kent@sparky.imd.sterling.com (Kent Landfield) (05/09/91)
Submitted-by: Kent Landfield <kent@sparky.imd.sterling.com> Posting-number: Volume 19, Issue 17 Archive-name: md4tools/part01 MD4 can be used to apply a fingerprint on an article posted to USENET. When run through a verification tool, MD4 will tell you whether an article has been corrupted. The use of MD4 does not detect or prevent the complete replacement of an article. Think of MD4 as a super-strong checksum. The header X-Md4-Signature: contains the value that will be checked against to determine if the article is intact. Starting with this posting, I am going to be using the X-Md4-Signature: on all articles posted to comp.sources.misc. While I don't think that this is worth doing for most general USENET articles, I think it will be extremely useful for archives. The X-Md4-Signature: header is going to replace the X-Checksum-Snefru: header previously used in this newsgroup. I would like to thank Ron Rivest (the author of RFC1186, "The MD4 Message Digest Algorithm") for the MD4 code and RSA Data Security, Inc. for giving me the permission to post it. I would also like to thank Rich Salz for the push to do it and for his snefru code that I hacked... -Kent+ --- #! /bin/sh # into a shell via "sh file" or similar. To overwrite existing files, # type "sh file -c". # The tool that generated this appeared in the comp.sources.unix newsgroup; # send mail to comp-sources-unix@uunet.uu.net if you want that tool. # Contents: README foo md4.c md4_def.h rfc1186 # Wrapped by kent@sparky on Tue May 7 23:23:53 1991 PATH=/bin:/usr/bin:/usr/ucb ; export PATH echo If this archive is complete, you will see the following message: echo ' "shar: End of archive 1 (of 2)."' if test -f 'README' -a "${1}" != "-c" ; then echo shar: Will not clobber existing file \"'README'\" else echo shar: Extracting \"'README'\" \(3116 characters\) sed "s/^X//" >'README' <<'END_OF_FILE' XAs stated in RFC1186: X X This note describes the MD4 message digest algorithm. The algorithm X takes as input an input message of arbitrary length and produces as X output a 128-bit "fingerprint" or "message digest" of the input. It X is conjectured that it is computationally infeasible to produce two X messages having the same message digest, or to produce any message X having a given prespecified target message digest. The MD4 algorithm X is thus ideal for digital signature applications, where a large file X must be "compressed" in a secure manner before being signed with the X RSA public-key cryptosystem. X XMD4 can be used to apply a fingerprint on an article posted to USENET that, Xwhen run through a verification tool, will tell you whether an article has Xbeen corrupted. It does not detect or prevent complete replacement of the Xarticle. Think of MD4 as a super-strong checksum. X XThis package is an "assembled" set of tools that uses the MD4 Message XDigest Algorithm specified in RFC1186. There are three parts to the package. X Xmd4 - This is the heart of the tools. The code was taken from RFC1186, X "The MD4 Message Digest Algorithm" authored by Ronald L. Rivest. X XI *completely* and shamefully stole the MD4 code from the RFC and as such it Xmust be distributed under the terms specified in md4.c, md4.h and md4driver.c. XThanks to Ron for the code and to RSA Data Security, Inc. for giving me the Xpermission to post it. X Xhashmd4 - This program is used to apply an MD4 digest on a specified X USENET article. A new header is added to the article. The X header X-Md4-Signature: contains the value that will be checked X against to determine if the article is intact. X Xcheckmd4 - This is the program used to check whether the MD4 digest X of a USENET article indicates whether or not the article X has been tampered with. X XI would like to thank Rich Salz for posting snefru and giving me something Xto "hack".. :-) checkmd4 and hashmd4 look almost exactly like Rich's code Xfor good reason. They are... I did the bare minimum to get this working. XThat's all I had to do... :-) For those of you currently using hashnews Xand checknews, this (other than the names) should be a plug amd play... X XThere are manual pages for checkmd4 and for hashmd4 supplied. The actual XRFC1186 is supplied as a reference for MD4. X XCode that does not contain an explicit RSA copyright was derived from XRich Salz's code that was already in the public domain. Feel free to do Xwhat you want with those pieces. X XThis code needs a 32bit machine; good luck if you've only got 16 bits! X XTo compile, edit the Makefile if you use strchr/strrchr rather than Xindex/rindex. Also edit md4.c to determine the byte order of your Xmachine. There is also a discussion of how to use the routines in Xmd4.c in your own programs. X XStarting with this posting, I am going to be using hashmd4 on all my Xc.s.m articles. While I don't think that this is worth doing for most Xgeneral USENET articles, I think it will be an interesting experiment for Xarchives. X XEnjoy! X -Kent+ END_OF_FILE if test 3116 -ne `wc -c <'README'`; then echo shar: \"'README'\" unpacked with wrong size! fi # end of 'README' fi if test -f 'foo' -a "${1}" != "-c" ; then echo shar: Will not clobber existing file \"'foo'\" else echo shar: Extracting \"'foo'\" \(4 characters\) sed "s/^X//" >'foo' <<'END_OF_FILE' Xabc END_OF_FILE if test 4 -ne `wc -c <'foo'`; then echo shar: \"'foo'\" unpacked with wrong size! fi # end of 'foo' fi if test -f 'md4.c' -a "${1}" != "-c" ; then echo shar: Will not clobber existing file \"'md4.c'\" else echo shar: Extracting \"'md4.c'\" \(10725 characters\) sed "s/^X//" >'md4.c' <<'END_OF_FILE' X/* X*--------------------------------------------------------------------------* X* (C) Copyright 1990, RSA Data Security, Inc. All rights reserved. * X* License to copy and use this software is granted provided it is * X* identified