Why doesn't send() in winsock guarantee delivery of the all bytes you request?
This is TCP and it's blocking sockets.
Similarly, this happens when non-blocking. How can you guarantee that you send everything?
I've noticed recv() does the same.
If it didn't send everything, just call send again on the rest. If blocking, you can do it immediately. If non-blocking, you can either wait or use a socket discovery method (like select or I/O completion ports). The same goes for recv. If you didn't get all you wanted, call recv again. This is one of the reasons both recv and send return the number of bytes sent or received.
The number of bytes you pass to send or recv is just a limit. It can send less than that (though, unless non-blocking, usually won't). But it can definitely receive less than that. (The OS has no control over how much data it receives or when it receives it.)
TCP is implemented for you. But if you have an application protocol that involves application-level messages, then the application has to implement them. It won't happen by magic. TCP doesn't "glue the bytes together" into a message for you. TCP is a byte-stream protocol, not a message protocol. If you want messages, you have to implement them.
This behaviour is "by design".
You can use an outer loop as shown in this example:
int sendBuffer (SOCKET ClientSocket, const char *buf, int len, int flags)
{
int num_left = len;
int num_sent;
int err = 0;
const char *cp = buf;
while (num_left > 0)
{
num_sent = send(ClientSocket, cp, num_left, flags);
if (num_sent < 0)
{
err = SOCKET_ERROR;
break;
}
assert(num_sent <= num_left);
num_left -= num_sent;
cp += num_sent;
}
return (err == SOCKET_ERROR ? SOCKET_ERROR : len);
}
send tells you what it was able to send via its return value.
Loop until send has cumulatively sent all the data or returns an error.
Related
I'm trying to send data to the connected client, even when the client did not send me a message first.
This is my current code:
while (true) {
// open a new socket to transmit data per connection
int sock;
if ((sock = accept(listen_sock, (sockaddr *) &client_address, &client_address_len)) < 0) {
logger.log(TYPE::ERROR, "server::could not open a socket to accept data");
exit(0);
}
int n = 0, total_received_bytes = 0, max_len = 4096;
std::vector<char> buffer(max_len);
logger.log(TYPE::SUCCESS,
"server::client connected with ip address: " + std::string(inet_ntoa(client_address.sin_addr)));
// keep running as long as the client keeps the connection open
while (true) {
n = recv(sock, &buffer[0], buffer.size(), 0);
if (n > 0) {
total_received_bytes += n;
std::string str(buffer.begin(), buffer.end());
KV key_value = kv_from(vector_from(str));
messaging.set_command(key_value);
}
std::string message = "hmc::" + messaging.get_value("hmc") + "---" + "sonar::" + messaging.get_value("sonar") + "\n";
send(sock, message.c_str(), message.length(), 0);
}
logger.log(TYPE::INFO, "server::connection closed");
close(sock);
}
I thought by moving the n = recv(sock, &buffer[0], buffer.size(), 0); outside the while condition that it would send the data indefinitely, but that is not what happened.
Thanks in advance.
Solution
Adding MSG_DONTWAIT to the recv function enabled non-blocking operations which I was looking for.
First I will explain, why it does not work, then I will make a proposal for solutions. Basically you will find the answer in the man7.org > Linux > man-pages and for recv specifially here.
When the function "recv" is called, then it will not return, until data is available and can be read. This behavior of functions is called "blocking". Means, the current execution thread is blocked until data has been read.
So, calling the function
n = recv(sock, &buffer[0], buffer.size(), 0);
as you did, causes the trouble. You need also to check the return code. 0 means, connection closed, -1 means error and you must check errno for further information.
You can modify the socket to work in non-blocking mode with the function fnctl and the O_NONBLOCK flag, for the lifetime of the socket. You can also use the the flag MSG_DONTWAIT as 4th parameter (flags), to unblock the function on a per-function-call base.
In both cases, if no data is available, the functions returns a -1 and you need to check errno for EAGAIN or EWOULDBLOCK.
return value 0 indicates that the connection has been closed.
But from the architecture point of view, I would not recommend to use this approach. You could use multiple threads for receiving and sending data, or, using Linux, one of select, poll or similar functions. There is even a common design pattern for this. It is called "reactor", There are also related patterns like "Acceptor/Connector" and "Proactor"/"ACT" available. If you plan to write a more robust application, then you may consider those.
