Received-SPF: pass (google.com: best guess record for domain of linux-ext4-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
From:   Eric Biggers <ebiggers@kernel.org>
To:     fstests@vger.kernel.org
Cc:     linux-fscrypt@vger.kernel.org, linux-ext4@vger.kernel.org,
        linux-f2fs-devel@lists.sourceforge.net
Subject: [RFC PATCH 2/7] fscrypt-crypt-util: add utility for reproducing fscrypt encrypted data
Date:   Fri, 26 Apr 2019 13:41:48 -0700
Message-Id: <20190426204153.101861-3-ebiggers@kernel.org>
In-Reply-To: <20190426204153.101861-1-ebiggers@kernel.org>
References: <20190426204153.101861-1-ebiggers@kernel.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Sender: linux-ext4-owner@vger.kernel.org
Precedence: bulk

From: Eric Biggers <ebiggers@google.com>

Add a utility program that can reproduce encrypted contents and
filenames.  It implements all encryption algorithms currently supported
by fscrypt (a.k.a. ext4, f2fs, and ubifs encryption), and it generates
IVs in the same way.  The program takes the algorithm and master key on
the command line, and encrypts stdin to stdout.

A file nonce may also be passed on the command line, and the program
will "tweak" the encryption using this nonce in the same way the kernel
does -- either by deriving a subkey, or by including the nonce in the
IVs.  The block size and padding amount may also be specified.

No dependencies are added, as all algorithms implemented from scratch.

Signed-off-by: Eric Biggers <ebiggers@google.com>
---
 .gitignore               |    1 +
 src/Makefile             |    3 +-
 src/fscrypt-crypt-util.c | 1645 ++++++++++++++++++++++++++++++++++++++
 3 files changed, 1648 insertions(+), 1 deletion(-)
 create mode 100644 src/fscrypt-crypt-util.c

diff --git a/.gitignore b/.gitignore
index c13fb713..11232be7 100644
--- a/.gitignore
+++ b/.gitignore
@@ -73,6 +73,7 @@
 /src/fill
 /src/fill2
 /src/fs_perms
+/src/fscrypt-crypt-util
 /src/fssum
 /src/fstest
 /src/fsync-err
diff --git a/src/Makefile b/src/Makefile
index 4057d7b3..9d3d2529 100644
--- a/src/Makefile
+++ b/src/Makefile
@@ -27,7 +27,8 @@ LINUX_TARGETS = xfsctl bstat t_mtab getdevicesize preallo_rw_pattern_reader \
 	renameat2 t_getcwd e4compact test-nextquota punch-alternating \
 	attr-list-by-handle-cursor-test listxattr dio-interleaved t_dir_type \
 	dio-invalidate-cache stat_test t_encrypted_d_revalidate \
-	attr_replace_test swapon mkswap t_attr_corruption t_open_tmpfiles
+	attr_replace_test swapon mkswap t_attr_corruption t_open_tmpfiles \
+	fscrypt-crypt-util
 
 SUBDIRS = log-writes perf
 
diff --git a/src/fscrypt-crypt-util.c b/src/fscrypt-crypt-util.c
new file mode 100644
index 00000000..715e2bf2
--- /dev/null
+++ b/src/fscrypt-crypt-util.c
@@ -0,0 +1,1645 @@
+// SPDX-License-Identifier: GPL-2.0+
+/*
+ * fscrypt-crypt-util.c - utility for verifying fscrypt-encrypted data
+ *
+ * Copyright 2019 Google LLC
+ */
+
+/*
+ * This program implements all crypto algorithms supported by fscrypt (a.k.a.
+ * ext4, f2fs, and ubifs encryption), for the purpose of verifying the
+ * correctness of the ciphertext stored on-disk.  See usage() below.
+ *
+ * All algorithms are implemented in portable C code to avoid depending on
+ * libcrypto (OpenSSL), and because some fscrypt-supported algorithms aren't
+ * available in libcrypto anyway (e.g. Adiantum).  For simplicity, all crypto
+ * code here tries to follow the mathematical definitions directly, without
+ * optimizing for performance or worrying about following security best
+ * practices such as mitigating side-channel attacks.  So, only use this program
+ * for testing!
+ */
+
+#include <asm/byteorder.h>
+#include <errno.h>
+#include <getopt.h>
+#include <limits.h>
+#include <linux/types.h>
+#include <stdarg.h>
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+
+#define PROGRAM_NAME "fscrypt-crypt-util"
+
+/*
+ * Define to enable the tests of the crypto code in this file.  If enabled, you
+ * must link this program with OpenSSL (-lcrypto) v1.1.0 or later, and your
+ * kernel needs CONFIG_CRYPTO_USER_API_SKCIPHER=y and CONFIG_CRYPTO_ADIANTUM=y.
+ */
+#undef ENABLE_ALG_TESTS
+
+#define NUM_ALG_TEST_ITERATIONS	10000
+
+static void usage(FILE *fp)
+{
+	fputs(
+"Usage: " PROGRAM_NAME " [OPTION]... CIPHER MASTER_KEY\n"
+"\n"
+"Utility for verifying fscrypt-encrypted data.  This program encrypts\n"
+"(or decrypts) the data on stdin using the given CIPHER with the given\n"
+"MASTER_KEY (or a key derived from it, if a KDF is specified), and writes the\n"
+"resulting ciphertext (or plaintext) to stdout.\n"
+"\n"
+"CIPHER can be AES-256-XTS, AES-256-CTS-CBC, AES-128-CBC-ESSIV, AES-128-CTS-CBC,\n"
+"or Adiantum.  MASTER_KEY must be a hex string long enough for the cipher.\n"
+"\n"
+"WARNING: this program is only meant for testing, not for \"real\" use!\n"
+"\n"
+"Options:\n"
+"  --block-size=BLOCK_SIZE     Encrypt each BLOCK_SIZE bytes independently.\n"
+"                                Default: 4096 bytes\n"
+"  --decrypt                   Decrypt instead of encrypt\n"
+"  --file-nonce=NONCE          File's nonce as a 32-character hex string\n"
+"  --kdf=KDF                   Key derivation function to use: AES-128-ECB\n"
+"                                or none.  Default: none\n"
+"  --help                      Show this help\n"
+"  --padding=PADDING           If last block is partial, zero-pad it to next\n"
+"                                PADDING-byte boundary.  Default: BLOCK_SIZE\n"
+	, fp);
+}
+
+/*----------------------------------------------------------------------------*
+ *                                 Utilities                                  *
+ *----------------------------------------------------------------------------*/
+
+#define ARRAY_SIZE(A)		(sizeof(A) / sizeof((A)[0]))
+#define MIN(x, y)		((x) < (y) ? (x) : (y))
+#define MAX(x, y)		((x) > (y) ? (x) : (y))
+#define ROUND_DOWN(x, y)	((x) & ~((y) - 1))
+#define ROUND_UP(x, y)		(((x) + (y) - 1) & ~((y) - 1))
+#define DIV_ROUND_UP(n, d)	(((n) + (d) - 1) / (d))
+#define STATIC_ASSERT(e)	((void)sizeof(char[1 - 2*!(e)]))
+
+typedef __u8			u8;
+typedef __u16			u16;
+typedef __u32			u32;
+typedef __u64			u64;
+
+#define cpu_to_le32		__cpu_to_le32
+#define cpu_to_be32		__cpu_to_be32
+#define cpu_to_le64		__cpu_to_le64
+#define cpu_to_be64		__cpu_to_be64
+#define le32_to_cpu		__le32_to_cpu
+#define be32_to_cpu		__be32_to_cpu
+#define le64_to_cpu		__le64_to_cpu
+#define be64_to_cpu		__be64_to_cpu
+
+#define DEFINE_UNALIGNED_ACCESS_HELPERS(type, native_type)	\
+static inline native_type __attribute__((unused))		\
+get_unaligned_##type(const void *p)				\
+{								\
+	__##type x;						\
+								\
+	memcpy(&x, p, sizeof(x));				\
+	return type##_to_cpu(x);				\
+}								\
+								\
+static inline void __attribute__((unused))			\
+put_unaligned_##type(native_type v, void *p)			\
+{								\
+	__##type x = cpu_to_##type(v);				\
+								\
+	memcpy(p, &x, sizeof(x));				\
+}
+
+DEFINE_UNALIGNED_ACCESS_HELPERS(le32, u32)
+DEFINE_UNALIGNED_ACCESS_HELPERS(be32, u32)
+DEFINE_UNALIGNED_ACCESS_HELPERS(le64, u64)
+DEFINE_UNALIGNED_ACCESS_HELPERS(be64, u64)
+
+static inline bool is_power_of_2(unsigned long v)
+{
+	return v != 0 && (v & (v - 1)) == 0;
+}
+
+static inline u32 rol32(u32 v, int n)
+{
+	return (v << n) | (v >> (32 - n));
+}
+
+static inline u32 ror32(u32 v, int n)
+{
+	return (v >> n) | (v << (32 - n));
+}
+
+static inline u64 ror64(u64 v, int n)
+{
+	return (v >> n) | (v << (64 - n));
+}
+
+static inline void xor(u8 *res, const u8 *a, const u8 *b, size_t count)
+{
+	while (count--)
+		*res++ = *a++ ^ *b++;
+}
+
+static void __attribute__((noreturn, format(printf, 2, 3)))
+do_die(int err, const char *format, ...)
