I want to send a zero copy in tcp. But I do not know how to send the data in the sendfile.
Please tell me how the tcp zero copy of the socket communication.
#define BUFSIZE 10240
#define NUMBER 2
int main(void)
{
struct sockaddr_in server;
int s, n;
char buf[BUFSIZE];
char ada[BUFSIZE];
int optval = 1;
int i=0;
int j;
memset((void *) ada, (int)'a', sizeof(ada));
s = socket(AF_INET, SOCK_STREAM, 0);
setsockopt(s, SOL_TCP, TCP_CORK, &optval, sizeof(int));
memset((char *)&server, 0, sizeof(server));
server.sin_family = AF_INET;
server.sin_addr.s_addr = inet_addr("127.0.0.1");
server.sin_port = htons(5320);
connect(s, (struct sockaddr *)&server, sizeof(server));
for(j = 0; j <NUMBER ; j++){
memset((void *) ada,i+64, sizeof(ada));
sendto(s, ada, strlen(ada), 0, NULL, 0); // want to send zero-copy
i++;
sleep(1);
}
optval = 0;
setsockopt(s, SOL_TCP, TCP_CORK, &optval, sizeof(int));
close(s);
exit(0);
}
Also, is there a zero-copy send in other ways?
Zero-copy means no data copying between user space and kernel space. In this program, we are copying data from the user space buffer to kernel space to transmit that data using socket.
When we need to transfer data out from a file using a socket, we can use sendfile system call, passing it the socket and file descriptor.
Please refer to the link http://www.linuxjournal.com/article/6345?page=0,0 which gives useful information on zero copy.
Related
In my code, I have this snippet:
char temp_buff[2048] = "";
strcpy(temp_buff, json_object_to_json_string(hb));
printf("%s\n", temp_buff);
char *str;
int fd = 0;
struct sockaddr_in demoserverAddr, cliaddr;
fd = socket(AF_INET, SOCK_STREAM, 0);
if (fd < 0)
{
debug_level > 0 && printf("[SKT]\tError creating socket\n");
}
else
{
demoserverAddr.sin_family = AF_INET;
demoserverAddr.sin_port = htons(9100);
demoserverAddr.sin_addr.s_addr = inet_addr("127.0.0.1");
memset(demoserverAddr.sin_zero, '\0', sizeof(demoserverAddr.sin_zero));
}
memset(&cliaddr, 0, sizeof(cliaddr));
int len=sizeof(cliaddr);
sendto(fd, temp_buff, strlen(temp_buff),MSG_CONFIRM, (const struct sockaddr *)&cliaddr, len);
On the other side, I write netcat -u -l 9100 in a terminal to see the incoming message, but nothing happens. Why?
You have two issues here.
First, by using SOCK_STREAM in the call to socket you're creating a TCP socket, but you're using sendto to and your netcat call is using the -u option indicating that you want to use UDP. So use SOCK_DGRAM instead.
Second, you're specifying cliaddr as the address to send to, but that variable was zero'ed out by memset. The demoserverAddr variable contains the IP and port of the remote server, so pass that to sendto. Also, be sure to check the return value.
int len=sizeof(demoserverAddr);
int rval = sendto(fd, temp_buff, strlen(temp_buff),MSG_CONFIRM,
(const struct sockaddr *)&demoserverAddr, len);
if (rval < 0) perror("sendto failed");
The below code is from Git. It joins a multicast group and receives packets.
Here we loop and receive the data in a buffer called msgbuf:
while (1)
{
char msgbuf[MSGBUFSIZE];
const int addrlen = sizeof(addr);
const int nbytes = recvfrom(fd, msgbuf, MSGBUFSIZE, 0, (struct sockaddr *) &addr, &addrlen);
How do I choose the size for the buffer msgBuf? Does it just have to be the max packet size? Or do I need to store multiple packets whilst I process the first?