You will find an implementation of Acceptor, Connector, Reactor, Proactor, ACT here
Hope this helps
I'm sending and receiving info with a unix socket, but I do not completely understand how it works. Basically, I send a message like this:
int wr_bytes = write(sock, msg.c_str(), msg.length());
And receive message like this:
int rd_bytes = read(msgsock, buf, SOCKET_BUFFER_SIZE);
This code works perfectly with thousands of bytes, what I don't understand is, how does the read function knows when the other part is done sending the message? I tried to read the read documentation and, on my understanding read will return once it reaches EOF or the SOCKET_BUFFER_SIZE, is that correct?
So I'm guessing that when I give my string to the write function, it adds an EOF at the end of my content so the read function knows when to stop.
I'm asking this question because, I did not add any code that checks whether the other part finished sending the message, however, I'm receiving big messages (thousands of bytes) without any problem, why is that happening, why am I not getting only parts of the message?
Here is the full function I'm using to send a message to a unix socket server:
string sendSocketMessage(string msg) {
int sock;
struct sockaddr_un server;
char buf[1024];
sock = socket(AF_UNIX, SOCK_STREAM, 0);
if (sock < 0) {
throw runtime_error("opening stream socket");
}
server.sun_family = AF_UNIX;
strcpy(server.sun_path, "socket");
if (connect(sock, (struct sockaddr *) &server, sizeof(struct sockaddr_un)) < 0) {
close(sock);
throw runtime_error("connecting stream socket");
}
if (write(sock, msg.c_str(), msg.length()) < 0){
throw runtime_error("writing on stream socket");
close(sock);
}
bzero(buf, sizeof(buf));
int rval = read(sock, buf, 1024);
return string( reinterpret_cast< char const* >(buf), rval );
}
And here is my server function (a little bit more complicated, the type vSocketHandler represents a function that I call to handle requests):
void UnixSocketServer::listenRequests(vSocketHandler requestHandler){
int sock, msgsock, rval;
struct sockaddr_un server;
char buf[SOCKET_BUFFER_SIZE];
sock = socket(AF_UNIX, SOCK_STREAM, 0);
if (sock < 0) {
throw runtime_error("opening stream socket");
}
server.sun_family = AF_UNIX;
strcpy(server.sun_path, SOCKET_FILE_PATH);
if (bind(sock, (struct sockaddr *) &server, sizeof(struct sockaddr_un))) {
throw runtime_error("binding stream socket");
}
listen(sock, SOCKET_MAX_CONNECTIONS);
while(true) {
msgsock = accept(sock, 0, 0);
if (msgsock == -1){
throw runtime_error("accept socket");
} else {
bzero(buf, sizeof(buf));
if((rval = read(msgsock, buf, SOCKET_BUFFER_SIZE)) < 0)
throw runtime_error("reading stream message");
else if (rval == 0){
//do nothing, client closed socket
break;
} else {
string msg = requestHandler(string( reinterpret_cast< char const* >(buf), rval ));
if(write(msgsock, msg.c_str(), msg.length()) < 0)
throw runtime_error("sending stream message");
}
close(msgsock);
}
}
close(sock);
unlink(SOCKET_FILE_PATH);
}
what I don't understand is, how does the read function knows when the other part is done sending the message?
For a stream-type socket, such as you're using, it doesn't. For a datagram-type socket, communication is broken into distinct chunks, but if a message spans multiple datagrams then the answer is again "it doesn't". This is indeed one of the key things to understand about the read() and write() (and send() and recv()) functions in general, and about sockets more specifically.
For the rest of this answer I'll focus on stream oriented sockets, since that's what you're using. I'll also suppose that the socket is not in non-blocking mode. If you intend for your data transmitted over such a socket to be broken into distinct messages, then it is up to you to implement an application-level protocol by which the other end can recognize message boundaries.
I tried to read the read documentation and, on my understanding read will return once it reaches EOF or the SOCKET_BUFFER_SIZE, is that correct?
Not exactly. read() will return if it reaches the end of the file, which happens when the peer closes its socket (or at least shuts down the write side of it) so that it is certain that no more data will be sent. read() will also return in the event of any of a variety of error conditions. And read() may return under other unspecified circumstances, provided that it has transferred at least one byte. In practice, this last case is generally invoked if the socket buffer fills, but it may also be invoked under other circumstances, such as when the buffer empties.
