Linux Notification 机制的分析

1. 基本机制
1）数据结构
struct notifier_block
{
int (*notifier_call)(struct notifier_block *self, unsigned long, void *);
struct notifier_block *next;
int priority; /*用于对注册者进行优先级排队，高优先级的处理例程将被优先执行,由注册者自己指定 */
};
2）基本例程
extern int notifier_chain_register(struct notifier_block **list, struct notifier_block *n);
说明：注册到某个notifier_block链；这时的n可以只要初始化(*notifier_call)指针；
extern int notifier_chain_unregister(struct notifier_block **nl, struct notifier_block *n);
说明：从某个notifier_block链中移去n；
extern int notifier_call_chain(struct notifier_block **n, unsigned long val, void *v);
说明：轮循执行某个notifier_block链中的所有notifier_block，对其(*notifier_call)传入参数val和*v；
其中val应该是EVENT NUMBER，而*v是导致这个事件的数据结构，比如某个网络设备UP，则val=NETDEV_UP，v=dev；
3）返回值
#define NOTIFY_DONE 0x0000 /* Don't care */
#define NOTIFY_OK 0x0001 /* Suits me */
#define NOTIFY_STOP_MASK 0x8000 /* Don't call further */
#define NOTIFY_BAD (NOTIFY_STOP_MASK|0x0002) /* Bad/Veto action */

4）已定义事件
/*
* Declared notifiers so far. I can imagine quite a few more chains
* over time (eg laptop power reset chains, reboot chain (to clean
* device units up), device [un]mount chain, module load/unload chain,
* low memory chain, screenblank chain (for plug in modular screenblankers)
* VC switch chains (for loadable kernel svgalib VC switch helpers) etc...
*/

/* netdevice notifier chain */
#define NETDEV_UP 0x0001 /* For now you can't veto a device up/down */
#define NETDEV_DOWN 0x0002
#define NETDEV_REBOOT 0x0003 /* Tell a protocol stack a network interface
detected a hardware crash and restarted
- we can use this eg to kick tcp sessions
once done */
#define NETDEV_CHANGE 0x0004 /* Notify device state change */
#define NETDEV_REGISTER 0x0005
#define NETDEV_UNREGISTER 0x0006
#define NETDEV_CHANGEMTU 0x0007
#define NETDEV_CHANGEADDR 0x0008
#define NETDEV_GOING_DOWN 0x0009
#define NETDEV_CHANGENAME 0x000A

#define SYS_DOWN 0x0001 /* Notify of system down */
#define SYS_RESTART SYS_DOWN
#define SYS_HALT 0x0002 /* Notify of system halt */
#define SYS_POWER_OFF 0x0003 /* Notify of system power off */


2. 举例分析
以网络设备的通知信息块netdev_chain为例来说明如何使用notification机制。
在net/core/dev.c中定义了netdev_chain链：
static struct notifier_block *netdev_chain=NULL;
提供别的模块的接口，以便它们使用netdev_chain链：
/*
* Device change register/unregister. These are not inline or static
* as we export them to the world.
*/

/**
* register_netdevice_notifier - register a network notifier block
* @nb: notifier
*
* Register a notifier to be called when network device events occur.
* The notifier passed is linked into the kernel structures and must
* not be reused until it has been unregistered. A negative errno code
* is returned on a failure.
*/

int register_netdevice_notifier(struct notifier_block *nb)
{
return notifier_chain_register(&netdev_chain, nb);
}

/**
* unregister_netdevice_notifier - unregister a network notifier block
* @nb: notifier
*
* Unregister a notifier previously registered by
* register_netdevice_notifier(). The notifier is unlinked into the
* kernel structures and may then be reused. A negative errno code
* is returned on a failure.
*/

int unregister_netdevice_notifier(struct notifier_block *nb)
{
return notifier_chain_unregister(&netdev_chain,nb);
}

以X25为例来说明使用者。
在af_x25.c中，定义了：
struct notifier_block x25_dev_notifier = {
notifier_call: x25_device_event,
};
然后模块初始化时向netdev_chain注册：
static int __init x25_init(void)
{
... ...
register_netdevice_notifier(&x25_dev_notifier);
... ...
}
比如当NETDEV_UP事件发生时，调用到：
notifier_call_chain(&netdev_chain, NETDEV_UP, dev);
就会执行到x25_dev_notifier中注册的处理例程：x25_device_event，至于对相应的事件（event number）是不是感兴趣，
需要处理例程自己来判断。
static int x25_device_event(struct notifier_block *this, unsigned long event, void *ptr)
{
struct net_device *dev = (struct net_device *)ptr;
struct x25_neigh *neigh;

if (dev->type == ARPHRD_X25
#if defined(CONFIG_LLC) || defined(CONFIG_LLC_MODULE)
|| dev->type == ARPHRD_ETHER
#endif
) {
switch (event) {
case NETDEV_UP:
x25_link_device_up(dev);
break;
case NETDEV_GOING_DOWN:
if ((neigh = x25_get_neigh(dev)))
x25_terminate_link(neigh);
break;
case NETDEV_DOWN:
x25_kill_by_device(dev);
x25_route_device_down(dev);
x25_link_device_down(dev);
break;
}
}

return NOTIFY_DONE;
}

所有关于网络设备的事件全部在net/core/dev.c中发生，从而引发notifier_call_chain(&netdev_chain，val, dev)的调用：
Dev.c (linux\net\core): notifier_call_chain(&netdev_chain, NETDEV_CHANGE, dev);
Dev.c (linux\net\core): notifier_call_chain(&netdev_chain, NETDEV_UP, dev);
Dev.c (linux\net\core): notifier_call_chain(&netdev_chain, NETDEV_GOING_DOWN, dev);
Dev.c (linux\net\core): notifier_call_chain(&netdev_chain, NETDEV_DOWN, dev);
Dev.c (linux\net\core): notifier_call_chain(&netdev_chain, NETDEV_CHANGE, dev);
Dev.c (linux\net\core): notifier_call_chain(&netdev_chain, NETDEV_CHANGEMTU, dev);
Dev.c (linux\net\core): notifier_call_chain(&netdev_chain, NETDEV_CHANGEADDR, dev);
Dev.c (linux\net\core): notifier_call_chain(&netdev_chain, NETDEV_CHANGEADDR, dev);
Dev.c (linux\net\core): notifier_call_chain(&netdev_chain, NETDEV_CHANGENAME, dev);
Dev.c (linux\net\core): notifier_call_chain(&netdev_chain, NETDEV_REGISTER, dev);
Dev.c (linux\net\core): notifier_call_chain(&netdev_chain, NETDEV_UNREGISTER, dev);
Dev.c (linux\net\core): notifier_call_chain(&netdev_chain, NETDEV_UNREGISTER, dev);

3. 总结
从上面的分析可以看出，Linux下的Notification机制不是通过消息的方式实现的，而是一旦外部事件发生，所以对这个事件感兴趣的模块都会立即响应这个事件。但是，这个通知机制的效率不是很高，因为它的粒度不够细，比如A对E1、E2事件感兴趣，B对E2、E3感兴趣，但是E1～E3都是由N链来管理的，这样当发生E1事件时，A、B的处理例程都会被调用一次。如果能够区分对待不同模块感兴趣的事件集，然后只把事件发送到感兴趣的模块，效率会更高一些。另外，对优先级的处理是必须得，但是如何利用这个优先级似乎没有很好的说明和例证。