A long time ago I added code to block ports for a given protocol.

See :

https://bugzilla.mozilla.org/show_bug.cgi?id=83401
http://lxr.mozilla.org/seamonkey/source/netwerk/base/src/nsIOService.cpp#84

Basically this prevented you from constructing cross-protocol attacks.

I was recently questioned about where this check happens. Right now, this check happens in necko -- all urls that get loaded, are explicitly checked. You can globally override these settings.

However, I believe that we should make this check only for urls that come from content. In doing so, you would allow extensions to be able to construct a URL to a port that is restricted while allowing the rest of the browser to enjoy the protection this check brings. It would also allow a technical user to type in a url to service they which to debug.

I am worried about people copying and pasting urls from spam into the url to avoid this this check. Maybe there needs to be a permissions style dialog similar to cookies.

Right now, can necko tell where a URL has come from? Does necko know if a url has come from user content verse chrome? I think I know the answer, but in asking it someone might find the bug number which addresses this shortcoming.

Thoughts?
_______________________________________________
dev-tech-network mailing list
[email protected]
https://lists.mozilla.org/listinfo/dev-tech-network

Reply via email to