Magic SysRq Keys for Frozen Linux Systems

When a Linux operating system completely freezes and stops responding to mouse clicks, keyboard shortcuts, or terminal commands, the Magic SysRq key combination provides a safe, low-level mechanism to regain control or reboot without corrupting data. Because the key combination communicates directly with the Linux kernel rather than relying on user-space applications or the display server, it remains functional even during severe system lockups. Understanding how to use this feature allows users and administrators to cleanly flush cached data, unmount drives, and reboot safely instead of resorting to an abrupt, damaging hard power cutoff.

Why Magic SysRq Matters During a Freeze

During an operating system lockup, standard interfaces like the desktop environment or virtual consoles (TTY) often fail because the user-space processes governing them have crashed or starved for CPU cycles. Pressing the physical power button or cutting power to the machine abruptly terminates input/output operations, which can lead to filesystem corruption, damaged database files, and permanent data loss.

The Magic System Request (SysRq) mechanism bypasses the user space entirely. Implemented directly within the Linux kernel's input handling layer, it triggers predetermined kernel-level debugging and recovery routines. As long as the kernel itself has not suffered a total panic or hardware-level hang, it will immediately process these commands, ensuring critical data integrity procedures are performed prior to a reset.

Prerequisites and Enabling SysRq

On modern Linux distributions, full Magic SysRq access is often restricted by default for security purposes. The feature is controlled via the /proc/sys/kernel/sysrq virtual file.

To check the current status, run:

cat /proc/sys/kernel/sysrq

A value of 1 enables all functions, while a value of 0 disables them. Other numerical values represent a bitmask permitting only specific subsets of commands. To enable all commands immediately, use:

echo "1" | sudo tee /proc/sys/kernel/sysrq

To persist this setting across reboots, add kernel.sysrq = 1 to /etc/sysctl.conf or an appropriate file in /etc/sysctl.d/.

The Safe Recovery Sequence: REISUB

To invoke a Magic SysRq command on standard desktop keyboards, hold down Alt and Print Screen (SysRq), then sequentially press the desired command key. On laptops, holding the Fn key may also be required.

The standard diagnostic sequence for safely recovering or rebooting a frozen system is commonly remembered by the mnemonic "Raising Elephants Is So Utterly Boring":

  1. R (Raw): Switches keyboard control from the display server (X11 or Wayland) back to raw mode (K_RAW). This allows keyboard input to bypass frozen graphical environments.
  2. E (Terminate): Sends the SIGTERM signal to all running processes except init (PID 1), prompting them to save data and shut down cleanly.
  3. I (Kill): Sends the SIGKILL signal to any processes that failed to terminate after the initial signal, clearing locked resources.
  4. S (Sync): Writes all dirty data currently cached in RAM back to the physical disks. This is the most crucial step for preventing data loss and file corruption.
  5. U (Unmount): Remounts all mounted filesystems into a read-only state. This guarantees that no further write operations can occur while the drive is active.
  6. B (Reboot): Immediately reboots the system safely without executing user-space shutdown scripts.

Allow a brief pause of a few seconds between each keypress—particularly after E, S, and U—to give the kernel sufficient time to execute the tasks, write out data, and remount storage media properly before the final reboot command is issued.