Linux uevent分析、用户接收uevent以及mdev分析【转】
转自：https://www.cnblogs.com/arnoldlu/p/11246204.html
关键词：uevent、netlink、ADD/REMOVE/CHANGE、uevent_helper、hotplug、usermode helper、mdev、mdev.conf等等。
本文从三方面了解uevent相关内容：内核中uevent如何传送、用户空间如何处理uevent、如何通过mdev实现热插拔功能。
1. Linux uevent分析
kobject_action定义了 Linux下的uevent类型；struct kerenl_uevent_env表示一个待发送的uevent。
uevent_net_init()创建发送uevent所需要的socket等信息。
内核驱动通过kobject_uevent()/kobject_uevent_env()发送uevent到用户空间，主要包括两部分工作：一是通过netlink_broadcast_filtered()发送netlink消息；另一是通过call_usermodehelper_setup()/call_usermodehelper_exec()调用用户空间程序处理uevent消息。
1.1 uevent数据结构
kobject_action定义了kobject的动作，包括ADD、REMOVE、CHANGE等等。用户空间根据ADD或者REMOVE处理热插拔时间，电池模块根据CHANGE处理电量更新。
kobj_uevent_env用于表示一个kobject事件，argv是用户空间执行的helper参数；envp和buf组成发送uevent字符串信息。
enum kobject_action { KOBJ_ADD,------------------------ADD/REMOVE添加/移除事件。 KOBJ_REMOVE, KOBJ_CHANGE,---------------------设备状态或者内容发生改变。 KOBJ_MOVE,-----------------------更改名称或者更改parent，即更改了目录结构。 KOBJ_ONLINE,---------------------设备上线/下线事件，常表示使能或者去使能。 KOBJ_OFFLINE, KOBJ_MAX }; static const char *kobject_actions[] = { [KOBJ_ADD] = "add", [KOBJ_REMOVE] = "remove", [KOBJ_CHANGE] = "change", [KOBJ_MOVE] = "move", [KOBJ_ONLINE] = "online", [KOBJ_OFFLINE] = "offline", }; struct kobj_uevent_env { char *argv[3];------------------------------用户空间可执行文件路径，以及参数等。 char *envp[UEVENT_NUM_ENVP];----------------指针数组，保存每个环境变量的地址。 int envp_idx; char buf[UEVENT_BUFFER_SIZE];---------------环境变量内容。 int buflen; };
1.2 uevent初始化
uevent_net_init()创建类型为NETLINK_KOBJECT_UEVENT的socket，并将其放入uevent_sock_list链表上。uevent_net_exit()则将其从uevent_socket_list中摘除，并且释放socket相关资源。
static int uevent_net_init(struct net *net) { struct uevent_sock *ue_sk; struct netlink_kernel_cfg cfg = { .groups = 1, .flags = NL_CFG_F_NONROOT_RECV, }; ue_sk = kzalloc(sizeof(*ue_sk), GFP_KERNEL); if (!ue_sk) return -ENOMEM; ue_sk->sk = netlink_kernel_create(net, NETLINK_KOBJECT_UEVENT, &cfg);------------创建NETLINK_KOBJECT_UEVENT类型的socket。 if (!ue_sk->sk) { printk(KERN_ERR "kobject_uevent: unable to create netlink socket!\n"); kfree(ue_sk); return -ENODEV; } mutex_lock(&uevent_sock_mutex); list_add_tail(&ue_sk->list, &uevent_sock_list);-----------------------------------将创建的uevent_sock加入到uevent_sock_list中。 mutex_unlock(&uevent_sock_mutex); return 0; } static void uevent_net_exit(struct net *net) { struct uevent_sock *ue_sk; mutex_lock(&uevent_sock_mutex); list_for_each_entry(ue_sk, &uevent_sock_list, list) { if (sock_net(ue_sk->sk) == net) goto found; } mutex_unlock(&uevent_sock_mutex); return; found: list_del(&ue_sk->list); mutex_unlock(&uevent_sock_mutex); netlink_kernel_release(ue_sk->sk); kfree(ue_sk); } static struct pernet_operations uevent_net_ops = { .init = uevent_net_init, .exit = uevent_net_exit, }; static int __init kobject_uevent_init(void) { return register_pernet_subsys(&uevent_net_ops);-----------将uevent网络协议模块添加到新的命名空间子系统中，并且调用init初始化函数。 } postcore_initcall(kobject_uevent_init);
1.3 对uevent_helper设置
对uevent_helper设置，可以对/proc/sys/kernel/hotplug写可执行文件路径即可。
然后在内核触发uevent事件的之后调用相关可执行文件进行处理。
static struct ctl_table kern_table[] = {
...
