blocking select() from multiple sockets - c

Unix/C question here.
I have multiple sockets that I am trying to poll for periodic data. I don't want select to wait indefinitely so I have a timeout in place and I'm running in a loop. I have found that once a socket is ready to read, it is always ready to read. As in, I cannot have select go to sleep when there is no data to be read from any of the sockets.
for (i = 0; i < n_connections; i++) {
FD_SET( sockfd[i], &master );
if (sockfd[i] > fdmax)
fdmax = sockfd[i];
}
for(;;) {
int nready = 0;
timeout.tv_sec = 1;
timeout.tv_usec = 0;
read_fds = master;
if ( (nready = select(fdmax+1, &read_fds, NULL, NULL, NULL)) == -1 ) {
fprintf( stderr, "Select Error\n" );
return FAILURE;
}
printf( "Number of ready descriptors: %d\n", nready );
for (i = 0; i <= fdmax; i++) {
if (FD_ISSET(i, &read_fds)) {
if (( nbytes = recv(i, buf, sizeof(buf), 0)) <= 0 ) {
if (nbytes == 0) {
//connection closed
printf("Socket %d hung up\n", i );
}
else {
fprintf( stderr, "Recv Error %d\n", nbytes);
}
}
else {
printf( "Data Received on %d: %s\n", i, buf );
}
}
} // end file descriptor loop
It seems that after my first read, the 1 second timeout no longer applies and the socket is always "ready to read", even if there are 0 bytes available. How can I get select to sleep until data comes in (for the one second, or by switching the final argument to NULL, indefinitely waiting for data to come in on the socket?)
Output:
Number of Ready Descriptors: 2
Data Received on 4: GreetingsChap
Data Received on 5: HiMatengsChap
Loop...
Number of Ready Descriptors: 2
Socket 4 hung up
Socket 5 hung up
Loop...
Number of Ready Descriptors: 2
Socket 4 hung up
Socket 5 hung up
Loop...
Thank you,
Note: Code updated for clarity
Updated based on #yvesBraumes suggestions - still doesn't work.

If you detect that a connection is closed, remove the socket from the fd set, otherwise select is going to report them (Socket 4 hung up).. select is not edge triggered, if you don't handle the event, it's going to report it again.

Indeed, if recv returns 0 (and not -1, with errno=EWOULDBLOCK), the socket is closed. You should call close() on it as well, and take it out of the select() call. Otherwise it will remain in WAIT1 and release select() each time.

You are using FD_ISSET incorrectly. You need to be passing a socket ID to the "fd" parameter, not an index:
if (FD_ISSET(i, &read_fds))...
needs to be
if (FD_ISSET(sockfd[i], &read_fds))...
Likewise with recv.

Related

recv() does not get value from network

I Have a typical client server C socket program.
SERVER:
if(ndp.cmd == 11)
{
//ack1 = 0;
puts("Query Command for Light 2");
pthread_mutex_lock(&lock);
// ..Some critical stuff
pthread_mutex_unlock(&lock);
printf("ID: %d, Level: %d\n", new.address, new.level);
ack1 = new.level; //This result is showing correct on server
ack1 = htonl(ack1);
send(client_sock, &ack1, sizeof(ack1), 0);
}
CLIENT:
printf("Query Actual level Light2\n");
dp.id = 2;dp.cmd = 11; dp.active = 0; dp.level = 0; dp.group = 0;
if( send(sock , &dp , sizeof(dp) , 0) < 0) { puts("Send failed"); }
sleep(1);
ret = recv(sock , &level1 , sizeof(level1) , 0);
printf("Number of bytes received: %d\n", ret); //Always gives 4
fflush(stdout);
printf("Light level %x\n", ntohl(level1) ); //This prints 0 (incorrect)
fflush(stdout);
sleep(5);
printf("Query Actual level Light2\n");
dp.id = 2;dp.cmd = 11; dp.active = 0; dp.level = 0; dp.group = 0;
if( send(sock , &dp , sizeof(dp) , 0) < 0) { puts("Send failed"); }
sleep(1);
ret = recv(sock , &level2 , sizeof(level2) , 0);
printf("Number of bytes received: %d\n", ret); // Always gives 4
fflush(stdout);
printf("Light level %x\n", ntohl(level2) ); //This prints correct value
fflush(stdout);
close(sock);
Expected output:
Light level 32 (hex value for 50);
Light level 32
Actual output:
Light level 0
Light level 32
So the problem is on same requests, 1st send from server is not received by client, but next recv() gives correct value.
Why is the data from 1st recv() getting lost, is it getting buffered and maybe I am getting the previous value in 2nd recv() ?
Please help.
Your assumption is not wrong. It may be buffered. Also note that when the receiver reads data faster than the sender does send it, some recv calls may read no data since the sender did not send any data yet. That could be the cause of what you're describing.
UPDATE: If you plan to always send/receive the same amount of data, for example a 20-byte packet containing what you defined as a command, then you can wrap your send and recv in a loop. This loop will write (or read) a single command at a time. You cannot assume a call to send or recv will succeed, so you MUST check for errors as #Joachim mentioned in comments. Another thing to note is that recv MAY read less bytes than the specified, so you may have to read multiple times in order to receive the complete command. The same applies to send.
You may use select/poll mechanism to check if there's new data has arrived to socket and waiting to be read properly. You can also check the amount of data has come to socket by invoking an ioctl call as the following:
{
int size = 0;
ioctl(socket, FIONREAD, &size);
if(size > 0)
{
// do recv operation(s)
}
}