+{
+	va_list va;
+
+	va_start(va, format);
+	fputs("[" PROGRAM_NAME "] ERROR: ", stderr);
+	vfprintf(stderr, format, va);
+	if (err)
+		fprintf(stderr, ": %s", strerror(errno));
+	putc('\n', stderr);
+	va_end(va);
+	exit(1);
+}
+
+#define die(format, ...)	do_die(0,     (format), ##__VA_ARGS__)
+#define die_errno(format, ...)	do_die(errno, (format), ##__VA_ARGS__)
+
+static __attribute__((noreturn)) void
+assertion_failed(const char *expr, const char *file, int line)
+{
+	die("Assertion failed: %s at %s:%d", expr, file, line);
+}
+
+#define ASSERT(e) ({ if (!(e)) assertion_failed(#e, __FILE__, __LINE__); })
+
+static void *xmalloc(size_t size)
+{
+	void *p = malloc(size);
+
+	ASSERT(p != NULL);
+	return p;
+}
+
+static int hexchar2bin(char c)
+{
+	if (c >= 'a' && c <= 'f')
+		return 10 + c - 'a';
+	if (c >= 'A' && c <= 'F')
+		return 10 + c - 'A';
+	if (c >= '0' && c <= '9')
+		return c - '0';
+	return -1;
+}
+
+static int hex2bin(const char *hex, u8 *bin, int max_bin_size)
+{
+	size_t len = strlen(hex);
+	size_t i;
+
+	if (len & 1)
+		return -1;
+	len /= 2;
+	if (len > max_bin_size)
+		return -1;
+
+	for (i = 0; i < len; i++) {
+		int high = hexchar2bin(hex[2 * i]);
+		int low = hexchar2bin(hex[2 * i + 1]);
+
+		if (high < 0 || low < 0)
+			return -1;
+		bin[i] = (high << 4) | low;
+	}
+	return len;
+}
+
+static size_t xread(int fd, void *buf, size_t count)
+{
+	const size_t orig_count = count;
+
+	while (count) {
+		ssize_t res = read(fd, buf, count);
+
+		if (res < 0)
+			die_errno("read error");
+		if (res == 0)
+			break;
+		buf += res;
+		count -= res;
+	}
+	return orig_count - count;
+}
+
+static void full_write(int fd, const void *buf, size_t count)
+{
+	while (count) {
+		ssize_t res = write(fd, buf, count);
+
+		if (res < 0)
+			die_errno("write error");
+		buf += res;
+		count -= res;
+	}
+}
+
+#ifdef ENABLE_ALG_TESTS
+static void rand_bytes(u8 *buf, size_t count)
+{
+	while (count--)
+		*buf++ = rand();
+}
+#endif
+
+/*----------------------------------------------------------------------------*
+ *                          Finite field arithmetic                           *
+ *----------------------------------------------------------------------------*/
+
+/* Multiply a GF(2^8) element by the polynomial 'x' */
+static inline u8 gf2_8_mul_x(u8 b)
+{
+	return (b << 1) ^ ((b & 0x80) ? 0x1B : 0);
+}
+
+/* Multiply four packed GF(2^8) elements by the polynomial 'x' */
+static inline u32 gf2_8_mul_x_4way(u32 w)
+{
+	return ((w & 0x7F7F7F7F) << 1) ^ (((w & 0x80808080) >> 7) * 0x1B);
+}
+
+/* Element of GF(2^128) */
+typedef struct {
+	__le64 lo;
+	__le64 hi;
+} ble128;
+
+/* Multiply a GF(2^128) element by the polynomial 'x' */
+static inline void gf2_128_mul_x(ble128 *t)
+{
+	u64 lo = le64_to_cpu(t->lo);
+	u64 hi = le64_to_cpu(t->hi);
+
+	t->hi = cpu_to_le64((hi << 1) | (lo >> 63));
+	t->lo = cpu_to_le64((lo << 1) ^ ((hi & (1ULL << 63)) ? 0x87 : 0));
+}
+
+/*----------------------------------------------------------------------------*
+ *                             Group arithmetic                               *
+ *----------------------------------------------------------------------------*/
+
+/* Element of Z/(2^{128}Z)  (a.k.a. the integers modulo 2^128) */
+typedef struct {
+	__le64 lo;
+	__le64 hi;
+} le128;
+
+static inline void le128_add(le128 *res, const le128 *a, const le128 *b)
+{
+	u64 a_lo = le64_to_cpu(a->lo);
+	u64 b_lo = le64_to_cpu(b->lo);
+
+	res->lo = cpu_to_le64(a_lo + b_lo);
+	res->hi = cpu_to_le64(le64_to_cpu(a->hi) + le64_to_cpu(b->hi) +
+			      (a_lo + b_lo < a_lo));
+}
+
+static inline void le128_sub(le128 *res, const le128 *a, const le128 *b)
+{
+	u64 a_lo = le64_to_cpu(a->lo);
+	u64 b_lo = le64_to_cpu(b->lo);
+
+	res->lo = cpu_to_le64(a_lo - b_lo);
+	res->hi = cpu_to_le64(le64_to_cpu(a->hi) - le64_to_cpu(b->hi) -
+			      (a_lo - b_lo > a_lo));
+}
+
+/*----------------------------------------------------------------------------*
+ *                              AES block cipher                              *
+ *----------------------------------------------------------------------------*/
+
+/*
+ * Reference: "FIPS 197, Advanced Encryption Standard"
+ *	https://nvlpubs.nist.gov/nistpubs/fips/nist.fips.197.pdf
+ */
+
+#define AES_BLOCK_SIZE		16
+#define AES_128_KEY_SIZE	16
+#define AES_192_KEY_SIZE	24
+#define AES_256_KEY_SIZE	32
+
+static inline void AddRoundKey(u32 state[4], const u32 *rk)
+{
+	int i;
+
+	for (i = 0; i < 4; i++)
+		state[i] ^= rk[i];
+}
+
+static const u8 aes_sbox[256] = {
+	0x63, 0x7c, 0x77, 0x7b, 0xf2, 0x6b, 0x6f, 0xc5, 0x30, 0x01, 0x67, 0x2b,
+	0xfe, 0xd7, 0xab, 0x76, 0xca, 0x82, 0xc9, 0x7d, 0xfa, 0x59, 0x47, 0xf0,
+	0xad, 0xd4, 0xa2, 0xaf, 0x9c, 0xa4, 0x72, 0xc0, 0xb7, 0xfd, 0x93, 0x26,
+	0x36, 0x3f, 0xf7, 0xcc, 0x34, 0xa5, 0xe5, 0xf1, 0x71, 0xd8, 0x31, 0x15,
+	0x04, 0xc7, 0x23, 0xc3, 0x18, 0x96, 0x05, 0x9a, 0x07, 0x12, 0x80, 0xe2,
+	0xeb, 0x27, 0xb2, 0x75, 0x09, 0x83, 0x2c, 0x1a, 0x1b, 0x6e, 0x5a, 0xa0,
+	0x52, 0x3b, 0xd6, 0xb3, 0x29, 0xe3, 0x2f, 0x84, 0x53, 0xd1, 0x00, 0xed,
+	0x20, 0xfc, 0xb1, 0x5b, 0x6a, 0xcb, 0xbe, 0x39, 0x4a, 0x4c, 0x58, 0xcf,
+	0xd0, 0xef, 0xaa, 0xfb, 0x43, 0x4d, 0x33, 0x85, 0x45, 0xf9, 0x02, 0x7f,
+	0x50, 0x3c, 0x9f, 0xa8, 0x51, 0xa3, 0x40, 0x8f, 0x92, 0x9d, 0x38, 0xf5,
+	0xbc, 0xb6, 0xda, 0x21, 0x10, 0xff, 0xf3, 0xd2, 0xcd, 0x0c, 0x13, 0xec,
+	0x5f, 0x97, 0x44, 0x17, 0xc4, 0xa7, 0x7e, 0x3d, 0x64, 0x5d, 0x19, 0x73,
+	0x60, 0x81, 0x4f, 0xdc, 0x22, 0x2a, 0x90, 0x88, 0x46, 0xee, 0xb8, 0x14,
+	0xde, 0x5e, 0x0b, 0xdb, 0xe0, 0x32, 0x3a, 0x0a, 0x49, 0x06, 0x24, 0x5c,
+	0xc2, 0xd3, 0xac, 0x62, 0x91, 0x95, 0xe4, 0x79, 0xe7, 0xc8, 0x37, 0x6d,
+	0x8d, 0xd5, 0x4e, 0xa9, 0x6c, 0x56, 0xf4, 0xea, 0x65, 0x7a, 0xae, 0x08,
+	0xba, 0x78, 0x25, 0x2e, 0x1c, 0xa6, 0xb4, 0xc6, 0xe8, 0xdd, 0x74, 0x1f,
+	0x4b, 0xbd, 0x8b, 0x8a, 0x70, 0x3e, 0xb5, 0x66, 0x48, 0x03, 0xf6, 0x0e,
+	0x61, 0x35, 0x57, 0xb9, 0x86, 0xc1, 0x1d, 0x9e, 0xe1, 0xf8, 0x98, 0x11,
+	0x69, 0xd9, 0x8e, 0x94, 0x9b, 0x1e, 0x87, 0xe9, 0xce, 0x55, 0x28, 0xdf,
+	0x8c, 0xa1, 0x89, 0x0d, 0xbf, 0xe6, 0x42, 0x68, 0x41, 0x99, 0x2d, 0x0f,
+	0xb0, 0x54, 0xbb, 0x16,
+};
+
+static u8 aes_inverse_sbox[256];
+
+static void aes_init(void)
+{
+	int i;
+
+	for (i = 0; i < 256; i++)
+		aes_inverse_sbox[aes_sbox[i]] = i;
+}
+
+static inline u32 DoSubWord(u32 w, const u8 sbox[256])
+{
+	return ((u32)sbox[(u8)(w >> 24)] << 24) |
+	       ((u32)sbox[(u8)(w >> 16)] << 16) |