#ifdef CONFIG_UEVENT_HELPER
{
.procname = "hotplug",
.data = &uevent_helper,
.maxlen = UEVENT_HELPER_PATH_LEN,
.mode = 0644,
.proc_handler = proc_dostring,
},
#endif...
{ }
};
或者还可以对/proc/kernel/uevent_helper写入可执行文件路径。
static ssize_t uevent_helper_show(struct kobject *kobj,
struct kobj_attribute *attr, char *buf)
{
return sprintf(buf, "%s\n", uevent_helper);
}
static ssize_t uevent_helper_store(struct kobject *kobj,
struct kobj_attribute *attr,
const char *buf, size_t count)
{
if (count+1 > UEVENT_HELPER_PATH_LEN)
return -ENOENT;
memcpy(uevent_helper, buf, count);
uevent_helper[count] = '\0';
if (count && uevent_helper[count-1] == '\n')
uevent_helper[count-1] = '\0';
return count;
}
KERNEL_ATTR_RW(uevent_helper);
1.4 usermode helper
usermode helper用于帮助在内核空间启动一个用户空间程序。首先通过call_usermodehelper_setup()初始化一个struct subprocess_info实例；然后调用call_usermodehelper_exec()执行，通过kernel_thread()创建线程，入口函数call_usermodehelper_exec_async()调用do_execve()加载用户空间程序。
这里不同等待程序运行结束的方式，UMH_NO_WAIT在将work放入system_unbound_wq之后，不等待直接退出；UMH_KILLABLE则会等待进程变为TASK_KILLABLE。UMH_WAIT_PROC等待进程执行完毕，UMH_WAIT_EXEC只是等待do_exec()执行完毕，而不是进程结束。
struct subprocess_info表示一个usermode helper执行的实例。
#define UMH_NO_WAIT 0 /* don't wait at all */ #define UMH_WAIT_EXEC 1 /* wait for the exec, but not the process */ #define UMH_WAIT_PROC 2 /* wait for the process to complete */ #define UMH_KILLABLE 4 /* wait for EXEC/PROC killable */ struct subprocess_info { struct work_struct work;---------------将usermode helper作为一个work放入system_unbound_wq中。 struct completion *complete; char *path;----------------------------用户空间可执行文件路径。 char **argv;---------------------------可执行文件所需参数。 char **envp;---------------------------可执行文件所需环境变量。 int wait;------------------------------等待标志。 int retval; int (*init)(struct subprocess_info *info, struct cred *new);---执行产需之前的初始化函数。 void (*cleanup)(struct subprocess_info *info);-----------------释放struct subprocess_info是的清理程序。 void *data; };
call_usermodehelper()首先创建struct subprocess_info，然后执行用户空间程序。
int call_usermodehelper(char *path, char **argv, char **envp, int wait) { struct subprocess_info *info; gfp_t gfp_mask = (wait == UMH_NO_WAIT) ? GFP_ATOMIC : GFP_KERNEL; info = call_usermodehelper_setup(path, argv, envp, gfp_mask, NULL, NULL, NULL);-------------------------------需要用户空间执行的程序路径以及参数，内存分配gfp_mask等等，填充倒struc subprocess_info中。 if (info == NULL) return -ENOMEM; return call_usermodehelper_exec(info, wait);----------------------将subprocess_info->work放入system_unbound_eq执行。 }
call_usermodehelper_setup()初始化struct subprocess_info实例，包括程序路径、参数等等，还有初始化一个work，对应的执行函数式call_usermodehelper_exec_work()。
