OSDI lab I

Objective

• Understand version control system and makefile project
• Learn how to use QEMU and GDB to debug Linux 0.11 kernel
• Learn how to commit to GIT system

---

Lab1.1 Linux 0.11 Development Environment Setup

Fix syntax error of Makefile

From Rule Syntax:

the recipe must start with a tab character

...

.c.s:
	@$(CC) $(CFLAGS) -S -o $*.s $<
.s.o:
	@$(AS)  -o $*.o $<
.c.o:
	@$(CC) $(CFLAGS) -c -o $*.o $<

...

Give execute permission to tools/build.sh

chmod +x tools/build.sh

Change GCC version to 4.8

# This file is the Makefile Header for every sub Makefile, which designed to
# simplfy the porting and maintaining operation
# author: falcon <wuzhangjin@gmail.com>
# update: 2008-10-29

AS	= as --32
LD	= ld
#LDFLAGS = -m elf_i386 -x 
LDFLAGS = -m elf_i386
CC	= gcc-4.8
CFLAGS  = -g -m32 -fno-builtin -fno-stack-protector -fomit-frame-pointer -fstrength-reduce #-Wall

CPP	= cpp-4.8 -nostdinc
AR	= ar

# we should use -fno-stack-protector with gcc 4.3
gcc_version=$(shell ls -l `which $(CC)` | tr '-' '\n' | tail -1)

Compile and install newest QEMU

On Ubuntu 18.04

Setup the dependencies

sudo apt-get install checkinstall
sudo apt-get install python
sudo apt-get install build-essential zlib1g-dev pkg-config libglib2.0-dev binutils-dev libboost-all-dev autoconf libtool libssl-dev libpixman-1-dev libpython-dev python-pip python-capstone virtualenv
sudo apt-get install flex bison g++-4.8

ref

Start to install qemu

git clone https://git.qemu.org/git/qemu.git
cd qemu
git submodule init
git submodule update --recursive
./configure
make
sudo checkinstall make install

ref

Validate the version must be larger than 2.5.x

Run the Linux 0.11

qemu-system-i386 -m 16M -boot a -fda Image -hda ../osdi.img

Lab 1-2 Debug kernel

Look the problem of panic

The kernel stuck on the panic("") at init/maic.c.

Look at kernel/panic.c.

/*
 *  linux/kernel/panic.c
 *
 *  (C) 1991  Linus Torvalds
 */

/*
 * This function is used through-out the kernel (includeinh mm and fs)
 * to indicate a major problem.
 */
#define PANIC

#include <linux/kernel.h>
#include <linux/sched.h>

void sys_sync(void);	/* it's really int */

void panic(const char * s)
{
	printk("Kernel panic: %s\n\r",s);
	if (current == task[0])
		printk("In swapper task - not syncing\n\r");
	else
		sys_sync();
	for(;;);
}

Due to the description of panic, it may not suitable to put panic at init/main.c.

void main(void){
...
	mem_init(main_memory_start,memory_end);
	trap_init();
	blk_dev_init();
	chr_dev_init();
	tty_init();
	time_init();
	sched_init();
	buffer_init(buffer_memory_end);
	hd_init();
	floppy_init();
	sti();
	// panic(""); 
	move_to_user_mode();
	if (!fork()) {		/* we count on this going ok */
		init();
	}
...
}

Uncomment it and try it.

Here it is, fix next bug.

Look the problem of out of memory

Actually, there are several warnings relating to task at sched.c.

struct task_struct * task[NR_TASKS] = {&(init_task.task), };

task has at leat one task_struct, checkout the NR_TASKS.

#define NR_TASKS 0

The value of NR_TASKS is zero, which is unreasonable.

Check out document at P994.

Therefore, change 0 to 64, and try it.

Fortunately, no error happened!

Work perfectly!

Print your student id

We need to print out the student ID before shell startup, so look at main.c.

...
	printf("%d buffers = %d bytes buffer space\n\r",NR_BUFFERS,
		NR_BUFFERS*BLOCK_SIZE);
	printf("Free mem: %d bytes\n\r",memory_end-main_memory_start);
	if (!(pid=fork())) {
		close(0);
		if (open("/etc/rc",O_RDONLY,0))
			_exit(1);
		execve("/bin/sh",argv_rc,envp_rc);
		_exit(2);
	}
	if (pid>0)
		while (pid != wait(&i))
			/* nothing */;
	while (1) {
		if ((pid=fork())<0) {
			printf("Fork failed in init\r\n");
			continue;
		}
		if (!pid) {
			close(0);close(1);close(2);
			setsid();
			(void) open("/dev/tty0",O_RDWR,0);
			(void) dup(0);
			(void) dup(0);
			_exit(execve("/bin/sh",argv,envp));
		}
        ...
...

There are two shell startup, if we print out the student ID before first shell.

The OK message will print before student ID.

So print student ID before second shell.

Work great! All done!!!

Questions

Q1: QEMU

qemu-system-i386 -m 16M -boot a -fda Image -hda ../osdi.img -s -S -serial stdio

According to the above command: Q1.1: What’s the difference between -S and -s in the command?

According to man page of qemu-system-i386: -s: Wait gdb connection to port 1234. -S: Do not start CPU at startup (you must type 'c' in the monitor).

Q1.2: What are -boot, -fda and -hda used for? If I want to boot with ../osdi.img(supposed it’s a bootable image) what should I do?

-boot a: Boot from floppy. -fda Image: Use Image as floppy. -hda ../osdi.img: Use ../osdi.img as hard disk. -boot c -hda ../osdi.img.

Q2: Git

Q2.1: Please explain all the flags and options used in below command:

git checkout -b lab1 origin/lab1

There is a remote repo called origin and has a branch named lab1, we want to create a local branch called lab1 attach on it and move the HEAD on it.

Q2.2 What are the differences among git add, git commit, and git push? What’s the timing you will use them?

There are three working directory, stagin area and repository. When you add something, you move these into from working direcotory to staging area, and when you commit it, you move to repository. At last, when you push it, you push your code from local repo to remote repo.

Q3: Makefile

Q3.1: What happened when you run the below command? Please explain it according to the Makefile.

make clean && make

According to Makefile, make clean will remove all object file, image etc. make will follow the instruction on Makefile and build it.

Q3.2: I did edit the include/linux/sched.h file and run make command successfully but the Image file remains the same. However, if I edit the init/main.c file and run make command. My Image will be recompile. What’s the difference between these two operations?

The Makefile didn't write carefully, you should execute make clean && make to rebuild.

Q4: After making, what does the kernel Image ‘Image’ look like?

According to build.sh, bootsect will put first and setup will be the second. Finally, the system will be at last.

# Write bootsect (512 bytes, one sector) to stdout
[ ! -f "$bootsect" ] && echo "there is no bootsect binary file there" && exit -1
dd if=$bootsect bs=512 count=1 of=$IMAGE 2>&1 >/dev/null

# Write setup(4 * 512bytes, four sectors) to stdout
[ ! -f "$setup" ] && echo "there is no setup binary file there" && exit -1
dd if=$setup seek=1 bs=512 count=4 of=$IMAGE 2>&1 >/dev/null

# Write system(< SYS_SIZE) to stdout
[ ! -f "$system" ] && echo "there is no system binary file there" && exit -1
system_size=`wc -c $system |cut -d" " -f1`
[ $system_size -gt $SYS_SIZE ] && echo "the system binary is too big" && exit -1
dd if=$system seek=5 bs=512 count=$((2888-1-4)) of=$IMAGE 2>&1 >/dev/null