+	       ((u32)sbox[(u8)(w >>  8)] <<  8) |
+	       ((u32)sbox[(u8)(w >>  0)] <<  0);
+}
+
+static inline u32 SubWord(u32 w)
+{
+	return DoSubWord(w, aes_sbox);
+}
+
+static inline u32 InvSubWord(u32 w)
+{
+	return DoSubWord(w, aes_inverse_sbox);
+}
+
+static inline void SubBytes(u32 state[4])
+{
+	int i;
+
+	for (i = 0; i < 4; i++)
+		state[i] = SubWord(state[i]);
+}
+
+static inline void InvSubBytes(u32 state[4])
+{
+	int i;
+
+	for (i = 0; i < 4; i++)
+		state[i] = InvSubWord(state[i]);
+}
+
+static inline void DoShiftRows(u32 state[4], int direction)
+{
+	u32 newstate[4];
+	int i;
+
+	for (i = 0; i < 4; i++)
+		newstate[i] = (state[(i + direction*0) & 3] & 0xff) |
+			      (state[(i + direction*1) & 3] & 0xff00) |
+			      (state[(i + direction*2) & 3] & 0xff0000) |
+			      (state[(i + direction*3) & 3] & 0xff000000);
+	memcpy(state, newstate, 16);
+}
+
+static inline void ShiftRows(u32 state[4])
+{
+	DoShiftRows(state, 1);
+}
+
+static inline void InvShiftRows(u32 state[4])
+{
+	DoShiftRows(state, -1);
+}
+
+/*
+ * Mix one column by doing the following matrix multiplication in GF(2^8):
+ *
+ *     | 2 3 1 1 |   | w[0] |
+ *     | 1 2 3 1 |   | w[1] |
+ *     | 1 1 2 3 | x | w[2] |
+ *     | 3 1 1 2 |   | w[3] |
+ *
+ * a.k.a. w[i] = 2*w[i] + 3*w[(i+1)%4] + w[(i+2)%4] + w[(i+3)%4]
+ */
+static inline u32 MixColumn(u32 w)
+{
+	u32 _2w0_w2 = gf2_8_mul_x_4way(w) ^ ror32(w, 16);
+	u32 _3w1_w3 = ror32(_2w0_w2 ^ w, 8);
+
+	return _2w0_w2 ^ _3w1_w3;
+}
+
+/*
+ *           ( | 5 0 4 0 |   | w[0] | )
+ *          (  | 0 5 0 4 |   | w[1] |  )
+ * MixColumn(  | 4 0 5 0 | x | w[2] |  )
+ *           ( | 0 4 0 5 |   | w[3] | )
+ */
+static inline u32 InvMixColumn(u32 w)
+{
+	u32 _4w = gf2_8_mul_x_4way(gf2_8_mul_x_4way(w));
+
+	return MixColumn(_4w ^ w ^ ror32(_4w, 16));
+}
+
+static inline void MixColumns(u32 state[4])
+{
+	int i;
+
+	for (i = 0; i < 4; i++)
+		state[i] = MixColumn(state[i]);
+}
+
+static inline void InvMixColumns(u32 state[4])
+{
+	int i;
+
+	for (i = 0; i < 4; i++)
+		state[i] = InvMixColumn(state[i]);
+}
+
+struct aes_key {
+	u32 round_keys[15 * 4];
+	int nrounds;
+};
+
+/* Expand an AES key */
+static void aes_setkey(struct aes_key *k, const u8 *key, int keysize)
+{
+	const int N = keysize / 4;
+	u32 * const rk = k->round_keys;
+	u8 rcon = 1;
+	int i;
+
+	ASSERT(keysize == 16 || keysize == 24 || keysize == 32);
+	k->nrounds = 6 + N;
+	for (i = 0; i < 4 * (k->nrounds + 1); i++) {
+		if (i < N) {
+			rk[i] = get_unaligned_le32(&key[i * sizeof(__le32)]);
+		} else if (i % N == 0) {
+			rk[i] = rk[i - N] ^ SubWord(ror32(rk[i - 1], 8)) ^ rcon;
+			rcon = gf2_8_mul_x(rcon);
+		} else if (N > 6 && i % N == 4) {
+			rk[i] = rk[i - N] ^ SubWord(rk[i - 1]);
+		} else {
+			rk[i] = rk[i - N] ^ rk[i - 1];
+		}
+	}
+}
+
+/* Encrypt one 16-byte block with AES */
+static void aes_encrypt(const struct aes_key *k, const u8 src[AES_BLOCK_SIZE],
+			u8 dst[AES_BLOCK_SIZE])
+{
+	u32 state[4];
+	int i;
+
+	for (i = 0; i < 4; i++)
+		state[i] = get_unaligned_le32(&src[i * sizeof(__le32)]);
+
+	AddRoundKey(state, k->round_keys);
+	for (i = 1; i < k->nrounds; i++) {
+		SubBytes(state);
+		ShiftRows(state);
+		MixColumns(state);
+		AddRoundKey(state, &k->round_keys[4 * i]);
+	}
+	SubBytes(state);
+	ShiftRows(state);
+	AddRoundKey(state, &k->round_keys[4 * i]);
+
+	for (i = 0; i < 4; i++)
+		put_unaligned_le32(state[i], &dst[i * sizeof(__le32)]);
+}
+
+/* Decrypt one 16-byte block with AES */
+static void aes_decrypt(const struct aes_key *k, const u8 src[AES_BLOCK_SIZE],
+			u8 dst[AES_BLOCK_SIZE])
+{
+	u32 state[4];
+	int i;
+
+	for (i = 0; i < 4; i++)
+		state[i] = get_unaligned_le32(&src[i * sizeof(__le32)]);
+
+	AddRoundKey(state, &k->round_keys[4 * k->nrounds]);
+	InvShiftRows(state);
+	InvSubBytes(state);
+	for (i = k->nrounds - 1; i >= 1; i--) {
+		AddRoundKey(state, &k->round_keys[4 * i]);
+		InvMixColumns(state);
+		InvShiftRows(state);
+		InvSubBytes(state);
+	}
+	AddRoundKey(state, k->round_keys);
+
+	for (i = 0; i < 4; i++)
+		put_unaligned_le32(state[i], &dst[i * sizeof(__le32)]);
+}
+
+#ifdef ENABLE_ALG_TESTS
+#include <openssl/aes.h>
+static void test_aes_keysize(int keysize)
+{
+	unsigned long num_tests = NUM_ALG_TEST_ITERATIONS;
+
+	while (num_tests--) {
+		struct aes_key k;
+		AES_KEY ref_k;
+		u8 key[AES_256_KEY_SIZE];
+		u8 ptext[AES_BLOCK_SIZE];
+		u8 ctext[AES_BLOCK_SIZE];
+		u8 ref_ctext[AES_BLOCK_SIZE];
+		u8 decrypted[AES_BLOCK_SIZE];
+
+		rand_bytes(key, keysize);
+		rand_bytes(ptext, AES_BLOCK_SIZE);
+
+		aes_setkey(&k, key, keysize);
+		aes_encrypt(&k, ptext, ctext);
+
+		ASSERT(AES_set_encrypt_key(key, keysize*8, &ref_k) == 0);
+		AES_encrypt(ptext, ref_ctext, &ref_k);
+
+		ASSERT(memcmp(ctext, ref_ctext, AES_BLOCK_SIZE) == 0);
+
+		aes_decrypt(&k, ctext, decrypted);
+		ASSERT(memcmp(ptext, decrypted, AES_BLOCK_SIZE) == 0);
+	}
+}
+
+static void test_aes(void)
+{
+	test_aes_keysize(AES_128_KEY_SIZE);
+	test_aes_keysize(AES_192_KEY_SIZE);
+	test_aes_keysize(AES_256_KEY_SIZE);
+}
+#endif /* ENABLE_ALG_TESTS */
+
+/*----------------------------------------------------------------------------*
+ *                                  SHA-256                                   *
+ *----------------------------------------------------------------------------*/
+
+/*
+ * Reference: "FIPS 180-2, Secure Hash Standard"
+ *	https://csrc.nist.gov/csrc/media/publications/fips/180/2/archive/2002-08-01/documents/fips180-2withchangenotice.pdf
+ */
+
+#define SHA256_DIGEST_SIZE	32
+#define SHA256_BLOCK_SIZE	64
+
+#define Ch(x, y, z)	(((x) & (y)) ^ (~(x) & (z)))
+#define Maj(x, y, z)	(((x) & (y)) ^ ((x) & (z)) ^ ((y) & (z)))
+#define Sigma256_0(x)	(ror32((x),  2) ^ ror32((x), 13) ^ ror32((x), 22))
+#define Sigma256_1(x)	(ror32((x),  6) ^ ror32((x), 11) ^ ror32((x), 25))
+#define sigma256_0(x)	(ror32((x),  7) ^ ror32((x), 18) ^ ((x) >>  3))
+#define sigma256_1(x)	(ror32((x), 17) ^ ror32((x), 19) ^ ((x) >> 10))
+
+static const u32 sha256_iv[8] = {
+	0x6a09e667, 0xbb67ae85, 0x3c6ef372, 0xa54ff53a, 0x510e527f, 0x9b05688c,
+	0x1f83d9ab, 0x5be0cd19,
+};
+
+static const u32 sha256_round_constants[64] = {
+	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,
+};
+
+/* Compute the SHA-256 digest of the given buffer */
+static void sha256(const u8 *in, size_t inlen, u8 out[SHA256_DIGEST_SIZE])
+{
+	const size_t msglen = ROUND_UP(inlen + 9, SHA256_BLOCK_SIZE);
