When Not to Use chroot

Hyena writes "Linux guru Alan Cox is quoted as saying 'chroot is not and never has been a security tool' in a KernelTrap article summarizing a lengthy thread on the Linux Kernel mailing list. The discussion began with a patch attempting to 'fix a security hole' in the Unix chroot command, trying to improve the ability of chroot to contain a process. When it was pointed out that people have been using chroot as a security tool for years, another kernel hacker retorted, 'incompetent people implementing security solutions are a real problem.' A quick search on the terms 'chroot+security' quickly reveals that many people have long thought (wrongly) that chroot's purpose was for improving security."

misleading...

[onemorehour (162028)]

This summary is truly and terribly misleading--the discussion simply says that a root user can break out of a chroot jail. Is this news? chroot can still be effectively used to contain processes that do not run as root.

Or is it?

[phorm (591458)]

The problem is that - for many root-running processes - running chroot has often been recommended as a security practice. This has often been the recommendation of the daemon authors, in the documentation, as a way to improve security.

I think that this was once (or may currently be) the case with bind and various MTA's. Standard practice for many daemons now is to start as root and fork as another non-privileged user, but not every daemon has this option.

Re:For daemons that don't run as root

[TheRaven64 (641858)]

Yes you can. As I said, you launch them with:

# chroot directory su nobody daemon

This will chroot into directory, and run daemon as the nobody user. As long as the version of su inside your jail doesn't have any security holes, you will be fine. If you don't trust it, I've written a modified version of chroot, which calls setuid() and setgid() to a named user and group before executing the named process. This eliminates the need for a working su inside the chroot, typically removing the need for any setuid programs in the chroot.

Just because you can only run a command as a superuser doesn't mean that all of the child processes of that command have to be run as the superuser. If this were the case, since init runs as root you would not have a multiuser system.

Re:Or is it?

[h4rm0ny (722443)]

as well as removing all the advice that's currently around to chroot processes, if chroot isn't actually for security.

Well a good place to start would be at http://en.wikipedia.org/wiki/Chroot [wikipedia.org] because one of the first things I did when this article came up was to double-check my understanding of chroot right there. FTFW:

This provides a convenient way to sandbox an untrusted, untested or otherwise dangerous program. It is also a simple kind of jail mechanism.

Re:misleading...

[nine-times (778537)]

Honestly, I might be in the classification of people who don't understand, but I resent the implication of "incompetent". I really hate the idea that you have to be an all-knowledgeable ubergeek, or else stay completely away from computers.

It seems like a simple issue: people have obviously felt the need to jail users for security reasons. They've been lead by someone to believe that chroot is a solution. If chroot isn't the solution, then why not give people a better solution instead of calling them incompetent.

It reminds me of a discussion that I was involved in a while back. I'll tell the story:

I posted to a forum asking what the best method was to jail SFTP users. I wanted something like FTP, but secure, and I didn't want users to be able to browse the whole filesystem. Some security expert chimed in basically calling me a moron, that if I didn't want people to browse the whole filesystem, I should use FTP and jail people. A lot of people in the forum agreed.

I tried to explain that I didn't want to use FTP because authentication wasn't encrypted, but if I must use FTP did anyone know how to get encryption on the login. The same security expert chimed in again to inform me that there wasn't actually a good implementation for SSL on FTP. A lot of people in the forum agreed.

I replied again asking more general advice. I wanted some kind of FTP-type login where authentication was encrypted and users were jailed. Again, it was implied that I was a moron. I was told I didn't understand security at all. I was told: If you trust your users, you shouldn't need to keep them from browsing the filesystem. If I didn't trust my users, then I should only worry about protecting my system from users, and jailed FTP logins were a good solution.

I tried to explain again that I didn't want to trust my users, but I wanted to protect my users' information by providing a secure method for login. The reply again was that I was stupid and incompetent, didn't understand security, and shouldn't be running a server anyway. Many people in the forum agreed.