struct subprocess_info *call_usermodehelper_setup(char *path, char **argv, char **envp, gfp_t gfp_mask, int (*init)(struct subprocess_info *info, struct cred *new), void (*cleanup)(struct subprocess_info *info), void *data) { struct subprocess_info *sub_info; sub_info = kzalloc(sizeof(struct subprocess_info), gfp_mask); if (!sub_info) goto out; INIT_WORK(&sub_info->work, call_usermodehelper_exec_work); sub_info->path = path; sub_info->argv = argv; sub_info->envp = envp; sub_info->cleanup = cleanup; sub_info->init = init; sub_info->data = data; out: return sub_info; } static void call_usermodehelper_exec_work(struct work_struct *work) { struct subprocess_info *sub_info = container_of(work, struct subprocess_info, work); if (sub_info->wait & UMH_WAIT_PROC) { call_usermodehelper_exec_sync(sub_info); } else { pid_t pid; /* * Use CLONE_PARENT to reparent it to kthreadd; we do not * want to pollute current->children, and we need a parent * that always ignores SIGCHLD to ensure auto-reaping. */ pid = kernel_thread(call_usermodehelper_exec_async, sub_info, CLONE_PARENT | SIGCHLD);--------------------------CLONE_PARENT让新创建的进程与创建它的进程成了‘兄弟’而不是‘父子’。 if (pid < 0) { sub_info->retval = pid; umh_complete(sub_info); } } }
call_usermode_herlper_exec_async()和call_usermodehelper_exec_sync()最大的区别是 创建进程的flags，前者CLONE_PARENT导致新创建的进程和创建它的进程编程兄弟关系，而后者还保持父子关系。
static void call_usermodehelper_exec_sync(struct subprocess_info *sub_info) { pid_t pid; /* If SIGCLD is ignored sys_wait4 won't populate the status. */ kernel_sigaction(SIGCHLD, SIG_DFL); pid = kernel_thread(call_usermodehelper_exec_async, sub_info, SIGCHLD); if (pid < 0) { sub_info->retval = pid; } else { int ret = -ECHILD; sys_wait4(pid, (int __user *)&ret, 0, NULL);-------------------------等待子进程退出，这也是async和sync最大的区别所在。 if (ret) sub_info->retval = ret; } kernel_sigaction(SIGCHLD, SIG_IGN); umh_complete(sub_info); } static int call_usermodehelper_exec_async(void *data) { struct subprocess_info *sub_info = data; struct cred *new; int retval; spin_lock_irq(¤t->sighand->siglock); flush_signal_handlers(current, 1);------------------------------------------进行signal、nice、credential准备工作。 spin_unlock_irq(¤t->sighand->siglock); set_user_nice(current, 0); retval = -ENOMEM; new = prepare_kernel_cred(current); if (!new) goto out; spin_lock(&umh_sysctl_lock); new->cap_bset = cap_intersect(usermodehelper_bset, new->cap_bset); new->cap_inheritable = cap_intersect(usermodehelper_inheritable, new->cap_inheritable); spin_unlock(&umh_sysctl_lock); if (sub_info->init) { retval = sub_info->init(sub_info, new);--------------------------------为进程创建进行初始化工作。 if (retval) { abort_creds(new); goto out; } } commit_creds(new); retval = do_execve(getname_kernel(sub_info->path), (const char __user *const __user *)sub_info->argv, (const char __user *const __user *)sub_info->envp);------------调用usermode程序替代当前进程。 ... }
call_usermodehelper_exec()最主要的工作就是将一个usermode helper命令放入system_unbound_wq执行，然后根据wait类型进行不同条件的等待。