+	u8 * const msg = xmalloc(msglen);
+	u32 H[8];
+	int i;
+
+	/* super naive way of handling the padding */
+	memcpy(msg, in, inlen);
+	memset(&msg[inlen], 0, msglen - inlen);
+	msg[inlen] = 0x80;
+	put_unaligned_be64((u64)inlen * 8, &msg[msglen - sizeof(__be64)]);
+	in = msg;
+
+	memcpy(H, sha256_iv, sizeof(H));
+	do {
+		u32 a = H[0], b = H[1], c = H[2], d = H[3],
+		    e = H[4], f = H[5], g = H[6], h = H[7];
+		u32 W[64];
+
+		for (i = 0; i < 16; i++)
+			W[i] = get_unaligned_be32(&in[i * sizeof(__be32)]);
+		for (; i < ARRAY_SIZE(W); i++)
+			W[i] = sigma256_1(W[i - 2]) + W[i - 7] +
+			       sigma256_0(W[i - 15]) + W[i - 16];
+		for (i = 0; i < ARRAY_SIZE(W); i++) {
+			u32 T1 = h + Sigma256_1(e) + Ch(e, f, g) +
+				 sha256_round_constants[i] + W[i];
+			u32 T2 = Sigma256_0(a) + Maj(a, b, c);
+
+			h = g; g = f; f = e; e = d + T1;
+			d = c; c = b; b = a; a = T1 + T2;
+		}
+		H[0] += a; H[1] += b; H[2] += c; H[3] += d;
+		H[4] += e; H[5] += f; H[6] += g; H[7] += h;
+	} while ((in += SHA256_BLOCK_SIZE) != &msg[msglen]);
+
+	for (i = 0; i < ARRAY_SIZE(H); i++)
+		put_unaligned_be32(H[i], &out[i * sizeof(__be32)]);
+	free(msg);
+}
+
+#ifdef ENABLE_ALG_TESTS
+#include <openssl/sha.h>
+static void test_sha2(void)
+{
+	unsigned long num_tests = NUM_ALG_TEST_ITERATIONS;
+
+	while (num_tests--) {
+		u8 in[4096];
+		u8 digest[SHA256_DIGEST_SIZE];
+		u8 ref_digest[SHA256_DIGEST_SIZE];
+		const size_t inlen = rand() % (1 + sizeof(in));
+
+		rand_bytes(in, inlen);
+
+		sha256(in, inlen, digest);
+		SHA256(in, inlen, ref_digest);
+		ASSERT(memcmp(digest, ref_digest, SHA256_DIGEST_SIZE) == 0);
+	}
+}
+#endif /* ENABLE_ALG_TESTS */
+
+/*----------------------------------------------------------------------------*
+ *                            AES encryption modes                            *
+ *----------------------------------------------------------------------------*/
+
+static void aes_256_xts_crypt(const u8 key[2 * AES_256_KEY_SIZE],
+			      const u8 iv[AES_BLOCK_SIZE], const u8 *src,
+			      u8 *dst, size_t nbytes, bool decrypting)
+{
+	struct aes_key tweak_key, cipher_key;
+	ble128 t;
+	size_t i;
+
+	ASSERT(nbytes % AES_BLOCK_SIZE == 0);
+	aes_setkey(&cipher_key, key, AES_256_KEY_SIZE);
+	aes_setkey(&tweak_key, &key[AES_256_KEY_SIZE], AES_256_KEY_SIZE);
+	aes_encrypt(&tweak_key, iv, (u8 *)&t);
+	for (i = 0; i < nbytes; i += AES_BLOCK_SIZE) {
+		xor(&dst[i], &src[i], (const u8 *)&t, AES_BLOCK_SIZE);
+		if (decrypting)
+			aes_decrypt(&cipher_key, &dst[i], &dst[i]);
+		else
+			aes_encrypt(&cipher_key, &dst[i], &dst[i]);
+		xor(&dst[i], &dst[i], (const u8 *)&t, AES_BLOCK_SIZE);
+		gf2_128_mul_x(&t);
+	}
+}
+
+static void aes_256_xts_encrypt(const u8 key[2 * AES_256_KEY_SIZE],
+				const u8 iv[AES_BLOCK_SIZE], const u8 *src,
+				u8 *dst, size_t nbytes)
+{
+	aes_256_xts_crypt(key, iv, src, dst, nbytes, false);
+}
+
+static void aes_256_xts_decrypt(const u8 key[2 * AES_256_KEY_SIZE],
+				const u8 iv[AES_BLOCK_SIZE], const u8 *src,
+				u8 *dst, size_t nbytes)
+{
+	aes_256_xts_crypt(key, iv, src, dst, nbytes, true);
+}
+
+#ifdef ENABLE_ALG_TESTS
+#include <openssl/evp.h>
+static void test_aes_256_xts(void)
+{
+	unsigned long num_tests = NUM_ALG_TEST_ITERATIONS;
+	EVP_CIPHER_CTX *ctx = EVP_CIPHER_CTX_new();
+
+	ASSERT(ctx != NULL);
+	while (num_tests--) {
+		u8 key[2 * AES_256_KEY_SIZE];
+		u8 iv[AES_BLOCK_SIZE];
+		u8 ptext[512];
+		u8 actual_ctext[sizeof(ptext)];
+		u8 expected_ctext[sizeof(ptext)];
+		u8 decrypted[sizeof(ptext)];
+		const size_t datalen = ROUND_DOWN(rand() % (1 + sizeof(ptext)),
+						  AES_BLOCK_SIZE);
+		int outl;
+
+		rand_bytes(key, sizeof(key));
+		rand_bytes(iv, sizeof(iv));
+		rand_bytes(ptext, datalen);
+
+		aes_256_xts_encrypt(key, iv, ptext, actual_ctext, datalen);
+		ASSERT(EVP_EncryptInit_ex(ctx, EVP_aes_256_xts(),
+					  NULL, key, iv) > 0);
+		ASSERT(EVP_EncryptUpdate(ctx, expected_ctext, &outl,
+					 ptext, datalen) > 0);
+		ASSERT(outl == datalen);
+		ASSERT(memcmp(actual_ctext, expected_ctext, datalen) == 0);
+
+		aes_256_xts_decrypt(key, iv, actual_ctext, decrypted, datalen);
+		ASSERT(memcmp(ptext, decrypted, datalen) == 0);
+	}
+	EVP_CIPHER_CTX_free(ctx);
+}
+#endif /* ENABLE_ALG_TESTS */
+
+static void aes_cbc_encrypt(const struct aes_key *k,
+			    const u8 iv[AES_BLOCK_SIZE],
+			    const u8 *src, u8 *dst, size_t nbytes)
+{
+	size_t i;
+
+	ASSERT(nbytes % AES_BLOCK_SIZE == 0);
+	for (i = 0; i < nbytes; i += AES_BLOCK_SIZE) {
+		xor(&dst[i], &src[i], (i == 0 ? iv : &dst[i - AES_BLOCK_SIZE]),
+		    AES_BLOCK_SIZE);
+		aes_encrypt(k, &dst[i], &dst[i]);
+	}
+}
+
+static void aes_cbc_decrypt(const struct aes_key *k,
+			    const u8 iv[AES_BLOCK_SIZE],
+			    const u8 *src, u8 *dst, size_t nbytes)
+{
+	size_t i = nbytes;
+
+	ASSERT(i % AES_BLOCK_SIZE == 0);
+	while (i) {
+		i -= AES_BLOCK_SIZE;
+		aes_decrypt(k, &src[i], &dst[i]);
+		xor(&dst[i], &dst[i], (i == 0 ? iv : &src[i - AES_BLOCK_SIZE]),
+		    AES_BLOCK_SIZE);
+	}
+}
+
+static void aes_cts_cbc_encrypt(const u8 *key, int keysize,
+				const u8 iv[AES_BLOCK_SIZE],
+				const u8 *src, u8 *dst, size_t nbytes)
+{
+	const size_t offset = ROUND_DOWN(nbytes - 1, AES_BLOCK_SIZE);
+	const size_t final_bsize = nbytes - offset;
+	struct aes_key k;
+	u8 *pad;
+	u8 buf[AES_BLOCK_SIZE];
+
+	ASSERT(nbytes >= AES_BLOCK_SIZE);
+
+	aes_setkey(&k, key, keysize);
+
+	if (nbytes == AES_BLOCK_SIZE)
+		return aes_cbc_encrypt(&k, iv, src, dst, nbytes);
+
+	aes_cbc_encrypt(&k, iv, src, dst, offset);
+	pad = &dst[offset - AES_BLOCK_SIZE];
+
+	memcpy(buf, pad, AES_BLOCK_SIZE);
+	xor(buf, buf, &src[offset], final_bsize);
+	memcpy(&dst[offset], pad, final_bsize);
+	aes_encrypt(&k, buf, pad);
+}
+
+static void aes_cts_cbc_decrypt(const u8 *key, int keysize,
+				const u8 iv[AES_BLOCK_SIZE],
+				const u8 *src, u8 *dst, size_t nbytes)
+{
+	const size_t offset = ROUND_DOWN(nbytes - 1, AES_BLOCK_SIZE);
+	const size_t final_bsize = nbytes - offset;
+	struct aes_key k;
+	u8 *pad;
+
+	ASSERT(nbytes >= AES_BLOCK_SIZE);
+
+	aes_setkey(&k, key, keysize);
+
+	if (nbytes == AES_BLOCK_SIZE)
+		return aes_cbc_decrypt(&k, iv, src, dst, nbytes);
+
+	pad = &dst[offset - AES_BLOCK_SIZE];
+	aes_decrypt(&k, &src[offset - AES_BLOCK_SIZE], pad);
+	xor(&dst[offset], &src[offset], pad, final_bsize);
+	xor(pad, pad, &dst[offset], final_bsize);
+
+	aes_cbc_decrypt(&k, (offset == AES_BLOCK_SIZE ?