So all I wanted was to know how to do something, and everyone thought it was a lot of fun to tell me how incompetent I was. If the answer is so obvious, why not explain it? More to the point, if you're such a fricken genius, why not figure out a way to get people the functionality they want in a form they'll understand? I still don't understand why secure authentication is a silly thing to want.

Assuming that everyone running a server is going to be a super-genius who wants to spend all day researching everything-- having that expectation is retarded. I've been working in IT for a while, and I'll tell you right now that there are an awful lot of admins that are way dumber than I am. A solution that only super-geniuses can figure out isn't a practical solution because no one will use it.

So if a lot of people want to jail users into a specific directory for various reasons, why can't we have that functionality? If one particular method (in this case, chroot) doesn't do a good job of jailing users, then can one of the super-geniuses out there come up with a good/real/practical/secure method for accomplishing that?

If you can't, please refrain from name-calling because they want to do something that you can't figure out how to accomplish.

Re:misleading...

[nine-times (778537)]

People don't tend to maintain a list of links to every subject they've ever discussed. So somebody has to do the searching, rightfully it should be the one who wants to know the answer...

Weren't you the one who just asked me elsewhere to post a link to the thread I was referring to?

Re:misleading...

[NNKK (218503)]

It can be used as a safety mechanism for testing non-malicious code. It can also be used to setup an environment that uses different versions of key libraries or utilities from the host to run code that might not otherwise work. Linux distributions tend to make use of the chroot facility for installation tasks, as well.

Using chroot on a process is like handing a person a map with an X on the destination. You've shown them where they're supposed to go, you haven't really done anything to prevent them from running off in another direction.

Re:misleading...Re:Asshole Stereotype

[visualight (468005)]

I think his comment was directed specifically at people who do not have enough understanding to implement a security solution on linux but think they do. Would the same comment coming from an official MS authority on security make you not want to use Vista?

Anyway, I do understand the perspective behind your reaction, but it doesn't fit in this specific case.

Re:misleading...Re:Asshole Stereotype

[Jah-Wren Ryel (80510)]

'incompetent people implementing security solutions are a real problem.'

Man, things like this make me want to NOT switch to Linux... Even though I had a better experience with Ubuntu that I did Vista.

What's your problem with that statement?
It's absolutely true and it is not limited to linux.

Let's take it a few more steps further as an example:

'incompetent people designing bridges are a real problem.'

'incompetent people performing surgery are a real problem.'

'incompetent people running the government are a real problem.'

Do you have a problem with any of those statements?

If you don't even know what chroot() is, then you are not the target of the man's complaint.

Re:misleading...Re:Asshole Stereotype

[bfields (66644)]

Man, things like this make me want to NOT switch to Linux... Even though I had a better experience with Ubuntu that I did Vista.

Go ahead. One of the (many) differences between Vista and Linux is that if you want to, you can march up to any of the core Linux kernel architects and tell them they have some fundamental long-standing unix interface totally wrong. The flip side of that is that they also won't stop anyone from flaming you if you do that.

And that's exactly what happened here. This guy wasn't posting a question on a local LUG. He was posting to the Linux kernel mailing list--the place where people actually meet to do kernel development. And he wasn't asking a question, he was arguing with people like Al Viro, a primary architect of the Linux filesystem api's. Which would be great if he was correct. But in fact he was totally wrong. And even that would be OK if he took the time to do his homework and to listen carefully when people explained the issue to him.

But he didn't really, so as a result he got a few flames. Some of the posters to lkml aren't polite in such a situation. I think that's kind of understandable, though actually agree that that's a problem. Are the core Vista kernel developers any better? Who knows? Does the general public doesn't have the option of participating in their development forums?

chroot, security and other uses

[jmorris42 (1458)]

> A process running as a regular user can break out too.

Only if it can get access to a broken SUID program, etc. I always though the point of a chroot in a security context was that the chroot only had the absolute minimum environment the task being isolated had to have, thus there shouldn't be too much in there to worry about getting exploited. Which is very useful.

