RC6 Algorithm Optimized

[John Gleason]

'  Note: BELOW CODE NOT YET CONFIRMED. I'm testing against known correct test vectors
'        and this code may need modification to conform. The multiply overflows may need
'        to be reinstated. Stay tuned....
'  RC6 algorithm, AES finalist.
'  I corrected f_rnd and i_rnd macro multiply overflows and significantly optimized
'  this RC6 algorithm that was ported to PB from "C" by Aleksandr Dobrev, with testbed
'  code by Mike Trader.
'  Using it, my laptop averages 58 ticks/byte encoding, 53 ticks/byte decoding.
'  This demo will encode, decode and verify all files included on a list of files
'  stored in c:\fListCrypt.txt--one file per record--and display statistics after
'  each file. Individual files can be tested also.
'  Discussion here: http://www.powerbasic.com/support/pb...ad.php?t=35257
'  John Gleason, Oct. 22, 07.
' ====================================================================================

'  In cryptography, RC6 is a symmetric key block cipher derived from RC5.
'  It was designed by Ron Rivest, Matt Robshaw, Ray Sidney, and Yiqun Lisa Yin
'  to meet the requirements of the Advanced Encryption Standard (AES) competition.
'
'  The algorithm was one of the five finalists, and was also submitted to the NESSIE
'  and CRYPTREC projects. It is a proprietary algorithm, patented by RSA Security.
'
'  RC6 proper has a block size of 128 bits and supports key sizes of 128, 192 and 256 bits,
'  but, like RC5, it can be parameterised to support a wide variety of word-lengths,
'  key sizes and number of rounds. RC6 is very similar to RC5 in structure,
'  using data-dependent rotations, modular addition and XOR operations; in fact,
'  RC6 could be viewed as interweaving two parallel RC5 encryption processes.
'
'  However, RC6 does use an extra multiplication operation not present in RC5 in
'  order to make the rotation dependent on every bit in a word, and not just the
'  least significant few bits.
'
'
'  As RC6 has not been selected for the AES, it is not guaranteed that RC6
'  is royalty-free. As of January 2007, a web page on the official web site
'  of the designers of RC6, RSA Laboratories, states the following:
'  "We emphasize that if RC6 is selected for the AES, RSA Security will
'  not require any licensing or royalty payments for products using the algorithm".
'  The emphasis on the word "if" suggests that RSA Security Inc. may now require
'  licensing and royalty payments for any products using the RC6 algorithm.
'  RC6 is a patented encryption algorithm


' RC6, an encryption algorithm designed by Ron Rivest (RSA Labs) and submitted as a candidate algorithm for the
' Advanced Encryption Standard programme of the US National Institute of Standards and Technology."

' Ported to PB from "C" by Aleksandr Dobrev - Thanks to wonsky Lu for provide source on C
' Ver. 0.0.1  December 03, 2006

' Padding and testbed Mike Trader 2007


#COMPILE EXE
#DIM ALL
#REGISTER NONE


#INCLUDE "win32api.inc"


%LITTLE_ENDIAN = 1 ' For Little Endian Machines



' Utility macros
MACRO zbs(xx) = STRING$(xx, 0)


MACRO FUNCTION rotr(x,n)
   MACROTEMP s
   LOCAL s AS DWORD
   s = (x)
   ROTATE RIGHT s,n
END MACRO = s


'#define rotl(x,n)   (((x) << ((INT)((n) & 0x1f))) | ((x) >> ((INT)((32 - ((n) & 0x1f))))))
MACRO FUNCTION rotl(x,n)
   MACROTEMP s
   LOCAL s AS DWORD
   s = (x)
   ROTATE LEFT s,n
END MACRO = s
'--------------------------------added 2 macros
MACRO rotr2(x,n)
   MACROTEMP s
   LOCAL s AS DWORD
    s = x
   !mov eax, s
   !mov ecx, n
   !ror eax, cl
END MACRO

MACRO rotl2(x,n)
   !mov ecx, n
   !mov eax, x
   !rol eax, cl
END MACRO
'----------------------------end added 2 macros


