Instead of going public, how about getting a hand from someone who can help you fully understand the bug and its ramifications? One problem vendors have to deal with is that only about 5% of vulnerabilities reported to them are for real -- the rest are pranks, misunderstandings, or people with creative definitions of what security ought to be. For every Weld Pond or Rain Forest Puppy who provides detailed, verifiable reports, there's another 19 Steve Gibsons telling anyone who'll listen that support for raw sockets is a vulnerability. I'm sure someone here on vuln-dev would be willing to help you explore the technical issue in a private forum. If it turns out to be for real, then you can notify the vendor and expect them to respond to your updated report in a timely manner. (Though I would consider the date of notification to be when you serve them the updated report, and not the day when you first contacted them with only preliminary data.) On the other hand, shame on this vendor for not following up on the DoS aspect of this bug. If you've demonstrated to the vendor that you've got a DoS attack attack, then you wouldn't be too out-of-line in publishing it as such. pty -----Original Message----- From: Josha Bronson [mailto:dmuzat_private] Sent: Wednesday, January 16, 2002 7:01 PM To: vuln-devat_private Subject: Complicated Disclosure Scenario Greetings fellow security folk, I would like to gather some opinions on a not so theoretical disclosure scenario. Please for the sake of focused discussion keep your replies related to the specific scenario that I am proposing and not alternate opinions on disclosure in general. The situation is thus. I have discovered a bug in a major software vendors application. Initially the bug presented itself as a way to crash the application, i.e. a DoS condition. Upon further research I determined that I was able to overwrite some return addresses by formating the overflow in a specific way. As we all know this means that there is the possibility that this could allow code to be executed on the remote system. At this point I contacted the vendor to alert them to the existence of this problem. After exchanging multiple emails, in which I tediously outlined the DoS condition and *potential* exploit situation I was told that they would wait until I determined if code could be exploited before they began creating an advisory or even working on a patch. I informed this vendor, who is by no means short on resources, that I might not be able to successfully make that determination due to constraints on my time (after all I do this for fun) and ability, as this problem exists on an architecture that I have very little experience with. I encouraged the vendor to begin their own investigation. They ignored this, and again stated that they would await my results. This is the problem as it sits. If I reach out to "the community" for additional assistance with researching this bug I might as well just send out an advisory. If I release an advisory the vendor will most likely not have a patch ready, they will feel violated and the user base will be left open to exploitation with no fix. If I do nothing, the problem persists and nothing gets accomplished, and maybe someone with not so good intentions discovers the same bug and uses it to do harm. So, what would you do? -- Josha Bronson dmuzat_private AngryPacket Security
This archive was generated by hypermail 2b30 : Thu Jan 17 2002 - 23:15:31 PST