as the "RSA Data Security, Inc. MD4 message digest algorithm" * X* in all material mentioning or referencing this software or function. * X* * X* License is also granted to make and use derivative works provided such * X* works are identified as "derived from the RSA Data Securitry, Inc. MD4 * X* message digest algorithm" in all material mentioning or referencing the * X* derived work. * X* * X* RSA Data Security, Inc. makes no representations concerning the * X* merchantability of this software or the suitability of the software * X* for any particular purpose. It is provided "as is" without express * X* or implied warranty of any kind. * X* * X* These notices must be retained in any copies of any part of this * X* documentation and/or software. * X*--------------------------------------------------------------------------* X** ******************************************************************** X** md4.c -- Implementation of MD4 Message Digest Algorithm ** X** Updated: 2/16/90 by Ronald L. Rivest ** X** (C) 1990 RSA Data Security, Inc. ** X** ******************************************************************** X*/ X X/* X** To use MD4: X** -- Include md4.h in your program X** -- Declare an MDstruct MD to hold the state of the digest X** computation. X** -- Initialize MD using MDbegin(&MD) X** -- For each full block (64 bytes) X you wish to process, call X** MDupdate(&MD,X,512) X** (512 is the number of bits in a full block.) X** -- For the last block (less than 64 bytes) you wish to process, X** MDupdate(&MD,X,n) X** where n is the number of bits in the partial block. A partial X** block terminates the computation, so every MD computation X** should terminate by processing a partial block, even if it X** has n = 0. X** -- The message digest is available in MD.buffer[0] ... X** MD.buffer[3]. (Least-significant byte of each word X** should be output first.) X** -- You can print out the digest using MDprint(&MD) X*/ X X/* Implementation notes: X** This implementation assumes that ints are 32-bit quantities. X** If the machine stores the least-significant byte of an int in the X** least-addressed byte (e.g., VAX and 8086), then LOWBYTEFIRST X** should be set to TRUE. Otherwise (e.g., SUNS), LOWBYTEFIRST X** should be set to FALSE. Note that on machines with LOWBYTEFIRST X** FALSE the routine MDupdate modifies has a side-effect on its input X** array (the order of bytes in each word are reversed). If this is X** undesired a call to MDreverse(X) can reverse the bytes of X back X** into order after each call to MDupdate. X*/ X X/* Compile-time includes X*/ X#include <stdio.h> X#include "md4.h" X X#define TRUE 1 X#define FALSE 0 X#define LOWBYTEFIRST FALSE X X X/* Compile-time declarations of MD4 "magic constants". X*/ X#define I0 0x67452301 /* Initial values for MD buffer */ X#define I1 0xefcdab89 X#define I2 0x98badcfe X#define I3 0x10325476 X#define C2 013240474631 /* round 2 constant = sqrt(2) in octal */ X#define C3 015666365641 /* round 3 constant = sqrt(3) in octal */ X/* C2 and C3 are from Knuth, The Art of Programming, Volume 2 X** (Seminumerical Algorithms), Second Edition (1981), Addison-Wesley. X** Table 2, page 660. X*/ X#define fs1 3 /* round 1 shift amounts */ X#define fs2 7 X#define fs3 11 X#define fs4 19 X#define gs1 3 /* round 2 shift amounts */ X#define gs2 5 X#define gs3 9 X#define gs4 13 X#define hs1 3 /* round 3 shift amounts */ X#define hs2 9 X#define hs3 11 X#define hs4 15 X X/* Compile-time macro declarations for MD4. X** Note: The "rot" operator uses the variable "tmp". X** It assumes tmp is declared as unsigned int, so that the >> X** operator will shift in zeros rather than extending the sign bit. X*/ X#define f(X,Y,Z) ((X&Y) | ((~X)&Z)) X#define g(X,Y,Z) ((X&Y) | (X&Z) | (Y&Z)) X#define h(X,Y,Z) (X^Y^Z) X#define rot(X,S) (tmp=X,(tmp<<S) | (tmp>>(32-S))) X#define ff(A,B,C,D,i,s) A = rot((A + f(B,C,D) + X[i]),s) X#define gg(A,B,C,D,i,s) A = rot((A + g(B,C,D) + X[i] + C2),s) X#define hh(A,B,C,D,i,s) A = rot((A + h(B,C,D) + X[i] + C3),s) X X/* MDprint(MDp) X** Print message digest buffer MDp as 32 hexadecimal digits. X** Order is from low-order byte of buffer[0] to high-order byte of X** buffer[3]. X** Each byte is printed with high-order hexadecimal digit first. X** This is a user-callable routine. X*/ Xvoid XMDprint(MDp) XMDptr MDp; X{ int i,j; X for (i=0;i<4;i++) X for (j=0;j<32;j=j+8) X printf("%02x",(MDp->buffer[i]>>j) & 0xFF); X} X X/* MDbegin(MDp) X** Initialize message digest buffer MDp. X** This is a user-callable routine. X*/ Xvoid XMDbegin(MDp) XMDptr MDp; X{ int i; X MDp->buffer[0] = I0; X MDp->buffer[1] = I1; X MDp->buffer[2] = I2; X MDp->buffer[3] = I3; X for (i=0;i<8;i++) MDp->count[i] = 0; X MDp->done = 0; X} X X/* MDreverse(X) X** Reverse the byte-ordering of every int in X. X** Assumes X is an array of 16 ints. X** The macro revx reverses the byte-ordering of the next word of X. X*/ X#define revx { t = (*X << 16) | (*X >> 16); \ X *X++ = ((t & 0xFF00FF00) >> 8) | ((t & 0x00FF00FF) << 8); } XMDreverse(X) Xunsigned int *X; X{ register unsigned int t; X revx; revx; revx; revx; revx; revx; revx; revx; X revx; revx; revx; revx; revx; revx; revx; revx; X} X X/* MDblock(MDp,X) X** Update message digest buffer MDp->buffer using 16-word data block X. X** Assumes all 16 words of X are full of data. X** Does not update MDp->count. X** This routine is not user-callable. X*/ Xstatic void XMDblock(MDp,X) XMDptr MDp; Xunsigned int *X; X{ X register unsigned int tmp, A, B, C, D; X#if LOWBYTEFIRST == FALSE X MDreverse(X); X#endif X A = MDp->buffer[0]; X B = MDp->buffer[1]; X C = MDp->buffer[2]; X D = MDp->buffer[3]; X /* Update the message digest buffer */ X ff(A , B , C , D , 0 , fs1); /* Round 1 */ X ff(D , A , B , C , 1 , fs2); X ff(C , D , A , B , 2 , fs3); X ff(B , C , D , A , 3 , fs4); X ff(A , B , C , D , 4 , fs1); X ff(D , A , B , C , 5 , fs2); X ff(C , D , A , B , 6 , fs3); X ff(B , C , D , A , 7 , fs4); X ff(A , B , C , D , 8 , fs1); X ff(D , A , B , C , 9 , fs2); X ff(C , D , A , B , 10 , fs3); X ff(B , C , D , A , 11 , fs4); X ff(A , B , C , D , 12 , fs1); X ff(D , A , B , C , 13 , fs2); X ff(C , D , A , B , 14 , fs3); X ff(B , C , D , A , 15 , fs4); X gg(A , B , C , D , 0 , gs1); /* Round 2 */ X gg(D , A , B , C , 4 , gs2); X gg(C , D , A , B , 8 , gs3); X gg(B , C , D , A , 12 , gs4); X gg(A , B , C , D , 1 , gs1); X gg(D , A , B , C , 5 , gs2); X gg(C , D , A , B , 9 , gs3); X gg(B , C , D , A , 13 , gs4); X gg(A , B , C , D , 2 , gs1); X gg(D , A , B , C , 6 , gs2); X gg(C , D , A , B , 10 , gs3); X gg(B , C , D , A , 14 , gs4); X gg(A , B , C , D , 3 , gs1); X gg(D , A , B , C , 7 , gs2); X gg(C , D , A , B , 11 , gs3); X gg(B , C , D , A , 15 , gs4); X hh(A , B , C , D , 0 , hs1); /* Round 3 */ X hh(D , A , B , C , 8 , hs2); X hh(C , D , A , B , 4 , hs3); X hh(B , C , D , A , 12 , hs4); X hh(A , B , C , D , 2 , hs1); X hh(D , A , B , C , 10 , hs2); X hh(C , D , A , B , 6 , hs3); X hh(B , C , D , A , 14 , hs4); X hh(A , B , C , D , 1 , hs1); X hh(D , A , B , C , 9 , hs2); X hh(C , D , A , B , 5 , hs3); X hh(B , C , D , A , 13 , hs4); X hh(A , B , C , D , 3 , hs1); X hh(D , A , B , C , 11 , hs2); X hh(C , D , A , B , 7 , hs3); X hh(B , C , D , A , 15 , hs4); X MDp->buffer[0] += A; X MDp->buffer[1] += B; X MDp->buffer[2] += C; X MDp->buffer[3] += D; X} X X/* MDupdate(MDp,X,count) X** Input: MDp -- an MDptr X** X -- a pointer to an array of unsigned characters. X** count -- the number of bits of X to use. X** (if not a multiple of 8, uses high bits of last byte.) X** Update MDp using the number of bits of X given by count. X** This is the basic input routine for an MD4 user. X** The routine completes the MD computation when count < 512, so X** every MD computation should end with one call to MDupdate with a X** count less than 512. A call with count 0 will be ignored if the X** MD has already been terminated (done != 0), so an extra call with X** count 0 can be given as a "courtesy close" to force termination X** if desired. X*/ Xvoid XMDupdate(MDp,X,count) XMDptr MDp; Xunsigned char *X; Xunsigned int count; X{ unsigned int i, tmp, bit, byte, mask; X unsigned char XX[64]; X unsigned char *p; X /* return with no error if this is a courtesy close with count X ** zero and MDp->done is true. X */ X if (count == 0 && MDp->done) return; X /* check to see if MD is already done and report error */ X if (MDp->done) X { printf("\nError: MDupdate MD already done."); return; } X /* Add count to MDp->count */ X tmp = count; X p = MDp->count; X while (tmp) X { tmp += *p; X *p++ = tmp; X tmp = tmp >> 8; X } X /* Process data */ X if (count == 512) X { /* Full block of data to handle */ X MDblock(MDp,(unsigned int *)X); X } X else if (count > 512) /* Check for count too large */ X { printf("\nError: MDupdate called with illegal count value %d." X ,count); X return; X } X else /* partial block -- must be last block so finish up */ X { /* Find out how many bytes and residual bits there are */ X byte = count >> 3; X bit = count & 7; X /* Copy X into XX since we need to modify it */ X for (i=0;i<=byte;i++) XX[i] = X[i]; X for (i=byte+1;i<64;i++) XX[i] = 0; X /* Add padding '1' bit and low-order zeros in last byte */ X mask = 1 << (7 - bit); X XX[byte] = (XX[byte] | mask) & ~( mask - 1); X /* If room for bit count, finish up with this block */ X if (byte <= 55) X { for (i=0;i<8;i++) XX[56+i] = MDp->count[i]; X MDblock(MDp,(unsigned int *)XX); X } X else /* need to do two blocks to finish up */ X { MDblock(MDp,(unsigned int *)XX); X for (i=0;i<56;i++) XX[i] = 0; X for (i=0;i<8;i++) XX[56+i] = MDp->count[i]; X MDblock(MDp,(unsigned int *)XX); X } X /* Set flag saying we're done with MD computation */ X MDp->done = 1; X } X} X X/* X** End of md4.c X*/ END_OF_FILE if test 10725 -ne `wc -c <'md4.c'`; then echo shar: \"'md4.c'\" unpacked with wrong size! fi # end of 'md4.c' fi if test -f 'md4_def.h' -a "${1}" != "-c" ; then echo shar: Will not clobber existing file \"'md4_def.h'\" else echo shar: Extracting \"'md4_def.h'\" \(194 characters\) sed "s/^X//" >'md4_def.h' <<'END_OF_FILE' X/* X** Include file for hashmd4/checkmd4 X*/ X X#define CHECKSUMHDR "X-Md4-Signature" X#define HDRFIRSTCHAR 'X' X#define TRUE 1 X#define FALSE 0 X#define HDRTEXTSIZE 32 X END_OF_FILE if test 194 -ne `wc -c <'md4_def.h'`; then echo shar: \"'md4_def.h'\" unpacked with wrong size! fi # end of 'md4_def.h' fi if test -f 'rfc1186' -a "${1}" != "-c" ; then echo shar: Will not clobber existing file \"'rfc1186'\" else echo shar: Extracting \"'rfc1186'\" \(34579 characters\) sed "s/^X//" >'rfc1186' <<'END_OF_FILE' X X X X X X XNetwork Working Group R. Rivest XRequest for Comments: 1186 MIT Laboratory for Computer Science X October 1990 X X X The MD4 Message Digest Algorithm X XStatus of this Memo X X This RFC is the specification of the MD4 Digest Algorithm. If you X are going to implement MD4, it is suggested you do it this way. This X memo is for informational use and does not constitute a standard. X Distribution of this memo is unlimited. X XTable of Contents X X 1. Abstract .................................................... 1 X 2. Terminology and Notation .................................... 2 X 3. MD4 Algorithm Description ................................... 2 X 4. Extensions .................................................. 6 X 5. Summary ..................................................... 7 X 6. Acknowledgements ............................................ 7 X APPENDIX - Reference Implementation ............................. 7 X Security Considerations.......................................... 18 X Author's Address................................................. 