summaryrefslogtreecommitdiffstats
path: root/src/crypto
diff options
context:
space:
mode:
Diffstat (limited to 'src/crypto')
-rw-r--r--src/crypto/md4.c280
1 files changed, 280 insertions, 0 deletions
diff --git a/src/crypto/md4.c b/src/crypto/md4.c
new file mode 100644
index 00000000..f4a8d78d
--- /dev/null
+++ b/src/crypto/md4.c
@@ -0,0 +1,280 @@
+/*
+ * Copyright (C) 2017 Michael Brown <mbrown@fensystems.co.uk>.
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation; either version 2 of the
+ * License, or any later version.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ * General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
+ * 02110-1301, USA.
+ *
+ * You can also choose to distribute this program under the terms of
+ * the Unmodified Binary Distribution Licence (as given in the file
+ * COPYING.UBDL), provided that you have satisfied its requirements.
+ */
+
+FILE_LICENCE ( GPL2_OR_LATER_OR_UBDL );
+
+/** @file
+ *
+ * MD4 algorithm
+ *
+ */
+
+#include <stdint.h>
+#include <string.h>
+#include <byteswap.h>
+#include <assert.h>
+#include <ipxe/rotate.h>
+#include <ipxe/crypto.h>
+#include <ipxe/asn1.h>
+#include <ipxe/md4.h>
+
+/** MD4 variables */
+struct md4_variables {
+ /* This layout matches that of struct md4_digest_data,
+ * allowing for efficient endianness-conversion,
+ */
+ uint32_t a;
+ uint32_t b;
+ uint32_t c;
+ uint32_t d;
+ uint32_t w[16];
+} __attribute__ (( packed ));
+
+/** MD4 shift amounts */
+static const uint8_t r[3][4] = {
+ { 3, 7, 11, 19 },
+ { 3, 5, 9, 13 },
+ { 3, 9, 11, 15 },
+};
+
+/**
+ * f(b,c,d,w) for steps 0 to 15
+ *
+ * @v v MD4 variables
+ * @v i Index within round
+ * @ret f f(b,c,d,w)
+ */
+static uint32_t md4_f_0_15 ( struct md4_variables *v, unsigned int i ) {
+ return ( ( ( v->b & v->c ) | ( ~v->b & v->d ) ) + v->w[i] );
+}
+
+/**
+ * f(b,c,d,w) for steps 16 to 31
+ *
+ * @v v MD4 variables
+ * @v i Index within round
+ * @ret f f(b,c,d,w)
+ */
+static uint32_t md4_f_16_31 ( struct md4_variables *v, unsigned int i ) {
+ return ( ( ( v->b & v->c ) | ( v->b & v->d ) | ( v->c & v->d ) ) +
+ v->w[ ( ( i << 2 ) | ( i >> 2 ) ) % 16 ] );
+}
+
+/**
+ * f(b,c,d,w) for steps 32 to 47
+ *
+ * @v v MD4 variables
+ * @v i Index within round
+ * @ret f f(b,c,d,w)
+ */
+static uint32_t md4_f_32_47 ( struct md4_variables *v, unsigned int i ) {
+ static const uint8_t reverse[16] = {
+ 0, 8, 4, 12, 2, 10, 6, 14, 1, 9, 5, 13, 3, 11, 7, 15
+ };
+ return ( ( v->b ^ v->c ^ v->d ) + v->w[reverse[i]] );
+}
+
+/** An MD4 step function */
+struct md4_step {
+ /**
+ * Calculate f(b,c,d,w)
+ *
+ * @v v MD4 variables
+ * @v i Index within round
+ * @ret f f(b,c,d,w)
+ */
+ uint32_t ( * f ) ( struct md4_variables *v, unsigned int i );
+ /** Constant */
+ uint32_t constant;
+};
+
+/** MD4 steps */
+static struct md4_step md4_steps[4] = {
+ /** 0 to 15 */
+ { .f = md4_f_0_15, .constant = 0x00000000UL },
+ /** 16 to 31 */
+ { .f = md4_f_16_31, .constant = 0x5a827999UL },
+ /** 32 to 47 */
+ { .f = md4_f_32_47, .constant = 0x6ed9eba1UL },
+};
+
+/**
+ * Initialise MD4 algorithm
+ *
+ * @v ctx MD4 context
+ */
+static void md4_init ( void *ctx ) {
+ struct md4_context *context = ctx;
+
+ context->ddd.dd.digest.h[0] = cpu_to_le32 ( 0x67452301 );
+ context->ddd.dd.digest.h[1] = cpu_to_le32 ( 0xefcdab89 );
+ context->ddd.dd.digest.h[2] = cpu_to_le32 ( 0x98badcfe );
+ context->ddd.dd.digest.h[3] = cpu_to_le32 ( 0x10325476 );
+ context->len = 0;
+}
+
+/**
+ * Calculate MD4 digest of accumulated data
+ *
+ * @v context MD4 context
+ */
+static void md4_digest ( struct md4_context *context ) {
+ union {
+ union md4_digest_data_dwords ddd;
+ struct md4_variables v;
+ } u;
+ uint32_t *a = &u.v.a;
+ uint32_t *b = &u.v.b;
+ uint32_t *c = &u.v.c;
+ uint32_t *d = &u.v.d;