' PUT OR GET a 32 BIT WORD (v) IN machine order FROM a BYTE address IN (x)
#IF %DEF(%LITTLE_ENDIAN)
   '#define u4byte_in(x)        (*(u4byte*)(x))
   MACRO FUNCTION u4byte_in(x)
     MACROTEMP pdwZmacro
     DIM pdwZmacro AS DWORD PTR
     pdwZmacro = (x)
   END MACRO = @pdwZmacro
   '#define u4byte_out(x,v)    (*(u4byte*)(x) = (v))
   MACRO u4byte_out(x,v)
     MACROTEMP pdwZmacro
     DIM pdwZmacro AS DWORD PTR
     pdwZmacro = (x)
     @pdwZmacro = v
   END MACRO

#ELSE
   '#define u4byte_in(x)        bswap(*(u4byte)(x))
   MACRO u4byte_in(x)
     MACROTEMP dwZmacro
     DIM pdwZmacro AS DWORD PTR
     pdwZmacro = (x)
     @pdwZmacro = bswap(@z)
   END MACRO
   '#define u4byte_out(x, v)    (*(u4byte*)(x) = bswap(v))
   MACRO u4byte_out(x, v)
     MACROTEMP pdwZmacro
     DIM pdwZmacro AS DWORD PTR
     pdwZmacro = (x)
     @pdwZmacro = bswap(v)
   END MACRO
#ENDIF



MACRO f_rnd(i,a,b,c,d)           'changed 2 macros------------------------
  MACROTEMP u,t,r,d2,b2
  LOCAL u, t, r AS LONG, d2, b2 AS DWORD
        !mov eax, d
        !mov ecx, d
        !add eax, eax
        !add eax, 1
        !mul ecx
        !mov d2,  eax
        rotl2(d2, 5)
        !mov u, eax

        !mov eax, b
        !mov ecx, b
        !add eax, eax
        !add eax, 1
        !mul ecx
        !mov b2,  eax
        rotl2(b2, 5)
        !mov t, eax

        !xor eax, a
        !mov r, eax
        rotl2(r      , u)
        !mov a, eax
        a = a + RC6Key(i)

        !mov eax, c
        !xor u,   eax
        rotl2(u      , t)
        !mov c, eax
        c = c + RC6Key(i + 1)
END MACRO

MACRO i_rnd(i,a,b,c,d)
  MACROTEMP u,t,d2,b2
  LOCAL u, t, d2, b2 AS DWORD
        !mov eax, d
        !mov ecx, d
        !add eax, eax
        !add eax, 1
        !mul ecx
        !mov d2,  eax
        rotl2(d2, 5)       'rotl2(d * (d + d + 1), 5)
        !mov u, eax

        !mov eax, b
        !mov ecx, b
        !add eax, eax
        !add eax, 1
        !mul ecx
        !mov b2,  eax
        rotl2(b2, 5)       'rotl2(b * (b + b + 1), 5)
        !mov t, eax

        rotr2(c - RC6Key(i + 1), t)
        !xor eax, u
        !mov c, eax
        rotr2(a - RC6Key(i), u)
        !xor eax, t
        !mov a, eax
END MACRO                        'end change------------------------------


UNION ConvertTwoByteTYPE          ' Converts a Two Byte Value
    str               AS STRING * 02
    igr               AS INTEGER
    wrd               AS WORD
    byt(1 TO 2)       AS BYTE
END UNION
GLOBAL CvtTwoByte AS ConvertTwoByteTYPE





'¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤'
SUB time_stamp_count(tick AS QUAD) ' CPU Clock count    Charles E V Pegge

  '---------------------------'
  '                           ' approx because it is not a serialised instruction
  '                           ' it may execute before or after other instructions
  '                           ' in the pipeline.
  ! mov ebx,tick              ' var address where count is to be stored.
  ! db  &h0f,&h31             ' RDTSC read time-stamp counter into edx:eax hi lo.
  ! mov [ebx],eax             ' save low order 4 bytes.
  ! mov [ebx+4],edx           ' save high order 4 bytes.
  '---------------------------'

END SUB


'¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤
FUNCTION StrToHex( sBin AS STRING ) AS STRING