18 X X1. Abstract X X This note describes the MD4 message digest algorithm. The algorithm X takes as input an input message of arbitrary length and produces as X output a 128-bit "fingerprint" or "message digest" of the input. It X is conjectured that it is computationally infeasible to produce two X messages having the same message digest, or to produce any message X having a given prespecified target message digest. The MD4 algorithm X is thus ideal for digital signature applications, where a large file X must be "compressed" in a secure manner before being signed with the X RSA public-key cryptosystem. X X The MD4 algorithm is designed to be quite fast on 32-bit machines. X On a SUN Sparc station, MD4 runs at 1,450,000 bytes/second. On a DEC X MicroVax II, MD4 runs at approximately 70,000 bytes/second. On a X 20MHz 80286, MD4 runs at approximately 32,000 bytes/second. In X addition, the MD4 algorithm does not require any large substitution X tables; the algorithm can be coded quite compactly. X X The MD4 algorithm is being placed in the public domain for review and X possible adoption as a standard. X X X X XRivest [Page 1] X XRFC 1186 MD4 Message Digest Algorithm October 1990 X X X (Note: The document supersedes an earlier draft. The algorithm X described here is a slight modification of the one described in the X draft.) X X2. Terminology and Notation X X In this note a "word" is a 32-bit quantity and a byte is an 8-bit X quantity. A sequence of bits can be interpreted in a natural manner X as a sequence of bytes, where each consecutive group of 8 bits is X interpreted as a byte with the high-order (most significant) bit of X each byte listed first. Similarly, a sequence of bytes can be X interpreted as a sequence of 32-bit words, where each consecutive X group of 4 bytes is interpreted as a word with the low-order (least X significant) byte given first. X X Let x_i denote "x sub i". If the subscript is an expression, we X surround it in braces, as in x_{i+1}. Similarly, we use ^ for X superscripts (exponentiation), so that x^i denotes x to the i-th X power. X X Let the symbol "+" denote addition of words (i.e., modulo- 2^32 X addition). Let X <<< s denote the 32-bit value obtained by circularly X shifting (rotating) X left by s bit positions. Let not(X) denote the X bit-wise complement of X, and let X v Y denote the bit-wise OR of X X and Y. Let X xor Y denote the bit-wise XOR of X and Y, and let XY X denote the bit-wise AND of X and Y. X X3. MD4 Algorithm Description X X We begin by supposing that we have a b-bit message as input, and that X we wish to find its message digest. Here b is an arbitrary X nonnegative integer; b may be zero, it need not be a multiple of 8, X and it may be arbitrarily large. We imagine the bits of the message X written down as follows: X X m_0 m_1 ... m_{b-1} . X X The following five steps are performed to compute the message digest X of the message. X X Step 1. Append padding bits X X The message is "padded" (extended) so that its length (in bits) X is congruent to 448, modulo 512. That is, the message is X extended so that it is just 64 bits shy of being a multiple of X 512 bits long. Padding is always performed, even if the length X of the message is already congruent to 448, modulo 512 (in X which case 512 bits of padding are added). X X X XRivest [Page 2] X XRFC 1186 MD4 Message Digest Algorithm October 1990 X X X Padding is performed as follows: a single "1" bit is appended X to the message, and then enough zero bits are appended so that X the length in bits of the padded message becomes congruent to X 448, modulo 512. X X Step 2. Append length X X A 64-bit representation of b (the length of the message before X the padding bits were added) is appended to the result of the X previous step. In the unlikely event that b is greater than X 2^64, then only the low-order 64 bits of b are used. (These X bits are appended as two 32-bit words and appended low-order X word first in accordance with the previous conventions.) X X At this point the resulting message (after padding with bits X and with b) has a length that is an exact multiple of 512 bits. X Equivalently, this message has a length that is an exact X multiple of 16 (32-bit) words. Let M[0 ... N-1] denote the X words of the resulting message, where N is a multiple of 16. X X Step 3. Initialize MD buffer X X A 4-word buffer (A,B,C,D) is used to compute the message X digest. Here each of A,B,C,D are 32-bit registers. These X registers are initialized to the following values in X hexadecimal, low-order bytes first): X X word A: 01 23 45 67 X word B: 89 ab cd ef X word C: fe dc ba 98 X word D: 76 54 32 10 X X Step 4. Process message in 16-word blocks X X We first define three auxiliary functions that each take X as input three 32-bit words and produce as output one X 32-bit word. X X f(X,Y,Z) = XY v not(X)Z X g(X,Y,Z) = XY v XZ v YZ X h(X,Y,Z) = X xor Y xor Z X X In each bit position f acts as a conditional: if x then y else X z. (The function f could have been defined using + instead of X v since XY and not(X)Z will never have 1's in the same bit X position.) In each bit position g acts as a majority function: X if at least two of x, y, z are on, then g has a one in that bit X position, else g has a zero. It is interesting to note that if X X X XRivest [Page 3] X XRFC 1186 MD4 Message Digest Algorithm October 1990 X X X the bits of X, Y, and Z are independent and unbiased, the each X bit of f(X,Y,Z) will be independent and unbiased, and similarly X each bit of g(X,Y,Z) will be independent and unbiased. The X function h is the bit-wise "xor" or "parity" function; it has X properties similar to those of f and g. X X Do the following: X X For i = 0 to N/16-1 do /* process each 16-word block */ X For j = 0 to 15 do: /* copy block i into X */ X Set X[j] to M[i*16+j]. X end /* of loop on j */ X Save A as AA, B as BB, C as CC, and D as DD. X X [Round 1] X Let [A B C