+ uint32_t *w = u.v.w;
+ uint32_t f;
+ uint32_t temp;
+ struct md4_step *step;
+ unsigned int round;
+ unsigned int i;
+
+ /* Sanity checks */
+ assert ( ( context->len % sizeof ( context->ddd.dd.data ) ) == 0 );
+ linker_assert ( &u.ddd.dd.digest.h[0] == a, md4_bad_layout );
+ linker_assert ( &u.ddd.dd.digest.h[1] == b, md4_bad_layout );
+ linker_assert ( &u.ddd.dd.digest.h[2] == c, md4_bad_layout );
+ linker_assert ( &u.ddd.dd.digest.h[3] == d, md4_bad_layout );
+ linker_assert ( &u.ddd.dd.data.dword[0] == w, md4_bad_layout );
+
+ DBGC ( context, "MD4 digesting:\n" );
+ DBGC_HDA ( context, 0, &context->ddd.dd.digest,
+ sizeof ( context->ddd.dd.digest ) );
+ DBGC_HDA ( context, context->len, &context->ddd.dd.data,
+ sizeof ( context->ddd.dd.data ) );
+
+ /* Convert h[0..3] to host-endian, and initialise a, b, c, d,
+ * and x[0..15]
+ */
+ for ( i = 0 ; i < ( sizeof ( u.ddd.dword ) /
+ sizeof ( u.ddd.dword[0] ) ) ; i++ ) {
+ le32_to_cpus ( &context->ddd.dword[i] );
+ u.ddd.dword[i] = context->ddd.dword[i];
+ }
+
+ /* Main loop */
+ for ( i = 0 ; i < 48 ; i++ ) {
+ round = ( i / 16 );
+ step = &md4_steps[round];
+ f = step->f ( &u.v, ( i % 16 ) );
+ temp = *d;
+ *d = *c;
+ *c = *b;
+ *b = rol32 ( ( *a + f + step->constant ), r[round][ i % 4 ] );
+ *a = temp;
+ DBGC2 ( context, "%2d : %08x %08x %08x %08x\n",
+ i, *a, *b, *c, *d );
+ }
+
+ /* Add chunk to hash and convert back to little-endian */
+ for ( i = 0 ; i < 4 ; i++ ) {
+ context->ddd.dd.digest.h[i] =
+ cpu_to_le32 ( context->ddd.dd.digest.h[i] +
+ u.ddd.dd.digest.h[i] );
+ }
+
+ DBGC ( context, "MD4 digested:\n" );
+ DBGC_HDA ( context, 0, &context->ddd.dd.digest,
+ sizeof ( context->ddd.dd.digest ) );
+}
+
+/**
+ * Accumulate data with MD4 algorithm
+ *
+ * @v ctx MD4 context
+ * @v data Data
+ * @v len Length of data
+ */
+static void md4_update ( void *ctx, const void *data, size_t len ) {
+ struct md4_context *context = ctx;
+ const uint8_t *byte = data;
+ size_t offset;
+
+ /* Accumulate data a byte at a time, performing the digest
+ * whenever we fill the data buffer
+ */
+ while ( len-- ) {
+ offset = ( context->len % sizeof ( context->ddd.dd.data ) );
+ context->ddd.dd.data.byte[offset] = *(byte++);
+ context->len++;
+ if ( ( context->len % sizeof ( context->ddd.dd.data ) ) == 0 )
+ md4_digest ( context );
+ }
+}
+
+/**
+ * Generate MD4 digest
+ *
+ * @v ctx MD4 context
+ * @v out Output buffer
+ */
+static void md4_final ( void *ctx, void *out ) {
+ struct md4_context *context = ctx;
+ uint64_t len_bits;
+ uint8_t pad;
+
+ /* Record length before pre-processing */
+ len_bits = cpu_to_le64 ( ( ( uint64_t ) context->len ) * 8 );
+
+ /* Pad with a single "1" bit followed by as many "0" bits as required */
+ pad = 0x80;
+ do {
+ md4_update ( ctx, &pad, sizeof ( pad ) );
+ pad = 0x00;
+ } while ( ( context->len % sizeof ( context->ddd.dd.data ) ) !=
+ offsetof ( typeof ( context->ddd.dd.data ), final.len ) );
+
+ /* Append length (in bits) */
+ md4_update ( ctx, &len_bits, sizeof ( len_bits ) );
+ assert ( ( context->len % sizeof ( context->ddd.dd.data ) ) == 0 );
+
+ /* Copy out final digest */
+ memcpy ( out, &context->ddd.dd.digest,
+ sizeof ( context->ddd.dd.digest ) );
+}
+
+/** MD4 algorithm */
+struct digest_algorithm md4_algorithm = {
+ .name = "md4",
+ .ctxsize = sizeof ( struct md4_context ),
+ .blocksize = sizeof ( union md4_block ),
+ .digestsize = sizeof ( struct md4_digest ),
+ .init = md4_init,
+ .update = md4_update,
+ .final = md4_final,
+};
+
+/** "md4" object identifier */
+static uint8_t oid_md4[] = { ASN1_OID_MD4 };
+
+/** "md4" OID-identified algorithm */
+struct asn1_algorithm oid_md4_algorithm __asn1_algorithm = {
+ .name = "md4",
+ .digest = &md4_algorithm,
+ .oid = ASN1_OID_CURSOR ( oid_md4 ),
+};