+			     iv : &src[offset - 2 * AES_BLOCK_SIZE]),
+			pad, pad, AES_BLOCK_SIZE);
+	aes_cbc_decrypt(&k, iv, src, dst, offset - AES_BLOCK_SIZE);
+}
+
+static void aes_256_cts_cbc_encrypt(const u8 key[AES_256_KEY_SIZE],
+				    const u8 iv[AES_BLOCK_SIZE],
+				    const u8 *src, u8 *dst, size_t nbytes)
+{
+	aes_cts_cbc_encrypt(key, AES_256_KEY_SIZE, iv, src, dst, nbytes);
+}
+
+static void aes_256_cts_cbc_decrypt(const u8 key[AES_256_KEY_SIZE],
+				    const u8 iv[AES_BLOCK_SIZE],
+				    const u8 *src, u8 *dst, size_t nbytes)
+{
+	aes_cts_cbc_decrypt(key, AES_256_KEY_SIZE, iv, src, dst, nbytes);
+}
+
+#ifdef ENABLE_ALG_TESTS
+#include <openssl/modes.h>
+static void aes_block128_f(const unsigned char in[16],
+			   unsigned char out[16], const void *key)
+{
+	aes_encrypt(key, in, out);
+}
+
+static void test_aes_256_cts_cbc(void)
+{
+	unsigned long num_tests = NUM_ALG_TEST_ITERATIONS;
+
+	while (num_tests--) {
+		u8 key[AES_256_KEY_SIZE];
+		u8 iv[AES_BLOCK_SIZE];
+		u8 iv_copy[AES_BLOCK_SIZE];
+		u8 ptext[512];
+		u8 actual_ctext[sizeof(ptext)];
+		u8 expected_ctext[sizeof(ptext)];
+		u8 decrypted[sizeof(ptext)];
+		const size_t datalen = 16 + (rand() % (sizeof(ptext) - 15));
+		struct aes_key k;
+
+		rand_bytes(key, sizeof(key));
+		rand_bytes(iv, sizeof(iv));
+		rand_bytes(ptext, datalen);
+
+		aes_256_cts_cbc_encrypt(key, iv, ptext, actual_ctext, datalen);
+
+		/* OpenSSL doesn't allow datalen=AES_BLOCK_SIZE; Linux does */
+		if (datalen != AES_BLOCK_SIZE) {
+			aes_setkey(&k, key, sizeof(key));
+			memcpy(iv_copy, iv, sizeof(iv));
+			ASSERT(CRYPTO_cts128_encrypt_block(ptext,
+							   expected_ctext,
+							   datalen, &k, iv_copy,
+							   aes_block128_f)
+			       == datalen);
+			ASSERT(memcmp(actual_ctext, expected_ctext,
+				      datalen) == 0);
+		}
+		aes_256_cts_cbc_decrypt(key, iv, actual_ctext, decrypted,
+					datalen);
+		ASSERT(memcmp(ptext, decrypted, datalen) == 0);
+	}
+}
+#endif /* ENABLE_ALG_TESTS */
+
+static void essiv_generate_iv(const u8 orig_key[AES_128_KEY_SIZE],
+			      const u8 orig_iv[AES_BLOCK_SIZE],
+			      u8 real_iv[AES_BLOCK_SIZE])
+{
+	u8 essiv_key[SHA256_DIGEST_SIZE];
+	struct aes_key essiv;
+
+	/* AES encrypt the original IV using a hash of the original key */
+	STATIC_ASSERT(SHA256_DIGEST_SIZE == AES_256_KEY_SIZE);
+	sha256(orig_key, AES_128_KEY_SIZE, essiv_key);
+	aes_setkey(&essiv, essiv_key, AES_256_KEY_SIZE);
+	aes_encrypt(&essiv, orig_iv, real_iv);
+}
+
+static void aes_128_cbc_essiv_encrypt(const u8 key[AES_128_KEY_SIZE],
+				      const u8 iv[AES_BLOCK_SIZE],
+				      const u8 *src, u8 *dst, size_t nbytes)
+{
+	struct aes_key k;
+	u8 real_iv[AES_BLOCK_SIZE];
+
+	aes_setkey(&k, key, AES_128_KEY_SIZE);
+	essiv_generate_iv(key, iv, real_iv);
+	aes_cbc_encrypt(&k, real_iv, src, dst, nbytes);
+}
+
+static void aes_128_cbc_essiv_decrypt(const u8 key[AES_128_KEY_SIZE],
+				      const u8 iv[AES_BLOCK_SIZE],
+				      const u8 *src, u8 *dst, size_t nbytes)
+{
+	struct aes_key k;
+	u8 real_iv[AES_BLOCK_SIZE];
+
+	aes_setkey(&k, key, AES_128_KEY_SIZE);
+	essiv_generate_iv(key, iv, real_iv);
+	aes_cbc_decrypt(&k, real_iv, src, dst, nbytes);
+}
+
+static void aes_128_cts_cbc_encrypt(const u8 key[AES_128_KEY_SIZE],
+				    const u8 iv[AES_BLOCK_SIZE],
+				    const u8 *src, u8 *dst, size_t nbytes)
+{
+	aes_cts_cbc_encrypt(key, AES_128_KEY_SIZE, iv, src, dst, nbytes);
+}
+
+static void aes_128_cts_cbc_decrypt(const u8 key[AES_128_KEY_SIZE],
+				    const u8 iv[AES_BLOCK_SIZE],
+				    const u8 *src, u8 *dst, size_t nbytes)
+{
+	aes_cts_cbc_decrypt(key, AES_128_KEY_SIZE, iv, src, dst, nbytes);
+}
+
+/*----------------------------------------------------------------------------*
+ *                           XChaCha12 stream cipher                          *
+ *----------------------------------------------------------------------------*/
+
+/*
+ * References:
+ *   - "XChaCha: eXtended-nonce ChaCha and AEAD_XChaCha20_Poly1305"
+ *	https://tools.ietf.org/html/draft-arciszewski-xchacha-03
+ *
+ *   - "ChaCha, a variant of Salsa20"
+ *	https://cr.yp.to/chacha/chacha-20080128.pdf
+ *
+ *   - "Extending the Salsa20 nonce"
+ *	https://cr.yp.to/snuffle/xsalsa-20081128.pdf
+ */
+
+#define CHACHA_KEY_SIZE		32
+#define XCHACHA_KEY_SIZE	CHACHA_KEY_SIZE
+#define XCHACHA_NONCE_SIZE	24
+
+static void chacha_init_state(u32 state[16], const u8 key[CHACHA_KEY_SIZE],
+			      const u8 iv[16])
+{
+	static const u8 consts[16] = "expand 32-byte k";
+	int i;
+
+	for (i = 0; i < 4; i++)
+		state[i] = get_unaligned_le32(&consts[i * sizeof(__le32)]);
+	for (i = 0; i < 8; i++)
+		state[4 + i] = get_unaligned_le32(&key[i * sizeof(__le32)]);
+	for (i = 0; i < 4; i++)
+		state[12 + i] = get_unaligned_le32(&iv[i * sizeof(__le32)]);
+}
+
+#define CHACHA_QUARTERROUND(a, b, c, d)		\
+	do {					\
+		a += b; d = rol32(d ^ a, 16);	\
+		c += d; b = rol32(b ^ c, 12);	\
+		a += b; d = rol32(d ^ a, 8);	\
+		c += d; b = rol32(b ^ c, 7);	\
+	} while (0)
+
+static void chacha_permute(u32 x[16], int nrounds)
+{
+	do {
+		/* column round */
+		CHACHA_QUARTERROUND(x[0], x[4], x[8], x[12]);
+		CHACHA_QUARTERROUND(x[1], x[5], x[9], x[13]);
+		CHACHA_QUARTERROUND(x[2], x[6], x[10], x[14]);
+		CHACHA_QUARTERROUND(x[3], x[7], x[11], x[15]);
+
+		/* diagonal round */
+		CHACHA_QUARTERROUND(x[0], x[5], x[10], x[15]);
+		CHACHA_QUARTERROUND(x[1], x[6], x[11], x[12]);
+		CHACHA_QUARTERROUND(x[2], x[7], x[8], x[13]);
+		CHACHA_QUARTERROUND(x[3], x[4], x[9], x[14]);
+	} while ((nrounds -= 2) != 0);
+}
+
+static void xchacha(const u8 key[XCHACHA_KEY_SIZE],
+		    const u8 nonce[XCHACHA_NONCE_SIZE],
+		    const u8 *src, u8 *dst, size_t nbytes, int nrounds)
+{
+	u32 state[16];
+	u8 real_key[CHACHA_KEY_SIZE];
+	u8 real_iv[16] = { 0 };
+	size_t i, j;
+
+	/* Compute real key using original key and first 128 nonce bits */
+	chacha_init_state(state, key, nonce);
+	chacha_permute(state, nrounds);
+	for (i = 0; i < 8; i++) /* state words 0..3, 12..15 */
+		put_unaligned_le32(state[(i < 4 ? 0 : 8) + i],
+				   &real_key[i * sizeof(__le32)]);
+
+	/* Now do regular ChaCha, using real key and remaining nonce bits */
+	memcpy(&real_iv[8], nonce + 16, 8);
+	chacha_init_state(state, real_key, real_iv);
+	for (i = 0; i < nbytes; i += 64) {
+		u32 x[16];
+		__le32 keystream[16];
+
+		memcpy(x, state, 64);
+		chacha_permute(x, nrounds);
+		for (j = 0; j < 16; j++)
+			keystream[j] = cpu_to_le32(x[j] + state[j]);
+		xor(&dst[i], &src[i], (u8 *)keystream, MIN(nbytes - i, 64));
+		if (++state[12] == 0)
+			state[13]++;
+	}
+}
+
+static void xchacha12(const u8 key[XCHACHA_KEY_SIZE],
+		      const u8 nonce[XCHACHA_NONCE_SIZE],
+		      const u8 *src, u8 *dst, size_t nbytes)
+{
+	xchacha(key, nonce, src, dst, nbytes, 12);
+}
+
+/*----------------------------------------------------------------------------*
+ *                                 Poly1305                                   *
+ *----------------------------------------------------------------------------*/
+
+/*
+ * Note: this is only the Poly1305 ε-almost-∆-universal hash function, not the
+ * full Poly1305 MAC.  I.e., it doesn't add anything at the end.
