operating system - From Kernel Space to User Space: Inner-workings of Interrupts -

i have been trying understand how h/w interrupts end in user space code, through kernel.

my research led me understand that:

1- external device needs attention cpu

2- signals cpu raising interrupt (h/w trance cpu or bus)

3- cpu asserts, saves current context, looks address of isr in interrupt descriptor table (vector)

4- cpu switches kernel (privileged) mode , executes isr.

question #1: how did kernel store isr address in interrupt vector table? might done sending cpu piece of assembly described in cpus user manual? more detail on subject better please.

---

in user space how can programmer write piece of code listens h/w device notifications?

this understand far.

5- kernel driver specific device has message device , executing isr.

question #3:if programmer in user space wanted poll device, assume done through system call (or @ least understood far). how done? how can driver tell kernel called upon specific systemcall can execute request user? , happens, how driver gives requested data user space?

i might off track here, guidance appreciated. not looking specific details answers, trying understand general picture.

question #1: how did kernel store isr address in interrupt vector table?

driver calls request_irq kernel function (defined in include/linux/interrupt.h , in kernel/irq/manage.c), , linux kernel register in right way according current cpu/arch rules.

it might done sending cpu piece of assembly described in cpus user manual?

in x86 linux kernel stores isr in interrupt descriptor table (idt), format described vendor (intel - volume 3) , in many resources http://en.wikipedia.org/wiki/interrupt_descriptor_table , http://wiki.osdev.org/idt , http://phrack.org/issues/59/4.html , http://en.wikibooks.org/wiki/x86_assembly/advanced_interrupts.

pointer idt table registered in special cpu register (idtr) special assembler commands: lidt , sidt.

if programmer in user space wanted poll device, assume done through system call (or @ least understood far). how done? how can driver tell kernel called upon specific systemcall can execute request user? , happens, how driver gives requested data user space?

driver registers device special file in /dev; pointers several driver functions registered file "file operations". user-space program opens file (syscall open), , kernels calls device's special code open; program calls poll or read syscall on fd, kernel call *poll or *read of driver's file operations (http://www.makelinux.net/ldd3/chp-3-sect-7.shtml). driver may put caller sleep (wait_event*) , irq handler wake (wake_up* - http://www.makelinux.net/ldd3/chp-6-sect-2 ).

you can read more linux driver creation in book linux device drivers (2005) jonathan corbet, alessandro rubini, , greg kroah-hartman: https://lwn.net/kernel/ldd3/

• chapter 3: char drivers https://lwn.net/images/pdf/ldd3/ch03.pdf
• chapter 10: interrupt handling https://lwn.net/images/pdf/ldd3/ch10.pdf