Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Wed, 12 Dec 2018 11:02:52 -0800
From: Tavis Ormandy <>
Subject: Re: Multiple telnet.c overflows

On Wed, Dec 12, 2018 at 10:08 AM Hacker Fantastic
<> wrote:
> Hi Tavis, thanks for the input - I referenced Mikrotik as a vendor using a vulnerable implementation that can be used to escape restricted shells. This is just one example of a instance where a restricted shell could be escaped when using inetutils, or when the vulnerable code path reached unexpected systems (like NetBSD).

Yes, the bug exists on NetBSD, but in order for it to be a security
issue, there has to be an example of this bug being used to cross a
privilege boundary. I assume we agree that not every bug is a security
bug, there has to be some sort of supported security boundary that the
bug allows an attacker to violate. The question I'm asking is can you
elaborate on which security boundary is being crossed? I don't dispute
the bug exists and that NetBSD are shipping the code.

> As Mikrotik case is not an oss security issue I did not post the advisory here, but as I shared to you already on social media:
> (The overflows are present in those devices as well, several million of them, in case this isn't clear in our advisory)

That part is clear, but it's not clear to me that Mikrotik intend for
this to be a security boundary. Do you get unintended privileges from
exploiting this? Either way, RouterOS is not open source, so
oss-security isn't the right place to discuss it.

> The heap overflow occurs in ANY environment variables (an example instead of DISPLAY, use USER which maybe reachable via telnet://user@ip), yes the stack sprintf might not be remotely reachable which is why the advisory states "multiple overflows". If instances of telnet being called with a username via a URI handler the this would reach the heap overflow code path as described in the advisory. Thankfully, most modern browsers no longer implement telnet URI handlers anymore.

You say "most", but do you have an example of anyone invoking GNU
inetutils via untrusted telnet URIs? I think any example in a security
supported open-source project would be enough to justify calling this
a security issue.

> You are welcome to dismiss client side environment handling vulnerabilities as none-security issues or feel free to patch the referenced vulnerabilities as stated in the advisory. Thanks for your input I hope the comments above with the referenced advisory are clear enough and that the issue can be addressed by projects still using inetutils.

It's not that environment handling is a non-issue, I've reported
dozens over the years, it's just that it requires a privilege
boundary. For example, setuid binaries are the classic example.


Powered by blists - more mailing lists

Please check out the Open Source Software Security Wiki, which is counterpart to this mailing list.

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.