6 ms·
Nix 2.24 is vulnerable to (remote) privilege escalation
- johnklos 2y agoI don't get why someone who responded to this calls this something other than responsible disclosure. puck clearly writes, "at this point the disclosure timeline has passed". It's on Nix to meet the disclosure timeline. Missing that is Nix being irresponsible, not puck. I see this often, and it's not a good look when people would rather punish the reporter of an issue rather than the organization that was given plenty of time to do something about it.
- jaculabilis 2y ago> the organization that was given plenty of time One week doesn't seem like "plenty of time" to me. The guy who ack'd the initial report and created the vulnerability tracker in GitHub was on vacation.
- hkmaxpro 2y agoReporter’s defense: https://lobste.rs/s/ixb3v7/nix_2_24_is_vulnerable_remote_privilege#c_v6ogj5 https://lobste.rs/s/ixb3v7/nix_2_24_is_vulnerable_remote_pri...
- johnklos 2y agoSomeone who writes a response to a security vulnerability report, includes nobody else in the discussion, then leaves for vacation within 15 minutes of sending that report is irresponsible. Giving someone a week to respond is not unreasonable. If nobody responds in a week, it can safely be assumed that they don't take security seriously, and the responsible thing is to let the community know. Spin it how you like, but the "we're too important to respond" shtick is old and tired.
- exe34 2y agothat's a bit entitled. I'm coming over for dinner, I hope you're prepared.
- saghm 2y agoThis isn't "I'm entitled to free dinner", this is "your dinner guests are getting sick from food poisoning, and you're demanding I don't tell them".
- elorm 2y agoThis is a very poor take. The security team is composed of unpaid volunteers who work on numerous time-sensitive projects simultaneously. You may not be aware, but since a group of maintainers and contributors left earlier this year to form their own fork called "Lix," there have been many vacant positions across several Nix teams. They actually held a meeting about the security issue earlier in the day before the disclosure and had reached out to the reporter(0). The sense of entitlement here is pretty much rank because any animosity between the Lix team and Nix teams going forward will only be to the detriment of the Lix team. Everyone's really tight at the moment and no one is paid for this much drama every couple of months. https://discourse.nixos.org/t/2024-09-09-nix-team-meeting-minutes-176-175/51852 https://discourse.nixos.org/t/2024-09-09-nix-team-meeting-mi...
- johnklos 2y ago> before the disclosure and had reached out to the reporter(0). First, the only link you provided doesn't look to be related to this issue. Edit: I see it bizarrely redirects to "https://discourse.nixos.org/t/iohk-hiring-devops-with-nix-experience/2024 https://discourse.nixos.org/t/iohk-hiring-devops-with-nix-ex...". What happened to the minutes? Second, I understand that it's run by volunteers, that they might not have the humanpower they need, and so on - as a volunteer who spends a good bit of time working on an open source project, I get it - but if someone's about to go on vacation, they shouldn't just fire off an email with no information and leave. They should reply with information: "I'm leaving for vacation and we have no other people to handle this", or "I can't do anything myself for the next week, but let's cc someone else", or something. Also, when they finally did reach out, they didn't say it was being worked on, nor did they ask for an extension, nor give any kind of timeframe. Creating a point release means volunteers were working on things, and releasing it without the fix means they didn't take the security report seriously. Unless someone shows me something that doesn't point to a completely opaque process, I have to say I would likely've done the same thing. After all, if I reported something to an organization, and the organization didn't assure me they were working on it and offer some kind of time frame, and in the meanwhile released an update that didn't have a fix, I'd take that at face value: they just don't care about security (or don't understand the security implications, which is even worse - there's nothing wrong with being ignorant about a thing, but deciding to do nothing about a thing because of ignorance is inexcusable). So was puck being malicious by releasing this information? I don't think so. If I were a Nix user, I'd want to know about a security issue that might affect me, so I'd welcome this as someone trying to help Nix users. If it hurts the Nix organization, then tough cookies. They should've taken action and communicated better. Is the issue even fixed yet?
- soraminazuki 2y ago> If nobody responds in a week, it can safely be assumed that they don't take security seriously, and the responsible thing is to let the community know. That's an if that did not happen: > Eelco is working on it, there's a patch on the GitHub advisory, we plan to get it out on Monday, but no promises yet if everything will get done by then https://matrix.to/#/!VRULIdgoKmKPzJZzjj:nixos.org/$tJgEBGqKsyTUyCN9V_QvL21_rIt2zahFvw3KNujN8aI https://matrix.to/#/!VRULIdgoKmKPzJZzjj:nixos.org/$tJgEBGqKs...
