nginx2root
85 points by MountainRose
85 points by MountainRose
This discovery was around the same time this same researcher found another critical vulnerability in a different project that we cannot talk about yet due to the same concerns as nginx2root.
Fun times lie ahead!
As of the time of this writing, the Nginx version in Debian stable is vulnerable and no security advisories / updates are available.
why on earth are they revealing all this information BEFORE the patch is available????
May be due to this: https://is-a.cat/@ar/117634569518544368
Enough of the details were already circulating that the usual coordinated disclosure circus wouldn't matter.
I expect at the point where five different places found at least some of the same issues, you have to assume that the moment the release is cut, people will figure out what the vulnerability is. Holding it any longer just means people might not realize the update is important on time.
(Also it does seem the releases have been published, so a patch is "available". It may not make it to your distro on time, but again, once the release is cut, it's effectively public already.)
sure, but even with AI and so on, surely it'd take would-be attackers at least a little time to develop a working exploit. doesn't it make sense to give the distros at least an hour or two to get this ready first?
It's probably true that it would take a little time, but aiui here they didn't necessarily give an immediately-usable exploit. You still have to turn it into an actual exploit itself. Nowadays, I think the capabilities gap between "the commit is out" and "the write-up is out", when the commit clearly showcases the problem, is specifically much shorter.
EDIT: ...and apparently some details of it had already leaked out, so perhaps there was some belief that it could already be exploited. This does appear to have been a coordinated disclosure with F5, at least, so perhaps F5 themselves felt there was no need to hold it back.
I suppose because it’s relatively easy to prevent before patches are available (by disabling HTTP/2), making it a bit less of a race between updating and attackers reverse-engineering the vulnerability?
There are already proof-of-concept scripts available, and it can be mitigated by disabling http 2. The point seems to be to get people to mitigate.
I’ve been watching H2O, my choice reverse proxy, push so many patches recently (finally slowed down this week after a month of tons of patches on a project that said it was largely feature complete & in maintenance mode until the mythical HTTP/4 is released). I’ve been trying to stay on top of it since the reverse proxy is exposed to the internet & I’m the maintainer of the Nixpkgs package. One advantage of not choosing the mainstream option is being less of a target—I doubt H2O2root would get 3 reposts like NGINX did here 😅
I'm now thankful that CertBot doesn't enable HTTP/2 and I never bothered to enable it myself. Sometimes procrastination is a good security strategy after all.
For anyone running nginx on Ubuntu or Debian, https://n.wtf/ provides latest nginx packages which have worked well for me. (1.31.9 is already available with fixes for nxinx2root.)