This is a security fix for CVE-2024-21626. See the advisory[1] for more
details.
Aleksa Sarai (6):
init: don't special-case logrus fds
libcontainer: mark all non-stdio fds O_CLOEXEC before spawning init
cgroup: plug leaks of /sys/fs/cgroup handle
init: close internal fds before execve
setns init: do explicit lookup of execve argument early
init: verify after chdir that cwd is inside the container
Hang Jiang (1):
Fix File to Close
[1]: https://github.com/opencontainers/runc/security/advisories/GHSA-xr7r-f8xq-vfvv
Fixes: GHSA-xr7r-f8xq-vfvv CVE-2024-21626
LGTMs: cyphar AkihiroSuda kolyshkin lifubang
We close the logfd before execve so there's no need to special case it.
In addition, it turns out that (*os.File).Fd() doesn't handle the case
where the file was closed and so it seems suspect to use that kind of
check.
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
Given the core issue in GHSA-xr7r-f8xq-vfvv was that we were unknowingly
leaking file descriptors to "runc init", it seems prudent to make sure
we proactively prevent this in the future. The solution is to simply
mark all non-stdio file descriptors as O_CLOEXEC before we spawn "runc
init".
For libcontainer library users, this could result in unrelated files
being marked as O_CLOEXEC -- however (for the same reason we are doing
this for runc), for security reasons those files should've been marked
as O_CLOEXEC anyway.
Fixes: GHSA-xr7r-f8xq-vfvv CVE-2024-21626
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
We auto-close this file descriptor in the final exec step, but it's
probably a good idea to not possibly leak the file descriptor to "runc
init" (we've had issues like this in the past) especially since it is a
directory handle from the host mount namespace.
In practice, on runc 1.1 this does leak to "runc init" but on main the
handle has a low enough file descriptor that it gets clobbered by the
ForkExec of "runc init".
OPEN_TREE_CLONE would let us protect this handle even further, but the
performance impact of creating an anonymous mount namespace is probably
not worth it.
Also, switch to using an *os.File for the handle so if it goes out of
scope during setup (i.e. an error occurs during setup) it will get
cleaned up by the GC.
Fixes: GHSA-xr7r-f8xq-vfvv CVE-2024-21626
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
If we leak a file descriptor referencing the host filesystem, an
attacker could use a /proc/self/fd magic-link as the source for execve
to execute a host binary in the container. This would allow the binary
itself (or a process inside the container in the 'runc exec' case) to
write to a host binary, leading to a container escape.
The simple solution is to make sure we close all file descriptors
immediately before the execve(2) step. Doing this earlier can lead to very
serious issues in Go (as file descriptors can be reused, any (*os.File)
reference could start silently operating on a different file) so we have
to do it as late as possible.
Unfortunately, there are some Go runtime file descriptors that we must
not close (otherwise the Go scheduler panics randomly). The only way of
being sure which file descriptors cannot be closed is to sneakily
go:linkname the runtime internal "internal/poll.IsPollDescriptor"
function. This is almost certainly not recommended but there isn't any
other way to be absolutely sure, while also closing any other possible
files.
In addition, we can keep the logrus forwarding logfd open because you
cannot execve a pipe and the contents of the pipe are so restricted
(JSON-encoded in a format we pick) that it seems unlikely you could even
construct shellcode. Closing the logfd causes issues if there is an
error returned from execve.
In mainline runc, runc-dmz protects us against this attack because the
intermediate execve(2) closes all of the O_CLOEXEC internal runc file
descriptors and thus runc-dmz cannot access them to attack the host.
Fixes: GHSA-xr7r-f8xq-vfvv CVE-2024-21626
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
(This is a partial backport of a minor change included in commit
dac41717465462b21fab5b5942fe4cb3f47d7e53.)
This mirrors the logic in standard_init_linux.go, and also ensures that
we do not call exec.LookPath in the final execve step.
While this is okay for regular binaries, it seems exec.LookPath calls
os.Getenv which tries to emit a log entry to the test harness when
running in "go test" mode. In a future patch (in order to fix
CVE-2024-21626), we will close all of the file descriptors immediately
before execve, which would mean the file descriptor for test harness
logging would be closed at execve time. So, moving exec.LookPath earlier
is necessary.