- abhinavk 2y agoMore context: > The vulnerability report doesn’t mention it directly, but the discussion thread about it on Fedi gives some more context on that deadline: the reporter has had several previous vulnerability reports completely ignored by the Nix development team, including one open since February and still untriaged. The Nix development team received and acknowledged this new Nix 2.24 vulnerability on August 30th (so, > 9 days ago) and they seem to have mostly sat on it until today (the reporter received no further comms), to the extent that a new point release of Nix was released a few days after the vuln was reported and did not contain a fix. Source: https://lobste.rs/s/ixb3v7/nix_2_24_is_vulnerable_remote_privilege#c_qbpdlv https://lobste.rs/s/ixb3v7/nix_2_24_is_vulnerable_remote_pri... The first one from Feb: https://matrix-client.matrix.org/_matrix/media/v3/download/puck.moe/d3352f2cf9d9a0e84b9d551121335a043cc9971898783d711c266638be5a2e67?allow_redirect=true https://matrix-client.matrix.org/_matrix/media/v3/download/p... It's a community project run by volunteers but I don't think such response ("Impact: blabla") to a vulnerability gives a good impression to your users.
- myst1c 2y agoThis is a "both things are true" situation. There is a group working to smear Nix, Eelco, and everyone related to the project who are now playing the innocent victims and pretending they didn't sign a letter that started with the words "Eelco Dolstra’s leadership is corrosive to the Nix project." This is where people citing GH issue templates that contain "blabla" reinforces the narrative that the people who were working with Puck to solve the issue before she dropped a zero-day in public are somehow irresponsible. It is just that - a narrative. Everyone cannot stop staring at the sheer amount of narcissism on display in support of this narrative. The grain of truth is that it's probably not Puck's fault, and the security disclosure process could also be improved since it's evident a ball was accidentally dropped somewhere. More points of constant contact involved in solving these problems are good for everyone.
- soraminazuki 2y agoIf you know that much about the ongoing disagreements, you know very well that the volunteers weren’t given “plenty of time.” The author deliberately chose a short deadline and admits that. That deadline wasn’t known to Nix core team members because it wasn’t mentioned where the main discussion took place. This is all public information available in the linked Mastodon thread. If you’re going to choose a short deadline, you should at least make sure that the other side acknowledges it. Or at least warn before posting 0 days on Mastodon. That’s what “responsible” people do. The reason this didn’t happen is because of the ongoing drama between Lix and Nix. Very unfortunate.
- nrabulinski 2y agoThe deadline was literally publicly available in the public matrix channel which anyone can read even without a matrix account. The reporter also said they were willing to extend the deadline, if the nix team reached out. They didn’t and chose to ignore the publicly available messages. Given past experiences (it’s not the first vulnerability that was outright ignored) I think it’s fair to say “you have a week to respond, otherwise I’m dropping the vuln”. If only they responded and said “hey, we’re working on this but we need more time” nothing would’ve happened. But they didn’t.
- soraminazuki 2y agoThat's just flat out false. Both parties admit that the author and Nix maintainers were in touch regarding this vulnerability. Public accounts and meeting minutes prove that Nix maintainers were preparing a fix. The short deadline was only mentioned once, again separate from where the main discussion took place. There are dozens of Nix Matrix channels and so many messages being exchanged there every day. It's easy to miss. The author isn't a newcomer to Nix and most certainly knows this too. So the way the author dropped the 0 day on social media was outright and needlessly hostile considering that the vulnerability was acknowledged and being worked on.
- nrabulinski 2y agoWe must have different definitions of “being in touch”. Sending an email and going on vacation, then actively ignoring incoming messages isn’t being in touch. And I don’t blame the person from the Nix team who went on vacation, but not forwarding the researcher to anyone else is solely on the nix team. They responded to the message in the matrix channel which included the deadline so they were well aware of it. If they knew they wouldn’t be able to release the fix before the deadline, they could’ve (and should’ve) asked for an extension. Which they did not
- cookiengineer 2y agoI reported to Alpine that the security tracker has an incorrect approach of correlating packages and CVEs. At the time the security tracker had incorrect data for over 6 months without anyone caring about it. I forked it on their gitlab, patched it, and merge requested it. Nothing happened 3 years later, still the same. And they still doxx me from time to time. Mostly on github and shitter, so I don't see it early enough anyways to react to it. Assholes are always gonna be assholes online. The only way to deal with this is to walk away and not care about those toxic communities. There's better places on the internet and your time is just wasted effort in those chan-like areas.
- blueflow 2y ago> And they still doxx me from time to time. Mostly on github and shitter, so I don't see it early enough anyways to react to it. I'm somewhat glad I'm not the only one with problems like this in the Alpine community.
- cookiengineer 2y agoFrom my perspective it's just sad observing how a project with such huge potential derailed so quickly because of the wrong (read as: very insecure in their ego) people managing it. It's just childish behavior at this point.
- andrewchambers 2y agoIt seems like a binary cache can already get root on your system - it is serving you binaries to run, often as root. Don't authorize a binary cache you don't trust.