Waiting for child processes when using select() for multiplexing

I am facing some trouble dealing with zombie processes. I wrote a simple server which creates tic tac toe matches between players. I am using select() to multiplex between multiple connected clients. Whenever there are two clients, the server will fork another process which execs a match arbiter program.
The problem is that select() blocks. So therefore, say if there is a match arbiter program running as a child process and it exits, the parent will never wait for the child if there are no incoming connections because select() is blocking.
I have my code here, apologies since it is quite messy.
while(1) {
if (terminate)
terminate_program();
FD_ZERO(&rset);
FD_SET(tcp_listenfd, &rset);
FD_SET(udpfd, &rset);
maxfd = max(tcp_listenfd, udpfd);
/* add child connections to set */
for (i = 0; i < MAXCLIENTS; i++) {
sd = tcp_confd_lst[i];
if (sd > 0)
FD_SET(sd, &rset);
if (sd > maxfd)
maxfd = sd;
}
/* Here select blocks */
if ((nready = select(maxfd + 1, &rset, NULL, NULL, NULL)) < 0) {
if (errno == EINTR)
continue;
else
perror("select error");
}
/* Handles incoming TCP connections */
if (FD_ISSET(tcp_listenfd, &rset)) {
len = sizeof(cliaddr);
if ((new_confd = accept(tcp_listenfd, (struct sockaddr *) &cliaddr, &len)) < 0) {
perror("accept");
exit(1);
}
/* Send connection message asking for handle */
writen(new_confd, handle_msg, strlen(handle_msg));
/* adds new_confd to array of connected fd's */
for (i = 0; i < MAXCLIENTS; i++) {
if (tcp_confd_lst[i] == 0) {
tcp_confd_lst[i] = new_confd;
break;
}
}
}
/* Handles incoming UDP connections */
if (FD_ISSET(udpfd, &rset)) {
}
/* Handles receiving client handles */
/* If client disconnects without entering their handle, their values in the arrays will be set to 0 and can be reused. */
for (i = 0; i < MAXCLIENTS; i++) {
sd = tcp_confd_lst[i];
if (FD_ISSET(sd, &rset)) {
if ((valread = read(sd, confd_handle, MAXHANDLESZ)) == 0) {
printf("Someone disconnected: %s\n", usr_handles[i]);
close(sd);
tcp_confd_lst[i] = 0;
usr_in_game[i] = 0;
} else {
confd_handle[valread] = '\0';
printf("%s\n", confd_handle); /* For testing */
fflush(stdout);
strncpy(usr_handles[i], confd_handle, sizeof(usr_handles[i]));
for (j = i - 1; j >= 0; j--) {
if (tcp_confd_lst[j] != 0 && usr_in_game[j] == 0) {
usr_in_game[i] = 1; usr_in_game[j] = 1;
if ((child_pid = fork()) == 0) {
close(tcp_listenfd);
snprintf(fd_args[0], sizeof(fd_args[0]), "%d", tcp_confd_lst[i]);
snprintf(fd_args[1], sizeof(fd_args[1]), "%d", tcp_confd_lst[j]);
execl("nim_match_server", "nim_match_server", usr_handles[i], fd_args[0], usr_handles[j], fd_args[1], (char *) 0);
}
close(tcp_confd_lst[i]); close(tcp_confd_lst[j]);
tcp_confd_lst[i] = 0; tcp_confd_lst[j] = 0;
usr_in_game[i] = 0; usr_in_game[j] = 0;
}
}
}
}
}
}
Is there a method which allows wait to run even when select() is blocking? Preferably without signal handling since they are asynchronous.
EDIT: Actually, I found out that select has a timeval data structure which we can specify the timeout. Would using that be a good idea?
I think your options are:
Save all your child descriptors in a global array and call wait() from a signal handler. If you don't need the exit status of your children in your main loop, I think this is the easiest.
Instead of select, use pselect -- it will return upon receiving a specified (set of) signal(s), in your case, SIGCHLD. Then call wait/WNOHANG on all child PIDs. You will need to block/unblock SIGCHLD at the right moments before/after pselect(), see here: http://pubs.opengroup.org/onlinepubs/9699919799/functions/pselect.html
Wait on/cleanup child PIDs from a secondary thread. I think this is the most complicated solution (re. synchronization between threads), but since you asked, it's technically possible.
If you just want to prevent zombie processes, you could set up a SIGCHLD signal handler. If you want to actually wait for the return status, you could write bytes into a pipe (non-blocking, just in case) from the signal handler and then read those bytes in the select loop.
For how to handle SIGCHLD, see http://www.microhowto.info/howto/reap_zombie_processes_using_a_sigchld_handler.html -- you want to do something like while (waitpid((pid_t)(-1), 0, WNOHANG) > 0) {}
Perhaps the best approach is sending a single byte from the SIGCHLD signal handler to the main select loop (non-blocking, just in case) and doing the waitpid loop in the select loop when bytes can be read from the pipe.
You could also use a signalfd file descriptor to read the SIGCHLD signal, although that works only on Linux.