int call_usermodehelper_exec(struct subprocess_info *sub_info, int wait) { DECLARE_COMPLETION_ONSTACK(done); int retval = 0; if (!sub_info->path) { call_usermodehelper_freeinfo(sub_info); return -EINVAL; } helper_lock(); if (usermodehelper_disabled) { retval = -EBUSY; goto out; } sub_info->complete = (wait == UMH_NO_WAIT) ? NULL : &done; sub_info->wait = wait; queue_work(system_unbound_wq, &sub_info->work);---------------将usermode helper进程放入system_unbound_wq上调度，即不绑定到任何CPU上，尽快得到执行。 if (wait == UMH_NO_WAIT) /* task has freed sub_info */-----对于UMH_NO_WAIT类型，跳过下面的completion同步等待步骤。 goto unlock; if (wait & UMH_KILLABLE) { retval = wait_for_completion_killable(&done);-------------等待进程属性变为TASK_KILLABLE。 if (!retval) goto wait_done; /* umh_complete() will see NULL and free sub_info */ if (xchg(&sub_info->complete, NULL)) goto unlock; /* fallthrough, umh_complete() was already called */ } wait_for_completion(&done); wait_done: retval = sub_info->retval; out: call_usermodehelper_freeinfo(sub_info); unlock: helper_unlock(); return retval; } static void umh_complete(struct subprocess_info *sub_info) { struct completion *comp = xchg(&sub_info->complete, NULL); if (comp) complete(comp); else call_usermodehelper_freeinfo(sub_info); } static void call_usermodehelper_freeinfo(struct subprocess_info *info) { if (info->cleanup) (*info->cleanup)(info); kfree(info); }
1.5 uevent发送
uevent发送可以通过kobject_uevent()，或者通过kobject_uevent_env()附加更多uevent信息。
kobject_uevent_env()主要分为两部分，一是通过netlink_broadcast_filtered()将socket信息发出去；另一个是通过uevent helper将uevent调用指定的uevent_helper进行处理，通常是热插拔程序mdev、udevd等。
int kobject_uevent(struct kobject *kobj, enum kobject_action action) { return kobject_uevent_env(kobj, action, NULL); } int kobject_uevent_env(struct kobject *kobj, enum kobject_action action, char *envp_ext[]) { struct kobj_uevent_env *env; const char *action_string = kobject_actions[action];------------将action转换成字符串。 const char *devpath = NULL; const char *subsystem; struct kobject *top_kobj; struct kset *kset; const struct kset_uevent_ops *uevent_ops; int i = 0; int retval = 0; #ifdef CONFIG_NET struct uevent_sock *ue_sk; #endif top_kobj = kobj; while (!top_kobj->kset && top_kobj->parent) top_kobj = top_kobj->parent; ... kset = top_kobj->kset; uevent_ops = kset->uevent_ops; ... /* originating subsystem */ if (uevent_ops && uevent_ops->name) subsystem = uevent_ops->name(kset, kobj); else subsystem = kobject_name(&kset->kobj); ... env = kzalloc(sizeof(struct kobj_uevent_env), GFP_KERNEL); if (!env) return -ENOMEM; /* complete object path */ devpath = kobject_get_path(kobj, GFP_KERNEL); if (!devpath) { retval = -ENOENT; goto exit; } /* default keys */ retval = add_uevent_var(env, "ACTION=%s", action_string);-----------默认添加ACTION、DEVPATH、SUBSYSTEM三个键值。 if (retval) goto exit; ... if (envp_ext) {-----------------------------------------------------将自定义的键值附上。 for (i = 0; envp_ext[i]; i++) { retval = add_uevent_var(env, "%s", envp_ext[i]); if (retval) goto exit; } } /* let the kset specific function add its stuff */ if (uevent_ops && uevent_ops->uevent) { retval = uevent_ops->uevent(kset, kobj, env); if (retval) { pr_debug("kobject: '%s' (%p): %s: uevent() returned " "%d\n", kobject_name(kobj), kobj, __func__, retval); goto exit; } } if (action == KOBJ_ADD) kobj->state_add_uevent_sent = 1; else if (action == KOBJ_REMOVE) kobj->state_remove_uevent_sent = 1; mutex_lock(&uevent_sock_mutex); ... #if defined(CONFIG_NET) /* send netlink message */ list_for_each_entry(ue_sk, &uevent_sock_list, list) {------------------遍历uevent_sock_list上所有的socket。 