So I'm guessing that when I give my string to the write function, it adds an EOF at the end of my content so the read function knows when to stop.
No, it does no such thing. On success, the write() function sends some or all of the bytes you asked it to send, and nothing else. Note that it is not guaranteed even to send all the requested bytes; its return value tells you how many of them it actually did send. If that's fewer than "all", then ordinarily you should simply perform another write() to transfer the rest. You may need to do this multiple times to send the whole message. In any event, only the bytes you specify are sent.
I'm asking this question because, I did not add any code that checks whether the other part finished sending the message, however, I'm receiving big messages (thousands of bytes) without any problem, why is that happening, why am I not getting only parts of the message?
More or less because you're getting lucky, but the fact that you're using UNIX-domain sockets (as opposed to network sockets) helps. Your data are transferred very efficiently from sending process to receiving process through the kernel, and it is not particularly surprising that large writes() are received by single read()s. You cannot safely rely on that always to happen, however.
I am writing the client side of the Socket. When there is something to read my code works fine but when there is nothing to read, the recv never returns. Help please.
Code:
m_socket = socket(AF_INET, SOCK_STREAM, 0);
struct sockaddr_in dest;
if ( m_socket )
{
memset(&dest, 0, sizeof(dest)); /* zero the struct */
dest.sin_family = AF_INET;
dest.sin_addr.s_addr = inet_addr(address); /* set destination IP number */
dest.sin_port = htons(port);
if (connect(m_socket, (struct sockaddr *)&dest, sizeof(struct sockaddr)) == SOCKET_ERROR)
{
return false;
}
else
{
std::vector<char> inStartup1(2);
int recvReturn = recv(Socket, &inStartup1.at(0), inStartup1.size(), 0);
}
recv is a blocking call. This would help you:-
The recv() call is normally used only on a connected socket.It returns the length of the message on successful completion. If a message is too long to fit in the supplied buffer, excess bytes may be discarded DEPENDING on the type of socket the message is received from.
If no messages are available at the socket, the receive calls wait for a message to arrive, unless the socket is nonblocking, in which case the value -1 is returned and the external variable errno is set to EAGAIN or EWOULDBLOCK. The receive calls normally return any data available, up to the requested amount, rather than waiting for receipt of the full amount requested.
Taking this one step further, on a server this is how you would correctly handle a connection (socket or serial port does not matter):
make the socket/port non-blocking: this is the first important step; it means that recv() will read what is available (if anything) and return the number of read bytes or -1 in case of an error.
use select(), with a timeout, to find out when data becomes available. So now you wait for a certain amount of time for data to become available and than read it.
The next problem to handle is making sure you read the full message. Since there is no guarantee that the whole message will be available when you call recv(), you need to save whatever is available and go back to select() and wait for the next data to become available.
Put everything in a while(cond) construct to make sure you read all the data.
The condition in the while is the only thing left to figure out - you either know the length of the expected message or you use some delimiters to mark the end of the message.
Hope this helps!
In got the following problem:
I made a server which is able to handle multiple connection by using select(). But select returns a client(index of FD_SET) also if the socket just got an error like "client disconnect" or whatever.
Is it possible to check a socket without calling recv(). Because to receive I need to get a buffer out of my "BufferPool"
Sample code:
int ret = recv(client, buffer_pool->get(), BUFFER_SIZE, 0);
if(ret == -1) ... // something went wrong
Well then I have to release the buffer again, and it was pretty much a waste of one buffer in my pool. (for a short time)
So isn't it possible to check the socket without calling recv()
I am not sure about the Windows, but using getsockopt() works like a charm on POSIX-compliant systems. Though before you use it - make sure that getting your buffer from the pool is more expensive than making an extra system call. Here is a code snippet:
int my_get_socket_error(int fd)
{
int err_code;
socklen_t len = sizeof(err_code);
if (getsockopt(fd, SOL_SOCKET, SO_ERROR, &err_code, &len) != 0)
err_code = errno;
else
errno = err_code;
return err_code;
}
UPDATE:
According to this document, it seems like Windows supports it too.