  REGISTER i AS LONG, j AS LONG
  LOCAL a, b AS BYTE PTR
  LOCAL sHex AS STRING

    sHex  = STRING$( LEN(sBin)*2, " " ) ' make a string the right size
    a     = STRPTR(sHex)
    b     = STRPTR(sBin)
    j     = 0
    FOR i = 0 TO LEN(sBin)-1
      CvtTwoByte.str = HEX$(@b[i],2)  ' NO String concatenation
      @a[j] = CvtTwoByte.byt(1)
      INCR j
      @a[j] = CvtTwoByte.byt(2)
      INCR j
    NEXT i

  FUNCTION = sHex ' return string

END FUNCTION


'¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤'
' Invert BYTE order IN a 32 BIT variable
'#define bswap(x)    (rotl(x, 8) & 0x00ff00ff | rotr(x, 8) & 0xff00ff00)
FUNCTION bswap( BYVAL dwX AS DWORD ) AS DWORD

  LOCAL XXdwl,XXdwr AS DWORD

    XXdwl = dwX : XXdwr = dwX
    ROTATE RIGHT XXdwr,8 : XXdwr = XXdwr AND &Hff00ff00???
    ROTATE LEFT  XXdwl,8 : XXdwl = XXdwl AND &H00ff00ff???

  FUNCTION = XXdwl OR XXdwr

END FUNCTION


'¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤'
' initialise the key schedule FROM the USER supplied key
SUB RC6_set_key( UserKey() AS BYTE, BYVAL key_len AS DWORD, RC6Key() AS DWORD )

  LOCAL i, j, k, a, b, t AS DWORD
  DIM l(7) AS DWORD

  REDIM RC6Key(43)  ' STATIC u4byte RC6(l_key[44]);
  RC6Key(0) = &Hb7e15163???

  LOCAL pdwUserKey AS DWORD PTR
  pdwUserKey = VARPTR(UserKey(0))

    k = 1
    WHILE k < 44
        RC6Key(k) = RC6Key(k - 1) + &H9e3779b9???
        INCR k
    WEND

    k = 0
    WHILE k < key_len / 32
        l(k) = u4byte_in(pdwUserKey + (4 * k))
        INCR k
    WEND

    t = (key_len / 32) - 1 ' t = (key_len / 32);

    a = 0: b = 0: i = 0: j = 0

    k = 0
    WHILE k < 132
       a = rotl(RC6Key(i) + a + b, 3): b = b+a
       b = rotl(l(j) + b, b)
       RC6Key(i) = a : l(j) = b
       i = IIF(i = 43, 0, i + 1) ' i = (i + 1) % 44;
       j = IIF(j = t , 0, j + 1) ' j = (j + 1) % t;
       INCR k
    WEND

END SUB



'¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤'

MACRO mRC6_EnCrypt(pdw_InBlk, pdw_OutBlk) ' encrypt a 16byte block OF TEXT
  MACROTEMP a,b,c,d
  LOCAL  a,b,c,d AS DWORD

    a = u4byte_in(pdw_InBlk    ): b = u4byte_in(pdw_InBlk +  4) + RC6Key(0)
    c = u4byte_in(pdw_InBlk + 8): d = u4byte_in(pdw_InBlk + 12) + RC6Key(1)

    f_rnd( 2,a,b,c,d): f_rnd( 4,b,c,d,a)
    f_rnd( 6,c,d,a,b): f_rnd( 8,d,a,b,c)
    f_rnd(10,a,b,c,d): f_rnd(12,b,c,d,a)
    f_rnd(14,c,d,a,b): f_rnd(16,d,a,b,c)
    f_rnd(18,a,b,c,d): f_rnd(20,b,c,d,a)
    f_rnd(22,c,d,a,b): f_rnd(24,d,a,b,c)
    f_rnd(26,a,b,c,d): f_rnd(28,b,c,d,a)
    f_rnd(30,c,d,a,b): f_rnd(32,d,a,b,c)
    f_rnd(34,a,b,c,d): f_rnd(36,b,c,d,a)
    f_rnd(38,c,d,a,b): f_rnd(40,d,a,b,c)

    u4byte_out(pdw_OutBlk,     a + RC6Key(42))  : u4byte_out(pdw_OutBlk +  4, b)
    u4byte_out(pdw_OutBlk + 8, c + RC6Key(43))  : u4byte_out(pdw_OutBlk + 12, d)

END MACRO



'¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤'

