Re: Trust issues with hooks and config files
On 03/09/2014 10:57 PM, Julian Brost wrote: On 07.03.2014 22:04, Jeff King wrote: Yes, this is a well-known issue. The only safe operation on a repository for which somebody else controls hooks and config is to fetch from it (upload-pack on the remote repository does not respect any dangerous config or hooks). I'm a little bit surprised that you and some other people I asked see this as such a low-priority problem as this makes social engineering attacks on multi-user systems, like they are common at universities, really easy (this is also how I noticed the problem). This, and a lot more control related issues, are solved by tools like gitolite (which I maintain) and Gerrit (from Google), and also many GUI based access control tools like gitlab, gitorious, etc. In these schemes the user does not have a "unix" account on the server, and any hooks that run will run as the "hosting user". Access is either via ssh pub keys (most commonly) or http auth. It is my belief that most multi-user systems have installed one of these systems, and therefore the situation you speak of does not arise. They probably didn't install them to solve *this* problem, but to keep some semblance of control over who can access what repo, but as a side-effect they solve this problem also. sitaram It has been discussed, but nobody has produced patches. I think that nobody is really interested in doing so because: 1. It introduces complications into previously-working setups (e.g., a daemon running as "nobody" serving repos owned by a "git" user needs to mark "git" as trusted). 2. In most cases, cross-server boundaries provide sufficient insulation (e.g., I might not push to an evil person's repo, but rather to a shared repo whose hooks and config are controlled by root on the remote server). If you want to work on it, I think it's an interesting area. But any development would need to think about the transition plan for existing sites that will be broken. I can understand the problem with backward compatibility but in my opinion the default behavior should definitely be to ignore untrusted config files and hooks as it would otherwise only protect users that are already aware of the issue anyways and manually enable this option. Are there any plans for some major release in the future that would allow introducing backward incompatible changes? I would definitely spend some time working on a patch but so far I have no idea of git's internals and never looked at the source before. At the very least, the current trust model could stand to be documented much better (I do not think the rule of "fetching is safe, everything else is not" is mentioned anywhere explicitly). Good point but not enough in my opinion as I haven't read every git manpage before running it for the first time. -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html signature.asc Description: OpenPGP digital signature
Re: Trust issues with hooks and config files
Julian Brost writes: > On 07.03.2014 22:04, Jeff King wrote: >> >> If you want to work on it, I think it's an interesting area. But >> any development would need to think about the transition plan for >> existing sites that will be broken. > > I can understand the problem with backward compatibility but in my > opinion the default behavior should definitely be to ignore untrusted > config files and hooks as it would otherwise only protect users that > are already aware of the issue anyways and manually enable this option. > > Are there any plans for some major release in the future that would > allow introducing backward incompatible changes? Git 2.0 has been in the planning for quite some months, and I am inclined to merge these topics prepared for that release to 'master' during this cycle. Anything new like this one is way too late for it, but that does not mean we can never do 3.0 in the future. Perhaps going this way might be possible: * Introduce a configuration that is read only from $HOME/.gitconfig (or its xdg equivalent) to enable or disable the "unsafe" behaviour. By default (i.e. when the above configuration is not set), allow "unsafe" read; when instructed by the above configuration to forbid "unsafe" read, ignore configuration files that are not owned by the owner of the process. People can toggle the "unsafe" read to experiment with the above (~gitdaemon/.gitconfig can perhaps be used to restrict the daemon access) Keep it that way for a few releases. * After a few releases, start warning people who do not have the "unsafe" option in their $HOME/.gitconfig about a future default change, to force them to explicitly set it. Keep it that way for a few releases. * Flip the default, perhaps still keeping the warning on the flipped default to help people who have not been following along. Keep it that way for a few releases. * Then finally remove the warning. A release cycle usually last 10-12 weeks on average. -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: Trust issues with hooks and config files
