Martin raised a good point. However, the problem is not the APIs itself
(all the APIs have this weakness), but how the U2 environment relies on
the underlying OS for user authentication and security. This is the
fundamental basic flaw of the U2 environment.

Improve this and then you can build a better and more robust secure
environment. Perhaps including encrption and some form authentication
that the object hasn't been altered except by an internal command -- as
another level for critical system tables (such as the global catdir).
Obviously, putting all of these system critical files as SQL tables
would be a start. A radical departure from what is there now. But it's
what is required for today's environment.

Such security enhancements need to include the logging of whom, what,
when and how any entity has connected to the U2 environment. All of
which is sadly lacking presently, again relying on the OS and "homebrew"
methods.

As for the global catdir issue...there's an enhancement requested for
G23072 in U2TechConnect from 15 May 1998. It says:

SHORT DESCRIPTION
 Wants to remove the write permissions from catdir
 
FULL DESCRIPTION
 Customer would like to be able to remove the write permissions from the
programs in catdir. He would like to be able to only place certain users
in a unix group that would catalog programs. 
  
Regards,
David

PS., There other "security holes" too. This isn't the only one out
there. Nor is it the worse one.


-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Stevenson,
Charles
Sent: Saturday, 17 December 2005 2:47 AM
To: [email protected]
Subject: RE: [U2] global catdir question - security hole

> David Wolverton
> As a 'security risk', has IBM explicitly been asked to fix this item 
> and said they'd prefer just to leave a gaping hole?
> Or is it like many things, everyone knows it, but everyone thinks 
> someone else has followed up on it, and it must just be 'the way it 
> must be'...  Remember, IBM does not monitor this list for bugs to 
> fix... At least, I'm not expecting them to!
> 
> IBM seems to respond to TechConnect issues -- Log it!

I first _formally_ reported it in 1996, although I can't prove that at
this point.  I think there was a GTAR.
I have also had personal conversations about it with several
Vmark/Ardent/Informix/IBM people who were in a position to care or take
action. I remember asking about it in a question/answer panel during the
Ft. Lauderdale, 1998 national conference. So it has been a conscious
decision to leave it as is for about a decade. (When was UV first
implemented on NT? I do not remember how catdir's REF counter is
implemented there.)

I cannot imagine I am the only one who has ever complained.  It is a
glaring hole that everyone sees when they do the "ls -lt uv/catdir" that
John Reid mentioned at the top of this thread.  Or everyone who wondered
how the &MAP&'s REF counter was incremented.

I have not vigorously pursued it because those paying my bills, whose
DBs I would be protecting, have not cared enough.   I don't think the
majority of companies worry about malicious attacks (from their own
staff or contractors).  Even SJ+'s PRC, the premier U2 software control
tool, does not prevent malicious attempts to circumvent it.  My own
UV/RCS-based SCM effort tightens things down pretty well, but I haven't
figure out how to protect catdir.  I can only log changes to it.

I'll take it to U2UG's Enhancement committee.

cds
-------
-------
u2-users mailing list
[email protected]
To unsubscribe please visit http://listserver.u2ug.org/

Reply via email to