struct sock *uevent_sock = ue_sk->sk; struct sk_buff *skb; size_t len; if (!netlink_has_listeners(uevent_sock, 1)) continue; /* allocate message with the maximum possible size */ len = strlen(action_string) + strlen(devpath) + 2; skb = alloc_skb(len + env->buflen, GFP_KERNEL);--------------------为下面消息发送创建sk_buff实例。 if (skb) { char *scratch; /* add header */ scratch = skb_put(skb, len); sprintf(scratch, "%s@%s", action_string, devpath);-------------在已有键值基础上添加action_string@devpath。 /* copy keys to our continuous event payload buffer */ for (i = 0; i < env->envp_idx; i++) { len = strlen(env->envp[i]) + 1; scratch = skb_put(skb, len); strcpy(scratch, env->envp[i]); } NETLINK_CB(skb).dst_group = 1; retval = netlink_broadcast_filtered(uevent_sock, skb, 0, 1, GFP_KERNEL, kobj_bcast_filter, kobj);--------------------------------------通过netlink_broadcast_filtered()发送skb数据。 /* ENOBUFS should be handled in userspace */ if (retval == -ENOBUFS || retval == -ESRCH) retval = 0; } else retval = -ENOMEM; } #endif mutex_unlock(&uevent_sock_mutex); #ifdef CONFIG_UEVENT_HELPER /* call uevent_helper, usually only enabled during early boot */ if (uevent_helper[0] && !kobj_usermode_filter(kobj)) { struct subprocess_info *info; retval = add_uevent_var(env, "HOME=/"); if (retval) goto exit; retval = add_uevent_var(env, "PATH=/sbin:/bin:/usr/sbin:/usr/bin"); if (retval) goto exit; retval = init_uevent_argv(env, subsystem); if (retval) goto exit; retval = -ENOMEM; info = call_usermodehelper_setup(env->argv[0], env->argv, env->envp, GFP_KERNEL, NULL, cleanup_uevent_env, env); if (info) { retval = call_usermodehelper_exec(info, UMH_NO_WAIT); env = NULL; /* freed by cleanup_uevent_env */ } } #endif... } int add_uevent_var(struct kobj_uevent_env *env, const char *format, ...) { va_list args; int len; if (env->envp_idx >= ARRAY_SIZE(env->envp)) { WARN(1, KERN_ERR "add_uevent_var: too many keys\n"); return -ENOMEM; } va_start(args, format); len = vsnprintf(&env->buf[env->buflen], sizeof(env->buf) - env->buflen, format, args); va_end(args); if (len >= (sizeof(env->buf) - env->buflen)) { WARN(1, KERN_ERR "add_uevent_var: buffer size too small\n"); return -ENOMEM; } env->envp[env->envp_idx++] = &env->buf[env->buflen]; env->buflen += len + 1; return 0; } static int kobj_bcast_filter(struct sock *dsk, struct sk_buff *skb, void *data) { struct kobject *kobj = data, *ksobj; const struct kobj_ns_type_operations *ops; ops = kobj_ns_ops(kobj); if (!ops && kobj->kset) { ksobj = &kobj->kset->kobj; if (ksobj->parent != NULL) ops = kobj_ns_ops(ksobj->parent); } if (ops && ops->netlink_ns && kobj->ktype->namespace) { const void *sock_ns, *ns; ns = kobj->ktype->namespace(kobj); sock_ns = ops->netlink_ns(dsk); return sock_ns != ns; } return 0; } static int kobj_usermode_filter(struct kobject *kobj) { const struct kobj_ns_type_operations *ops; ops = kobj_ns_ops(kobj); if (ops) { const