Of course there are lots of uses for chroot that have nothing to do with security. Like keeping a whole 32bit environment seperate from the main 64bit install, wonderful tools like mock which allow keeping multiple distros on multiple arches installed and usable for building packages, etc.

Yeah, okay.

[by Anonymous Coward]

And RAID isn't for safety of your data either, hey?

Locks on your house aren't for security, they're just to keep the door closed if a cat pushes on it, right?

Seatbelts aren't to prevent you from flying through a windshield, they're just there so you don't slide around while taking corners.

Sorry, chroot *is* a security tool; it's very much useful for security. Maybe it wasn't written as one - maybe it was never intended to be one, but it *is* one now, no matter what Alan Cox says.

Software, especially open source software, is a lot like language. Despite the best efforts of nitpicking English teachers everywhere, the meaning of both words and code are whatever the vast majority agrees upon. And regardless of that, you may call me crazy, but the ability to restrict what a user can and can't access; what a process can or can't access, sounds like a security tool to me.

Chroot as a non-security tool

[BurningSpiral (413606)]

The purpose of chroot is to change the root directory. Chroot is particularly useful for recovery and diagnostics.

If you system that won't boot due to a boot sector problem Boot from a CD, mount your partitions, chroot to your root partition and run lilo/grub/... to rewrite your boot sector.

If you system that won't boot due to init script problems Boot from a CD, mount your partitions, chroot to your root partition and run run your full init process. If you run into problems, rerun your init scripts rather than rebooting.

Unfortunately, many people think chroot is a security tool so many people don't think it in non-security contexts.

Re:Chroot as a non-security tool

[xenocide2 (231786)]

Chroot is also used in building Debian packages. pbuilder sets up a seperate install to chroot into and ensure that build dependencies are correct by copying a base install, installing the build deps and attempting to build the package, eliminating some of those "works for me" FTBFS problems.

What the hell is wrong with these people?

[Epistax (544591)]

Do all OS developers become assholes? I've done a lot with VxWorks and I hope I don't become as twisted as these folk. I better just stay away from authoring my own kernel.

Are they serious?

[GiMP (10923)]

Please tell me that none of those bone-heads on LKVM advocating that chroot should be 'root proof' haven't had any patches accepted!

Of course chroot() doesn't do any good if a process inside of it is running as root. This is very well known. However, that doesn't make chroot() useless, it is still plenty useful. If you execute chroot() and then a seteuid(uid) where uid>0, then you prevent a hole/bug in your program from being exploited in a way that will allow file access/execution outside the chroot. That *is* a security advantage.

The point of "chroot security", cases where chroot is used to improve security, isn't to contain a malicious root user. The point is to prevent privilege escalation. You can create a chroot without any directories with mode 7771 privileges (a la /tmp), that is free of any setuid binaries, and without "useful" utilities like wget or curl that can make exploiting the system child's play. If your program runs inside of a chroot as a non-root user, and your chroot has no setuid binaries, and your kernel has no privilege escalation vulns, then you can be reasonably sure that nobody will break the chroot or achieve privilege escalation. Without a chroot, you would have to clear your entire server of setuid binaries and mode 7771 directories -- not to mention the potential for intentionally world-readable files that can lead to information exposure. Quite simply, a chroot prevents an arbitrary-execution vulnerability in bind (or other process) from exploiting a privilege escalation vulnerability in apache (or other process).

What some people think, apparently due to pure ignorance, is that chroot() is an end-all solution that will prevent even a root-owned process from accessing files outside the chroot, or worse, thinking that it protects the memory subsytem in any way. It doesn't. Even if the discussed patch was applied to the kernel, a root-owned process could still alter kernel memory, access raw devices, etc.

Improvements in ACLs under Linux minimize some of the needs for a chroot, other than the fact that a chroot is still much easier to configure and ACLs do not handle all of the use-cases that a chroot can solve. (and visa-versa, chroot cannot solve all of the problems solved by ACLs) Additionally, a chroot *and* ACLs can be used together for further-improved security.

