What is interrupt in assembly language?

An interrupt (for example, Intel-based opcode 0xCD) causes the CPU to pause its current execution, store the state of the registers on the stack, then process a defined subroutine. When this subroutine completes, the interrupt finishes, the registers are restored from the stack, and the previous execution state resumes. Interrupts come in two basic flavors: hardware and software.

A hardware interrupt occurs when a connected piece of hardware raises an IRQ signal. This mechanism tells the CPU that the hardware has information that needs to be processed. This is more efficient than polling for devices that have relatively few inputs or that have a latency that the CPU shouldn't wait for. Keyboards and mice are often interrupt-driven; the CPU only needs to know when something interesting has happened (a key was pressed or the mouse was moved, for example). Hard drives are also interrupt-driven; this allows the hard drive to locate the requested data and then return that data some CPU cycles later without the CPU having to request the status of the operation repeatedly.

In contrast, a software interrupt is usually triggered by software as a means of invoking operating system code in a reliable manner. For example, in MS-DOS, a developer would call INT 0x21 in order to invoke typical DOS commands, such as reading or writing a file, obtaining keyboard input, and other basic operations. Most older operating systems running on Intel-compatible processors use software interrupts to provide a way for the program to access system functions. Newer operating systems use new opcodes that circumvent the need for software interrupts, although most platforms still provide this mechanism for legacy software.