UNIX for Dummies Questions & Answers

Well, I don't know where exactly to ask this doubt so I'm asking in the newbie section. I was reading about traps and interrupts when I thought of traps as something that cease the control of the OS from the user and interrupts that cease the control yet provide support for multitasking. Am I right in thinking that? In that case, I think we can generate traps using our programs but why would we want to do it anyways? I mean when interrupts are proving better ways of handling the transfer of control, what use would we have for a trap?

The terminology changes between processors and operating systems, this is one interpretation:

1. an interrupt is a signal from a hardware device to a processor that some even has occured, this allows the processor to service the device, an example may be a serial port that generates an interrupt every time a character is received or it's transmit buffer is empty.

2. a trap is a mechanism to get from user space to system space, typically used in a protected operating system to transfer control from a user program to the kernel.

Well, when you were defining interrupts, I believe you defined only about one type right? I mean, I've read about software interrupts too besides hardware interrupts. And who generates a trap anyways? Is it the user or the system?

And in the case of a serial port, instead of an interrupt, can a trap be executed to service the device?

As I said different processors use different terminology, MC68000 terminology says traps, Intel 8086 says software interrupts, Intel 80286 also talks about gates and rings.

The point of a hardware interrupt is that the device is telling the CPU when some event has occurred rather than the CPU having to poll.

Look up "interrupt driven" compared to "polling".

There is then another mechanism, called DMA, which by passes the CPU entirely.

Also some devices have their own CPU, such as the standard PC keyboard controller or some SCSI cards.