kernelpop is a framework for performing automated kernel vulnerability enumeration and exploitation on the following operating systems:
-
Linux
-
Mac
It is designed to be python
version-agnostic, meaning that it should work with both python2
and python3
- please let me know if you find that it doesn't
- run from project source
$ git clone https://github.com/spencerdodd/kernelpop
$ cd kernelpop
$ python kernelpop.py || python3 kernelpop.py
- build binary
$ git clone https://github.com/spencerdodd/kernelpop
$ cd kernelpop
$ ./create_executable.sh
$ ./kernelpop
- steps to build binary without script
- install
pyinstaller
(steps vary per build system) $ pyinstaller kernelpop.py --onefile
$ cp dist/kernelpop .
both of the binary build steps should create a binary kernelpop
in the project root.
user@debian:~/Desktop/kernelpop$ python3 kernelpop.py
##########################
# welcome to kernelpop #
# #
# let's pop some kernels #
##########################
[*] grabbing distro version and release from underlying OS (linuxdebian7)
[*] grabbing kernel version from 'uname -a'
[+] kernel (Linux debian 3.2.0-4-686-pae #1 SMP Debian 3.2.41-2 i686 GNU/Linux) identified as:
[base]
type: linux
distro: linuxdebian7
version: 3.2.0-4
architecture: i686
[specific]
type: linux
distro: linuxdebian7
version: 3.2.41-2
architecture: i686
[*] matching kernel to known exploits
[+] discovered 11 possible exploits !
[[ distro kernel matched exploit available ]]
CVE20165195_32 Dirty COW race condition root priv esc for 32 bit
[[ distro kernel version vulnerable ]]
CVE20144699 Exploitable race condition in linux before 3.15.4
CVE20143153 `futex_requeue` vulnerability before 3.14.6 allows for priv esc
CVE20162384 Double free vulnerability in the `snd_usbmidi_create` (requires physical proximity)
CVE20140196 `n_tty_write` vuln before 3.14.4 allows priv esc to root
CVE20132094_semtex perf_swevent_init Local root exploit (32 bit)
CVE20176074 `dccp_rcv_state_process` in net/dccp/input.c mishandles structs and can lead to local root
CVE20132094_32 perf_swevent_init Local root exploit (32 bit)
[[ base linux kernel vulnerable ]]
CVE20144014 `chmod` restriction bypass allows users to get root before 3.14.8
CVE20177308 `packet_set_ring` in net/packet/af_packet.c can gain privileges via crafted system calls.
CVE20171000112 ip_ufo_append_data() memory corruption flaw can be exploited to gain root privileges.
In the output, there are a few categories. This is what each means
[[ distro kernel matched exploit available ]]
- there is a distro version specific matched kernel exploit in the project that you can use to exploit the kernel. High likelihood of successful exploitation.
[[ distro kernel version vulnerable ]]
- the distro kernel version is vulnerable to the vulnerability listed, but was not explicitly stated as tested in the PoC exploit in the project. Exploitation without modification of the exploit may work, but has a lower likelihood of success.
[[ base linux kernel vulnerable ]]
- it is unknown if the distro kernel version is vulnerable to the vulnerability, but the base linux kernel is in the vulnerable range for the exploit. Exploitation without modification of the exploit is either unlikely or unknown, but may still work.
-
python2
orpython3
-
pyinstaller
(if you want to build a binary)
run modes:
(default) python3 kernelpop.py
(exploit-mode) python3 kernelpop.py -e {exploit name}
(dump-source) python3 kernelpop.py -e {exploit name} -d
(uname-mode) python3 kernelpop.py -u {uname -a output}
(interactive-mode) python3 kernelpop.py -i # LEGACY option (same as uname-mode)
other:
(playground path) -p {new path}
(json output file) --digest json
The default
mode processes information about the host kernel and compares it to the known kernel exploits available
to the program. It then outputs a list of potentially useful vulnerabilities and attached exploits.
-e {exploit name}
The exploit
mode dynamically compiles and runs the exploit source code with stdio interactions inside the program.
It can catch interrupts from short-stopped attempts as well. Use the name of the exploit that comes up in the
initial kernelpop
enumeration run in default
mode.
-
dump-source (option)
-d
- This is a modifier for
exploit
mode. It dumps the source-files for the given exploit to their respective files inPLAYGROUND_PATH
(default/tmp
). This is useful for modifying exploit source on a box or working with exploits that require manual interaction or hard-coding of values. It is especially useful when running from a binary, because you won't have access to the project source code (i.e. exploit source).
- This is a modifier for
-u {uname -a output}
This option allows you to pass the output of a uname -a
command to the program inline, which makes it use-able for
integration into automated scripts or for any other forseeable reason. Replacement for interactive mode
.
-i {uname -a output}
The interactive
mode allows you to perform enumeration with just the output of a uname -a
command,
which makes it useful as a host-side only enumeration tool. This run by starting kernelpop
and with the -i
flag
and then passing the output of uname -a
when requested. This is a legacy feature and replaced by uname mode
, but remains
supported as -u
does not support Mac enumeration.
-p {new PLAYGROUND_PATH dir}
The -p
option sets the value of the global variable PLAYGROUND_PATH
. This is the location where all exploit source
files are written to, and exploits compiled to. This is set, by default, to /tmp
. However, you can set it to any
directory with this modifier in case you do not have write access in /tmp
.
--digest json
This option allows you to dump the results of a kernelpop run to a digestible json file for later processing. So
far, I have just implemented the json
dump, but I will work on an XML version if it is requested.
-
add more exploits! (src/to_add if anyone wants to get cracking on a few of these, be my guest!)
-
include patch levels in vulnerable window comparisons
-
add way to override detected settings in case of incorrect parsing or adversarial settings
-
google: CVE-XXXX-XXXX "ubuntu"
-
click the canonical link (https://people.canonical.com/~ubuntu-security/cve/2016/CVE-XXXX-XXXX.html)
-
click all linked advisories at usn.ubuntu.com and parse info
-
grab patch versions for filling in vulnerability windows per version
-
-
-
google: CVE-XXXX-XXXX "debian"
-
security-tracker.debian.org link (https://security-tracker.debian.org/tracker/CVE-XXXX-XXXX)
- pull information from the bottom of the page that relates to the patched versions for the specific cve
-
-
google: CVE-XXXX-XXXX "mitre"
- links to other distros to pull info from
-
CVE-2017-1000379
-
CVE-2017-1000373
-
CVE-2017-1000372
-
CVE-2017-1000371
-
CVE-2017-1000370
-
CVE-2017-1000367
-
CVE-2017-1000112
-
CVE-2017-7308
-
CVE-2017-6074
-
CVE-2017-5123
-
CVE-2016-5195
-
CVE-2016-2384
-
CVE-2016-0728
-
CVE-2015-1328
-
CVE-2014-4699
-
CVE-2014-4014
-
CVE-2014-3153
-
CVE-2014-0196
-
CVE-2014-0038
-
CVE-2013-2094
-
CVE-2010-4347
-
CVE-2010-2959
-
CVE-2009-1185
https://github.com/SecWiki/linux-kernel-exploits
http://exploit-db.com/
https://github.com/lucyoa/kernel-exploits
https://github.com/SecWiki/windows-kernel-exploits
Debian