+ */
+
+#define POLY1305_KEY_SIZE	16
+#define POLY1305_BLOCK_SIZE	16
+
+static void poly1305(const u8 key[POLY1305_KEY_SIZE],
+		     const u8 *msg, size_t msglen, le128 *out)
+{
+	const u32 limb_mask = 0x3ffffff;	/* limbs are base 2^26 */
+	const u64 r0 = (get_unaligned_le32(key +  0) >> 0) & 0x3ffffff;
+	const u64 r1 = (get_unaligned_le32(key +  3) >> 2) & 0x3ffff03;
+	const u64 r2 = (get_unaligned_le32(key +  6) >> 4) & 0x3ffc0ff;
+	const u64 r3 = (get_unaligned_le32(key +  9) >> 6) & 0x3f03fff;
+	const u64 r4 = (get_unaligned_le32(key + 12) >> 8) & 0x00fffff;
+	u32 h0 = 0, h1 = 0, h2 = 0, h3 = 0, h4 = 0;
+	u32 g0, g1, g2, g3, g4, ge_p_mask;
+
+	/* Partial block support is not necessary for Adiantum */
+	ASSERT(msglen % POLY1305_BLOCK_SIZE == 0);
+
+	while (msglen) {
+		u64 d0, d1, d2, d3, d4;
+
+		/* h += *msg */
+		h0 += (get_unaligned_le32(msg +  0) >> 0) & limb_mask;
+		h1 += (get_unaligned_le32(msg +  3) >> 2) & limb_mask;
+		h2 += (get_unaligned_le32(msg +  6) >> 4) & limb_mask;
+		h3 += (get_unaligned_le32(msg +  9) >> 6) & limb_mask;
+		h4 += (get_unaligned_le32(msg + 12) >> 8) | (1 << 24);
+
+		/* h *= r */
+		d0 = h0*r0 + h1*5*r4 + h2*5*r3 + h3*5*r2 + h4*5*r1;
+		d1 = h0*r1 + h1*r0   + h2*5*r4 + h3*5*r3 + h4*5*r2;
+		d2 = h0*r2 + h1*r1   + h2*r0   + h3*5*r4 + h4*5*r3;
+		d3 = h0*r3 + h1*r2   + h2*r1   + h3*r0   + h4*5*r4;
+		d4 = h0*r4 + h1*r3   + h2*r2   + h3*r1   + h4*r0;
+
+		/* (partial) h %= 2^130 - 5 */
+		d1 += d0 >> 26;		h0 = d0 & limb_mask;
+		d2 += d1 >> 26;		h1 = d1 & limb_mask;
+		d3 += d2 >> 26;		h2 = d2 & limb_mask;
+		d4 += d3 >> 26;		h3 = d3 & limb_mask;
+		h0 += (d4 >> 26) * 5;	h4 = d4 & limb_mask;
+		h1 += h0 >> 26;		h0 &= limb_mask;
+
+		msg += POLY1305_BLOCK_SIZE;
+		msglen -= POLY1305_BLOCK_SIZE;
+	}
+
+	/* fully carry h */
+	h2 += (h1 >> 26);	h1 &= limb_mask;
+	h3 += (h2 >> 26);	h2 &= limb_mask;
+	h4 += (h3 >> 26);	h3 &= limb_mask;
+	h0 += (h4 >> 26) * 5;	h4 &= limb_mask;
+	h1 += (h0 >> 26);	h0 &= limb_mask;
+
+	/* if (h >= 2^130 - 5) h -= 2^130 - 5; */
+	g0 = h0 + 5;
+	g1 = h1 + (g0 >> 26);	g0 &= limb_mask;
+	g2 = h2 + (g1 >> 26);	g1 &= limb_mask;
+	g3 = h3 + (g2 >> 26);	g2 &= limb_mask;
+	g4 = h4 + (g3 >> 26);	g3 &= limb_mask;
+	ge_p_mask = ~((g4 >> 26) - 1); /* all 1's if h >= 2^130 - 5, else 0 */
+	h0 = (h0 & ~ge_p_mask) | (g0 & ge_p_mask);
+	h1 = (h1 & ~ge_p_mask) | (g1 & ge_p_mask);
+	h2 = (h2 & ~ge_p_mask) | (g2 & ge_p_mask);
+	h3 = (h3 & ~ge_p_mask) | (g3 & ge_p_mask);
+	h4 = (h4 & ~ge_p_mask) | (g4 & ge_p_mask & limb_mask);
+
+	/* h %= 2^128 */
+	out->lo = cpu_to_le64(((u64)h2 << 52) | ((u64)h1 << 26) | h0);
+	out->hi = cpu_to_le64(((u64)h4 << 40) | ((u64)h3 << 14) | (h2 >> 12));
+}
+
+/*----------------------------------------------------------------------------*
+ *                          Adiantum encryption mode                          *
+ *----------------------------------------------------------------------------*/
+
+/*
+ * Reference: "Adiantum: length-preserving encryption for entry-level processors"
+ *	https://tosc.iacr.org/index.php/ToSC/article/view/7360
+ */
+
+#define ADIANTUM_KEY_SIZE	32
+#define ADIANTUM_IV_SIZE	32
+#define ADIANTUM_HASH_KEY_SIZE	((2 * POLY1305_KEY_SIZE) + NH_KEY_SIZE)
+
+#define NH_KEY_SIZE		1072
+#define NH_KEY_WORDS		(NH_KEY_SIZE / sizeof(u32))
+#define NH_BLOCK_SIZE		1024
+#define NH_HASH_SIZE		32
+#define NH_MESSAGE_UNIT		16
+
+static u64 nh_pass(const u32 *key, const u8 *msg, size_t msglen)
+{
+	u64 sum = 0;
+
+	ASSERT(msglen % NH_MESSAGE_UNIT == 0);
+	while (msglen) {
+		sum += (u64)(u32)(get_unaligned_le32(msg +  0) + key[0]) *
+			    (u32)(get_unaligned_le32(msg +  8) + key[2]);
+		sum += (u64)(u32)(get_unaligned_le32(msg +  4) + key[1]) *
+			    (u32)(get_unaligned_le32(msg + 12) + key[3]);
+		key += NH_MESSAGE_UNIT / sizeof(key[0]);
+		msg += NH_MESSAGE_UNIT;
+		msglen -= NH_MESSAGE_UNIT;
+	}
+	return sum;
+}
+
+/* NH ε-almost-universal hash function */
+static void nh(const u32 *key, const u8 *msg, size_t msglen,
+	       u8 result[NH_HASH_SIZE])
+{
+	size_t i;
+
+	for (i = 0; i < NH_HASH_SIZE; i += sizeof(__le64)) {
+		put_unaligned_le64(nh_pass(key, msg, msglen), &result[i]);
+		key += NH_MESSAGE_UNIT / sizeof(key[0]);
+	}
+}
+
+/* Adiantum's ε-almost-∆-universal hash function */
+static void adiantum_hash(const u8 key[ADIANTUM_HASH_KEY_SIZE],
+			  const u8 iv[ADIANTUM_IV_SIZE],
+			  const u8 *msg, size_t msglen, le128 *result)
+{
+	const u8 *header_poly_key = key;
+	const u8 *msg_poly_key = header_poly_key + POLY1305_KEY_SIZE;
+	const u8 *nh_key = msg_poly_key + POLY1305_KEY_SIZE;
+	u32 nh_key_words[NH_KEY_WORDS];
+	u8 header[ADIANTUM_IV_SIZE + POLY1305_BLOCK_SIZE];
+	const size_t num_nh_blocks = DIV_ROUND_UP(msglen, NH_BLOCK_SIZE);
+	u8 *nh_hashes = xmalloc(num_nh_blocks * NH_HASH_SIZE);
+	const size_t padded_msglen = ROUND_UP(msglen, NH_MESSAGE_UNIT);
+	u8 *padded_msg = xmalloc(padded_msglen);
+	le128 hash1, hash2;
+	size_t i;
+
+	for (i = 0; i < NH_KEY_WORDS; i++)
+		nh_key_words[i] = get_unaligned_le32(&nh_key[i * sizeof(u32)]);
+
+	/* Hash tweak and message length with first Poly1305 key */
+	put_unaligned_le64((u64)msglen * 8, header);
+	put_unaligned_le64(0, &header[sizeof(__le64)]);
+	memcpy(&header[POLY1305_BLOCK_SIZE], iv, ADIANTUM_IV_SIZE);
+	poly1305(header_poly_key, header, sizeof(header), &hash1);
+
+	/* Hash NH hashes of message blocks using second Poly1305 key */
+	/* (using a super naive way of handling the padding) */
+	memcpy(padded_msg, msg, msglen);
+	memset(&padded_msg[msglen], 0, padded_msglen - msglen);
+	for (i = 0; i < num_nh_blocks; i++) {
+		nh(nh_key_words, &padded_msg[i * NH_BLOCK_SIZE],
+		   MIN(NH_BLOCK_SIZE, padded_msglen - (i * NH_BLOCK_SIZE)),
+		   &nh_hashes[i * NH_HASH_SIZE]);
+	}
+	poly1305(msg_poly_key, nh_hashes, num_nh_blocks * NH_HASH_SIZE, &hash2);
+