void *init_ns, *ns; ns = kobj->ktype->namespace(kobj); init_ns = ops->initial_ns(); return ns != init_ns; } return 0; } static int init_uevent_argv(struct kobj_uevent_env *env, const char *subsystem) { int len; len = strlcpy(&env->buf[env->buflen], subsystem, sizeof(env->buf) - env->buflen); if (len >= (sizeof(env->buf) - env->buflen)) { WARN(1, KERN_ERR "init_uevent_argv: buffer size too small\n"); return -ENOMEM; } env->argv[0] = uevent_helper; env->argv[1] = &env->buf[env->buflen]; env->argv[2] = NULL; env->buflen += len + 1; return 0; } static void cleanup_uevent_env(struct subprocess_info *info) { kfree(info->data); }
kobject_uevent_env()详细解释参考《设备模型的uevent机制》。
2. 用户空间处理uevent
2.1 kernel发送uevent
通过内核发送uevent很简单，将数据代表环境变量的字符串组装好后，选择合适的action，指定对应的kobject设备即可。
static int user_cooling_set_cur_state(struct thermal_cooling_device *cdev,
unsigned long new_target_ratio)
{
int ret = 0, i = 0, temperature = 0;
char *thermal_prop[4];
struct thermal_instance *instance;
list_for_each_entry(instance, &cdev->thermal_instances, cdev_node) {
if (instance->tz->temperature > temperature)
temperature = instance->tz->temperature;
}
user_cooling_state = new_target_ratio;
thermal_prop[0] = kasprintf(GFP_KERNEL, "NAME=%s", cdev->type);
thermal_prop[1] = kasprintf(GFP_KERNEL, "STATE=%lu", new_target_ratio);
thermal_prop[2] = kasprintf(GFP_KERNEL, "TEMP=%d", temperature);
thermal_prop[3] = NULL;
kobject_uevent_env(&cdev->device.kobj, KOBJ_CHANGE, thermal_prop);
for (i = 0; i < 3; ++i)
kfree(thermal_prop[i]);
return ret;
}
通过kobject_uevent_env()可以添加自定义环境变量，用户空间就会收到如下uevent消息。
change@/devices/virtual/thermal/cooling_device0 ACTION=change DEVPATH=/devices/virtual/thermal/cooling_device0 SUBSYSTEM=thermal NAME=user_cooling STATE=1 TEMP=90 SEQNUM=747
2.2 用户空间uevent处理
用户空间首先创建一个socket，并绑定到AF_NETLINK上，然后recv()接收消息，在处理字符串。
#include <stdio.h>
#include <string.h>
#include <sys/types.h>
#include <unistd.h>
#include <stdlib.h>
#include <sys/socket.h>
#include <linux/netlink.h>
#define UEVENT_MSG_LEN 2048
#define USER_COOLING_DEV "/devices/virtual/thermal/cooling_device0"
struct cooling_device {
const char *name;
const char *action;
const char *path;
int state;
int temp;
};
static int open_uevent_socket(void);
static void parse_uevent(const char *msg, struct cooling_device *cdev);
int main(int argc, char* argv[])
{
int socket_fd = -1;
char msg[UEVENT_MSG_LEN+2];
int n;
socket_fd = open_uevent_socket();--------------------------------------创建socket。
printf("socket_fd = %d\n", socket_fd);
do {
while((n = recv(socket_fd, msg, UEVENT_MSG_LEN, 0)) > 0) {---------接收uevent信息。
struct cooling_device cdev;
memset(&cdev, 0x0, sizeof(cdev));
if(n == UEVENT_MSG_LEN)
continue;
msg[n] = '\0';
msg[n+1] = '\0';
parse_uevent(msg, &cdev);---------------------------------------解析收到的uevent字符。
}
} while(1);
}
static int open_uevent_socket(void)
{
struct sockaddr_nl addr;
int sz = 64*1024;
int s = 0;
memset(&addr, 0, sizeof(addr));
addr.nl_family = AF_NETLINK;
addr.nl_pid = getpid();
addr.nl_groups = 0xffffffff;