D i s] denote the operation X A = (A + f(B,C,D) + X[i]) <<< s . X Do the following 16 operations: X [A B C D 0 3] X [D A B C 1 7] X [C D A B 2 11] X [B C D A 3 19] X [A B C D 4 3] X [D A B C 5 7] X [C D A B 6 11] X [B C D A 7 19] X [A B C D 8 3] X [D A B C 9 7] X [C D A B 10 11] X [B C D A 11 19] X [A B C D 12 3] X [D A B C 13 7] X [C D A B 14 11] X [B C D A 15 19] X X [Round 2] X Let [A B C D i s] denote the operation X A = (A + g(B,C,D) + X[i] + 5A827999) <<< s . X (The value 5A..99 is a hexadecimal 32-bit X constant, written with the high-order digit X first. This constant represents the square X root of 2. The octal value of this constant X is 013240474631. See Knuth, The Art of X Programming, Volume 2 (Seminumerical X Algorithms), Second Edition (1981), X Addison-Wesley. Table 2, page 660.) X Do the following 16 operations: X [A B C D 0 3] X X X XRivest [Page 4] X XRFC 1186 MD4 Message Digest Algorithm October 1990 X X X [D A B C 4 5] X [C D A B 8 9] X [B C D A 12 13] X [A B C D 1 3] X [D A B C 5 5] X [C D A B 9 9] X [B C D A 13 13] X [A B C D 2 3] X [D A B C 6 5] X [C D A B 10 9] X [B C D A 14 13] X [A B C D 3 3] X [D A B C 7 5] X [C D A B 11 9] X [B C D A 15 13] X X [Round 3] X Let [A B C D i s] denote the operation X A = (A + h(B,C,D) + X[i] + 6ED9EBA1) <<< s . X (The value 6E..A1 is a hexadecimal 32-bit X constant, written with the high-order digit X first. This constant represents the square X root of 3. The octal value of this constant X is 015666365641. See Knuth, The Art of X Programming, Volume 2 (Seminumerical X Algorithms), Second Edition (1981), X Addison-Wesley. Table 2, page 660.) X Do the following 16 operations: X [A B C D 0 3] X [D A B C 8 9] X [C D A B 4 11] X [B C D A 12 15] X [A B C D 2 3] X [D A B C 10 9] X [C D A B 6 11] X [B C D A 14 15] X [A B C D 1 3] X [D A B C 9 9] X [C D A B 5 11] X [B C D A 13 15] X [A B C D 3 3] X [D A B C 11 9] X [C D A B 7 11] X [B C D A 15 15] X X Then perform the following additions: X A = A + AA X B = B + BB X X X XRivest [Page 5] X XRFC 1186 MD4 Message Digest Algorithm October 1990 X X X C = C + CC X D = D + DD X (That is, each of the four registers is incremented by X the value it had before this block was started.) X X end /* of loop on i */ X X Step 5. Output X X The message digest produced as output is A,B,C,D. That is, we X begin with the low-order byte of A, and end with the high-order X byte of D. X X This completes the description of MD4. A reference X implementation in C is given in the Appendix. X X4. Extensions X X If more than 128 bits of output are required, then the following X procedure is recommended to obtain a 256-bit output. (There is no X provision made for obtaining more than 256 bits.) X X Two copies of MD4 are run in parallel over the input. The first copy X is standard as described above. The second copy is modified as X follows. X X The initial state of the second copy is: X word A: 00 11 22 33 X word B: 44 55 66 77 X word C: 88 99 aa bb X word D: cc dd ee ff X X The magic constants in rounds 2 and 3 for the second copy of MD4 are X changed from sqrt(2) and sqrt(3) to cuberoot(2) and cuberoot(3): X X Octal Hex X Round 2 constant 012050505746 50a28be6 X Round 3 constant 013423350444 5c4dd124 X X Finally, after every 16-word block is processed (including the last X block), the values of the A registers in the two copies are X exchanged. X X The final message digest is obtaining by appending the result of the X second copy of MD4 to the end of the result of the first copy of MD4. X X X X X X XRivest [Page 6] X XRFC 1186 MD4 Message Digest Algorithm October 1990 X X X5. Summary X X The MD4 message digest algorithm is simple to implement, and provides X a "fingerprint" or message digest of a message of arbitrary length. X X It is conjectured that the difficulty of coming up with two messages X having the same message digest is on the order of 2^64 operations, X and that the difficulty of coming up with any message having a given X message digest is on the order of 2^128 operations. The MD4 X algorithm has been carefully scrutinized for weaknesses. It is, X however, a relatively new algorithm and further security analysis is X of course justified, as is the case with any new proposal of this X sort. The level of security provided by MD4 should be sufficient for X implementing very high security hybrid digital signature schemes X based on MD4 and the RSA public-key cryptosystem. X X6. Acknowledgements X X I'd like to thank Don Coppersmith, Burt Kaliski, Ralph Merkle, and X Noam Nisan for numerous helpful comments and suggestions. X XAPPENDIX - Reference Implementation X XThis appendix contains the following files: X X md4.h -- header file for using MD4 implementation X md4.c -- the source code for MD4 routines X md4driver.c -- a sample "user" routine X session -- sample results of running md4driver X X /* X ** ******************************************************************** X ** md4.h -- Header file for implementation of ** X ** MD4 Message Digest Algorithm ** X ** Updated: 2/13/90 by Ronald L. Rivest ** X ** (C) 1990 RSA Data Security, Inc. ** X ** ******************************************************************** X */ X X /* MDstruct is the data structure for a message digest computation. X */ X typedef struct { X unsigned int buffer[4]; /* Holds 4-word result of MD computation */ X unsigned char count[8]; /* Number of bits processed so far */ X unsigned int done; /* Nonzero means MD computation finished */ X } MDstruct, *MDptr; X X /* MDbegin(MD) X X X XRivest [Page 7] X XRFC 1186 MD4 Message Digest Algorithm October 1990 X X X ** Input: MD -- an MDptr X ** Initialize the MDstruct prepatory to doing a message digest X ** computation. X */ X extern void MDbegin(); X X /* MDupdate(MD,X,count) X ** Input: MD -- an MDptr X ** X -- a pointer to an array of unsigned characters. X ** count -- the number of bits of X to use (an unsigned int). X ** Updates MD using the first "count" bits of X. X ** The array pointed to by X is not modified. X ** If count is not a multiple of 8, MDupdate uses high bits of X ** last byte. X ** This is the basic input routine for a user. X ** The routine terminates the MD computation when count < 512, so X ** every MD computation should end with one call to MDupdate with a X ** count less than 512. Zero is OK for a count. X */ X extern void MDupdate(); X X /* MDprint(MD) X ** Input: MD -- an MDptr X ** Prints message digest buffer MD as 32 hexadecimal digits. X ** Order is from low-order byte of buffer[0] to high-order byte X ** of buffer[3]. X ** Each byte is printed with high-order hexadecimal digit first. X */ X extern void MDprint(); X X /* X ** End of md4.h X ****************************(cut)***********************************/ X X /* X ** ******************************************************************** X ** md4.c -- Implementation of MD4 Message Digest Algorithm ** X ** Updated: 2/16/90 by Ronald L. Rivest ** X ** (C) 1990 RSA Data Security, Inc. ** X ** ******************************************************************** X */ X X /* X ** To use MD4: X ** -- Include md4.h in your program X ** -- Declare an MDstruct MD to hold the state of the digest X ** computation. X ** -- Initialize MD using MDbegin(&MD) X X X XRivest [Page 8] X XRFC 1186 MD4 Message Digest Algorithm October 1990 X X X ** -- For each full block (64 bytes) X you wish to process, call X ** MDupdate(&MD,X,512) X ** (512 is the number of bits in a full block.) X ** -- For the last block (less than 64 bytes) you wish to process, X ** MDupdate(&MD,X,n) X ** where n is the number of bits in the partial block. A partial X ** block terminates the computation, so every MD computation X ** should terminate by processing a partial block, even if it X ** has n = 0. X ** -- The message digest is available in MD.buffer[0] ... X ** MD.buffer[3]. (Least-significant byte of each word X ** should be output first.) X ** -- You can print out the digest using MDprint(&MD) X */ X X /* Implementation notes: X ** This implementation assumes that ints are 32-bit quantities. X ** If the machine stores the least-significant byte of an int in the X ** least-addressed byte (e.g., VAX and 8086), then LOWBYTEFIRST X ** should be set to TRUE. Otherwise (e.g., SUNS), LOWBYTEFIRST X ** should be set to FALSE. Note that on machines with LOWBYTEFIRST X ** FALSE the routine MDupdate modifies has a side-effect on its input X ** array (the order of bytes in each word are reversed). If this is X ** undesired a call to MDreverse(X) can reverse the bytes of X back X ** into order after each call to MDupdate. X X */ X #define TRUE 1 X #define FALSE 0 X #define LOWBYTEFIRST FALSE X X /* Compile-time includes X */ X #include <stdio.h> X #include "md4.h" X X /* Compile-time declarations of MD4 "magic constants". X */ X #define I0 0x67452301 /* Initial values for MD buffer */ X #define I1 0xefcdab89 X #define I2 0x98badcfe X #define I3 0x10325476 X #define C2 013240474631 /* round 2 constant = sqrt(2) in octal */ X #define C3 015666365641 /* round 3 constant = sqrt(3) in octal */ X /* C2 and C3 are from Knuth, The Art of Programming, Volume 2 X ** (Seminumerical Algorithms), Second Edition (1981), Addison-Wesley. X ** Table 2, page 660. X */ X X X XRivest [Page 9] X XRFC 1186 MD4 Message Digest Algorithm October 1990 X X X #define fs1 3 /* round 1 shift amounts */ X #define fs2 7 X #define fs3 11 X #define fs4 19 X #define gs1 3 /* round 2 shift amounts */ X #define gs2 5 X #define gs3 9 X #define gs4 13 X #define hs1 3 /* round 3 shift amounts */ X #define hs2 9 X #define hs3 11 X #define hs4 15 X X /* Compile-time macro declarations for MD4. X ** Note: The "rot" operator uses the variable "tmp". X ** It assumes tmp is declared as unsigned int, so that the >> X ** operator will shift in zeros rather than extending the sign bit. X */ X #define f(X,Y,Z) ((X&Y) | ((~X)&Z)) X #define g(X,Y,Z) ((X&Y) | (X&Z) | (Y&Z)) X #define h(X,Y,Z) (X^Y^Z) X #define rot(X,S) (tmp=X,(tmp<<S) | (tmp>>(32-S))) X #define ff(A,B,C,D,i,s) A = rot((A + f(B,C,D) + X[i]),s) X #define gg(A,B,C,D,i,s) A = rot((A + g(B,C,D) + X[i] + C2),s) X #define hh(A,B,C,D,i,s) A = rot((A + h(B,C,D) + X[i] + C3),s) X X /* MDprint(MDp) X ** Print message digest buffer MDp as 32 hexadecimal digits. X ** Order is from low-order byte of buffer[0] to high-order byte of X ** buffer[3]. X ** Each byte is printed with high-order hexadecimal digit first. X ** This is a user-callable routine. X */ X void X MDprint(MDp) X MDptr MDp; X { int i,j; X for (i=0;i<4;i++) X for (j=0;j<32;j=j+8) X printf("%02x",(MDp->buffer[i]>>j) & 0xFF); X } X X /* MDbegin(MDp) X ** Initialize message digest buffer MDp. X ** This is a user-callable routine. X */ X void X MDbegin(MDp) X X X XRivest [Page 10] X XRFC 1186 MD4 Message Digest Algorithm October 1990 X X X MDptr MDp; X { int i; X MDp->buffer[0] = I0; X MDp->buffer[1] = I1; X MDp->buffer[2] = I2; X MDp->buffer[3] = I3; X for (i=0;i<8;i++) MDp->count[i] = 0; X MDp->done = 0; X } X X /* MDreverse(X) X ** Reverse the byte-ordering of every int in X. X ** Assumes X is an array of 16 ints. X ** The macro revx reverses the byte-ordering of the next word of X. X */ X #define revx { t = (*X << 16) | (*X >> 16); \ X *X++ = ((t & 0xFF00FF00) >> 8) | ((t & 0x00FF00FF) << 8); } X MDreverse(X) X unsigned int *X; X { register unsigned int t; X revx; revx; revx; revx; revx; revx; revx; revx; X revx; revx; revx; revx; revx; revx; revx; revx; X } X X /* MDblock(MDp,X) X ** Update message digest buffer MDp->buffer using 16-word data block X. X ** Assumes all 16 words of X are full of data. X ** Does not update MDp->count. X ** This routine is not user-callable. X */ X static void X MDblock(MDp,X) X MDptr MDp; X unsigned int *X; X { X register unsigned int tmp, A, B, C, D; X #if LOWBYTEFIRST == FALSE X MDreverse(X); X #endif X A = MDp->buffer[0]; X B = MDp->buffer[1]; X C = MDp->buffer[2]; X D = MDp->buffer[3]; X /* Update the message digest buffer */ X ff(A , B , C , D , 0 , fs1); /* Round 1 */ X ff(D , A , B , C , 1 , fs2); X ff(C , D , A , B , 2 , fs3); X ff(B , C , D , A , 3 , fs4); X X X XRivest [Page 11] X XRFC 1186 MD4 Message Digest Algorithm October 1990 X X X ff(A , B , C , D , 4 , fs1); X ff(D , A , B , C , 5 , fs2); X ff(C , D , A , B , 6 , fs3); X ff(B , C , D , A , 7 , fs4); X ff(A , B , C , D , 8 , fs1); X ff(D , A , B , C , 9 , fs2); X ff(C , D , A , B , 10 , fs3); X ff(B , C , D , A , 11 , fs4); X ff(A , B , C , D , 12 , fs1); X ff(D , A , B , C , 13 , fs2); X ff(C , D , A , B , 14 , fs3); X ff(B , C , D , A , 15 , fs4); X gg(A , B , C , D , 0 , gs1); /* Round 2 */ X gg(D , A , B , C , 4 , gs2); X gg(C , D , A , B , 8 , gs3); X gg(B , C , D , A , 12 , gs4); X gg(A , B , C , D , 1 , gs1); X gg(D , A , B , C , 5 , gs2); X gg(C , D , A , B , 9 , gs3); X gg(B , C , D , A , 13 , gs4); X gg(A , B , C , D , 2 , gs1); X gg(D , A , B , C , 6 , gs2); X gg(C , D , A , B , 10 , gs3); X gg(B , C , D , A , 14 , gs4); X gg(A , B , C , D , 3 , gs1); X gg(D , A , B , C , 7 , gs2); X gg(C , D , A , B , 11 , gs3); X gg(B , C , D , A , 15 , gs4); X hh(A , B , C , D , 0 , hs1); /* Round 3 */ X hh(D , A , B , C , 8 , hs2); X hh(C , D , A , B , 4 , hs3); X hh(B , C , D , A , 12 , hs4); X hh(A , B , C , D , 2 , hs1); X hh(D , A , B , C , 10 , hs2); X hh(C , D , A , B , 6 , hs3); X hh(B , C , D , A , 14 , hs4); X hh(A , B , C , D , 1 , hs1); X hh(D , A , B , C , 9 , hs2); X hh(C , D , A , B , 5 , hs3); X hh(B , C , D , A , 13 , hs4); X hh(A , B , C , D , 3 , hs1); X hh(D , A , B , C , 11 , hs2); X hh(C , D , A , B , 7 , hs3); X hh(B , C , D , A , 15 , hs4); X MDp->buffer[0] += A; X MDp->buffer[1] += B; X MDp->buffer[2] += C; X MDp->buffer[3] += D; X X X XRivest [Page 12] X XRFC 1186 MD4 Message Digest Algorithm October 1990 X X X } X X /* MDupdate(MDp,X,count) X ** Input: MDp -- an MDptr X ** X -- a pointer to an array of unsigned characters. X ** count -- the number of bits of X to use. X ** (if not a multiple of 8, uses high bits of last byte.) X ** Update MDp using the number of bits of X given by count. X ** This is the basic input routine for an MD4 user. X ** The routine completes the MD computation when count < 512, so X ** every MD computation should end with one call to MDupdate with a X ** count less than 512. A call with count 0 will be ignored if the X ** MD has already been terminated (done != 0), so an extra call with X ** count 0 can be given as a "courtesy close" to force termination X ** if desired. X */ X void X MDupdate(MDp,X,count) X MDptr MDp; X unsigned char *X; X unsigned int count; X { unsigned int i, tmp, bit, byte, mask; X unsigned char XX[64]; X unsigned char *p; X /* return with no error if this is a courtesy close with count X ** zero and MDp->done is true. X */ X if (count == 0 && MDp->done) return; X /* check to see if MD is already done and report error */ X if (MDp->done) X { printf("\nError: MDupdate MD already done."); return; } X /* Add count to MDp->count */ X tmp = count; X p = MDp->count; X while (tmp) X { tmp += *p; X *p++ = tmp; X tmp = tmp >> 8; X } X /* Process data */ X if (count == 512) X { /* Full block of data to handle */ X MDblock(MDp,(unsigned int *)X); X } X else if (count > 512) /* Check for count too large */ X { printf("\nError: MDupdate called with illegal count value %d." X ,count); X return; X X X XRivest [Page 13] X XRFC 1186 MD4 Message Digest Algorithm October 1990 X X X } X else /* partial block -- must be last block so finish up */ X { /* Find out how many bytes and residual bits there are */ X byte = count >> 3; X bit = count & 7; X /* Copy X into XX since we need to modify it */ X for (i=0;i<=byte;i++) XX[i] = X[i]; X for (i=byte+1;i<64;i++) XX[i] = 0; X /* Add padding '1' bit and low-order zeros in last byte */ X mask = 1 << (7 - bit); X XX[byte] = (XX[byte] | mask) & ~( mask - 1); X /* If room for bit count, finish up with this block */ X if (byte <= 55) X { for (i=0;i<8;i++) XX[56+i] = MDp->count[i]; X MDblock(MDp,(unsigned int *)XX); X } X else /* need to do two blocks to finish up */ X { MDblock(MDp,(unsigned int *)XX); X for (i=0;i<56;i++) XX[i] = 0; X for (i=0;i<8;i++) XX[56+i] = MDp->count[i]; X MDblock(MDp,(unsigned int *)XX); X } X /* Set flag saying we're done with MD computation */ X MDp->done = 1; X } X } X X /* X ** End of md4.c X ****************************(cut)***********************************/ X X /* X ** ******************************************************************** X ** md4driver.c -- sample routines to test ** X ** MD4 message digest algorithm. ** X ** Updated: 2/16/90 by Ronald L. Rivest ** X ** (C) 1990 RSA Data Security, Inc. ** X ** ******************************************************************** X */ X X #include <stdio.h> X #include "md4.h" X X /* MDtimetrial() X ** A time trial routine, to measure the speed of MD4. X ** Measures speed for 1M blocks = 64M bytes. X */ X MDtimetrial() X X X XRivest [Page 14] X XRFC 1186 MD4 Message Digest Algorithm October 1990 X X X { unsigned int X[16]; X MDstruct MD; X int i; X double t; X for (i=0;i<16;i++) X[i] = 0x01234567 + i; X printf X ("MD4 time trial. Processing 1 million 64-character blocks...\n"); X clock(); X MDbegin(&MD); X for (i=0;i<1000000;i++) MDupdate(&MD,X,512); X MDupdate(&MD,X,0); X t = (double) clock(); /* in microseconds */ X MDprint(&MD); printf(" is digest of 64M byte test input.\n"); X printf("Seconds to process test input: %g\n,t/1e6); X printf("Characters processed per second: %ld.