select returns 0 although packets returned, i did reset the readset every time before calling select

I wrote a ping program sending syn packet to destination ip address 1.1.1.1 port 0. I used wireshark that i saw packets returned with RST/ACK flags on. But the problem is when i use select() to try to read the socket, select() always returns 0. So I have no idea how to debug the program. I'm sure I reset the fd_set readset everytime. Is there anything i just missed ?
i discovered the reply from 1.1.1.1 returned back in 0.0003 second using wireshark
the timeout i set is 1 second(1000 ms)
Here is the code
void readloop() {
....
while(nsent < 4) {
send_v4();
wait_for_reply(1000); // wait for 1 second;
}
}
int wait_for_reply(long wait_time) {
....
result = recving_time(...);
if(result < 0) // because time out
return 0;
....
}
int recving_time(...) {
......
fd_set readset;
select_again:
set timeout value to structure *to*
FD_ZERO(&readset);
FD_SET(sockfd, &readset);
readable = select(sockfd+1, &readset, NULL, NULL, &to);
#ifdef DEBUG
fprintf(stderr, "readable is %d\n",readable); // **Why readable always be 0 although packets returned.**
#endif
if(readable < 0) {
if(errno == EINTR)
goto select_again;
else {
perror("select() error");
exit(1);
}
if(readable == 0) {
return -1;
}
......
}
I'm not sure there is enough code here to understand what I am trying to saying. I will be very very appreciate if you can help me
Select() returns the number of ready descriptors that are contained in the descriptor sets, or -1 if an error occurred. If the time limit expires, select() returns 0.
It would seem like your timeout (which you're not showing the setup of) is elapsing.

select() not responding?

The objective of my program is to use select to manage multiple sockets. However, I thought of trying it with one socket first. Now, the problem that I am facing is that initially client sends data to server, and server receives it and displays it, but then when client again sends some data, the server code remains still at select command.
here are some snippets that will give you an idea of how I am initializing the socket.
if((master_socket = socket(PF_INET, SOCK_STREAM, 0)) < 0)
{
exit(1);
}
if((bind(master_socket, (struct sockaddr *)&req, sizeof(req))) < 0)
{
exit(1);
}
listen(master_socket, 5);
FD_SET(master_socket,&listening);
/* wait for connection, then receive and print text */
len = sizeof(struct sockaddr);
while(1)
{
FD_ZERO(&listening); //Flush out everything in socket
FD_SET(master_socket,&listening); // Add master
if(f_client>0) // Add client if any
{
FD_SET(f_client,&listening);
}
printf("Checking for new connection \n");
//Timeout is null, so waiting indefinitely
rc = select(FD_SETSIZE, &listening, NULL, NULL, NULL);
if (FD_ISSET(master_socket, &listening))
{
printf("Master side invoked\n");
if((f_client = accept(master_socket, (struct sockaddr *)&req, &len)) < 0)
{
exit(1);
}
}
else if (FD_ISSET(f_client,&listening))
{
if ((valread = read( f_client , buf, 1024)) == 0)
{
close(f_client);
f_client=0;
}
else
{
fputs(buf, stdout);
}
}
}
Basically in above program, it connects to the server, maintains a file descriptor for client f_client and add it. And in every round, it clears the socket, add master socket, and client socket if any, and then checks. Problem here is, first time it works, but second time when client sends some data. it gets hang to rc = select(FD_SETSIZE, &listening, NULL, NULL, NULL);
I am not to understand what is wrong here. Can anyone help?
if ((valread = read( f_client , buf, 1024)) == 0)
{
close(f_client);
f_client=0;
}
else
{
fputs(buf, stdout);
}
This code is broken. The fputs function can only be used with a C-style string. You just have arbitrary data with no particular structure. Since you ignore valread, you also have no idea how many bytes you read. (Think about it, how could fputs possibly know how many bytes to output? That information is only in valread, and you don't pass it that information.)
You've already received the data, this broken code just threw it away. If you log valread, you'll see that you actually already read it in your last call to read before the call to select that hung.
instead of fputs, you could use something like this:
for (int i = 0; i < valread; ++i)
putchar(buf[i]);