+	/* Add the two hashes together to get the final hash */
+	le128_add(result, &hash1, &hash2);
+
+	free(nh_hashes);
+	free(padded_msg);
+}
+
+static void adiantum_crypt(const u8 key[ADIANTUM_KEY_SIZE],
+			   const u8 iv[ADIANTUM_IV_SIZE], const u8 *src,
+			   u8 *dst, size_t nbytes, bool decrypting)
+{
+	u8 subkeys[AES_256_KEY_SIZE + ADIANTUM_HASH_KEY_SIZE] = { 0 };
+	struct aes_key aes_key;
+	union {
+		u8 nonce[XCHACHA_NONCE_SIZE];
+		le128 block;
+	} u = { .nonce = { 1 } };
+	const size_t bulk_len = nbytes - sizeof(u.block);
+	le128 hash;
+
+	ASSERT(nbytes >= sizeof(u.block));
+
+	/* Derive subkeys */
+	xchacha12(key, u.nonce, subkeys, subkeys, sizeof(subkeys));
+	aes_setkey(&aes_key, subkeys, AES_256_KEY_SIZE);
+
+	/* Hash left part and add to right part */
+	adiantum_hash(&subkeys[AES_256_KEY_SIZE], iv, src, bulk_len, &hash);
+	memcpy(&u.block, &src[bulk_len], sizeof(u.block));
+	le128_add(&u.block, &u.block, &hash);
+
+	if (!decrypting) /* Encrypt right part with block cipher */
+		aes_encrypt(&aes_key, u.nonce, u.nonce);
+
+	/* Encrypt left part with stream cipher, using the computed nonce */
+	u.nonce[sizeof(u.block)] = 1;
+	xchacha12(key, u.nonce, src, dst, bulk_len);
+
+	if (decrypting) /* Decrypt right part with block cipher */
+		aes_decrypt(&aes_key, u.nonce, u.nonce);
+
+	/* Finalize right part by subtracting hash of left part */
+	adiantum_hash(&subkeys[AES_256_KEY_SIZE], iv, dst, bulk_len, &hash);
+	le128_sub(&u.block, &u.block, &hash);
+	memcpy(&dst[bulk_len], &u.block, sizeof(u.block));
+}
+
+static void adiantum_encrypt(const u8 key[ADIANTUM_KEY_SIZE],
+			     const u8 iv[ADIANTUM_IV_SIZE],
+			     const u8 *src, u8 *dst, size_t nbytes)
+{
+	adiantum_crypt(key, iv, src, dst, nbytes, false);
+}
+
+static void adiantum_decrypt(const u8 key[ADIANTUM_KEY_SIZE],
+			     const u8 iv[ADIANTUM_IV_SIZE],
+			     const u8 *src, u8 *dst, size_t nbytes)
+{
+	adiantum_crypt(key, iv, src, dst, nbytes, true);
+}
+
+#ifdef ENABLE_ALG_TESTS
+#include <linux/if_alg.h>
+#include <sys/socket.h>
+#define SOL_ALG 279
+static void af_alg_crypt(int algfd, int op, const u8 *key, size_t keylen,
+			 const u8 *iv, size_t ivlen,
+			 const u8 *src, u8 *dst, size_t datalen)
+{
+	size_t controllen = CMSG_SPACE(sizeof(int)) +
+			    CMSG_SPACE(sizeof(struct af_alg_iv) + ivlen);
+	u8 *control = xmalloc(controllen);
+	struct iovec iov = { .iov_base = (u8 *)src, .iov_len = datalen };
+	struct msghdr msg = {
+		.msg_iov = &iov,
+		.msg_iovlen = 1,
+		.msg_control = control,
+		.msg_controllen = controllen,
+	};
+	struct cmsghdr *cmsg;
+	struct af_alg_iv *algiv;
+	int reqfd;
+
+	memset(control, 0, controllen);
+
+	cmsg = CMSG_FIRSTHDR(&msg);
+	cmsg->cmsg_len = CMSG_LEN(sizeof(int));
+	cmsg->cmsg_level = SOL_ALG;
+	cmsg->cmsg_type = ALG_SET_OP;
+	*(int *)CMSG_DATA(cmsg) = op;
+
+	cmsg = CMSG_NXTHDR(&msg, cmsg);
+	cmsg->cmsg_len = CMSG_LEN(sizeof(struct af_alg_iv) + ivlen);
+	cmsg->cmsg_level = SOL_ALG;
+	cmsg->cmsg_type = ALG_SET_IV;
+	algiv = (struct af_alg_iv *)CMSG_DATA(cmsg);
+	algiv->ivlen = ivlen;
+	memcpy(algiv->iv, iv, ivlen);
+
+	if (setsockopt(algfd, SOL_ALG, ALG_SET_KEY, key, keylen) != 0)
+		die_errno("can't set key on AF_ALG socket");
+
+	reqfd = accept(algfd, NULL, NULL);
+	if (reqfd < 0)
+		die_errno("can't accept() AF_ALG socket");
+	if (sendmsg(reqfd, &msg, 0) != datalen)
+		die_errno("can't sendmsg() AF_ALG request socket");
+	if (xread(reqfd, dst, datalen) != datalen)
+		die("short read from AF_ALG request socket");
+	close(reqfd);
+
+	free(control);
+}
+
+static void test_adiantum(void)
+{
+	int algfd = socket(AF_ALG, SOCK_SEQPACKET, 0);
+	struct sockaddr_alg addr = {
+		.salg_type = "skcipher",
+		.salg_name = "adiantum(xchacha12,aes)",
+	};
+	unsigned long num_tests = NUM_ALG_TEST_ITERATIONS;
+
+	if (algfd < 0)
+		die_errno("can't create AF_ALG socket");
+	if (bind(algfd, (struct sockaddr *)&addr, sizeof(addr)) != 0)
+		die_errno("can't bind AF_ALG socket to Adiantum algorithm");
+
+	while (num_tests--) {
+		u8 key[ADIANTUM_KEY_SIZE];
+		u8 iv[ADIANTUM_IV_SIZE];
+		u8 ptext[4096];
+		u8 actual_ctext[sizeof(ptext)];
+		u8 expected_ctext[sizeof(ptext)];
+		u8 decrypted[sizeof(ptext)];
+		const size_t datalen = 16 + (rand() % (sizeof(ptext) - 15));
+
+		rand_bytes(key, sizeof(key));
+		rand_bytes(iv, sizeof(iv));
+		rand_bytes(ptext, datalen);
+
+		adiantum_encrypt(key, iv, ptext, actual_ctext, datalen);
+		af_alg_crypt(algfd, ALG_OP_ENCRYPT, key, sizeof(key),
+			     iv, sizeof(iv), ptext, expected_ctext, datalen);
+		ASSERT(memcmp(actual_ctext, expected_ctext, datalen) == 0);
+
+		adiantum_decrypt(key, iv, actual_ctext, decrypted, datalen);
+		ASSERT(memcmp(ptext, decrypted, datalen) == 0);
+	}
+	close(algfd);
+}
+#endif /* ENABLE_ALG_TESTS */
+
+/*----------------------------------------------------------------------------*
+ *                               Main program                                 *
+ *----------------------------------------------------------------------------*/
+
+#define FILE_NONCE_SIZE		16
+#define MAX_KEY_SIZE		64
+
+static const struct fscrypt_cipher {
+	const char *name;
+	void (*encrypt)(const u8 *key, const u8 *iv, const u8 *src,
+			u8 *dst, size_t nbytes);
+	void (*decrypt)(const u8 *key, const u8 *iv, const u8 *src,
+			u8 *dst, size_t nbytes);
+	int keysize;
+	int min_input_size;
+} fscrypt_ciphers[] = {
+	{
+		.name = "AES-256-XTS",
+		.encrypt = aes_256_xts_encrypt,
+		.decrypt = aes_256_xts_decrypt,
+		.keysize = 2 * AES_256_KEY_SIZE,
+		.min_input_size = AES_BLOCK_SIZE,
+	}, {
+		.name = "AES-256-CTS-CBC",
+		.encrypt = aes_256_cts_cbc_encrypt,
+		.decrypt = aes_256_cts_cbc_decrypt,
+		.keysize = AES_256_KEY_SIZE,
+		.min_input_size = AES_BLOCK_SIZE,
+	}, {
+		.name = "AES-128-CBC-ESSIV",
+		.encrypt = aes_128_cbc_essiv_encrypt,
+		.decrypt = aes_128_cbc_essiv_decrypt,