MACRO mRC6_DeCrypt(pdw_InBlk, pdw_OutBlk) ' decrypt a 16byte block OF TEXT
  MACROTEMP a,b,c,d
  LOCAL  a,b,c,d AS DWORD

    d = u4byte_in(pdw_InBlk + 12): c = u4byte_in(pdw_InBlk + 8) - RC6Key(43)
    b = u4byte_in(pdw_InBlk +  4): a = u4byte_in(pdw_InBlk    ) - RC6Key(42)

    i_rnd(40,d,a,b,c): i_rnd(38,c,d,a,b)
    i_rnd(36,b,c,d,a): i_rnd(34,a,b,c,d)
    i_rnd(32,d,a,b,c): i_rnd(30,c,d,a,b)
    i_rnd(28,b,c,d,a): i_rnd(26,a,b,c,d)
    i_rnd(24,d,a,b,c): i_rnd(22,c,d,a,b)
    i_rnd(20,b,c,d,a): i_rnd(18,a,b,c,d)
    i_rnd(16,d,a,b,c): i_rnd(14,c,d,a,b)
    i_rnd(12,b,c,d,a): i_rnd(10,a,b,c,d)
    i_rnd( 8,d,a,b,c): i_rnd( 6,c,d,a,b)
    i_rnd( 4,b,c,d,a): i_rnd( 2,a,b,c,d)

    u4byte_out(pdw_OutBlk + 12, d - RC6Key(1)) : u4byte_out(pdw_OutBlk + 8, c)
    u4byte_out(pdw_OutBlk +  4, b - RC6Key(0)) : u4byte_out(pdw_OutBlk,     a)

END MACRO




'¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤¤'
FUNCTION PBMAIN() AS LONG

  DIM RC6Key(0) AS STATIC DWORD
  LOCAL KeyLen, SrcLen AS LONG

  LOCAL sKey, sFile, bsFile, sCipher, t, sFileName AS STRING
  LOCAL FileLen, i, k, hFile, PadLen, RetVal, nLoops, nBlocks, cryptFile AS LONG
  LOCAL PerfFreq, t1, t2 AS QUAD
  LOCAL cBeg, cEnd AS QUAD
  LOCAL Elapsed AS SINGLE

  LOCAL pIn, pOut AS BYTE PTR



    RetVal = QueryPerformanceFrequency(PerfFreq) ' Checkhardware supports High Resolution Timer

  DO
    sFile = "Low voltage cutoff should not be of great concern to the average user." ' Default

    IF cryptFile = 0 THEN
       cryptFile = 1
       hFile = FREEFILE
       OPEN "C:\fListCrypt.txt" FOR INPUT LOCK SHARED AS #1 ' <<this is a list of all the files you want to en/decrypt
                                                            ' and check for correctness of en/decryption. Does not write
                                                            ' to disk, only memory. 10K min size, 85MB max size tested.
    END IF
   newFile:
    LINE INPUT #1, sfileName
'   sFileName = "D:\qencod2.dat" ' <<You can also test just a single file if uncommented -------  **   CHANGE THIS   **
    hFile = FREEFILE ' Save the Declarations
    OPEN sFileName FOR BINARY LOCK SHARED AS hFile
      IF ERR THEN MSGBOX "Problem opening file to be encrypted",48,"File Error"+STR$(ERRCLEAR): EXIT FUNCTION
      IF LOF(hFile) < 10000 OR LOF(hFile) > 85000000 THEN CLOSE hFile: GOTO newFile
      GET$ #hFile, LOF(hFile), sFile
    CLOSE hFile '



    bsFile = sFile                                          ' remove this line if not verification testing
    SrcLen = LEN(sFile)
    t = t + "Input File length :" + STR$(SrcLen) + $CRLF


'          "12345678901234567890123456789012"
    sKey = "RC6SixteenByteKe"

    nLoops = 1 ' 999 for short string

    KeyLen = LEN(sKey)
    t = t + "Key length:" + STR$(KeyLen) + " (max=32)" + $CRLF

    DIM UserKey(KeyLen) AS STATIC BYTE
    POKE$ VARPTR(UserKey(0)), sKey  ' string to byte array