Using sockets to read from client side

I basically have a server set up and I'm accepting new clients(UNIX) and i'm using select() command to wait for activity on file descriptor but I'm not sure how to write from the clients side and then read it on the servers side
FD_ZERO(&readfds);
FD_SET(server_sockfd, &readfds);
FD_SET(STDIN_FILENO, &readfds);
while (1) {
testfds = readfds;
select(4 + MAXCLIENTS, &testfds, NULL, NULL, NULL);
for (fd = 0; fd < 4 + MAX_CLIENTS; fd++) {
if (FD_ISSET(fd, &testfds)) {
if (fd == server_sockfd) { /* new connection request */
client_sockfd = accept(server_sockfd, NULL, NULL);
if (num_clients < MAXCLIENTS) {
FD_SET(client_sockfd, &readfds);
num_clients++;
} else {
sprintf(message, "XSorry, too many clients. Try again later.\n");
write(client_sockfd, message, strlen(message));
close(client_sockfd);
}
} else if (fd == STDIN_FILENO) {
fgets(kb_message, BUFSIZ + 1, stdin);
if (strcmp(kb_message, "quit\n") == 0) {
sprintf(message, "XServer is shutting down.\n");
for (fd2 = 4; fd2 < 4 + MAX_CLIENTS; fd2++) {
if (FD_ISSET(fd2, &readfds)) {
write(fd2, message, strlen(message));
close(fd2);
}
}
close(server_sockfd);
exit(EXIT_SUCCESS);
} else {
sprintf(message, "M%s", kb_message);
for (fd2 = 4; fd2 < 4 + MAX_CLIENTS; fd2++)
if (FD_ISSET(fd2, &readfds))
write(fd2, message, strlen(message));
}
} else { /* client leaving */
close(fd);
FD_CLR(fd, &readfds);
num_clients--;
}
}
}
}
How would I handle write request from clients and then write back to them, would it be under "else" and how can I check if client is exiting or writing.
Thanks
The most common mistake with select(2) is not re-initializing the descriptor sets, since second, third, and forth arguments are input-output parameters.
Setup an fd_set, for reading before the outer loop, add listening socket descriptor to it, enter the loop, make a copy of the this fd_set and give the copy to select(2). When new connection arrives, add its descriptor to the original fd_set. Same for closed socket (error or EOF on read(2)) - remove the descriptor from the original fd_set.
Hope this helps.
You are correct in thinking you need the read code in your 'else' block. If a file descriptor triggers and it isn't stdin or the 'connect' descriptor, then it is one of your clients attempting to send you data. When one of those file descriptors is triggered in the select, you need to call 'read' on that descriptor to read the data into the buffer. The read command will return you the number of bytes read. If this is a positive number, then it indicates the client has sent you data. If it is zero, then that indicates that the client has ended the TCP connection to your server.
The else block will look something like:
else
{
//Existing connection has data for us to read
if((nBytes = read(fd, buffer, MAXBUFFER)) <= 0)
{
if(nBytes == 0)
{
//Actually, its sending us zero bytes, connection closed
printf("Socket %d hung up\n", fd;
}
else
printf ("Read Error"\n)
}
Also, Follow Nikolai N Fetissov's advice above and make sure that when client's connect you store their fd in a permanent fd_set structure, as the one you are using is being modified by the select call.
Your problem might be that you have a variable called read. It's going to mask one of hte functions you need to use - the read() system call to get data out of the socket. The client puts it in with write(). You might also want to check the return value from select(), which will tell you how many of the file descriptors are ready for reading. Then you can check which ones using FD_ISSET(). It looks like you're doing that part already (except you seem to be checking the wrong variable?)... just call read() on that file descriptor to get out the data the client wrote.
else
{
bzero(buf,100);
n=read(i,buf,100); // Read the client message
buf[n]='\0';
if(n==0) // Check the client is closed or not
{
printf("%d is closed\n",i);
close(i);
FD_CLR(i, &master);
if(i==fdmax)
fdmax--;
}
else
{
n=strlen(buf);
write(1,buf,n);
fflush(stdout);
write(1,"Enter the message\n",18);
bzero(buf,100);
read(0,buf,100);
buf[n]='\0';
write(i,buf,n);
fflush(stdout);
}
}
Notes:
After accept the client, Add the client in the fd set.
Then read the message from the client
If the message is equal to 0, then the client is closed.
If you want to send the message to the client, using the client fd, you can send to the client

Resources