Re: Time to patch OpenVMS - DCE-RPC buffer overflow

From: Undisclosed (nomail_at_dontbeaweaselspammer.com)
Date: 07/29/04 

Date: Thu, 29 Jul 2004 01:00:41 -0400

David J Dachtera wrote:

> For our purposes, VMS runs quite happily without a network stack loaded.
> So, that's how *I* think of it. If you add a layered product,
> application, etc. and it breaks something, guess what?

ahhh.

I guess that's just a Unix/Windows-ism on my part, assuming the TCP/IP
stack is integrated with the kernel.

> An understanding of the system startup process might useful at this
> point.
>
> It is also interesing to note that the startup procedure on the bootable
> VMS CD is really just a severely abbreviated STARTUP.COM that does only
> what is necessary to perform certian very basic functions, like
> INITIALIZE-ing and MOUNTing disks and tapes, running BACKUP and PCSI,
> and so on. It never does enable interactive logins - the entire system
> is run under control of some special DCL scripts. The option to execute
> DCL commands and scripts runs as a subprocess of the "startup" process.
> No network stack, ...

> Common misnomer. Similarly, neither DCL nor the command programs it
> invokes are "VMS" - they are part of the operating environment, but they
> are not "VMS".

ok.

see, some people define the entire base package of the *BSD's, shell,
basic userland, and everything up to but not including X11, as "the
operating system", since they were designed to work together and provide
basic functionality.

in the post below this, Dan Foster used the analogy of DCL being like
the user shell in Unix... well, while the shell is not part of the
kernel, it is considered part of the OS under the *BSD's.

so, VMS uses more of a minimalist definition.

> ...except that VMS has a somewhat different structure as compared to
> UN*X.
>
>
>>that's a daemon running over TCP/IP.
>
>
> In VMS-ish, a detached process (i.e., not "VMS"). In VMS's case, should
> the image exit, the detached process will likely die, denying the
> would-be exploiter access to the system.

interesting.

> ...but still worthy of note. Sites sensitive to such issues will
> probably want to test to see the exact reuslt in their system of
> attempting such an exploit.
>
> D.J.D.

definitely.

thanks David.

I've got some reading to do.

incidental question - why does DCL not do tab-completion style
completion? I would have thought it would have been a perfect fit for
VMS, given that command names are longer than Unix.. that's great from a
newbie scrutability standpoint, but it gets annoying when you try to
type things fast.

I know that typing abbreviations works (e.g. DIR for DIRECTORY), but I'd
like to be sure that what I'm about to hit "Enter" on is the right thing.

Relevant Pages

  • Re: Paris Gun
    ... of about 30 miles a minute, is about the same as the cruise velocity ... at altitude on afterburners. ... You are going to need a 800 mm Dora/Gustav shell to get enough internal ... What made "Little David" fun was the way it was loaded...you swung it ...
    (sci.space.history)
  • Re: Class and instances (Stack overflow)
    ... but IMHO objects should normally be on the stack. ... > Using "new" makes you look like a Java or C# programmer! ... > class MyClass ... > David Wilkinson ...
    (microsoft.public.vc.language)
  • Re: DEC TCP/IP Services 5.3 for Hobbyist?
    ... I've been able to develop some DCL procedures that use the freeware LD ... OpenVMS Hobbyist program. ... Many thanks, David. ...
    (comp.os.vms)
  • Re: top (procps-2.0.7-25) vulnerability
    ... Sheesh what is with top and folks not fixing OLD arse bugs. ... You may wanna check with William LeFebvre and make sure this issue is ... >you will get a shell. ... >is allocated on the stack. ...
    (Vuln-Dev)
  • Re: Aluminum Shell and Record BB
    ... David N. Makinson wrote: ... > 1) Should I remove paint from the right face of BB shell? ... I haven't gone wrong yet with the Campy torque specs. ...
    (rec.bicycles.tech)