Operating System - OpenVMS
1748179 Members
4139 Online
108758 Solutions
New Discussion юеВ

Re: Comments on VMS TUD tour of Europe

 
SOLVED
Go to solution
Uwe Zessin
Honored Contributor

Re: Comments on VMS TUD tour of Europe

A stack overflow (what Willem was talking about) and a buffer overflow (be it within the bounds of a stack or not) is two different things to me. Perhaps Willem meant the second.
.
Karl Rohwedder
Honored Contributor

Re: Comments on VMS TUD tour of Europe

The last event was here in germany.
It was crowded, around 130 customer attended the event. It was excellent, especially the already mentioned presentation from Guy and Norm.
Willem Grooters
Honored Contributor

Re: Comments on VMS TUD tour of Europe

Uwe,

You're right - I meant the BUFFER overflow, but in context of stack. John explained the method of exploit. This is the exploit that caused the last 5 CERT reports on VMS - or, to be more precise, on OpenSource software ported to VMS.
Using this exploit requires execute acecss on the stack pages. By removing this, it won't be possible to execute pushed code, so it won't compromise security. Drawback, as stated, is that not just the offending process will crash (that is for sure), but quit possible, the whole system as well.

Willem
Willem Grooters
OpenVMS Developer & System Manager
Uwe Zessin
Honored Contributor

Re: Comments on VMS TUD tour of Europe

Well, VMS has always tried to minimize the impact when something went wrong. It first tries to rundown the currently executing image - I remember an incident in the old days when we had an access problem to a Massbus disk where a secondary pagefile was located. Lots of users were dropped to DCL, but the system was continuing.

The next step is trying to delete the process.

And finally, yes, the system will crash. But this is to protect the data - the system restarts with a new, fresh copy of OpenVMS instead of trying to continue from a possible compromised memory configuration.
.