\n,(int)(64e12/t)); X } X X /* MDstring(s) X ** Computes the message digest for string s. X ** Prints out message digest, a space, the string (in quotes) and a X ** carriage return. X */ X MDstring(s) X unsigned char *s; X { unsigned int i, len = strlen(s); X MDstruct MD; X MDbegin(&MD); X for (i=0;i+64<=len;i=i+64) MDupdate(&MD,s+i,512); X MDupdate(&MD,s+i,(len-i)*8); X MDprint(&MD); X printf(" \"%s\"\n",s); X } X X /* MDfile(filename) X ** Computes the message digest for a specified file. X ** Prints out message digest, a space, the file name, and a X ** carriage return. X */ X MDfile(filename) X char *filename; X { FILE *f = fopen(filename,"rb"); X unsigned char X[64]; X MDstruct MD; X int b; X if (f == NULL) X { printf("%s can't be opened.\n",filename); return; } X MDbegin(&MD); X while ((b=fread(X,1,64,f))!=0) MDupdate(&MD,X,b*8); X X X XRivest [Page 15] X XRFC 1186 MD4 Message Digest Algorithm October 1990 X X X MDupdate(&MD,X,0); X MDprint(&MD); X printf(" %s\n",filename); X fclose(f); X } X X /* MDfilter() X ** Writes the message digest of the data from stdin onto stdout, X ** followed by a carriage return. X */ X MDfilter() X { unsigned char X[64]; X MDstruct MD; X int b; X MDbegin(&MD); X while ((b=fread(X,1,64,stdin))!=0) MDupdate(&MD,X,b*8); X MDupdate(&MD,X,0); X MDprint(&MD); X printf("\n"); X } X X /* MDtestsuite() X ** Run a standard suite of test data. X */ X MDtestsuite() X { X printf("MD4 test suite results:\n"); X MDstring(""); X MDstring("a"); X MDstring("abc"); X MDstring("message digest"); X MDstring("abcdefghijklmnopqrstuvwxyz"); X MDstring X ("ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789"); X MDfile("foo"); /* Contents of file foo are "abc" */ X } X X main(argc,argv) X int argc; X char *argv[]; X { int i; X /* For each command line argument in turn: X ** filename -- prints message digest and name of file X ** -sstring -- prints message digest and contents of string X ** -t -- prints time trial statistics for 64M bytes X ** -x -- execute a standard suite of test data X ** (no args) -- writes messages digest of stdin onto stdout X */ X X X XRivest [Page 16] X XRFC 1186 MD4 Message Digest Algorithm October 1990 X X X if (argc==1) MDfilter(); X else X for (i=1;i<argc;i++) X if (argv[i][0]=='-' && argv[i][1]=='s') MDstring(argv[i]+2); X else if (strcmp(argv[i],"-t")==0) MDtimetrial(); X else if (strcmp(argv[i],"-x")==0) MDtestsuite(); X else MDfile(argv[i]); X } X X /* X ** end of md4driver.c X ****************************(cut)***********************************/ X X X -------------------------------------------------------------------- X --- Sample session. Compiling and using MD4 on SUN Sparcstation --- X -------------------------------------------------------------------- X >ls X total 66 X -rw-rw-r-- 1 rivest 3 Feb 14 17:40 abcfile X -rwxrwxr-x 1 rivest 24576 Feb 17 12:28 md4 X -rw-rw-r-- 1 rivest 9347 Feb 17 00:37 md4.c X -rw-rw-r-- 1 rivest 25150 Feb 17 12:25 md4.doc X -rw-rw-r-- 1 rivest 1844 Feb 16 21:21 md4.h X -rw-rw-r-- 1 rivest 3497 Feb 17 12:27 md4driver.c X > X >cc -o md4 -O4 md4.c md4driver.c X md4.c: X md4driver.c: X Linking: X > X >md4 -x X MD4 test suite results: X 31d6cfe0d16ae931b73c59d7e0c089c0 "" X bde52cb31de33e46245e05fbdbd6fb24 "a" X a448017aaf21d8525fc10ae87aa6729d "abc" X d9130a8164549fe818874806e1c7014b "message digest" X d79e1c308aa5bbcdeea8ed63df412da9 "abcdefghijklmnopqrstuvwxyz" X 043f8582f241db351ce627e153e7f0e4 X "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789" X a448017aaf21d8525fc10ae87aa6729d abcfile X > X >md4 -sabc -shi X a448017aaf21d8525fc10ae87aa6729d "abc" X cfaee2512bd25eb033236f0cd054e308 "hi" X > X >md4 * X a448017aaf21d8525fc10ae87aa6729d abcfile X X X XRivest [Page 17] X XRFC 1186 MD4 Message Digest Algorithm October 1990 X X X d316f994da0e951cf9502928a1f73300 md4 X 379adb39eada0dfdbbdfdcd0d9def8c4 md4.c X 9a3f73327c65954198b1f45a3aa12665 md4.doc X 37fe165ac177b461ff78b86d10e4ff33 md4.h X 7dcba2e2dc4d8f1408d08beb17dabb2a md4.o X 08790161bfddc6f5788b4353875cb1c3 md4driver.c X 1f84a7f690b0545d2d0480d5d3c26eea md4driver.o X > X >cat abcfile | md4 X a448017aaf21d8525fc10ae87aa6729d X > X >md4 -t X MD4 time trial. Processing 1 million 64-character blocks... X 6325bf77e5891c7c0d8104b64cc6e9ef is digest of 64M byte test input. X Seconds to process test input: 44.0982 X Characters processed per second: 1451305. X > X > X ------------------------ end of sample session -------------------- X X Note: A version of this document including the C source code is X available for FTP from THEORY.LSC.MIT.EDU in the file "md4.doc". X XSecurity Considerations X X The level of security discussed in this memo by MD4 is considered to X be sufficient for implementing very high security hybrid digital X signature schemes based on MD4 and the RSA public-key cryptosystem. X XAuthor's Address X X Ronald L. Rivest X Massachusetts Institute of Technology X Laboratory for Computer Science X NE43-324 X 545 Technology Square X Cambridge, MA 02139-1986 X X Phone: (617) 253-5880 X X EMail: rivest@theory.lcs.mit.edu X X X X X X X X X X XRivest [Page 18] X END_OF_FILE if test 34579 -ne `wc -c <'rfc1186'`; then echo shar: \"'rfc1186'\" unpacked with wrong size! fi # end of 'rfc1186' fi echo shar: End of archive 1 \(of 2\). cp /dev/null ark1isdone MISSING="" for I in 1 2 ; do if test ! -f ark${I}isdone ; then MISSING="${MISSING} ${I}" fi done if test "${MISSING}" = "" ; then echo You have unpacked both archives. rm -f ark[1-9]isdone else echo You still must unpack the following archives: echo " " ${MISSING} fi exit 0 exit 0 # Just in case... -- Kent Landfield INTERNET: kent@sparky.IMD.Sterling.COM Sterling Software, IMD UUCP: uunet!sparky!kent Phone: (402) 291-8300 FAX: (402) 291-4362 Please send comp.sources.misc-related mail to kent@uunet.uu.net. exit 0 # Just in case... -- Kent Landfield INTERNET: kent@sparky.IMD.Sterling.COM Sterling Software, IMD UUCP: uunet!sparky!kent Phone: (402) 291-8300 FAX: (402) 291-4362 Please send comp.sources.misc-related mail to kent@uunet.uu.net.