/* Lifted from the krb5 1.6 source tree and hacked slightly to fit in here Karl Ramm 12/21/08 */ /* * lib/des425/quad_cksum.c * * Copyright 1985, 1986, 1987, 1988,1990 by the Massachusetts Institute * of Technology. * All Rights Reserved. * * Export of this software from the United States of America may * require a specific license from the United States Government. * It is the responsibility of any person or organization contemplating * export to obtain such a license before exporting. * * WITHIN THAT CONSTRAINT, permission to use, copy, modify, and * distribute this software and its documentation for any purpose and * without fee is hereby granted, provided that the above copyright * notice appear in all copies and that both that copyright notice and * this permission notice appear in supporting documentation, and that * the name of M.I.T. not be used in advertising or publicity pertaining * to distribution of the software without specific, written prior * permission. Furthermore if you modify this software you must label * your software as modified software and not distribute it in such a * fashion that it might be confused with the original M.I.T. software. * M.I.T. makes no representations about the suitability of * this software for any purpose. It is provided "as is" without express * or implied warranty. * * * This routine does not implement: * * * Quadratic Congruential Manipulation Dectection Code * * ref: "Message Authentication" * R.R. Jueneman, S. M. Matyas, C.H. Meyer * IEEE Communications Magazine, * Sept 1985 Vol 23 No 9 p 29-40 * * This routine, part of the Athena DES library built for the Kerberos * authentication system, calculates a manipulation detection code for * a message. It is a much faster alternative to the DES-checksum * method. No guarantees are offered for its security. * * Implementation for 4.2bsd * by S.P. Miller Project Athena/MIT */ /* * Algorithm (per paper): * define: * message to be composed of n m-bit blocks X1,...,Xn * optional secret seed S in block X1 * MDC in block Xn+1 * prime modulus N * accumulator Z * initial (secret) value of accumulator C * N, C, and S are known at both ends * C and , optionally, S, are hidden from the end users * then * (read array references as subscripts over time) * Z[0] = c; * for i = 1...n * Z[i] = (Z[i+1] + X[i])**2 modulo N * X[n+1] = Z[n] = MDC * * Then pick * N = 2**31 -1 * m = 16 * iterate 4 times over plaintext, also use Zn * from iteration j as seed for iteration j+1, * total MDC is then a 128 bit array of the four * Zn; * * return the last Zn and optionally, all * four as output args. * * Modifications: * To inhibit brute force searches of the seed space, this * implementation is modified to have * Z = 64 bit accumulator * C = 64 bit C seed * N = 2**63 - 1 * S = S seed is not implemented here * arithmetic is not quite real double integer precision, since we * cant get at the carry or high order results from multiply, * but nontheless is 64 bit arithmetic. */ /* * This code purports to implement the above algorithm, but fails. * * First of all, there was an implicit mod 2**32 being done on the * machines where this was developed because of their word sizes, and * for compabitility this has to be done on machines with 64-bit * words, so we make it explicit. * * Second, in the squaring operation, I really doubt the carry-over * from the low 31-bit half of the accumulator is being done right, * and using a modulus of 0x7fffffff on the low half of the * accumulator seems completely wrong. And I challenge anyone to * explain where the number 83653421 comes from. * * --Ken Raeburn 2001-04-06 */ /* System include files */ #include #include #include #include /* Definitions for byte swapping */ /* vax byte order is LSB first. This is not performance critical, and is far more readable this way. */ #define four_bytes_vax_to_nets(x) ((((((x[3]<<8)|x[2])<<8)|x[1])<<8)|x[0]) #define vaxtohl(x) four_bytes_vax_to_nets(((const unsigned char *)(x))) #define two_bytes_vax_to_nets(x) ((x[1]<<8)|x[0]) #define vaxtohs(x) two_bytes_vax_to_nets(((const unsigned char *)(x))) /*** Routines ***************************************************** */ #ifdef HAVE_KRB5 unsigned long z_quad_cksum(const unsigned char *in, /* input block */ u_int32_t *out, /* optional longer output */ long length, /* original length in bytes */ int out_count, /* number of iterations */ unsigned char *c_seed /* secret seed, 8 bytes */ ) { /* * this routine both returns the low order of the final (last in * time) 32bits of the checksum, and if "out" is not a null * pointer, a longer version, up to entire 32 bytes of the * checksum is written unto the address pointed to. */ register u_int32_t z; register u_int32_t z2; register u_int32_t x; register u_int32_t x2; const unsigned char *p; register int32_t len; register int i; /* use all 8 bytes of seed */ z = vaxtohl(c_seed); z2 = vaxtohl((const char *)c_seed+4); if (out == NULL) out_count = 1; /* default */ /* This is repeated n times!! */ for (i = 1; i <=4 && i<= out_count; i++) { len = length; p = in; while (len) { /* * X = Z + Input ... sort of. Carry out from low half * isn't done, so we're using all 32 bits of x now. */ if (len > 1) { x = (z + vaxtohs(p)); p += 2; len -= 2; } else { x = (z + *(const unsigned char *)p++); len = 0; } x2 = z2; /* * I think this is supposed to be a squaring operation. * What it really is, I haven't figured out yet. * * Explicit mod 2**32 is for backwards compatibility. Why * mod 0x7fffffff and not 0x80000000 on the low half of * the (supposed) accumulator? And where does the number * 83653421 come from?? */ z = (((x * x) + (x2 * x2)) & 0xffffffff) % 0x7fffffff; z2 = ((x * (x2+83653421)) & 0xffffffff) % 0x7fffffff; /* modulo */ } if (out != NULL) { *out++ = z; *out++ = z2; } } /* return final z value as 32 bit version of checksum */ return z; } #endif