跳转至

Lab2

lab 2 kernal development

笔记

kernal / user mode

How would user actually override the bits that identify context if the instruction to change that bit is itself a privilege instruction? It seems user can never change to kernel mode.

  • use ecall to invoke kernel mode

The loader loads the XV_6 kernel into memory at physical address 0x80000000. The reason it places the kernel at 0x80000000 rather than 0x0 is because the address range 0x0:0x80000000 contains I/O devices

xv_6

The startup process for xv_6 system

为了实现系统调用，我们首先需要获取用户传递的参数。由于最初是在用户模式下调用的调用的系统调用函数 ，所以参数是储存在通用寄存器中。kernal trap code 会保存通用寄存器的值，kernal code 就可以利用 trapframe 来获取参数。xv6 为此封装了argint, argaddr, argfd三个函数。

  • 使用它们可以分别以 整形，指针，文件描述符来获取用户传递参数
  • 它们都借助 argraw 来从正确的寄存器获取参数

实验

trace system call

系统调用的执行过程

  1. user.h add function stub to pass compiling
  2. usys.pl generate corresponding assembly code
  3. syscall.c call corresponding function in sysproc.c
  4. add code in sycall.h 1 << xxx
  5. add function sys_trace in sysproc.c
  6. add extern uint64 sys_trace(void) function signature
  7. modify array syscalls

sysinfo system call

  • need record
  • freemem free memory, kalloc.c

  • nproc number of process, proc.c

  • copy struct info back to user space

  • sys_fstate in sysfile.c

  • filestat in file.c

  • quesion

  • why we need copy struct info to user space

  • because after the system call, we need read struct info in user space

  • why testmem test failed?

  • seems like the freemem is always 0

    p = p->next instead of p++

颜色主题调整

评论区~