Ref: dac4171746 ("runc-dmz: reduce memfd binary cloning cost with small C binary")
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
If a file descriptor of a directory in the host's mount namespace is
leaked to runc init, a malicious config.json could use /proc/self/fd/...
as a working directory to allow for host filesystem access after the
container runs. This can also be exploited by a container process if it
knows that an administrator will use "runc exec --cwd" and the target
--cwd (the attacker can change that cwd to be a symlink pointing to
/proc/self/fd/... and wait for the process to exec and then snoop on
/proc/$pid/cwd to get access to the host). The former issue can lead to
a critical vulnerability in Docker and Kubernetes, while the latter is a
container breakout.
We can (ab)use the fact that getcwd(2) on Linux detects this exact case,
and getcwd(3) and Go's Getwd() return an error as a result. Thus, if we
just do os.Getwd() after chdir we can easily detect this case and error
out.
In runc 1.1, a /sys/fs/cgroup handle happens to be leaked to "runc
init", making this exploitable. On runc main it just so happens that the
leaked /sys/fs/cgroup gets clobbered and thus this is only consistently
exploitable for runc 1.1.
Fixes: GHSA-xr7r-f8xq-vfvv CVE-2024-21626
Co-developed-by: lifubang <lifubang@acmcoder.com>
Signed-off-by: lifubang <lifubang@acmcoder.com>
[refactored the implementation and added more comments]
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
(This is a cherry-pick of 482e56379a1c2f8b2573f07c06013510e59f9eb8.)
Using ints for all of our mapping structures means that a 32-bit binary
errors out when trying to parse /proc/self/*id_map:
failed to cache mappings for userns: failed to parse uid_map of userns /proc/1/ns/user:
parsing id map failed: invalid format in line " 0 0 4294967295": integer overflow on token 4294967295
This issue was unearthed by commit 1912d5988b ("*: actually support
joining a userns with a new container") but the underlying issue has
been present since the docker/libcontainer days.
In theory, switching to uint32 (to match the spec) instead of int64
would also work, but keeping everything signed seems much less
error-prone. It's also important to note that a mapping might be too
large for an int on 32-bit, so we detect this during the mapping.
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
(This is a cherry-pick of ebcef3e651e61aeee96546301d8db9e92b505ce6.)
It turns out that the error added in commit 09822c3da8 ("configs:
disallow ambiguous userns and timens configurations") causes issues with
containerd and CRIO because they pass both userns mappings and a userns
path.
These configurations are broken, but to avoid the regression in this one
case, output a warning to tell the user that the configuration is
incorrect but we will continue to use it if and only if the configured
mappings are identical to the mappings of the provided namespace.
Fixes: 09822c3da8 ("configs: disallow ambiguous userns and timens configurations")
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
(This is a cherry-pick of 6fa8d068438ed47e8318448c34fc4587612a0740.)
Given we've had several bugs in this behaviour that have now been fixed,
add an integration test that makes sure that you can start a container
that joins all of the namespaces of a second container.
The only namespace we do not join is the mount namespace, because
joining a namespace that has been pivot_root'd leads to a bunch of
errors. In principle, removing everything from config.json that requires
a mount _should_ work, but the root.path configuration is mandatory and
we cannot just ignore setting up the rootfs in the namespace joining
scenario (if the user has configured a different rootfs, we need to use
it or error out, and there's no reasonable way of checking if if the
rootfs paths are the same that doesn't result in spaghetti logic).
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
(This is a cherry-pick of 09822c3da8ad8fa91b8796c5abf27ef06814a0c3.)
For userns and timens, the mappings (and offsets, respectively) cannot
be changed after the namespace is first configured. Thus, configuring a
container with a namespace path to join means that you cannot also
provide configuration for said namespace. Previously we would silently
ignore the configuration (and just join the provided path), but we
really should be returning an error (especially when you consider that
the configuration userns mappings are used quite a bit in runc with the
assumption that they are the correct mapping for the userns -- but in
this case they are not).
In the case of userns, the mappings are also required if you _do not_
specify a path, while in the case of the time namespace you can have a
container with a timens but no mappings specified.
