How does Linux remember its Kernel Stack Pointer?

Question

I know that there are two types of stack in Linux : user stack for each user threads and Kernel Stack for kernel threads (but 1 process). The interruptions, more precisely, the interruption procedures, are the bridges between this two modes, kernel (0) and user (3). The interrupt vector table let the processor loads the right instruction address in the PC register, but how is the stack pointer register changed when it switches in kernel mode? Does the subroutines indicate where's the kernel stack just before its first instruction? Or does the processor uses two stack pointer registers (I really doubt it)?

How does the "return from interrupt" knows where to return? Is the PCB saved in kernel stack or elsewhere?

Please don't hesitate to rectify it anything I've said to be true is not.

Thanks a lot for your help.

Answer 1

Kernel mode stack in Linux kernel is stored in task_struct->stack. Where and how it comes from is totally up to the platform. Some platforms might not be saving it like above. But then you can use task_stack_page() to find the stack.

And

While entering the interrupt handler, PC is stored on kernel stack. While returning from Interrupt this PC is loaded back from the kernel stack.