-BEGIN PGP SIGNED MESSAGE- Hash: SHA256 On 07.03.2014 22:04, Jeff King wrote: > Yes, this is a well-known issue. The only safe operation on a > repository for which somebody else controls hooks and config is to > fetch from it (upload-pack on the remote repository does not > respect any dangerous config or hooks). I'm a little bit surprised that you and some other people I asked see this as such a low-priority problem as this makes social engineering attacks on multi-user systems, like they are common at universities, really easy (this is also how I noticed the problem). > It has been discussed, but nobody has produced patches. I think > that nobody is really interested in doing so because: > > 1. It introduces complications into previously-working setups > (e.g., a daemon running as "nobody" serving repos owned by a "git" > user needs to mark "git" as trusted). > > 2. In most cases, cross-server boundaries provide sufficient > insulation (e.g., I might not push to an evil person's repo, but > rather to a shared repo whose hooks and config are controlled by > root on the remote server). > > If you want to work on it, I think it's an interesting area. But > any development would need to think about the transition plan for > existing sites that will be broken. I can understand the problem with backward compatibility but in my opinion the default behavior should definitely be to ignore untrusted config files and hooks as it would otherwise only protect users that are already aware of the issue anyways and manually enable this option. Are there any plans for some major release in the future that would allow introducing backward incompatible changes? I would definitely spend some time working on a patch but so far I have no idea of git's internals and never looked at the source before. > At the very least, the current trust model could stand to be > documented much better (I do not think the rule of "fetching is > safe, everything else is not" is mentioned anywhere explicitly). Good point but not enough in my opinion as I haven't read every git manpage before running it for the first time. -BEGIN PGP SIGNATURE- Version: GnuPG v2.0.22 (GNU/Linux) iQIcBAEBCAAGBQJTHKRfAAoJECLcYT6QIdBtGdkQALD24YhS2aHBzi/c6a/1+eOC xERUshDJvU/3picpTqqwFlqHB6X53vo7uXcCmJU8oaMsGLeLrTyrX77YOzrhlkuK kERkXGPgyoW1G4enATSURDbjW6kp7SG0lRvGTPNySqDt9FiZYTDP7CGsQRDDl8cL al50BoNFsx9K/kiPgbDJsenDi/MAclAYlHbHJnEB6aBo06G89zwC2tECFtXewnAD EbCKPI4tFrUZW/rWxHAEDVs+cI038nMzSNMi7I+HAMG48s+iMfFF69pkdOQjhIsc 3irisLQcKPVNRjSK/dGEKbqkAy9wziNza69tl0EgQn6ewju5NZ4xbAkWQG9LEWfZ Ux1safkumQsAKiYfn87t5YDXZ3vDYKbChKQv/UlicobVRm0YbhitY2AQAzu+wx1V mXmG6D91IxfR4B0+AA+3/E8huSD4JJ2laIUwIoYV+y4+ZAlxnT4cNdYiYAH4oEme wZ9R0wsxVfUm+uFdSBqsgEpv7Bp9PRcREuVXXz0GQH0wQ8zdwOeeNvA3v6ZtodRZ 0q7WOVJpd/3j4fQoWUXFAOZxDIZVorM0dQQvbhXiwOE9UY5Gwpq22lGKHM1t+8EO lOUCQBXjscfPgyLifdSbnaf11RGgVLERQlpz6hpEGht1J3AqF6tTZPjto+iVV97v nObYVtMXsAtrKbka2FF+ =ps1M -END PGP SIGNATURE- -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: Trust issues with hooks and config files
On Thu, Mar 06, 2014 at 10:47:43PM +0100, Julian Brost wrote: > I've noticed some behavior of git that might lead to some security > issues if the user is not aware of this. > > Assume we have an evil user on a system, let's call him eve. He > prepares a repository where he allows other user to push changes to. > If he now adds a post-receive hook, git will happly execute it as > whatever user pushes to this repository: Yes, this is a well-known issue. The only safe operation on a repository for which somebody else controls hooks and config is to fetch from it (upload-pack on the remote repository does not respect any dangerous config or hooks). > Something similiar might happen if eve adds some alias to the config > file in the repository and grants any other user read access to the > repository. These aliases will be executed when some other user is > running any git command in this repository. Even though git does not > allow defining aliases for existing commands, you might mistype > something, so adding an alias for "lg" instead of "log" might succeed: Much easier is to define an external diff or textconv tool; then the victim does not even have to typo. > I'd suggest taking a similar approach as Mercurial [1], i.e. ignoring > configuration files and hooks owned by another user unless the owner > is explicitly trusted It has been discussed, but nobody has produced patches. I think that nobody is really interested in doing so because: 1. It introduces complications into previously-working setups (e.g., a daemon running as "nobody" serving repos owned by a "git" user needs to mark "git" as trusted). 2. In most cases, cross-server boundaries provide sufficient insulation (e.g., I might not push to an evil person's repo, but rather to a shared repo whose hooks and config are controlled by root on the remote server). If you want to work on it, I think it's an interesting area. But any development would need to think about the transition plan for existing sites that will be broken. At the very least, the current trust model could stand to be documented much better (I do not think the rule of "fetching is safe, everything else is not" is mentioned anywhere explicitly). -Peff -- To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html