3 ms·
Before anyone suggests the X SECURITY extension (as in SSH's "untrusted" X forwarding): That only provides one level of confinement. Everything running behind t
by geofft 11y ago
Before anyone suggests the X SECURITY extension (as in SSH's "untrusted" X forwarding): That only provides one level of confinement. Everything running behind the SECURITY extension trusts each other. So if I run ssh -X somehost emacs, an "untrusted" XEvilTeddy won't be able to snoop on the terminal where I ran SSH, but it can snoop on my entire emacs window.
In college I spent a bit of time trying to write what's effectively a NATing proxy for the X protocol: clients get their own XIDs in their own namespace, and can't refer to other clients' resources. They see a virtual root window that only has their own windows. If X is going to stick around and people are going to attempt to care about security, it'd be worth implementing something like that.
- Zardoz84 11y agoCould do you release it ?
- geofft 11y agoI didn't get far enough to do anything past writing some X protocol parsing code in Haskell (for type etc. safety, since writing an X protocol parser in C sounded like a mistake), and there are better language choices today.
- lsaferite 11y agoThe subuser project is using Xpra to isolate and bridge the untrusted code into the trusted space. http://subuser.org/news/0.3.html http://subuser.org/news/0.3.html
- digi_owl 11y agoNow thats interesting.
- ashitlerferad 11y agoXACE is probably a better solution: http://www.x.org/releases/X11R7.5/doc/security/XACE-Spec.html http://www.x.org/releases/X11R7.5/doc/security/XACE-Spec.htm...