    RC6_set_key( UserKey(), KeyLen, RC6Key() ) ' Create password based key for crypt and decrypt




    t = t + sKey  + $CRLF + $CRLF
    t = t + "Input File: '"+sFileName+"', length=" + STR$(LEN(sFile)) + $CRLF




    PadLen = 16 - (SrcLen MOD 16)
    t = t + "PadLen :" + STR$(PadLen) + $CRLF

    nBlocks = (SrcLen+PadLen) / 16
    t = t + "nBlocks :" + STR$(nBlocks) + $CRLF


    sFile = sFile + zbs(LEN(PadLen))

    sCipher = zbs(SrcLen+PadLen) '

    t = t + "Cipher length :" + STR$(LEN(sCipher)) + $CRLF

    pIn  = STRPTR(sFile)
    pOut = STRPTR(sCipher)





' ----------------------------------------------------------  Encoding sCipher
    QueryPerformanceCounter t1

    time_stamp_count(cBeg) ' measuring cpu clock cycles. The overhead just for making this call is about 25 clocks

      LOCAL pi2, po2 AS LONG
      FOR k = 0 TO (nBlocks-1)*16 STEP 16
         pi2 = pIn+k
         po2 = pOut+k
         mRC6_EnCrypt(pi2, po2)
      NEXT

    time_stamp_count(cEnd) ' measuring cpu clock cycles. The overhead just for making this call is about 25 clocks
    t = t + "Encrypt Clock Cycles="+STR$( (cEnd-cBeg)\nLoops ) + $CRLF

    sCipher = sCipher + CHR$(PadLen) ' byte array to string

    QueryPerformanceCounter t2 : Elapsed = (t2-t1)/PerfFreq*1000 ' m/s

    t = t + "Encrypted in:"+STR$(Elapsed)+" m/s" + $CRLF
    t = t + "Encrypt Clock Cycles per Byte:"+STR$((cEnd-cBeg)\nLoops\LEN(sFile))+" c/b" + $CRLF + $CRLF
    sFile = "" ' Burn the Input.
    t = t + "Cipher length :" + STR$(LEN(sCipher)) + $CRLF
'   t = t + StrToHex(LEFT$(sCipher, 30)) + $CRLF + $CRLF + $CRLF






' ----------------------------------------------------------  Decoding sCipher
    QueryPerformanceCounter t1

    PadLen = ASC(RIGHT$(sCipher, 1))
    t = t + "Recovered PadLen :" + STR$(PadLen) + $CRLF

    SrcLen = LEN(sCipher) - 1

    nBlocks = SrcLen / 16
    t = t + "Recovered nBlocks :" + STR$(nBlocks) + $CRLF

    sFile = zbs(SrcLen)

    pIn  = STRPTR(sCipher)
    pOut = STRPTR(sFile)


    time_stamp_count(cBeg) ' measuring cpu clock cycles. The overhead just for making this call is about 25 clocks

      FOR k = 0 TO (nBlocks-1)*16 STEP 16
         pi2 = pIn+k
         po2 = pOut+k
         mRC6_DeCrypt(pi2, po2)
      NEXT

    time_stamp_count(cEnd) ' measuring cpu clock cycles. The overhead just for making this call is about 25 clocks
    t = t + "Decrypt Clock Cycles="+STR$( (cEnd-cBeg)\nLoops ) + $CRLF

    sFile = LEFT$(sFile, SrcLen-PadLen) ' convert byte array to string
    IF sFile = "" THEN MSGBOX "Decryption Operation Failed",64,"RC6" : EXIT FUNCTION
    QueryPerformanceCounter t2 : Elapsed = (t2-t1)/PerfFreq*1000 ' m/s
    IF sFile <> bsFile THEN ? "Error, decryption file <> original file."  ' for verification testing only

    sCipher = "" ' Burn the Cipher
    t = t + "Decrypted File length :" + STR$(LEN(sFile)) + $CRLF

    t = t + "Decrypt Completed in:"+STR$(Elapsed)+" m/s" + $CRLF
    t = t + "Decrypt Clock Cycles per Byte:"+STR$((cEnd-cBeg)\nLoops\LEN(sFile))+" c/b" + $CRLF+ $CRLF


    MSGBOX t,64,"RC6 Encryption"
    RESET t': BEEP
 LOOP UNTIL EOF(1)

END FUNCTION