+		.keysize = AES_128_KEY_SIZE,
+		.min_input_size = AES_BLOCK_SIZE,
+	}, {
+		.name = "AES-128-CTS-CBC",
+		.encrypt = aes_128_cts_cbc_encrypt,
+		.decrypt = aes_128_cts_cbc_decrypt,
+		.keysize = AES_128_KEY_SIZE,
+		.min_input_size = AES_BLOCK_SIZE,
+	}, {
+		.name = "Adiantum",
+		.encrypt = adiantum_encrypt,
+		.decrypt = adiantum_decrypt,
+		.keysize = ADIANTUM_KEY_SIZE,
+		.min_input_size = AES_BLOCK_SIZE,
+	}
+};
+
+static const struct fscrypt_cipher *find_fscrypt_cipher(const char *name)
+{
+	size_t i;
+
+	for (i = 0; i < ARRAY_SIZE(fscrypt_ciphers); i++) {
+		if (strcmp(fscrypt_ciphers[i].name, name) == 0)
+			return &fscrypt_ciphers[i];
+	}
+	return NULL;
+}
+
+struct fscrypt_iv {
+	union {
+		__le64 block_num;
+		u8 bytes[32];
+	};
+};
+
+static void crypt_loop(const struct fscrypt_cipher *cipher, const u8 *key,
+		       struct fscrypt_iv *iv, bool decrypting,
+		       size_t block_size, size_t padding)
+{
+	u8 *buf = xmalloc(block_size);
+	size_t res;
+
+	while ((res = xread(STDIN_FILENO, buf, block_size)) > 0) {
+		size_t crypt_len = block_size;
+
+		if (padding > 0)
+			crypt_len = MIN(ROUND_UP(res, padding), crypt_len);
+
+		crypt_len = MAX(crypt_len, cipher->min_input_size);
+
+		memset(&buf[res], 0, crypt_len - res);
+
+		if (decrypting)
+			cipher->decrypt(key, iv->bytes, buf, buf, crypt_len);
+		else
+			cipher->encrypt(key, iv->bytes, buf, buf, crypt_len);
+
+		full_write(STDOUT_FILENO, buf, crypt_len);
+
+		iv->block_num = cpu_to_le64(le64_to_cpu(iv->block_num) + 1);
+	}
+	free(buf);
+}
+
+/* The supported key derivation functions */
+enum kdf_algorithm {
+	KDF_NONE,
+	KDF_AES_128_ECB,
+};
+
+static enum kdf_algorithm parse_kdf_algorithm(const char *arg)
+{
+	if (strcmp(arg, "none") == 0)
+		return KDF_NONE;
+	if (strcmp(arg, "AES-128-ECB") == 0)
+		return KDF_AES_128_ECB;
+	die("Unknown KDF: %s", arg);
+}
+
+/*
+ * Get the key and starting IV with which the encryption will actually be done.
+ * If a KDF was specified, a subkey is derived from the master key and file
+ * nonce.  Otherwise, the master key is used directly.
+ */
+static void get_key_and_iv(const u8 *master_key, size_t master_key_size,
+			   enum kdf_algorithm kdf,
+			   const u8 nonce[FILE_NONCE_SIZE],
+			   u8 *real_key, size_t real_key_size,
+			   struct fscrypt_iv *iv)
+{
+	bool nonce_in_iv = false;
+	struct aes_key aes_key;
+	size_t i;
+
+	ASSERT(real_key_size <= master_key_size);
+
+	memset(iv, 0, sizeof(*iv));
+
+	switch (kdf) {
+	case KDF_NONE:
+		memcpy(real_key, master_key, real_key_size);
+		nonce_in_iv = true;
+		break;
+	case KDF_AES_128_ECB:
+		if (nonce == NULL)
+			die("--file-nonce is required with --kdf=AES-128-ECB");
+		STATIC_ASSERT(FILE_NONCE_SIZE == AES_128_KEY_SIZE);
+		ASSERT(real_key_size % AES_BLOCK_SIZE == 0);
+		aes_setkey(&aes_key, nonce, AES_128_KEY_SIZE);
+		for (i = 0; i < real_key_size; i += AES_BLOCK_SIZE)
+			aes_encrypt(&aes_key, &master_key[i], &real_key[i]);
+		break;
+	default:
+		ASSERT(0);
+	}
+
+	if (nonce_in_iv && nonce != NULL)
+		memcpy(&iv->bytes[8], nonce, FILE_NONCE_SIZE);
+}
+
+enum {
+	OPT_BLOCK_SIZE,
+	OPT_DECRYPT,
+	OPT_FILE_NONCE,
+	OPT_HELP,
+	OPT_KDF,
+	OPT_PADDING,
+};
+
+static const struct option longopts[] = {
+	{ "block-size",      required_argument, NULL, OPT_BLOCK_SIZE },
+	{ "decrypt",         no_argument,       NULL, OPT_DECRYPT },
+	{ "file-nonce",      required_argument, NULL, OPT_FILE_NONCE },
+	{ "help",            no_argument,       NULL, OPT_HELP },
+	{ "kdf",             required_argument, NULL, OPT_KDF },
+	{ "padding",         required_argument, NULL, OPT_PADDING },
+	{ NULL, 0, NULL, 0 },
+};
+
+int main(int argc, char *argv[])
+{
+	size_t block_size = 4096;
+	bool decrypting = false;
+	u8 _file_nonce[FILE_NONCE_SIZE];
+	u8 *file_nonce = NULL;
+	enum kdf_algorithm kdf = KDF_NONE;
+	size_t padding = 0;
+	const struct fscrypt_cipher *cipher;
+	u8 master_key[MAX_KEY_SIZE];
+	int master_key_size;
+	u8 real_key[MAX_KEY_SIZE];
+	struct fscrypt_iv iv;
+	char *tmp;
+	int c;
+
+	aes_init();
+
+#ifdef ENABLE_ALG_TESTS
+	test_aes();
+	test_sha2();
+	test_aes_256_xts();
+	test_aes_256_cts_cbc();
+	test_adiantum();
+#endif
+
+	while ((c = getopt_long(argc, argv, "", longopts, NULL)) != -1) {
+		switch (c) {
+		case OPT_BLOCK_SIZE:
+			block_size = strtoul(optarg, &tmp, 10);
+			if (block_size <= 0 || *tmp)
+				die("Invalid block size: %s", optarg);
+			break;
+		case OPT_DECRYPT:
+			decrypting = true;
+			break;
+		case OPT_FILE_NONCE:
+			if (hex2bin(optarg, _file_nonce, FILE_NONCE_SIZE) !=
+			    FILE_NONCE_SIZE)
+				die("Invalid file nonce: %s", optarg);
+			file_nonce = _file_nonce;
+			break;
+		case OPT_HELP:
+			usage(stdout);
+			return 0;
+		case OPT_KDF:
+			kdf = parse_kdf_algorithm(optarg);
+			break;
+		case OPT_PADDING:
+			padding = strtoul(optarg, &tmp, 10);
+			if (padding <= 0 || *tmp || !is_power_of_2(padding) ||
+			    padding > INT_MAX)
+				die("Invalid padding amount: %s", optarg);
+			break;
+		default:
+			usage(stderr);
+			return 2;
+		}
+	}
+	argc -= optind;
+	argv += optind;
+
+	if (argc != 2) {
+		usage(stderr);
+		return 2;
+	}
+
+	cipher = find_fscrypt_cipher(argv[0]);
+	if (cipher == NULL)
+		die("Unknown cipher: %s", argv[0]);
+
+	if (block_size < cipher->min_input_size)
+		die("Block size of %zu bytes is too small for cipher %s",
+		    block_size, cipher->name);
+
+	master_key_size = hex2bin(argv[1], master_key, MAX_KEY_SIZE);
+	if (master_key_size < 0)
+		die("Invalid master_key: %s", argv[1]);
+	if (master_key_size < cipher->keysize)
+		die("Master key is too short for cipher %s", cipher->name);
+
+	get_key_and_iv(master_key, master_key_size, kdf, file_nonce,
+		       real_key, cipher->keysize, &iv);
+
+	crypt_loop(cipher, real_key, &iv, decrypting, block_size, padding);
+	return 0;
+}
-- 
2.21.0.593.g511ec345e18-goog