linux 2.6.28 kernel之__initcal机制和module_init详解

include/linux/init.h

/* initcalls are now grouped by functionality into separate
 * subsections. Ordering inside the subsections is determined
 * by link order.
 * For backwards compatibility, initcall() puts the call in
 * the device init subsection.
 *
 * The `id' arg to __define_initcall() is needed so that multiple initcalls
 * can point at the same handler without causing duplicate-symbol build errors.
 */

#define __define_initcall(level,fn,id) \
    static initcall_t __initcall_##fn##id __used \
    __attribute__((__section__(".initcall" level ".init"))) = fn

 /*
 * A "pure" initcall has no dependencies on anything else, and purely
 * initializes variables that couldn't be statically initialized.
 *
 * This only exists for built-in code, not for modules.
 */
#define pure_initcall(fn)        __define_initcall("0",fn,0)

#define core_initcall(fn)        __define_initcall("1",fn,1)
#define core_initcall_sync(fn)        __define_initcall("1s",fn,1s)
#define postcore_initcall(fn)        __define_initcall("2",fn,2)
#define postcore_initcall_sync(fn)    __define_initcall("2s",fn,2s)
#define arch_initcall(fn)        __define_initcall("3",fn,3)
#define arch_initcall_sync(fn)        __define_initcall("3s",fn,3s)
#define subsys_initcall(fn)        __define_initcall("4",fn,4)
#define subsys_initcall_sync(fn)    __define_initcall("4s",fn,4s)
#define fs_initcall(fn)            __define_initcall("5",fn,5)
#define fs_initcall_sync(fn)        __define_initcall("5s",fn,5s)
#define rootfs_initcall(fn)        __define_initcall("rootfs",fn,rootfs)
#define device_initcall(fn)        __define_initcall("6",fn,6)
#define device_initcall_sync(fn)    __define_initcall("6s",fn,6s)
#define late_initcall(fn)        __define_initcall("7",fn,7)
#define late_initcall_sync(fn)        __define_initcall("7s",fn,7s)

#define __initcall(fn) device_initcall(fn)
/**
 * module_init() - driver initialization entry point
 * @x: function to be run at kernel boot time or module insertion
 *
 * module_init() will either be called during do_initcalls() (if
 * builtin) or at module insertion time (if a module).  There can only
 * be one per module.
 */
/include/linux/init.h
#define module_init(x)    __initcall(x);

可以发现这些*_initcall(fn)最终都是通过__define_initcall(level,fn)宏定义生成的。//这个版本少了id项
__define_initcall宏定义如下：
#define __define_initcall(level,fn,id) \
    static initcall_t __initcall_##fn##id __used \
    __attribute__((__section__(".initcall" level ".init"))) = fn

/include/linux/init.h
 typedef int (*initcall_t)(void);

initcall_t是个函数指针

__initcall_##fn##id = fn;
__initcall_##fn##id是放在__section__(".initcall" level ".init"))段里的。
   这句话的意思为定义一个 int (*initcall_t)(void)型的函数指针，函数存放在.initcall”level”.init section内。.initcall”level”.init section定义在vmlinux.lds内。
/* arch/arm/kernel/vmlinux.lds */


  __initcall_start = .;
   *(.initcallearly.init) __early_initcall_end = .; *(.initcall0.init) *(.initcall0s.init) *(.initcall1.init) *(.initcall1s.init) *(.initcall2.init) *(.initcall2s.init) *(.initcall3.init) *(.initcall3s.init) *(.initcall4.init) *(.initcall4s.init) *(.initcall5.init) *(.initcall5s.init) *(.initcallrootfs.init) *(.initcall6.init) *(.initcall6s.init) *(.initcall7.init) *(.initcall7s.init)
  __initcall_end = .;

正好包括了上面init.h里定义的从 core_initcall到late_initcall等7个level等级的.initcall”level”.init section. 因此通过不同的*_initcall声明的函数指针最终都会存放不同level等级的.initcall”level”.init section内。这些不同level的section按level等级高低依次存放。

下面我们再来看看，内核是什么时候调用存储在.initcall”level”.init section内的函数的。

内核是通过do_initcalls函数循环调用执行initcall.init section内的函数的，流程如下：
main.c
start_kernel -> rest_init -> kernel_thread -> kernel_init -> do_basic_setup -> do_initcalls


/*
 * We need to finalize in a non-__init function or else race conditions
 * between the root thread and the init thread may cause start_kernel to
 * be reaped by free_initmem before the root thread has proceeded to
 * cpu_idle.
 *
 * gcc-3.4 accidentally inlines this function, so use noinline.
 */
init/main.c
static void noinline __init_refok rest_init(void)
    __releases(kernel_lock)
{
    int pid;

    kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND);
    numa_default_policy();
    pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
    kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns);
    unlock_kernel();

    /*
     * The boot idle thread must execute schedule()
     * at least once to get things moving:
     */
    init_idle_bootup_task(current);
    preempt_enable_no_resched();
    schedule();
    preempt_disable();

    /* Call into cpu_idle with preempt disabled */
    cpu_idle();
}


/*
 * Ok, the machine is now initialized. None of the devices
 * have been touched yet, but the CPU subsystem is up and
 * running, and memory and process management works.
 *
 * Now we can finally start doing some real work..
 */
init/main.c
static void __init do_basic_setup(void)
{
    rcu_init_sched(); /* needed by module_init stage. */
    init_workqueues();
    usermodehelper_init();
    driver_init();
    init_irq_proc();
    do_initcalls();
}
////////////////////////////
init/main.c
extern initcall_t __initcall_start[], __initcall_end[], __early_initcall_end[];

static void __init do_initcalls(void)
{
    initcall_t *call;

//批量执行initcallx.init段中的函数指针 (arch/arm/kernel/vmlinux.lds链接脚本)
//执行了所有驱动的入口函数
    for (call = __early_initcall_end; call < __initcall_end; call++)
        do_one_initcall(*call);  //回调函数

    /* Make sure there is no pending stuff from the initcall sequence */
    flush_scheduled_work();
}