s = socket(PF_NETLINK, SOCK_DGRAM, NETLINK_KOBJECT_UEVENT);-------------地址族是AF_NETLINK类型的socket，协议类型是NETLINK_KOBJECT_UEVENT。
if (s < 0) {
return -1;
}
setsockopt(s, SOL_SOCKET, SO_RCVBUFFORCE, &sz, sizeof(sz));
if (bind(s, (struct sockaddr *) &addr, sizeof(addr)) < 0) {-------------将当前socket绑定到AF_NETLINK地址族，并且设置本进程为处理消息的进程。
close(s);
return -1;
}
return s;
}
static void parse_uevent(const char *msg, struct cooling_device *cdev)
{
while (*msg) {
//printf("%s\n", msg);
if (!strncmp(msg, "NAME=", 5)) {
msg += 5;
cdev->name = msg;
} else if (!strncmp(msg, "ACTION=", 7)) {
msg += 7;
cdev->action = msg;
} else if (!strncmp(msg, "DEVPATH=", 8)) {
msg += 8;
cdev->path = msg;
} else if (!strncmp(msg, "STATE=", 6)) {
msg += 6;
cdev->state = atoi(msg);
} else if (!strncmp(msg, "TEMP=", 5)) {
msg += 5;
cdev->temp = atoi(msg);
}
while(*msg++);
}
if(!strncmp(cdev->path, USER_COOLING_DEV, sizeof(USER_COOLING_DEV)) && !strncmp(cdev->action, "change", 5))
printf("event { name=%s, action=%s, path=%s, state=%d, temp=%d}\n",
cdev->name, cdev->action, cdev->path, cdev->state, cdev->temp);
}
3. mdev
3.1 buxybox下mdev分析
mdev一种是附加-s主动遍历/sys/dev下设备，另一种是作为hotplug处理程序，被内核uevent_helper调用到。
mdev作为hotplug程序处理时，从环境变量中获取参数，创建或者删除设备，或者加载firmware。
int mdev_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
int mdev_main(int argc UNUSED_PARAM, char **argv)
{
RESERVE_CONFIG_BUFFER(temp, PATH_MAX + SCRATCH_SIZE);
INIT_G();
#if ENABLE_FEATURE_MDEV_CONF
G.filename = "/etc/mdev.conf";
#endif
bb_sanitize_stdio();
umask(0);
xchdir("/dev");--------------------------------------------------当前工作目录切换到/dev下。
if (argv[1] && strcmp(argv[1], "-s") == 0) {---------------------mdev -s情况下遍历/sys/dev下面所有设备。
/*
* Scan: mdev -s
*/
struct stat st;
#if ENABLE_FEATURE_MDEV_CONF
/* Same as xrealloc_vector(NULL, 4, 0): */
G.rule_vec = xzalloc((1 << 4) * sizeof(*G.rule_vec));
#endif
xstat("/", &st);
G.root_major = major(st.st_dev);
G.root_minor = minor(st.st_dev);
putenv((char*)"ACTION=add");
/* Create all devices from /sys/dev hierarchy */
recursive_action("/sys/dev",
ACTION_RECURSE | ACTION_FOLLOWLINKS,
fileAction, dirAction, temp, 0);----------------这个函数是递归函数，扫描/sys/dev下所有文件，如果发现dev文件，则按照/etc/mdev.con文件进行相应的设置。
} else {
char *fw;
char *seq;
char *action;
char *env_devname;
char *env_devpath;
unsigned my_pid;
unsigned seqnum = seqnum; /* for compiler */
int seq_fd;
smalluint op;
/* Hotplug:
* env ACTION=... DEVPATH=... SUBSYSTEM=... [SEQNUM=...] mdev
* ACTION can be "add", "remove", "change"
* DEVPATH is like "/block/sda" or "/class/input/mice"
*/
env_devname = getenv("DEVNAME"); /* can be NULL */----------在内核的kobject_uevent_env()中已经将参数和环境变量作为参数传入do_execve()中。这里mdev可以通过getenv来解析。
G.subsystem = getenv("SUBSYSTEM");
action = getenv("ACTION");
env_devpath = getenv("DEVPATH");
if (!action || !env_devpath /*|| !G.subsystem*/)
bb_show_usage();
fw = getenv("FIRMWARE");
seq = getenv("SEQNUM");
op = index_in_strings(keywords, action);--------------------keywords仅包含add和remove，所以op也仅有OP_add和OP_remove。
...
