hard lockup
also: deadlock, kernel hang, watchdog timeout
A hard lockup occurs when the Linux kernel becomes completely unresponsive and stops servicing interrupts, typically requiring a hardware reset or power cycle to recover.
A hard lockup is a critical system failure where the kernel's main execution path (the timer interrupt handler or watchdog) hangs indefinitely, preventing any response to user input, network requests, or I/O operations. Unlike a soft lockup, a hard lockup cannot be detected or recovered by the kernel itself.
The kernel uses a watchdog timer to detect hard lockups: if the timer interrupt fails to execute within a specified window (usually seconds), the watchdog assumes a hard lockup has occurred. Common causes include CPU bugs, infinite loops in kernel code, deadlocks in critical sections, or faulty hardware.
When detected, the kernel may trigger a panic and reboot, or if NMI (non-maskable interrupt) watchdog is disabled, the system simply becomes frozen. Users will see a completely unresponsive machine where even SSH connections hang and the system doesn't respond to ICMP pings.