.edu> and works as follows:
+ *
+ * 1. state := InitState ()
+ * 2. state := ExpandKey (state, salt, password)
+ * 3. REPEAT rounds:
+ *      state := ExpandKey (state, 0, password)
+ *	state := ExpandKey (state, 0, salt)
+ * 4. ctext := "OrpheanBeholderScryDoubt"
+ * 5. REPEAT 64:
+ * 	ctext := Encrypt_ECB (state, ctext);
+ * 6. RETURN Concatenate (salt, ctext);
+ *
+ */
+
+#include <sys/types.h>
+#include <blf.h>
+#include <ctype.h>
+#include <errno.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include "pycabcrypt.h"
+
+/* This implementation is adaptable to current computing power.
+ * You can have up to 2^31 rounds which should be enough for some
+ * time to come.
+ */
+
+#define BCRYPT_VERSION '2'
+#define BCRYPT_MAXSALT 16	/* Precomputation is just so nice */
+#define BCRYPT_WORDS 6		/* Ciphertext words */
+#define BCRYPT_MINLOGROUNDS 4	/* we have log2(rounds) in salt */
+
+#define	BCRYPT_SALTSPACE	(7 + (BCRYPT_MAXSALT * 4 + 2) / 3 + 1)
+#define	BCRYPT_HASHSPACE	61
+
+char   *bcrypt_gensalt(u_int8_t);
+
+int encode_base64(char *, const u_int8_t *, size_t);
+static int decode_base64(u_int8_t *, size_t, const char *);
+
+/*
+ * the core bcrypt function
+ */
+int
+bcrypt_hashpass(const char *key, const char *salt, char *encrypted,
+    size_t encryptedlen)
+{
+	blf_ctx state;
+	u_int32_t rounds, i, k;
+	u_int16_t j;
+	size_t key_len;
+	u_int8_t salt_len, logr, minor;
+	u_int8_t ciphertext[4 * BCRYPT_WORDS] = "OrpheanBeholderScryDoubt";
+	u_int8_t csalt[BCRYPT_MAXSALT];
+	u_int32_t cdata[BCRYPT_WORDS];
+
+	if (encryptedlen < BCRYPT_HASHSPACE)
+		goto inval;
+
+	/* Check and discard "$" identifier */
+	if (salt[0] != '$')
+		goto inval;
+	salt += 1;
+
+	if (salt[0] != BCRYPT_VERSION)
+		goto inval;
+
+	/* Check for minor versions */
+	switch ((minor = salt[1])) {
+	case 'a':
+		key_len = (u_int8_t)(strlen(key) + 1);
+		break;
+	case 'b':
+		/* strlen() returns a size_t, but the function calls
+		 * below result in implicit casts to a narrower integer
+		 * type, so cap key_len at the actual maximum supported
+		 * length here to avoid integer wraparound */
+		key_len = strlen(key);
+		if (key_len > 72)
+			key_len = 72;
+		key_len++; /* include the NUL */
+		break;
+	default:
+		 goto inval;
+	}
+	if (salt[2] != '$')
+		goto inval;
+	/* Discard version + "$" identifier */
+	salt += 3;
+
+	/* Check and parse num rounds */
+	if (!isdigit((unsigned char)salt[0]) ||
+	    !isdigit((unsigned char)salt[1]) || salt[2] != '$')
+		goto inval;
+	logr = (salt[1] - '0') + ((salt[0] - '0') * 10);
+	if (logr < BCRYPT_MINLOGROUNDS || logr > 31)
+		goto inval;
+	/* Computer power doesn't increase linearly, 2^x should be fine */
+	rounds = 1U << logr;
+
+	/* Discard num rounds + "$" identifier */
+	salt += 3;
+
+	if (strlen(salt) * 3 / 4 < BCRYPT_MAXSALT)
+		goto inval;
+
+	/* We dont want the base64 salt but the raw data */
+	if (decode_base64(csalt, BCRYPT_MAXSALT, salt))
+		goto inval;
+	salt_len = BCRYPT_MAXSALT;
+
+	/* Setting up S-Boxes and Subkeys */
+	Blowfish_initstate(&state);
+	Blowfish_expandstate(&state, csalt, salt_len,
+	    (u_int8_t *) key, key_len);
+	for (k = 0; k < rounds; k++) {
+		Blowfish_expand0state(&state, (u_int8_t *) key, key_len);
+		Blowfish_expand0state(&state, csalt, salt_len);
+	}
+
+	/* This can be precomputed later */
