linux内核分析----初始化
作者:e4gle参考我提供的三个文件:bootsect.txt,head.txt,setup.txt
至于x86的引导无非如下步骤:
1,cpu初始化自身,在固定位置执行一条指令。
2,这条指令条转到bios中。
3,bios找到启动设备并获取mbr,该mbr指向我们的lilo
4,bios装载并把控制权交给lilo
5,压缩内核自解压,并把控制权转交给解压内核。
简单点讲,就是cpu成为内核引导程序的引导程序的引导程序的引导程序,西西。
这时内核将跳转到start_kernel是/init/main.c的重点函数,main.c函数很多定义都是为此函数服务的,这里我简要介绍一下这个函数的初始化流程。
初始化内核:
从start_kernel函数(/init/main.c)开始系统初始化工作,好,我们首先分析这个函数:
函数开始首先:
#ifdef __SMP__
static int boot_cpu = 1;
/* "current" has been set up, we need to load it now *//*定义双处理器用*/
if (!boot_cpu)
initialize_secondary();
boot_cpu = 0;
#endif
定义双处理器。
printk(linux_banner); /*打印linux banner*/
打印内核标题信息。
开始初始化自身的部分组件(包括内存,硬件终端,调度等),我来逐个分析其中的函数:
setup_arch(&command_line, &memory_start, &memory_end);/*初始化内存*/
返回内核参数和内核可用的物理地址范围
函数原型如下:
setup_arch(char **, unsigned long *, unsigned long *);
返回内存起始地址:
memory_start = paging_init(memory_start,memory_end);
看看paging_init的定义,是初始化请求页:
paging_init(unsigned long start_mem, unsigned long end_mem)(会在以后的内存管理子系统分析时详细介绍)
{
int i;
struct memclust_struct * cluster;
struct memdesc_struct * memdesc;
/* initialize mem_map[] */
start_mem = free_area_init(start_mem, end_mem);/*遍历查找内存的空闲页*/
/* find free clusters, update mem_map[] accordingly */
memdesc = (struct memdesc_struct *)
(hwrpb->mddt_offset + (unsigned long) hwrpb);
cluster = memdesc->cluster;
for (i = memdesc->numclusters ; i > 0; i--, cluster++) {
unsigned long pfn, nr;
/* Bit 0 is console/PALcode reserved. Bit 1 is
non-volatile memory -- we might want to mark
this for later */
if (cluster->usage & 3)
continue;
pfn = cluster->start_pfn;
if (pfn >= MAP_NR(end_mem)) /* if we overrode mem size */
continue;
nr = cluster->numpages;
if ((pfn + nr) > MAP_NR(end_mem)) /* if override in cluster */
nr = MAP_NR(end_mem) - pfn;
while (nr--)
clear_bit(PG_reserved, &mem_map[pfn++].flags);
}
memset((void *) ZERO_PAGE(0), 0, PAGE_SIZE);
return start_mem;
}
trap_init(); 初始化硬件中断
/arch/i386/kernel/traps.c文件里定义此函数
sched_init() 初始化调度
/kernel/sched.c文件里有详细的调度算法(这些会在以后进程管理和调度的结构分析中详细介绍)
parse_options(command_line) 分析传给内核的各种选项(随后再详细介绍)
memory_start = console_init(memory_start,memory_end) 初始化控制台
memory_start = kmem_cache_init(memory_start, memory_end) 初始化内核内存cache(同样,在以后的内存管理分析中介绍此函数)
sti();接受硬件中断
kernel_thread(init, NULL, CLONE_FS | CLONE_FILES | CLONE_SIGHAND);
current->need_resched = 1; need_resched标志增加,调用schedule(调度里面会详细说明)
cpu_idle(NULL) 进入idle循环以消耗空闲的cpu时间片
已经基本完成内核初始化工作,已经把需要完成的少量责任传递给了init,所身于地工作不过是进入idle循环以消耗空闲的cpu时间片。所以在这里调用了cpu_idle(NULL),它从不返回,所以当有实际工作好处理时,该函数就会被抢占。
parse_options函数:
static void __init parse_options(char *line)/*参数收集在一条长命令行中,内核被赋给指向该命令行头部的指针*/
{
char *next;
char *quote;
int args, envs;
if (!*line)
return;
args = 0;
envs = 1;/* TERM is set to 'linux' by default */
next = line;
while ((line = next) != NULL) {
quote = strchr(line,'"');
next = strchr(line, ' ');
while (next != NULL && quote != NULL && quote < next) {
next = strchr(quote+1, '"');
if (next != NULL) {
quote = strchr(next+1, '"');
next = strchr(next+1, ' ');
}
}
if (next != NULL)
*next++ = 0;
/*
* check for kernel options first..
*/
if (!strcmp(line,"ro")) {
root_mountflags |= MS_RDONLY;
continue;
}
if (!strcmp(line,"rw")) {
root_mountflags &= ~MS_RDONLY;
continue;
}
if (!strcmp(line,"debug")) {
console_loglevel = 10;
continue;
}
if (!strcmp(line,"quiet")) {
console_loglevel = 4;
continue;
}
if (!strncmp(line,"init=",5)) {
line += 5;
execute_command = line;
args = 0;
continue;
}
if (checksetup(line))
continue;
if (strchr(line,'=')) {
if (envs >= MAX_INIT_ENVS)
break;
envp_init[++envs] = line;
} else {
if (args >= MAX_INIT_ARGS)
break;
argv_init[++args] = line;
}
}
argv_init[args+1] = NULL;
envp_init[envs+1] = NULL;
}
////////////////////////////////////////////////////////////////////
// setup.txt
// Copyright(C) 2001, Feiyun Wang
////////////////////////////////////////////////////////////////////
// 易做图ysis on linux/arch/i386/boot/setup.S (for Linux 2.2.17)
////////////////////////////////////////////////////////////////////
////////////////////////////////////////////////////////////////////
// export the margin tags for .text, .data and .bss
{
.text
begtext:
.data
begdata:
.bss
begbss:
}
////////////////////////////////////////////////////////////////////
.text
start()
SYSSEG = 0x1000
SETUPSEG = 0x9020
modelist = end of .text:
{
// if loaded by bootsect.S,
// you can assume CS=SETUPSEG (=0x9020), otherwise...
goto start_of_setup();
/*
http://lxr.linux.no/source/Documentation/i386/boot.txt
Offset/Size Proto Name Meaning
01F1/1 ALL setup_sects The size of the setup in sectors
01F2/2 ALL root_flags If set, the root is mounted readonly
01F4/2 ALL syssize DO NOT USE - for bootsect.S use only
01F6/2 ALL swap_dev DO NOT USE - obsolete
01F8/2 ALL ram_size DO NOT USE - for bootsect.S use only