Syscalls

Description

System calls provide a way to execute kernel routines from userland, so they can get ressources like memory, I/O accesses… Their implementation always consists in two parts:

• kernel-side syscall handling
• user-side syscall wrapping (syscall user API)

A unique syscall gate, 0x80, is used for all the syscalls. The dispatch is done in the kernel through a table depending on a identifier. The syscall identifier is passed as a parameter. The parameter passing method is Linux-like (registers).

Syscalls parameters are given in %ebx through %edx, %eax is the syscall number.

Recommended methodology

• Ensure that the event manager is working properly
• Write the syscall dispatch table management functions:
  • allocate/clean the dispatch table
  • add a handler to the table depending on the syscall identifier
• Implement the syscall wrapper as a classic interrupts wrapper which:
  • fetches syscall parameters before they are corrupted by the wrapper execution
  • executes the handler depending on the syscall identifier
• Write the syscall initialization procedure:
  • initialize the syscall dispatch table
  • register every syscall
  • add the syscall gate to the IDT
• Write a fake syscall for testing purposes:
  • print the syscall identifier when executing int 0x80
• Fill the syscall dispatch table with the developed syscalls.

Notes

Syscall ids are defined in include/k/kstd.h. You must follow those values!