Overtaken by virualization

[flyingfsck (986395)]

Basically chroot was an early attempt at virtualization. It allowed one to keep servers separated and contained, which improved reliability and availability. It has a minor positive effect on security, but not really all that much. There is a good argument to be made for not using chroot since it increases the maintenance effort, which frequently results in chrooted servers being neglected which reduces security. As for myself, I avoid using it, due to the maintenance issues.

Re:Duh.

[nuzak (959558)]

The explanation of that exploit is a good one, but it still requires root.

Isn't it just easier to remount the device?

News flash: root can break security. World ends at 10:00. Film at 11.

Re:Not for security use?

[cperciva (102828)]

If chroot didn't require root privileges, the following exploit would be possible:

1. Create ~/etc/master.passwd with an empty root password.
2. Hard link /usr/bin/su to ~/usr/bin/su. (Yes, you can create hard links to files which you don't own.)
3. Copy /bin/sh, /bin/chmod, and the necessary libraries to the corresponding places under ~.
4. chroot ~ /usr/bin/su root /bin/chmod 4555 /bin/sh
5. ~/bin/sh is now an unrestricted root shell.

Re:Not for security use?

[kfstark (50638)]

$ cd /tmp
$ ls -l /bin/bash
-rwxr-xr-x 1 root root 700560 2007-04-11 09:32 /bin/bash
$ ln /bin/bash foo
$ ls -l foo
-rwxr-xr-x 2 root root 700560 2007-04-11 09:32 foo

Uhhh, why is a regular user allowed to create a file owned by root?

Apparently, you don't know what a hard link is.

You haven't created a file owned by root. You've created an i-node pointing to the data blocks of a file owned by root.

If root were to rm /bin/bash, the file would still exist and have the proper ownership and be accessed through /tmp/foo

Your way, I could do the following on a file with 600 permission:

cd /tmp
ln /sbin/protected_file mine
chmod 666 mine
cat mine

Nice and easy way to get around a 600 permission.

The behavior is correct, not a bug.

Regards,

--Keith

Re:Not for security use?

[kocsonya (141716)]

> Apparently, you don't know what a hard link is.
>
> You haven't created a file owned by root. You've created an i-node pointing to the
> data blocks of a file owned by root.

No, you didn't.
You created a directory entry pointing to the i-node of the file which is also pointed
by the 'sh' entry in the /bin directory file. In addition, the reference count in the i-node has been bumped up by 1. One file is one i-node and that inode contains the owner, permission and timestamp info as well as the disc blocks that the file uses. For all practical purposes the file and the i-node are the same thing. A dir entry contains a name for the file and a pointer to the inode. There can be as many dir entries in different directories to the same file as you wish. The i-node also contains a reference count of how many dir entries point to it. When you rm a file, it will remove the dir entry and decrement the refcount in the i-node. When the refcount reaches 0, the file is actually deleted (actually it will only be deleted if it is not open at the time, but that's a separate issue).

You can delete a hard link to a file owned by anyone if you have write permission on the directory that contains a link:

zoltan@gep:~> ln /bin/bash foo
zoltan@gep:~> ls -l foo
-rwxr-xr-x 2 root root 572200 2005-09-10 03:43 foo
zoltan@gep:~> rm foo
rm: remove write-protected regular file `foo'? y
zoltan@gep:~>

On the other hand, if you try it in /tmp:

zoltan@gep:~> cd /tmp
zoltan@gep:/tmp> ln /bin/bash foo
zoltan@gep:/tmp> ls -l foo
-rwxr-xr-x 2 root root 572200 2005-09-10 03:43 foo
zoltan@gep:/tmp> rm foo
rm: remove write-protected regular file `foo'? y
rm: cannot remove `foo': Operation not permitted
zoltan@gep:/tmp>

