Re: [PATCHv4 RESEND 0/3] syscalls,x86: Add execveat() system call

From: Andy Lutomirski
Date: Sun Oct 19 2014 - 15:12:35 EST


On Oct 18, 2014 5:21 PM, "Eric W. Biederman" <ebiederm@xxxxxxxxxxxx> wrote:
>
> Andy Lutomirski <luto@xxxxxxxxxxxxxx> writes:
>
> > [Added Eric Biederman, since I think your tree might be a reasonable
> > route forward for these patches.]
> >
> > On Thu, Jun 5, 2014 at 6:40 AM, David Drysdale <drysdale@xxxxxxxxxx> wrote:
> >> Resending, adding cc:linux-api.
> >>
> >> Also, it may help to add a little more background -- this patch is
> >> needed as a (small) part of implementing Capsicum in the Linux kernel.
> >>
> >> Capsicum is a security framework that has been present in FreeBSD since
> >> version 9.0 (Jan 2012), and is based on concepts from object-capability
> >> security [1].
> >>
> >> One of the features of Capsicum is capability mode, which locks down
> >> access to global namespaces such as the filesystem hierarchy. In
> >> capability mode, /proc is thus inaccessible and so fexecve(3) doesn't
> >> work -- hence the need for a kernel-space
> >
> > I just found myself wanting this syscall for another reason: injecting
> > programs into sandboxes or otherwise heavily locked-down namespaces.
> >
> > For example, I want to be able to reliably do something like nsenter
> > --namespace-flags-here toybox sh. Toybox's shell is unusual in that
> > it is more or less fully functional, so this should Just Work (tm),
> > except that the toybox binary might not exist in the namespace being
> > entered. If execveat were available, I could rig nsenter or a similar
> > tool to open it with O_CLOEXEC, enter the namespace, and then call
> > execveat.
> >
> > Is there any reason that these patches can't be merged more or less as
> > is for 3.19?
>
> Yes. There is a silliness in how it implements fexecve. The fexecve
> case should be use the empty string "" not a NULL pointer to indication
> that. That change will then harmonize execveat with the other ...at
> system calls and simplify the code and remove a special case. I believe
> using the empty string "" requires implementing the AT_EMPTY_PATH flag.
>

Sounds reasonable.

> For sandboxes execveat seems to make a great deal of sense. I can
> get the same functionality by passing in a directory file descriptor
> calling fchdir and execve so this should not introduce any new security
> holes. And using the final file descriptor removes a race.

The problem with that approach is that the execed program now has its
current directory outside the sandbox, which could be problematic if
you don't trust that program.

>
> AT_SYMLINK_NOFOLLOW seems to have some limited utility as well, although
> for exec I don't know what problems it can solve.

It can always be added later

>
> Until I am done moving I won't have time to pick this up, and the code
> clearly needs another revision but I will be happy to work to see that
> we get a sane execveat implemented.

Do you have an ETA? If it's likely to miss 3.19, but if you'll have
time to review before then, I can try to do it.

>
> Eric
>
> p.s. I don't believe there are any namespaces issues where doing
> something with execveat flags make sense.
>

OK, I'll bite. How feasible would it be to have a flag that activated
pid_ns_for_children? That would reduce a lot of the ugliness in tools
like nsenter that need to fork to enter a pid ns.

I always assume that the reason for the active vs. for_children
distinction was because a lot of userspace libraries, including glibc,
would malfunction if getpid(2) started returning a different value.
But, for exec, this doesn't matter.

--Andy
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/