+	j = 0;
+	for (i = 0; i < BCRYPT_WORDS; i++)
+		cdata[i] = Blowfish_stream2word(ciphertext, 4 * BCRYPT_WORDS, &j);
+
+	/* Now do the encryption */
+	for (k = 0; k < 64; k++)
+		blf_enc(&state, cdata, BCRYPT_WORDS / 2);
+
+	for (i = 0; i < BCRYPT_WORDS; i++) {
+		ciphertext[4 * i + 3] = cdata[i] & 0xff;
+		cdata[i] = cdata[i] >> 8;
+		ciphertext[4 * i + 2] = cdata[i] & 0xff;
+		cdata[i] = cdata[i] >> 8;
+		ciphertext[4 * i + 1] = cdata[i] & 0xff;
+		cdata[i] = cdata[i] >> 8;
+		ciphertext[4 * i + 0] = cdata[i] & 0xff;
+	}
+
+
+	snprintf(encrypted, 8, "$2%c$%2.2u$", minor, logr);
+	encode_base64(encrypted + 7, csalt, BCRYPT_MAXSALT);
+	encode_base64(encrypted + 7 + 22, ciphertext, 4 * BCRYPT_WORDS - 1);
+	explicit_bzero(&state, sizeof(state));
+	explicit_bzero(ciphertext, sizeof(ciphertext));
+	explicit_bzero(csalt, sizeof(csalt));
+	explicit_bzero(cdata, sizeof(cdata));
+	return 0;
+
+inval:
+	errno = EINVAL;
+	return -1;
+}
+
+/*
+ * internal utilities
+ */
+static const u_int8_t Base64Code[] =
+"./ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789";
+
+static const u_int8_t index_64[128] = {
+	255, 255, 255, 255, 255, 255, 255, 255, 255, 255,
+	255, 255, 255, 255, 255, 255, 255, 255, 255, 255,
+	255, 255, 255, 255, 255, 255, 255, 255, 255, 255,
+	255, 255, 255, 255, 255, 255, 255, 255, 255, 255,
+	255, 255, 255, 255, 255, 255, 0, 1, 54, 55,
+	56, 57, 58, 59, 60, 61, 62, 63, 255, 255,
+	255, 255, 255, 255, 255, 2, 3, 4, 5, 6,
+	7, 8, 9, 10, 11, 12, 13, 14, 15, 16,
+	17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27,
+	255, 255, 255, 255, 255, 255, 28, 29, 30,
+	31, 32, 33, 34, 35, 36, 37, 38, 39, 40,
+	41, 42, 43, 44, 45, 46, 47, 48, 49, 50,
+	51, 52, 53, 255, 255, 255, 255, 255
+};
+#define CHAR64(c)  ( (c) > 127 ? 255 : index_64[(c)])
+
+/*
+ * read buflen (after decoding) bytes of data from b64data
+ */
+static int
+decode_base64(u_int8_t *buffer, size_t len, const char *b64data)
+{
+	u_int8_t *bp = buffer;
+	const u_int8_t *p = b64data;
+	u_int8_t c1, c2, c3, c4;
+
+	while (bp < buffer + len) {
+		c1 = CHAR64(*p);
+		/* Invalid data */
+		if (c1 == 255)
+			return -1;
+
+		c2 = CHAR64(*(p + 1));
+		if (c2 == 255)
+			return -1;
+
+		*bp++ = (c1 << 2) | ((c2 & 0x30) >> 4);
+		if (bp >= buffer + len)
+			break;
+
+		c3 = CHAR64(*(p + 2));
+		if (c3 == 255)
+			return -1;
+
+		*bp++ = ((c2 & 0x0f) << 4) | ((c3 & 0x3c) >> 2);
+		if (bp >= buffer + len)
+			break;
+
+		c4 = CHAR64(*(p + 3));
+		if (c4 == 255)
+			return -1;
+		*bp++ = ((c3 & 0x03) << 6) | c4;
+
+		p += 4;
+	}
+	return 0;
+}
+
+/*
+ * Turn len bytes of data into base64 encoded data.
+ * This works without = padding.
+ */
+int
+encode_base64(char *b64buffer, const u_int8_t *data, size_t len)
+{
+	u_int8_t *bp = b64buffer;
+	const u_int8_t *p = data;
+	u_int8_t c1, c2;
+
+	while (p < data + len) {
+		c1 = *p++;
+		*bp++ = Base64Code[(c1 >> 2)];
+		c1 = (c1 & 0x03) << 4;
+		if (p >= data + len) {
+			*bp++ = Base64Code[c1];
+			break;
+		}
+		c2 = *p++;
+		c1 |= (c2 >> 4) & 0x0f;
+		*bp++ = Base64Code[c1];
+		c1 = (c2 & 0x0f) << 2;
+		if (p >= data + len) {
+			*bp++ = Base64Code[c1];
+			break;
+		}
+		c2 = *p++;
+		c1 |= (c2 >> 6) & 0x03;
+		*bp++ = Base64Code[c1];
+		*bp++ = Base64Code[c2 & 0x3f];
+	}
+	*bp = '\0';
+	return 0;
+}