contextswitch

Contextswitch, commonly referred to as a context switch, is the process by which a central processing unit stops executing one task and begins executing another. In multitasking operating systems, the CPU divides its time among multiple processes or threads, giving the illusion of parallelism even on single-processor systems.

During a context switch, the operating system saves the state of the currently running task so it can resume later. This saved state typically includes the program counter, stack pointer, and CPU registers, as well as memory-management information and other per-task data. The saved information is stored in a data structure such as a process control block (PCB) or a thread control block (TCB). The scheduler then loads the state of the next task and transfers control to its resume point.

Context switches can be triggered by interrupts, system calls, I/O completions, or timer events, and can be either preemptive or cooperative. In preemptive systems, the kernel may forcibly switch tasks based on scheduling policies, while cooperative systems require tasks to yield control voluntarily.

Performance considerations are central to context switching. Each switch incurs overhead from saving and restoring state, potentially causing cache misses and TLB reloads. In multi-core systems, coordination between cores may be required, including actions like TLB shootdowns to invalidate stale translations. Operating system designers optimize the path of saving/restoring state and minimize switching frequency through scheduling strategies, hardware support, and techniques such as lazy state saving for certain components.