源码剖析之poll
1. poll
从内核的角度看来，借助于VFS, 一切皆file
// 文件表示 include/linux/fs.h
struct file {
const struct file_operations *f_op;
spinlock_t f_lock;
// 文件内部实现细节
void *private_data;
#ifdef CONFIG_EPOLL
/* Used by fs/eventpoll.c to link all the hooks to this file */
struct list_head f_ep_links;
struct list_head f_tfile_llink;
#endif /* #ifdef CONFIG_EPOLL */
// 其他细节....
};
// 文件操作 include/linux/fs.h
struct file_operations {
// 文件提供给poll/select/epoll
// 调用poll_table_struct中指定的函数并获取文件的当前状态
unsigned int (*poll) (struct file *, struct poll_table_struct *);
// 其他方法read/write 等... ...
};
/**
* 通常文件poll方法的实现
* 调用poll_table_struct中指定的函数，并获得文件当前就绪事件的掩码
* @param flip 文件的指针
* @param wait 指向poll_table_struct的指针
* @return 返回文件当前就绪事件掩码
*/
unsigned int XXX_poll (struct file *filp, struct poll_table_struct *wait)
{
unsigned int mask = 0;
wait_queue_head_t * wait_queue;
// 1. 根据事件掩码wait->key_和文件实现filp->private_data 取得事件掩码对应的一个或多个wait queue head
some_code();
// 2. 调用poll_wait，目的是向获得的等待队列中添加等待队列项
poll_wait(filp, wait_queue, wait);
// 3. 取得文件当前就绪事件的掩码并保存到mask
some_code();
return mask;
}
// select/poll/epoll 向文件注册就绪后回调节点的接口结构
typedef struct poll_table_struct {
// 向指定等待队列(wait_queue_head)添加等待队列项的(wait_queue_t)的接口函数
poll_queue_proc _qproc;
// 关注的事件掩码, 文件的实现利用此掩码将对应的等待队列传递给_qproc
unsigned long _key;
} poll_table;
//
typedef void (*poll_queue_proc)(struct file *, wait_queue_head_t *, struct poll_table_struct *);
// 通用的poll_wait 函数, 文件的f_ops->poll 通常会调用此函数
static inline void poll_wait(struct file * filp, wait_queue_head_t * wait_address, poll_table *p)
{
if (p && p->_qproc && wait_address) {
// 调用poll_table_struct 中指定的函数_qproc
// qproc一般的作用是向指定事件等待队列中添加等待队列项
// 如果是select或poll 则是 __pollwait, 如果是 epoll 则是 ep_ptable_queue_proc
p->_qproc(filp, wait_address, p);
}
}
2. upd的poll
2.1 poll
net/ipv4/af_inet.c
const struct proto_ops inet_dgram_ops = {
// 其它细节
.poll = udp_poll,
// 其它细节
};2.2 udp_poll
net/ipv4/udp.c
/**
* udp_poll - wait for a UDP event.
* @file - file struct
* @sock - socket
* @wait - poll table
*
* This is same as datagram poll, except for the special case of
* blocking sockets. If application is using a blocking fd
* and a packet with checksum error is in the queue;
* then it could get return from select indicating data available
* but then block when reading it. Add special case code
* to work around these arguably broken applications.
*/
unsigned int udp_poll(struct file *file, struct socket *sock, poll_table *wait)
{
unsigned int mask = datagram_poll(file, sock, wait);
struct sock *sk = sock->sk;
sock_rps_record_flow(sk);
/* Check for false positives due to checksum errors */
if ((mask & POLLRDNORM) && !(file->f_flags & O_NONBLOCK) &&
!(sk->sk_shutdown & RCV_SHUTDOWN) && first_packet_length(sk) == -1)
mask &= ~(POLLIN | POLLRDNORM);
return mask;
}2.3 datagram_poll
net/core/datagram.c
/**
* datagram_poll - generic datagram poll
* @file: file struct
* @sock: socket
* @wait: poll table
*
* Datagram poll: Again totally generic. This also handles
* sequenced packet sockets providing the socket receive queue
* is only ever holding data ready to receive.
*
* Note: when you _don't_ use this routine for this protocol,
* and you use a different write policy from sock_writeable()
* then please supply your own write_space callback.
