[Cascavel-pm] Byte code compile parte 2

Christian Reichel chrisreichel em bol.com.br
Quinta Julho 24 15:39:11 CDT 2003


Ola amigos de novo!

Olha o que eu achei, pode ser de agrado de alguns.

Here is a repost of a message from John Macdonald on how
to encrypt a perl script.  I found it useful and thought
you might too.  
Happy coding
Marc
edgarm em rpi.edu

Re-Printed with permission of the authors.
______________________________________________________________
Marc Edgar wrote :
|| 
|| Greetings,
|| 
|| I read your replies about protecting commercial Perl code
|| with great interest.  Could you tell me where I can learn how
|| to do this?
|| 
|| Many thanks.
|| Marc Edgar
|| edgarm em rpi.edu

There are two parts involved in doing this.

Building Perl with the ability to decrypt scripts as it
executes them is pretty simple.  Take a look at the Perl
source files: perl.h toke.c and usersub.c
and look for the use of the manifests CRYPTSCRIPT and CRYPTLOCAL.
The first two files, perl.h and toke.c, need no change on your
part - you just have to make sure that toke.c gets compiled with
the option -DCRYPTSCRIPT.  The significant part is in usersub.c.
If you compile it with -DCRYPTSCRIPT, then it will compile in a
very simple decryption routine - every byte has the top bit
flipped, which for normal ascii text means that it is turned on.
This encryption scheme is not very hard to crack, obviously,
which brings us to the more difficult part of the job.

The hard part is choosing an encryption algorithm that is strong
enough for your needs.  Then you would write a piece of C code
and put it into a file called cryptlocal.h and encure that you
compile usersub.c with the options -DCRYPTSCRIPT -DCRYPTLOCAL.
Your cryptlocal.h file must do two things - it must define the
manifest constants CRYPT_MAGIC1 and CRYPT_MAGIC2 and it must
provide a subroutine cryptfilter which takes FILE * argument,
reads in the remainder of that file, and decrypts it.  (If you
look in usersub.c at what happens if CRYPTLOCAL is not defined,
it gives you a template for what your code in cryptlocal.h must
look like to usersub.c when it gets included.  You would also
have to write a separate program that takes a script, copies
some portion of it without encrypting (like the leading #! line
for instance) and then write out the two CRYPT_MAGIC characters
and then write out the rest of the script in encrypted form.
The code in usersub.c already provides the cryptswitch function
called from toke.c - its job is to check for the CRYPT_MAGIC
characters and if they are found, fork off a process to run
your cryptfilter subroutine to decrypt the rest of the input
file.  It prevents an obvious security breaking trick by just
quiting if debugging it turned on.  I won't guarantee there
aren't any other holes that would allow somebody to break the
security without applying crypto-analysis to the encrypted
scripts - in fact, I am certain that it can be done.  So, if your
scripts are going to be worth millions of dollars to someone who
breaks the encryption, then you probably should consider going
a different route.  However, given a good choice of encryption
algorithm, this mechanism provides a fairly good barrier to
reading your scripts.

The reason I chose such a trivial encryption routine to be the
standard one included in the Perl distribution is that the routine
in the distribution is publicly readable code and vulnerable to
analysis by many people.  If there was a "pretty good" routine
provided, then many people might choose to use it.  It is a lot
more tempting for a cracker to spend some time trying to break
a code for which he has the source and which he knows is going to
be used by many people, than it is for him to try and break a code
that he doesn't have source to and which is different from the
code used by any other organization.

By the way, Larry has some neat tools to support the make procedure
that lend themselves really well to supporting this sort of addition.

If you have a script that looks like:

    #!/bin/sh
    # cryptbuild
    toke_cflags='ccflags="$ccflags -DCRYPTSCRIPT"'
    ttoke_cflags='ccflags="$ccflags -DCRYPTSCRIPT"'
    usersub_cflags='ccflags="$ccflags -DCRYPTSCRIPT -DCRYPTLOCAL"'
    export toke_cflags ttoke_cflags usersub_cflags
    ./cflags
    make $*

then to have your own ready to go Perl source, you just take the
standard Perl source, and add this script (call it cryptbuild) and
the file cryptlocal.h that you made before.  When you are ready to
build Perl on your customer's site, copy on that source tree and
then whenever your would have typed "make something" you just use
"cryptbuild something" instead.  For example,

    cryptbuild depend all test

When you have finished compiling Perl, you just remove the cryptlocal.h
file and the usersub.o module and you can leave the rest of the source
behind for your customers to use as they wish.



Mais detalhes sobre a lista de discussão Cascavel-pm