- ElvishJerricco 2y agoThe problem is that arbitrary users can cause nix to unpack arbitrary nars and edit arbitrary files that user shouldn't have permissions for. The system doesn't have to be configured to trust any particular binary cache. This is just straight up persistent privilege escalation, plain and simple.
- otabdeveloper4 2y ago> It seems like a binary cache can already get root on your system No, rootless Nix is pretty well supported.
- forgotpwd16 2y agoWondering how vulnerability applies outside NixOS. Does Nix outside NixOS also has such access over the system?
- one-punch 2y agolix [1] might be less affected. Pierre Bourdon noticed that lix refactored surrounding code 4 months ago [2], and a comment claims that this Lix commit at least patched a different vulnerability GHSA-wf4c-57rh-9pjg [3]. To use lix instead of nix, set `nix.package = pkgs.lix` in your NixOS/home-manager configurations. [1]: https://lix.systems/ https://lix.systems/ [2]: https://mastodon.delroth.net/@delroth/113110218127456491 https://mastodon.delroth.net/@delroth/113110218127456491 [3]: https://lobste.rs/s/ixb3v7/nix_2_24_is_vulnerable_remote_privilege#c_v6ogj5 https://lobste.rs/s/ixb3v7/nix_2_24_is_vulnerable_remote_pri...
- takeda 2y agoWhat's the difference between Nix and Lix? The website is still not entirely clear. I mean as an user, why would I want to use it (besides avoiding this vulnerability)
- clhodapp 2y agoLix is a (nixpkgs-compatible) fork of Nix, led by Nix community members that don't get along with the core Nix team. At this point, the primary reason to switch to Lix would be if you trusted the Lix folks more than the core Nix team
- myst1c 2y agoThere are plenty of fantastic reasons not to, despite the prior issues with Eelco's leadership style. The NixOS Starknet situation[1] was very suspicious and involved many of the wrong incentives for an open source project. One of the primary people involved was trying to get the drops transferred to other people's accounts (getting around geofencing prohibiting US withdrawals too). So it is virtually guaranteed that is being transferred toward something other than the NixOS binary cache despite it being emphasised it would go there. I guess it will be going to "save Nix together" for the fork. All €20 000 or what-not of it. At least Lix is doing some interesting things with the language and fixing some long-standing regressions, but some of the people involved seem to enjoy standing next to others doing the work whilst they loudly take credit, and participating in cryptocurrency ponzi schemes using open source as the vessel. [1] https://discourse.nixos.org/t/starknet-cryptocurrency-contributions/40268 https://discourse.nixos.org/t/starknet-cryptocurrency-contri...
- remexre 2y agoPossibly of note, pkgs.nix on nixpkgs (stable and unstable) is from the 2.18 series, and so not vulnerable to this.
- forgotpwd16 2y agoCan choose to install `latest` (or `2_24` explicitly or even `git` that builds nix directly from master) version which before the submitted disclosure was set to 2.24. Has since been changed `latest` to install 2.23 instead and `2_24` and `git` have been marked as vulnerable.
- 0x69420 2y agotechnicals aside, as they are now resolved or being resolved in one manner or another, i would really like an understanding of precisely where the breakdown in communication occurred. i have thus far heard conflicting reports that indicate either the relevant nix org team dropped the ball, or was doing things right and got the rug pulled out from under them, which indicate respectively that puck was either exercising the best option available, or, well, not. can someone establish a proper timeline of events?
- abhinavk 2y agohttps://lobste.rs/s/ixb3v7/nix_2_24_is_vulnerable_remote_privilege#c_v6ogj5 https://lobste.rs/s/ixb3v7/nix_2_24_is_vulnerable_remote_pri...
- rstarast 2y agoThis is fixed in 2.24.6: https://github.com/NixOS/nix/releases/tag/2.24.6 https://github.com/NixOS/nix/releases/tag/2.24.6 See also https://discourse.nixos.org/t/vulnerability-in-nix-2-24/51902 https://discourse.nixos.org/t/vulnerability-in-nix-2-24/5190... for updates. Can someone link to the actual fix? It's a bit hard to navigate the git history for me...
- SSLy 2y agoI think it is https://github.com/NixOS/nix/commit/12fa019ae558641df0a23a7973d64e687b2d8ba8 https://github.com/NixOS/nix/commit/12fa019ae558641df0a23a79...
- pablo1107 2y agoHow can such a simple code change fixes such a big vulnerability?
- SSLy 2y agoOr maybe it's https://github.com/NixOS/nix/commit/35575873813f60fff26f27a65e09038986f17cb5 https://github.com/NixOS/nix/commit/35575873813f60fff26f27a6... the commit log is tad unclear and the GHSA writer didn't bother themselves with linking the offending code
- rstarast 2y agoThanks, that one seems likely. Hmm, though this seems to affect the case-hack thingy only, which seems like a macos-specific feature...