*/
unsigned int datagram_poll(struct file *file, struct socket *sock,
poll_table *wait)
{
struct sock *sk = sock->sk;
unsigned int mask;
sock_poll_wait(file, sk_sleep(sk), wait);
mask = 0;
/* exceptional events? */
if (sk->sk_err || !skb_queue_empty(&sk->sk_error_queue))
mask |= POLLERR |
(sock_flag(sk, SOCK_SELECT_ERR_QUEUE) ? POLLPRI : 0);
if (sk->sk_shutdown & RCV_SHUTDOWN)
mask |= POLLRDHUP | POLLIN | POLLRDNORM;
if (sk->sk_shutdown == SHUTDOWN_MASK)
mask |= POLLHUP;
/* readable? */
if (!skb_queue_empty(&sk->sk_receive_queue))
mask |= POLLIN | POLLRDNORM;
/* Connection-based need to check for termination and startup */
if (connection_based(sk)) {
if (sk->sk_state == TCP_CLOSE)
mask |= POLLHUP;
/* connection hasn't started yet? */
if (sk->sk_state == TCP_SYN_SENT)
return mask;
}
/* writable? */
if (sock_writeable(sk))
mask |= POLLOUT | POLLWRNORM | POLLWRBAND;
else
sk_set_bit(SOCKWQ_ASYNC_NOSPACE, sk);
return mask;
}2.4 sock_poll_wait
include/net/sock.c
/**
* sock_poll_wait - place memory barrier behind the poll_wait call.
* @filp: file
* @wait_address: socket wait queue
* @p: poll_table
*
* See the comments in the wq_has_sleeper function.
*/
static inline void sock_poll_wait(struct file *filp,
wait_queue_head_t *wait_address, poll_table *p)
{
if (!poll_does_not_wait(p) && wait_address) {
poll_wait(filp, wait_address, p);
/* We need to be sure we are in sync with the
* socket flags modification.
*
* This memory barrier is paired in the wq_has_sleeper.
*/
smp_mb();
}
}2.5 poll_wait
include/linux/poll.h
static inline void poll_wait(struct file * filp, wait_queue_head_t * wait_address, poll_table *p)
{
if (p && p->_qproc && wait_address)
p->_qproc(filp, wait_address, p);
}2.6 小结
poll -> udp_poll -> datagram_poll -> sock_poll_wait -> poll_wait
3. TCP的poll
3.1 poll
net/ipv4/af_inet.c
const struct proto_ops inet_stream_ops = {
.family = PF_INET,
.owner = THIS_MODULE,
.release = inet_release,
.bind = inet_bind,
.connect = inet_stream_connect,
.socketpair = sock_no_socketpair,
.accept = inet_accept,
.getname = inet_getname,
.poll = tcp_poll,
.ioctl = inet_ioctl,
.listen = inet_listen,
.shutdown = inet_shutdown,
.setsockopt = sock_common_setsockopt,
.getsockopt = sock_common_getsockopt,
.sendmsg = inet_sendmsg,
.recvmsg = inet_recvmsg,
.mmap = sock_no_mmap,
.sendpage = inet_sendpage,
.splice_read = tcp_splice_read,
.read_sock = tcp_read_sock,
.peek_len = tcp_peek_len,
#ifdef CONFIG_COMPAT
.compat_setsockopt = compat_sock_common_setsockopt,
.compat_getsockopt = compat_sock_common_getsockopt,
.compat_ioctl = inet_compat_ioctl,
#endif
};3.2 tcp_poll
net/ipv4/tcp.c
/*
* Wait for a TCP event.
*
* Note that we don't need to lock the socket, as the upper poll layers
* take care of normal races (between the test and the event) and we don't
* go look at any of the socket buffers directly.
*/
unsigned int tcp_poll(struct file *file, struct socket *sock, poll_table *wait)
{
unsigned int mask;
struct sock *sk = sock->sk;
const struct tcp_sock *tp = tcp_sk(sk);
int state;
sock_rps_record_flow(sk);
sock_poll_wait(file, sk_sleep(sk), wait);
state = sk_state_load(sk);
if (state == TCP_LISTEN)
return inet_csk_listen_poll(sk);
/* Socket is not locked. We are protected from async events
* by poll logic and correct handling of state changes
* made by other threads is impossible in any case.
