On Thu, Sep 19, 2002 at 08:15:23AM +0200, Giuseppe Ghib� wrote:
> Well, and if the patch doesn't exists yet? And furthermore I repeat,
> in most cases there aren't ANY sysadmin who worries about upgrades.

It exists it just hasn't been kept up to date with the current gcc...
Which I have to wonder why?

> well, not as nimda (btw, aften one year I still get apache logs full of
> nimda and red worm attempts, this means that sysadmins often upgrades
> their machines ASAP...), but the fact that there aren't so much (especially
> because on Linux hasn't the email as main vehicle for the propagation)
> doesn't mean they aren't dangerous.

I didn't say they weren't dangerous I just said they weren't as
prevalent.  And nimda/code red are both NT worms.  I'll note that I was
getting tons of nimda attempts when it first become well known on the
net.  The latest Linux worm has elicited may a few dozen possible (can't
say for sure because of the way it attempts to do it) attempts since it
showed up on slashdot.

Comparing the threat of a NT based worm to a Linux based worm is
specious at best.  I believe the Linux community is far better suited to
respond to threats.  How many email worms has Windows had?  How long did
it take Microsoft to get fixes that reliably stopped the propagation of
these worms?  How long does it take the Linux community to respond to
similar threats.  You do the math.

> when updates doesn't cause you to reconfigure your daemons... ;-)

Very rarely have we had to put out an update that requires
reconfiguration.  We almost always patch to fix security issues.  So
that's a really lame response.  I can think of only one case that
warranted that type of upgrade and that was an Apache upgrade because we
moved everyone to using the new style configuration setup
(commonhttpd.conf et al).

> tell me of the latest 100 security linux advisors for daemons having buffer
> overruns the possibility to obtain a remote shell (or root shell) and how 
> many of them would have been blocked (with a DOS for instance) if the 
> daemon have had libsafe active or was compiled with stackguard enabled.

What you're not taking in consideration here is the potential real
problems Mandrake would have in implementing stackguard.  Compiling
everything in the distro with it wouldn't probably be an option because
I'm positive there are at least a few things that it wouldn't work right
with.  Making it an option to use isn't exactly an easy thing to do
because it's a compile time option.  Not a run time option.  Mandrake is
already limited on space for new RPMS....  Where are we going to find
the MB if not GB's to put the duplicated RPMS?

> If we had based our security to what other distros do, then we wouldn't
> have had any msec, libsafe, kernel-secure, etc. and all our security tools 
> would
> have been tcp_wrappers...

You're certainly not the only distro shipping libsafe or secured
kernels.  And the stackguard patch to gcc isn't exactly new it's been
around for years.  There has to be a reason why other distros haven't
adopted it.   Or for that matter why Mandrake hasn't adopted it.

However the last fear I have about implementing it is that it will make
the very thing worse that you claim is a reason for implementing it.
Patches are necessary with or without stackguard.  As I've seen happen
time and time again firewalls became an excuse for poorly implemented
security.  It's not that firewalls don't have a place in a security
regime.  It's that many of the newbie type admins you are targeting
your issues to will think that's all they have to do.  So indeed I fear
that applying a bandaide (stackguard) and then the ensuing PR/marketing
that will surround it will create a false sense of security.  And will
in fact make the problem worse.

In the end between the difficulty in implementing this change and the
problems with the attitudes it might create I tend to think it wouldn't
benefit us in the long run.  I could be wrong.  But that's my opinion.

At any rate though you didn't answer my question.  *WHY* have other
distros ignored this.  It's not like it's something nobody has known
about (it's been on slashdot at least once).  There has to be a reason
for it.  Even if it is the attitude fear that I have about it.  

We certainly should take advantage of analysis that other distros have
made of the technology in determining if we should implement it.  Don't
you think?

-- 
Ben Reser <[EMAIL PROTECTED]>
http://ben.reser.org

Never take no as an answer from someone who isn't authorized to say yes.

Reply via email to