snprintf(temp, PATH_MAX, "/sys%s", env_devpath);
if (op == OP_remove) {
/* Ignoring "remove firmware". It was reported
* to happen and to cause erroneous deletion
* of device nodes. */
if (!fw)
make_device(env_devname, temp, op);-----------------在temp指定的目录下创建env_devnam名称的设备。
}
else {
make_device(env_devname, temp, op);---------------------删除temp目录下名称为env_devname的设备。
if (ENABLE_FEATURE_MDEV_LOAD_FIRMWARE) {
if (op == OP_add && fw)
load_firmware(fw, temp);------------------------将fw文件加载到temp路径中。
}
}
...
}
if (ENABLE_FEATURE_CLEAN_UP)
RELEASE_CONFIG_BUFFER(temp);
return EXIT_SUCCESS;
}
3.2 mdev.conf规则
下面是mdev.conf配置文件的基本格式：
<device regex> <uid>:<gid> <permissions> [=path] [@|$|*<command>] <device regex> <uid>:<gid> <permissions> [>path] [@|$|*<command>] <device regex> <uid>:<gid> <permissions> [!] [@|$|*<command>]
<device regex>：设备名称，支持正则表达式如hd[a-z][0-9]*等等。
<uid>:<gid>：用户ID和组ID。
<permissions>：表示设备的属性。
[=path]：如果path是个目录(比如drivers/)，则将设备节点移动到目录下；如果path是个名称，则将设备节点重命名为这个名称。
hda 0:3 660 =drivers/：移动hda到drivers目录下。
hdb 0:3 60 =cdrom：将hdb重命名为cdrom。
[>path]：重命名或者移动设备节点，类似于[=path]。但是同时会在/dev/下创建相关设备节点。
[!]：则不会创建设备节点。
[@<command>]：在创建设备节点之后执行command。
[$<command>]：在移动设备之前执行command。
[*<command>]：在创建设备之后以及移动设备之前都执行command。
上面的<command>通过system()调用执行，并且stdin/stdout/stderr都被重定向到/dev/null中。同时环境变量$MDEV指向匹配成功的设备节点名称，$ACTION表示uevent动作。
3.3 mdev和udev的区别
udev和mdev都是使用uevent机制处理热插拔的用户空间程序。
但是udev通过监听内核发送的uevent消息，解析后进行相应的热插拔擦欧洲哦，包括创建/删除设备节点，加载/卸载驱动程序，加载Firmware等等。
mdev则是基于uevent_helper机制，内核在发送uevent的时候，同时调用uevent_helper指向的用户空间程序进行热插拔处理。
另外udev是作为一个daemon常驻内存的，一直在监听uevent；mdev只是在需要的时候被调用。
4. 小结
uevent是内核发送消息到用户空间的一种途径，这种技术基于netlink实现。
内核中通过kobject_uevent()/kobject_uevent_env()发送uevent消息。
用户空间使用标准的socket接口，等待接收消息，然后进行解析处理；或者通过usermode helper调用用户空间进程mdev进行热插拔处理，处理的方式遵循mdev.conf规则。
联系方式:arnoldlu@qq.com
【作者】sky
本文版权归作者和博客园共有，欢迎转载，但未经作者同意必须保留此段声明，且在文章页面明显位置给出原文连接，否则保留追究法律责任的权利.