It should be noted that the case checking that the user has not
specified a userns path and a userns mapping needs to be handled in
specconv (as opposed to the configuration validator) because with this
patchset we now cache the mappings of path-based userns configurations
and thus the validator can't be sure whether the mapping is a cached
mapping or a user-specified one. So we do the validation in specconv,
and thus the test for this needs to be an integration test.
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
(This is a cherry-pick of 1912d5988bbb379189ea9ceb2e03945738c513dc.)
Our handling for name space paths with user namespaces has been broken
for a long time. In particular, the need to parse /proc/self/*id_map in
quite a few places meant that we would treat userns configurations that
had a namespace path as if they were a userns configuration without
mappings, resulting in errors.
The primary issue was down to the id translation helper functions, which
could only handle configurations that had explicit mappings. Obviously,
when joining a user namespace we need to map the ids but figuring out
the correct mapping is non-trivial in comparison.
In order to get the mapping, you need to read /proc/<pid>/*id_map of a
process inside the userns -- while most userns paths will be of the form
/proc/<pid>/ns/user (and we have a fast-path for this case), this is not
guaranteed and thus it is necessary to spawn a process inside the
container and read its /proc/<pid>/*id_map files in the general case.
As Go does not allow us spawn a subprocess into a target userns,
we have to use CGo to fork a sub-process which does the setns(2). To be
honest, this is a little dodgy in regards to POSIX signal-safety(7) but
since we do no allocations and we are executing in the forked context
from a Go program (not a C program), it should be okay. The other
alternative would be to do an expensive re-exec (a-la nsexec which would
make several other bits of runc more complicated), or to use nsenter(1)
which might not exist on the system and is less than ideal.
Because we need to logically remap users quite a few times in runc
(including in "runc init", where joining the namespace is not feasable),
we cache the mapping inside the libcontainer config struct. A future
patch will make sure that we stop allow invalid user configurations
where a mapping is specified as well as a userns path to join.
Finally, add an integration test to make sure we don't regress this again.
Signed-off-by: Aleksa Sarai <cyphar@cyphar.com>
This field reports swap-only usage. For cgroupv1, `Usage` and `Failcnt`
are set by subtracting memory usage from memory+swap usage. For cgroupv2,
`Usage`, `Limit`, and `MaxUsage` are set. This commit also export `MaxUsage`
of memory under cgroupv2 mode, using `memory.peak` introduced in kernel 5.19.
Signed-off-by: Heran Yang <heran55@126.com>
(cherry picked from commit 104b8dc951)
Signed-off-by: Harshal Patil <harpatil@redhat.com>
This prevents potential exploit of using "../" in cgroups.OpenFile
(as well as other methods that use OpenFile) to read or write to
other cgroups.
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
(cherry picked from commit 2c9598c886)
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
This adds support for hugetlb.<pagesize>.rsvd limiting and accounting.
The previous non-rsvd max/limit_in_bytes does not account for reserved
huge page memory, making it possible for a processes to reserve all the
huge page memory, without being able to allocate it (due to cgroup
restrictions).
In practice this makes it possible to successfully mmap more huge page
memory than allowed via the cgroup settings, but when using the memory
the process will get a SIGBUS and crash. This is bad for applications
trying to mmap at startup (and it succeeds), but the program crashes
when starting to use the memory. eg. postgres is doing this by default.
This also keeps writing to the old max/limit_in_bytes, for backward
compatibility.
More info can be found here: https://lkml.org/lkml/2020/2/3/1153
(commit message mostly written by Odin Ugedal)
[1.1 backport: check for CGROUP_UNIFIED in integration test]
Co-authored-by: Odin Ugedal <odin@ugedal.com>
(cherry picked from commit 4a7d3ae5cd)
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
Separate it out of get_cgroup_value. Needed for the next commit.
This function was initially introduced in main branch commit d4582ae2f.
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
Since today, the URL from download.opensuse.org started returning a
HTTP 302 redirect, so -L option for curl is needed to follow it.
While at it, remove apt-key as per its man page recommendation:
> Note: Instead of using this command a keyring should be placed
> directly in the /etc/apt/trusted.gpg.d/ directory with a descriptive
> name and either "gpg" or "asc" as file extension.
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
(cherry picked from commit f944d7b653)
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
Bump fileutils to v0.5.1, which fixes permissions of newly created directories
to not depend on the value of umask.