The reason for that strange behaviour is this:

zoltan@gep:/tmp> ls -ld /tmp /home/zoltan
drwxr-x--- 114 zoltan users 16248 2007-09-28 14:23 /home/zoltan
drwxrwxrwt 32 root root 3176 2007-09-28 14:25 /tmp

The /tmp can be written by anyone. Note the 't' at the end of the permissons of /tmp: that is the sticky bit. Normally, if you have write permission on a directory, you can add, delete and rename files in that directory, no matter who actually owns those files (since you do not want to touch the file, you only change the directory file, which you have right to). If the sticky is set, however, an entry can only be removed (or, for that matter, renamed) by the owner of the file pointed by that entry. This is actually a trick (kludge, if you like) to solve the problem that sometimes you need to be able to create files in a common place (e.g. lock files) but you must not let anyone but the owner delete them and the basic UNIX acces right model is not enough for that. One such place is /tmp, an other prominent example is /var/lock. Since creating a hard link is to insert an entry to the directory file you can do it in such a directory - anyone can add entries, as the directory is word-writable. However, if the sticky is set, you can only remove (or rename) an entry if the i-node pointed by that entry is actually owned by you - which is not the case with /bin/bash. It is not because /bin/bash is owned by root, it is because it is not owned by you. It has nothing to do with the i-nodes or hard links, but the way the sticky bit works, which was a really neat trick to solve a problem, but it has this side effect.

Is it a bug? Well, it's certainly a feature... Is it a security problem? I don't know, I am no security expert, but I haven't heard of an exploit based on links and the sticky bit yet.

Re:Not for security use?

[MSG (12810)]

Uhhh, why is a regular user allowed to create a file owned by root?

They're not. They are, however, allowed to create a name in a directory (a link) that points to an existing inode. Since the inode has all of the owner and permission data, and the user is not allowed to modify the inode, allowing them to create a link to it is not an inherent risk.

Awesome, the semantics of directory permissions are not even honored anymore.. anyone else smell a kludge here?

Yes, they are. /tmp has the "sticky bit" set in its permissions. That means that users aren't allowed to remove or rename a file (link) unless they are the owner. Arguably, that is an odd semantic; since links don't have owners -- only files do -- it's not possible to allow users to remove or rename links that they created. However, claiming that the semantics of directory permissions aren't honored simply shows that you're ignorant of what the semantics are.

Re:Then what is it for?

[larry bagina (561269)]

Actually, Bill Joy invented chroot as a hack to use a custom /usr/include directory in a compiler that didn't support alternate include paths.

Citation

[BobKagy (25820)]

Actually, Bill Joy invented <tt>chroot</tt> as a hack to use a custom /usr/include directory in a compiler that didn't support alternate include paths.

http://blogs.sun.com/chrisg/tags/chroot [sun.com]

Dr. Marshall Kirk Mckusick, private communication: ``According to the SCCS logs, the chroot call was added by Bill Joy on March 18, 1982 approximately 1.5 years before 4.2BSD was released. That was well before we had ftp servers of any sort (ftp did not show up in the source tree until January 1983). My best guess as to its purpose was to allow Bill to chroot into the /4.2BSD build directory and build a system using only the files, include files, etc contained in that tree. That was the only use of chroot that I remember from the early days.''

Re:Then what is it for?

[soloport (312487)]

What's it for?

https://portal.mytesting.org:8080/ (including)
* tinyHTTP (AppWeb, Apache, etc.)
* SQLite (MySQL, Postgres, etc.)
* [chroot-path-0]/www/html/*
* Other ([chroot-path-0]/usr/lib, [chroot-path-0]/bin, etc.)

and repeat...

https://my-test-env.org:8081/

https://my-test-env.org:8082/

https://my-test-env.org:8083/

https://my-test-env.org:8084/

Next, bind /proc to all 5. Then make a script to easily update them from SVN. Done.

Now you have 5 chroot'ed web environments to help your test team (of 5) speed up Alpha testing. May be fraught with bad security? That's not the point.