标签：art const walk 释放 技术 max put_user int etop
#include<poll.h>
int poll(struct pollfd* fds, nfds_t nfds, int timeout);
struct pollfd{
int fd;
short events; //注册的事件
short revents; //实际发生的事件
};fds是描述符数组，nfds是监听描述符的个数，timeout是以毫秒为单位的超时值ENOMEM：可用内存不足，无法完成请求
struct pollfd逐个拷贝进内核空间，构造成链表，然后进行轮训。有一个高级技术，使用链表混合使用堆与栈的空间。不过逐个将struct pollfd拷贝进内核地址空间，可能有点效率影响。内核为每一个关注的事件都会存在一个struct pollfd，与同一个fd存储在一个链表中，也就是说，链表的总长度= 文件描述符个数 × 关注的事件个数。将发生的事件从内核空间拷贝出来的时候也是逐个与拷贝。
基础数据结构：
struct poll_list {
struct poll_list *next;
int len;
struct pollfd entries[0];
};通用函数：
#define N_STACK_PPS ((sizeof(stack_pps) - sizeof(struct poll_list)) / sizeof(struct pollfd))
#define min_t(type,x,y) ({ type __x = (x); type __y = (y); __x < __y ? __x: __y; })
#define max_t(type,x,y) ({ type __x = (x); type __y = (y); __x > __y ? __x: __y; })Order 0：asmlinkage long sys_poll(struct pollfd __user *ufds, unsigned int nfds,long timeout_msecs)：接受用户空间传入的指针，经过拷贝后在内核空间进行操作
asmlinkage long sys_poll(struct pollfd __user *ufds, unsigned int nfds,long timeout_msecs)
{
s64 timeout_jiffies; //获得毫秒级的准确度
int ret;
if (timeout_msecs > 0) { //根据传入的参数对数据进行处理
#if HZ > 1000
/* We can only overflow if HZ > 1000 */
if (timeout_msecs / 1000 > (s64)0x7fffffffffffffffULL / (s64)HZ)
timeout_jiffies = -1;
else
#endif
timeout_jiffies = msecs_to_jiffies(timeout_msecs);
} else {
/* Infinite (< 0) or no (0) timeout */
timeout_jiffies = timeout_msecs;
}
ret = do_sys_poll(ufds, nfds, &timeout_jiffies); //做实际的操作
if (ret == -EINTR) { //如果被中断（也就是现实的告诉用户可以重复的进行调用），进行善后处理
struct restart_block *restart_block;
restart_block = ¤t_thread_info()->restart_block;
restart_block->fn = do_restart_poll;
restart_block->arg0 = (unsigned long)ufds;
restart_block->arg1 = nfds;
restart_block->arg2 = timeout_jiffies & 0xFFFFFFFF;
restart_block->arg3 = (u64)timeout_jiffies >> 32;
ret = -ERESTART_RESTARTBLOCK;
}
return ret; //返回有事件发生的描述符的个数
}Order 1：int do_sys_poll(struct pollfd __user *ufds, unsigned int nfds, s64 *timeout)
int do_sys_poll(struct pollfd __user *ufds, unsigned int nfds, s64 *timeout)
{
struct poll_wqueues table; //获得poll等待队列的操作
int err = -EFAULT, fdcount, len, size; //在还没有将用户空间的数据拷贝进来，将标志设置为默认出错
long stack_pps[POLL_STACK_ALLOC/sizeof(long)]; //在栈上分配一点空间，POLL_STACK_ALLOC = 256
struct poll_list *const head = (struct poll_list *)stack_pps; //使用链表将描述pollfd的数据结构
struct poll_list *walk = head;
unsigned long todo = nfds;
if (nfds > current->signal->rlim[RLIMIT_NOFILE].rlim_cur) //如果注册的描述符大小超过当前的进程的软资源限制，直接返回错误标志
return -EINVAL;
len = min_t(unsigned int, nfds, N_STACK_PPS); //获取用户传入struct pollfd的个数
for (;;) { //nb，挨个处理pollfd，混合栈空间和堆空间。。被秀了一脸。。
walk->next = NULL;
walk->len = len;
if (!len) //如果用户传入的len有0个，退出循环
break;
if (copy_from_user(walk->entries, ufds + nfds-todo,
sizeof(struct pollfd) * walk->len)) //一次从user空间中拷贝一个struct pollfd
