根据“男人选择”信息:
"On success, select() and pselect() return the number of file descrip‐
tors contained in the three returned descriptor sets which may be zero
if the timeout expires before anything interesting happens. On error,
-1 is returned, and errno is set appropriately; the sets and timeout become
undefined, so do not rely on their contents after an error."
选择将wakup因为:
1)read/write availability
2)select error
3)descriptoris closed.
但是,如果没有可用数据并且select仍在超时范围内,我们如何从另一个线程唤醒select()?
[更新]
伪代码
// Thread blocks on Select
void *SocketReadThread(void *param){
...
while(!(ReadThread*)param->ExitThread()) {
struct timeval timeout;
timeout.tv_sec = 60; //one minute
timeout.tv_usec = 0;
fd_set rds;
FD_ZERO(&rds);
FD_SET(sockfd, &rds)'
//actually, the first parameter of select() is
//ignored on windows, though on linux this parameter
//should be (maximum socket value + 1)
int ret = select(sockfd + 1, &rds, NULL, NULL, &timeout );
//handle the result
//might break from here
}
return NULL;
}
//main Thread
int main(){
//create the SocketReadThread
ReaderThread* rthread = new ReaderThread;
pthread_create(&pthreadid, NULL, SocketReaderThread,
NULL, (void*)rthread);
// do lots of things here
............................
//now main thread wants to exit SocketReaderThread
//it sets the internal state of ReadThread as true
rthread->SetExitFlag(true);
//but how to wake up select ??????????????????
//if SocketReaderThread currently blocks on select
}
[UPDATE]
1)@trojanfoe提供了一种实现此目的的方法,他的方法将套接字数据(可能是脏数据或退出消息数据)写入唤醒选择.我将进行测试并在那里更新结果.
2)另外要提一下,关闭套接字并不能保证唤醒select函数调用,请参见this post.
[UPDATE2]
做了很多测试后,这里有一些关于唤醒选择的事实:
1)如果select监视的套接字被另一个应用程序关闭,则select()调用
会立刻醒来.此后,读取或写入套接字将使用errno = 0获得返回值0
2)如果select监视的套接字被同一应用程序的另一个线程关闭,
如果没有要读取或写入的数据,则select()将在超时之前不会唤醒.选择超时后,使用errno = EBADF进行读/写操作会导致错误
(因为套接字在超时期间已被另一个线程关闭)
解决方法:
我使用基于pipe()的事件对象:
IoEvent.h:
#pragma once
class IoEvent {
protected:
int m_pipe[2];
bool m_ownsFDs;
public:
IoEvent(); // Creates a user event
IoEvent(int fd); // Create a file event
IoEvent(const IoEvent &other);
virtual ~IoEvent();
/**
* Set the event to signalled state.
*/
void set();
/**
* Reset the event from signalled state.
*/
void reset();
inline int fd() const {
return m_pipe[0];
}
};
IoEvent.cpp:
#include "IoEvent.h"
#include
#include
#include
#include
#include
using namespace std;
IoEvent::IoEvent() :
m_ownsFDs(true) {
if (pipe(m_pipe) < 0)
throw MyException("Failed to create pipe: %s (%d)", strerror(errno), errno);
if (fcntl(m_pipe[0], F_SETFL, O_NONBLOCK) < 0)
throw MyException("Failed to set pipe non-blocking mode: %s (%d)", strerror(errno), errno);
}
IoEvent::IoEvent(int fd) :
m_ownsFDs(false) {
m_pipe[0] = fd;
m_pipe[1] = -1;
}
IoEvent::IoEvent(const IoEvent &other) {
m_pipe[0] = other.m_pipe[0];
m_pipe[1] = other.m_pipe[1];
m_ownsFDs = false;
}
IoEvent::~IoEvent() {
if (m_pipe[0] >= 0) {
if (m_ownsFDs)
close(m_pipe[0]);
m_pipe[0] = -1;
}
if (m_pipe[1] >= 0) {
if (m_ownsFDs)
close(m_pipe[1]);
m_pipe[1] = -1;
}
}
void IoEvent::set() {
if (m_ownsFDs)
write(m_pipe[1], "x", 1);
}
void IoEvent::reset() {
if (m_ownsFDs) {
uint8_t buf;
while (read(m_pipe[0], &buf, 1) == 1)
;
}
}
你可以抛弃m_ownsFDs成员;我甚至不确定我是否会再使用它了.
标签:c-2,linux,tcp,multithreading,sockets
来源: https://codeday.me/bug/20190529/1177712.html