*/
mask = 0;
/*
* POLLHUP is certainly not done right. But poll() doesn't
* have a notion of HUP in just one direction, and for a
* socket the read side is more interesting.
*
* Some poll() documentation says that POLLHUP is incompatible
* with the POLLOUT/POLLWR flags, so somebody should check this
* all. But careful, it tends to be safer to return too many
* bits than too few, and you can easily break real applications
* if you don't tell them that something has hung up!
*
* Check-me.
*
* Check number 1. POLLHUP is _UNMASKABLE_ event (see UNIX98 and
* our fs/select.c). It means that after we received EOF,
* poll always returns immediately, making impossible poll() on write()
* in state CLOSE_WAIT. One solution is evident --- to set POLLHUP
* if and only if shutdown has been made in both directions.
* Actually, it is interesting to look how Solaris and DUX
* solve this dilemma. I would prefer, if POLLHUP were maskable,
* then we could set it on SND_SHUTDOWN. BTW examples given
* in Stevens' books assume exactly this behaviour, it explains
* why POLLHUP is incompatible with POLLOUT. --ANK
*
* NOTE. Check for TCP_CLOSE is added. The goal is to prevent
* blocking on fresh not-connected or disconnected socket. --ANK
*/
if (sk->sk_shutdown == SHUTDOWN_MASK || state == TCP_CLOSE)
mask |= POLLHUP;
if (sk->sk_shutdown & RCV_SHUTDOWN)
mask |= POLLIN | POLLRDNORM | POLLRDHUP;
/* Connected or passive Fast Open socket? */
if (state != TCP_SYN_SENT &&
(state != TCP_SYN_RECV || tp->fastopen_rsk)) {
int target = sock_rcvlowat(sk, 0, INT_MAX);
if (tp->urg_seq == tp->copied_seq &&
!sock_flag(sk, SOCK_URGINLINE) &&
tp->urg_data)
target++;
if (tp->rcv_nxt - tp->copied_seq >= target)
mask |= POLLIN | POLLRDNORM;
if (!(sk->sk_shutdown & SEND_SHUTDOWN)) {
if (sk_stream_is_writeable(sk)) {
mask |= POLLOUT | POLLWRNORM;
} else { /* send SIGIO later */
sk_set_bit(SOCKWQ_ASYNC_NOSPACE, sk);
set_bit(SOCK_NOSPACE, &sk->sk_socket->flags);
/* Race breaker. If space is freed after
* wspace test but before the flags are set,
* IO signal will be lost. Memory barrier
* pairs with the input side.
*/
smp_mb__after_atomic();
if (sk_stream_is_writeable(sk))
mask |= POLLOUT | POLLWRNORM;
}
} else
mask |= POLLOUT | POLLWRNORM;
if (tp->urg_data & TCP_URG_VALID)
mask |= POLLPRI;
}
/* This barrier is coupled with smp_wmb() in tcp_reset() */
smp_rmb();
if (sk->sk_err || !skb_queue_empty(&sk->sk_error_queue))
mask |= POLLERR;
return mask;
}3.3 sock_poll_wait
include/net/sock.c
/**
* sock_poll_wait - place memory barrier behind the poll_wait call.
* @filp: file
* @wait_address: socket wait queue
* @p: poll_table
*
* See the comments in the wq_has_sleeper function.
*/
static inline void sock_poll_wait(struct file *filp,
wait_queue_head_t *wait_address, poll_table *p)
{
if (!poll_does_not_wait(p) && wait_address) {
poll_wait(filp, wait_address, p);
/* We need to be sure we are in sync with the
* socket flags modification.
*
* This memory barrier is paired in the wq_has_sleeper.
*/
smp_mb();
}
}3.4 poll_wait
include/linux/poll.h
static inline void poll_wait(struct file * filp, wait_queue_head_t * wait_address, poll_table *p)
{
if (p && p->_qproc && wait_address)
p->_qproc(filp, wait_address, p);
}3.5 小结
poll -> tcp_poll -> sock_poll_wait -> poll_wait