goto out_fds;
todo -= walk->len;
if (!todo) //如果拷贝完成了，进行处理
break;
len = min(todo, POLLFD_PER_PAGE); //牛逼，竟然把栈空间和堆空间混合起来用了
size = sizeof(struct poll_list) + sizeof(struct pollfd) * len;
walk = walk->next = kmalloc(size, GFP_KERNEL);
if (!walk) {
err = -ENOMEM;
goto out_fds;
}
}
poll_initwait(&table); //初始化等待链表
fdcount = do_poll(nfds, head, &table, timeout); //获得发生事件的描述符个数
poll_freewait(&table); //释放等待链表
for (walk = head; walk; walk = walk->next) { //轮训所有描述符的链表
struct pollfd *fds = walk->entries; //获得fds数组
int j;
for (j = 0; j < walk->len; j++, ufds++) //每个描述符上观察的事件拷贝到用户空间，逐个字节拷贝
if (__put_user(fds[j].revents, &ufds->revents))
goto out_fds;
}
err = fdcount;
out_fds:
walk = head->next; //对之前分配的堆空间进行处理
while (walk) {
struct poll_list *pos = walk;
walk = walk->next;
kfree(pos);
}
return err;
}Order 2：static int do_poll(unsigned int nfds, struct poll_list *list,struct poll_wqueues *wait, s64 *timeout)：在链表组成的pollfd上进行操作
static int do_poll(unsigned int nfds, struct poll_list *list,struct poll_wqueues *wait, s64 *timeout)
{
int count = 0;
poll_table* pt = &wait->pt;
if (!(*timeout)) //如果时间为0，也就是说只轮训一遍不管有没有事件发生都直接返回
pt = NULL;
for (;;) {
struct poll_list *walk;
long __timeout;
set_current_state(TASK_INTERRUPTIBLE); //设置当前进程的状态
for (walk = list; walk != NULL; walk = walk->next) { //循环pollfd所在的链表
struct pollfd * pfd, * pfd_end;
pfd = walk->entries;
pfd_end = pfd + walk->len; //获得pollfd的描述
for (; pfd != pfd_end; pfd++) {
if (do_pollfd(pfd, pt)) { //调用文件系统例程，对pfd进行处理
count++;
pt = NULL;
}
}
}
pt = NULL;
if (!count) { //如果没有事件发生
count = wait->error;
if (signal_pending(current))
count = -EINTR; //将设置为可以重复调用的状态
}
if (count || !*timeout) //如果有事件发生或者事件用尽，打破循环，直接返回
break;
if (*timeout < 0) { //和select一样对设置时间的处理
__timeout = MAX_SCHEDULE_TIMEOUT;
} else if (unlikely(*timeout >= (s64)MAX_SCHEDULE_TIMEOUT-1))
__timeout = MAX_SCHEDULE_TIMEOUT - 1;
*timeout -= __timeout;
} else {
__timeout = *timeout;
*timeout = 0;
}
__timeout = schedule_timeout(__timeout);
if (*timeout >= 0)
*timeout += __timeout;
}
__set_current_state(TASK_RUNNING); //将进程设置为正常的状态
return count; //返回准备就绪的文件描述符的个数
}Order 3：static inline unsigned int do_pollfd(struct pollfd *pollfd, poll_table *pwait)：调用文件系统例程对pollfd进行处理
static inline unsigned int do_pollfd(struct pollfd *pollfd, poll_table *pwait)
{
unsigned int mask;
int fd;
mask = 0;
fd = pollfd->fd;
if (fd >= 0) {
int fput_needed;
struct file * file;
file = fget_light(fd, &fput_needed);
mask = POLLNVAL;
if (file != NULL) {
mask = DEFAULT_POLLMASK;
if (file->f_op && file->f_op->poll)
mask = file->f_op->poll(file, pwait); //文件系统例程
mask &= pollfd->events | POLLERR | POLLHUP;
fput_light(file, fput_needed);
}
}
pollfd->revents = mask;
return mask;
}标签：art const walk 释放 技术 max put_user int etop
原文地址：https://www.cnblogs.com/ukernel/p/9191087.html