No, there is no way to avoid the recv() call. If select() reports that the socket is readable, then you have to read from the socket to determine its new state. If the client disconnected gracefully, recv() will return 0, not -1. If you do not want to waste a pooled buffer, then you will have to read into a temporary local buffer first, and then if recv() returns any data, you can retrieve a pooled buffer and copy the read data into it.
Calling recv and similar function does not work directly with networking devices or something similar.
When you send or receive data, all you do is questioning OS for available data, or to put data in queue for sending. Then OS will do the other job when your code is already went further.
That is why you receive errors after next call of socket function that will "contact" OS networking layers.
It is normal to get errors on that point, and you have to deal with them.
But to prevent blocking sockets and wasting buffers, check out online techniques of implementing or ready libraries that gives you asynchronous way of working with sockets, that way you don't need to define anything before socket will trigger receive callback function where you have to do actual receiving.
As well, it is not good technique to receive big amount of data in one go, because you will face problems with merged or broken apart data through TCP layer, because it is stream based layer. It is recommended to have header in you packets (few bytes) and receive them, that way you don't need pull for header, but only after header you want to read rest of message based on length provided in header. This is just possible example.
After some minutes of work and your help I just receive 1byte before receiving the full amount:
SOCKET client = ...;
char temp = 0x00;
int len = recv(client, &temp, 1, 0);
if(len == 0)
{
// .. client error handling
return;
}
char* buffer = m_memory_pool->Get();
len = recv(client, buffer + 1, m_memory_pool->buffer_size() - 1, 0);
buffer[0] = temp;
// data handling
I tried also to set a timeout for recv() but seems that under Windows it does not work, this is my code:
...
long timeout_ms = 10;
struct timeval interval = {timeout_ms / 1000, (timeout_ms % 1000) * 1000};
if (interval.tv_sec < 0 || (interval.tv_sec == 0 && interval.tv_usec <= 0))
{
interval.tv_sec = 0;
interval.tv_usec = 10000;
}
setsockopt(s_sktIx, SOL_SOCKET, SO_RCVTIMEO, (char *)&interval, sizeof(struct timeval));
...
I'm having a problem where calling recv() system call does not block. I have a client-server structure setup at the moment, and the problem I am having is I send the server one message, while the server is set up so that it's something like:
while (1) {
char buf[1024];
recv(fd, buf, sizeof(buf), flags);
processMsg(buf);
}
It receives the first message correctly, but the recv() does not block and "receives" trash data which is not what is desired. I'd like to react to messages only when they are sent. Can anyone advise?
recv() does not necessarily block until the full request is fulfilled but can return a partial request. The return code will inform you of how many bytes were actually received which can be less than you requested. Even if you specify a MSG_WAITALL flag it can return less due to a signal, etc.
On posix systems, in blocking mode recv will only block until some data is present to be read. It will then return that data, which may be less than requested, up to the amount requested. In non-blocking mode recv will return immediately if there is zero bytes of data to be read and will return -1, setting errno to EAGAIN or EWOULDBLOCK.
The upshot is that normally you will call recv in a loop until you get the amount you want while also checking for return codes of 0 (other side disconnected) or -1 (some error).
I can't speak to windows behavior.
There's two possibilities: either an error is occurring, or the socket is set to non-blocking mode. To see if an error is occurring, check the return value of recv:
while() {
char buf[1024];
int ret = recv(,buf,,)
if(ret < 0) {
// handle error
printf("recv error: %s\n", strerror(errno));
} else {
// only use the first ret bytes of buf
processMsg(buf, ret);
}
}
To put the socket into non-blocking mode, or to query if a socket is in non-blocking mode, use fcntl(2) with the O_NONBLOCK flag:
// Test if the socket is in non-blocking mode:
if(fcntl(sockfd, F_GETFL) & O_NONBLOCK) {
// socket is non-blocking
}
// Put the socket in non-blocking mode:
if(fcntl(sockfd, F_SETFL, fcntl(sockfd, F_GETFL) | O_NONBLOCK) < 0) {
// handle error
}
Note that unless you're explicitly changing the blocking behavior, the socket should be blocking by default, so most likely an error is occurring.
If you're on windows, run wsagetlasterror() function and look at the return value.
http://msdn.microsoft.com/en-us/library/ms741580%28v=vs.85%29.aspx
If you're on a posix compliant system look at errno
http://pubs.opengroup.org/onlinepubs/009695399/functions/errno.html