Add a test case which fails like this before the fix:
mounts.bats
✗ runc run [tmpcopyup]
(in test file tests/integration/mounts.bats, line 28)
`[[ "${lines[0]}" == *'drwxrwxrwx'* ]]' failed
runc spec (status=0):
runc run test_busybox (status=0):
drwxr-xr-x 2 root root 40 Oct 4 22:35 /dir1/dir2
Fixes 3991.
(cherry picked from commit 730bc84418)
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
Unless the container's runtime config has intelRdt configuration set,
any checks for whether Intel RDT is supported or the resctrl filesystem
is mounted are a waste of time as, per the OCI Runtime Spec, "the
runtime MUST NOT manipulate any resctrl pseudo-filesystems." And in the
likely case where Intel RDT is supported by both the hardware and
kernel but the resctrl filesystem is not mounted, these checks can get
expensive as the intelrdt package needs to parse mountinfo to check
whether the filesystem has been mounted to a non-standard path.
Optimize for the common case of containers with no intelRdt
configuration by only performing the checks when the container has opted
in.
Signed-off-by: Cory Snider <csnider@mirantis.com>
(cherry picked from commit ea0bd78268)
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
The OCI runtime spec mandates "[i]f intelRdt is not set, the runtime
MUST NOT manipulate any resctrl pseudo-filesystems." Attempting to
delete files counts as manipulating, so stop doing that when the
container's RDT configuration is nil.
Signed-off-by: Cory Snider <csnider@mirantis.com>
(cherry picked from commit 56daf36be2)
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
The intelrdt package only needs to parse mountinfo to find the mount
point of the resctrl filesystem. Users are generally going to mount the
resctrl filesystem to the pre-created /sys/fs/resctrl directory, so
there is a common case where mountinfo parsing is not required. Optimize
for the common case with a fast path which checks both for the existence
of the /sys/fs/resctrl directory and whether the resctrl filesystem was
mounted to that path using a single statfs syscall.
Signed-off-by: Cory Snider <csnider@mirantis.com>
(cherry picked from commit c156bde7cc)
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
Reading /proc/cpuinfo is a surprisingly expensive operation. Since
kernel version 4.12 [1], opening /proc/cpuinfo on an x86 system can
block for around 20 milliseconds while the kernel samples the current
CPU frequency. There is a very recent patch [2] which gets rid of the
delay, but has yet to make it into the mainline kenel. Regardless,
kernels for which opening /proc/cpuinfo takes 20ms will continue to be
run in production for years to come. libcontainer only opens
/proc/cpuinfo to read the processor feature flags so all the delays to
get an accurate snapshot of the CPU frequency are just wasted time.
If we wanted to, we could interrogate the CPU features directly from
userspace using the `CPUID` instruction. However, Intel and AMD CPUs
have flags in different positions for their analogous sub-features and
there are CPU quirks [3] which would need to be accounted for. Some
Haswell server CPUs support RDT/CAT but are missing the `CPUID` flags
advertising their support; the kernel checks for support on that
processor family by probing the the hardware using privileged
RDMSR/WRMSR instructions [4]. This sort of probing could not be
implemented in userspace so it would not be possible to check for RDT
feature support in userspace without false negatives on some hardware
configurations.
It looks like libcontainer reads the CPU feature flags as a kind of
optimization so that it can skip checking whether the kernel supports an
RDT sub-feature if the hardware support is missing. As the kernel only
exposes subtrees in the `resctrl` filesystem for RDT sub-features with
hardware and kernel support, checking the CPU feature flags is redundant
from a correctness point of view. Remove the /proc/cpuinfo check as it
is an optimization which actually hurts performance.
[1]: https://unix.stackexchange.com/a/526679
[2]: https://lore.kernel.org/all/20220415161206.875029458@linutronix.de/
[3]: https://github.com/torvalds/linux/blob/7cf6a8a17f5b134b7e783c2d45c53298faef82a7/arch/x86/kernel/cpu/resctrl/core.c#L834-L851
[4]: https://github.com/torvalds/linux/blob/a6b450573b912316ad36262bfc70e7c3870c56d1/arch/x86/kernel/cpu/resctrl/core.c#L111-L153
Signed-off-by: Cory Snider <csnider@mirantis.com>
(cherry picked from commit 9f107489b0)
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
Aleksa Sarai