okoge-kaz/xv6-riscv

Tokyo Institute of Technology xv6 for RISC-V

System Software, Fall 2022

Tokyo Institute of Technology

Lecture Site URL: https://titech-os.github.io/

Overview

Directory

xv6-riscv/
├── Dockerfile
├── LICENSE
├── Makefile
├── README.md
├── docker-compose.yaml
├── fs.img
├── kernel
├── mkfs
└── user

• user/

  User programs, such as cat, echo, grep, init, kill, ln, ls, mkdir, rm, sh, stressfs, usertests, wc, zombie.

• kernel/

xv6 cPU mode transition

• M-mode(= Machine mode)
• S-mode(= Supervisor mode)
• U-mode(= User mode)

transition diagram

User process is running in U-mode, and kernel is running in S-mode.

Privileged instructions are not allowed in U-mode. (特権命令はユーザーモードでは許可されていない)

If you turn on the device, the CPU is in M-mode.(reset vector)

(some devices only have M-mode, such as built-in program) (組み込みプログラムなどは、M-mode しかないものもある)

Most interruptions are handled in S-mode. But some interruptions, such as timer interrupts, are handled in M-mode.

I/O interrupts occur U-mode to S-mode interrupts transitions.

xv6 trap

• initialization

  main 関数(os の main 関数)が実行される前に、行う初期化処理について以下では解説する。

  kernel/start.c

  ...
  // entry.S jumps here in machine mode on stack0.
  void
  start()                                                                 <-- 起動時は M-mode
  {
  // set M Previous Privilege mode to Supervisor, for mret.
  unsigned long x = r_mstatus();
  x &= ~MSTATUS_MPP_MASK;                                                 <-- mret 実行時に S-mode に遷移するように設定(上図参照)
  x |= MSTATUS_MPP_S;
  w_mstatus(x);
  
  // set M Exception Program Counter to main, for mret.
  // requires gcc -mcmodel=medany
  w_mepc((uint64)main);                                                   <-- mret 実行時に main 関数に遷移するように設定
  
  // disable paging for now.
  w_satp(0);
  
  // delegate all interrupts and exceptions to supervisor mode.
  w_medeleg(0xffff);                                                      <-- 割り込みと例外を S-mode で処理するように設定(mode遷移を行えるようにするための設定)
  w_mideleg(0xffff);
  w_sie(r_sie() | SIE_SEIE | SIE_STIE | SIE_SSIE);
  ...

• CSR (Control/ Status Register)

CPU の動作状態を参照、設定する特殊レジスタ

- M-mode CSR
    - mstatus: M-mode machine status
    - mepc: M-mode exception program counter(M-modeへの遷移が生じた際のProgram Counter)
    - medeleg/ mideleg: exceptions, interrupts delegation (例外、割り込みの移譲を設定)
    ...

• trap vector

トラップ発生時の制御の移動先(アドレス)

- mtvec: M-modeに遷移するトラップ用
- stvec: S-modeに遷移するトラップ用(U -> S, S -> S)

Environment

enter to xv6 environment

1. docker compose up -d
2. docker container exec -it xv6-riscv /bin/bsh
3. make
4. make qemu

quit from xv6 environment

1. type Ctrl + A and X

[Note] xv6 doesn't support Ctrl + C to quit.

run user tests

1. usertests (after make qemu)

How to add new command

user command

system call

• kalloc(void)

  Line: 51

  if(((uint64)pa % PGSIZE) != 0 || (char*)pa < end || (uint64)pa >= PHYSTOP)
  panic("kfree");

  (uint64)pa % PGSIZE) != 0 means pa(= page address) is not aligned to PGSIZE(= page size).

  (char*)pa < end means pa is not in the kernel memory.(kernel memory is from end to PHYSTOP)

  (uint64)pa >= PHYSTOP means pa is not in the kernel memory.(kernel memory is from end to PHYSTOP)

  

  [Note] end is defined in kernel/entry.S

  Line 55

  memset((char*)pa, 1, n);

  memset is defined in kernel/string.c

  argument 1: void* is the address of the memory to be set. argument 2: int is the value to be set. argument 3: uint is the number of bytes to be set.

  Line 59 - 62

  acquire(&kmem.lock);
  r->next = kmem.freelist;
  kmem.freelist = r;
  release(&kmem.lock);

  acquire is defined in kernel/spinlock.c r->next = kmem.freelist; means r->next is the next free page.

  kmem.freelist is the head of the free page list. address which pa points is now free. So, we add it to the free page list.

  kmem.freelist = r; means the above explanation.