BOUNCE mi400@xxxxxxxxxxxx: Invalid 'Approved:' header Non-member submission from [Gene_Gaunt/ReviewWorks@xxxxxxxxxxxxxxx] (fwd)

I'm forwarding this to the list.  But please folks, only submit from the
emails you get MI400 at.  HOWEVER, if you need to sub under more than one
address due to how your environment is configed, go for it...with in
reason..

Don in DC


---------- Forwarded message ----------
Date: Thu, 2 Dec 1999 19:41:35 -0600
From: owner-mi400@midrange.com
To: owner-mi400@midrange.com
Subject: BOUNCE mi400@midrange.com:    Invalid 'Approved:' header   Non-member 
submission from [Gene_Gaunt/ReviewWorks@reviewworks.com]   

>From mi400-owner@midrange.com  Thu Dec  2 19:41:31 1999
Received: from david (david [10.168.100.1])
        by midrange.com (8.9.3/8.9.3) with ESMTP id TAA16211
        for <mi400@midrange.com>; Thu, 2 Dec 1999 19:41:30 -0600
Message-Id: <4.2.2.19991202183624.00b14e90@midrange>
X-Sender: david@midrange
X-Mailer: QUALCOMM Windows Eudora Pro Version 4.2.2 
Date: Thu, 02 Dec 1999 18:36:42 -0600
To: mi400@midrange.com
From: owner-mi400@midrange.com (by way of David Gibbs <david@midrange.com>)
Subject: BOUNCE mi400@midrange.com:    Non-member submission from
  [Gene_Gaunt/ReviewWorks@reviewworks.com]   
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"; format=flowed

Approved: 115439
From: Gene_Gaunt/ReviewWorks@reviewworks.com
X-Lotus-FromDomain: REVIEWWORKS
To: MI400@midrange.com
Message-ID: <8525683B.004D6C3C.00@rwservernt>
Date: Thu, 2 Dec 1999 09:35:02 -0500
Subject: encryption
Mime-Version: 1.0
Content-type: text/plain; charset=us-ascii
Content-Disposition: inline

Here is a shorter password implementation, with no loops:

dcl dd     User        auto char(10);
dcl dd     Password    auto char(10);
dcl dd     Encrypted   auto char(16);
dcl dd     Text        auto char(8);
dcl spcptr ?Text       auto init(Text);
dcl spcptr ?Cipher     auto;
dcl dd     Control     auto char(32);
dcl dd     Function    def(Control) pos(1) bin(2) init(2);
dcl dd     Size        def(Control) pos(3) bin(2) init(8);
dcl dd     Key         def(Control) pos(6) char(8);

     cpybrep    Encrypted, " ";
     cpybla     Text, User(1:8);
     cmpbla(b)  User(9:2), "  " / eq(AA);
     cpybtlls   Text(1:1), User( 9:1), X"0000";
     cpybtlls   Text(2:1), User( 9:1), X"0002";
     cpybtlls   Text(3:1), User( 9:1), X"0004";
     cpybtlls   Text(4:1), User( 9:1), X"0006";
     cpybtlls   Text(5:1), User(10:1), X"0000";
     cpybtlls   Text(6:1), User(10:1), X"0002";
     cpybtlls   Text(7:1), User(10:1), X"0004";
     cpybtlls   Text(8:1), User(10:1), X"0006";
     and(s)     Text, X"C0C0C0C0C0C0C0C0";
     xor(s)     Text, User(1:8);
AA: xor        Key, Password(1:8), X"5555555555555555";
     cpybtlls   Key, Key, X"0001";
     setspp     ?Cipher, Encrypted;
     cipher     ?Cipher, Control, ?Text;
     cmpbla(b)  Password(9:2), "  " / eq(BB);
     cpyblap    Key, Password(9:2), " ";
     xor(s)     Key, X"5555555555555555";
     cpybtlls   Key, Key, X"0001";
     addspp     ?Cipher, ?Cipher, 8;
     cipher     ?Cipher, Control, ?Text;
BB: brk        "LOOK";

If this were iterated, we get better performance by replacing CIPHER
with hand-assembly of only the pertinent DES components.  The initial
and final permutations in DES cancel each other, so they and aren't
needed in password manager.  An old IBM System/370 assembler trick
can permute 64-bit DES blocks in just seven sequential instructions.
Here is MI pseudo-code:

1. XLATE the 8-byte block into 64 bytes using a table
2. AND(S) the 64 bytes with a table
3. XLATE each 64 bytes into X'00' or X'FF'
4. AND(S) the 64 bytes with a sequential mask
5. OR(S) the 64 bytes into 32 bytes
6. OR(S) the 32 bytes into 16 bytes
7. OR(S) the 16 bytes into 8 bytes

The cerebral challenge is to transform the decimal FIPS PUB 46 tables
into binary tables for steps (1) and (2) above.  A good tool for this is
REXX compound variables.  If you get the REXX procedure MAKECRYPT
from News/400  you can see an example of this compound variable
transformation from FIPS PUB 46 into MI.  (Be sure to get the United States
version.  The Italian version translated MAKECRYPT into Italian and
changed MI indicator keyword NEG into NEQ, which destroys decryption.)




+---
| This is the MI Programmers Mailing List!
| To submit a new message, send your mail to MI400@midrange.com.
| To subscribe to this list send email to MI400-SUB@midrange.com.
| To unsubscribe from this list send email to MI400-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: dr2@cssas400.com
+---