Full code:
int main(int argc, char *argv[])
{
if (argc != 3) {
printf("Command line args should be multicast group and port\n");
printf("(e.g. for SSDP, `listener 239.255.255.250 1900`)\n");
return 1;
}
char* group = argv[1]; // e.g. 239.255.255.250 for SSDP
int port = atoi(argv[2]); // 0 if error, which is an invalid port
if(port <= 0)
{
perror("Invalid port");
return 1;
}
// create what looks like an ordinary UDP socket
//
int fd = socket(AF_INET, SOCK_DGRAM, 0);
if (fd < 0)
{
perror("socket");
return 1;
}
// allow multiple sockets to use the same PORT number
//
u_int yes = 1;
if (setsockopt(fd, SOL_SOCKET, SO_REUSEADDR, (char*) &yes, sizeof(yes)) < 0)
{
perror("Reusing ADDR failed");
return 1;
}
// set up destination address
//
struct sockaddr_in addr;
memset(&addr, 0, sizeof(addr));
addr.sin_family = AF_INET;
addr.sin_addr.s_addr = htonl(INADDR_ANY); // differs from sender
addr.sin_port = htons(port);
// bind to receive address
//
if (bind(fd, (struct sockaddr*) &addr, sizeof(addr)) < 0)
{
perror("bind");
return 1;
}
// use setsockopt() to request that the kernel join a multicast group
//
struct ip_mreq mreq;
mreq.imr_multiaddr.s_addr = inet_addr(group);
mreq.imr_interface.s_addr = htonl(INADDR_ANY);
if (setsockopt(fd, IPPROTO_IP, IP_ADD_MEMBERSHIP, (char*) &mreq, sizeof(mreq)) < 0)
{
perror("setsockopt");
return 1;
}
// now just enter a read-print loop
//
while (1)
{
char msgbuf[MSGBUFSIZE];
const int addrlen = sizeof(addr);
const int nbytes = recvfrom(fd, msgbuf, MSGBUFSIZE, 0, (struct sockaddr *) &addr, &addrlen);
if (nbytes < 0)
{
perror("recvfrom");
return 1;
}
msgbuf[nbytes] = '\0';
puts(msgbuf);
}
return 0;
}
Unlike TCP which combines packets into a stream, UDP respects packet boundaries so recvfrom only gets one packet at a time.
So MSGBUFSIZE only needs to be as big as a single packet. If you're not using jumbo packets that would be 1500, otherwise it would be 9000.
as noted by #Ingo, in this code you should be using:
char msgbuf[MSGBUFSIZE + 1];
the + 1 is because recvfrom can write upto MSGBUFSIZE bytes into the array, and you then write another NUL byte at the end.
as far as choosing a value for MSGBUFSIZE, that would depend on the protocol specification. given that most physical networks would struggle to send more than 1500 bytes without fragmentation something like 2048 might be a reasonable value. you could also check for nbytes == MSGBUFSIZE (maybe also using MSG_TRUNC) and report a "packet truncated" warning, but this basically wouldn't happen for packets routed over the public internet
in response to:
do I need to store multiple packets whilst I process the first?
you'd normally let the network stack take care of that. recv maintains packet/datagram boundaries and hence will always start writing the next packet at the supplied buffer address. again it depends on the protocol how you detect and handle errors, e.g. missing or out-of-order packets, and timeouts
I would like to build a client - server model where, for any number of clients connected to the server, the server should send a broadcast Message. Below is the code which I implemented, it does not throw any error, but I could not see any communication happening between the server and the client. Please let me know, where is the problem in the code.
I would like to see the message passed from the server on the clients system.
/************* UDP SERVER CODE *******************/
#include <stdio.h>
#include <sys/socket.h>
#include <netinet/in.h>
#include <string.h>
#include <stdlib.h>
int main(int argc, char *argv[]){
int udpSocket, nBytes, buflen;
char buffer[1024];
char string[1024];
int portNum;
struct sockaddr_in serverAddr, clientAddr;
struct sockaddr_storage serverStorage;
socklen_t addr_size, client_addr_size;
int i;
/*Create UDP socket*/
udpSocket = socket(AF_INET, SOCK_DGRAM, 0);
portNum=atoi(argv[1]);
/*Configure settings in address struct*/
serverAddr.sin_family = AF_INET;
serverAddr.sin_port = htons(5000);
serverAddr.sin_addr.s_addr = inet_addr("192.168.1.117");
memset(serverAddr.sin_zero, '\0', sizeof serverAddr.sin_zero);
/*Bind socket with address struct*/
int bStatus=bind(udpSocket, (struct sockaddr *) &serverAddr, sizeof(serverAddr));
if (bStatus < 0) {
printf("error binding on server's port number\n");
exit(1);
}
/*Initialize size variable to be used later on*/
addr_size = sizeof serverStorage;
memset(&clientAddr, 0, sizeof(clientAddr));
clientAddr.sin_family = AF_INET;
clientAddr.sin_port = htons(portNum); // this is where client is bound
clientAddr.sin_addr.s_addr = inet_addr("192.168.1.114");
while(1){
sprintf(buffer,"Hello Client");
buflen=strlen(buffer);
int bytesSent = sendto(udpSocket,buffer,buflen,0,(struct sockaddr *)&clientAddr, sizeof(clientAddr));
if (bytesSent < 0) {
printf("Error sending/ communicating with client\n");
exit(1);
}
printf("%s\n",buffer);
/*
// ****************************************************
memset(buffer, 0, 1024);
buflen=65536;
recvfrom(udpSocket,buffer,buflen,0,(struct sockaddr *)&serverStorage, &addr_size);
printf("Received from server: %s\n",buffer);
*/
}
return 0;
}
Below is the client code :
/************* UDP CLIENT CODE *******************/
#include <stdio.h>
#include <sys/socket.h>
#include <netinet/in.h>
#include <string.h>
int main(int argc, char *argv[]){
int clientSocket, portNum, nBytes, buflen;
char buffer[1024];
char string[1024];
struct sockaddr_in serverAddr,clientAddr;
socklen_t addr_size;
/*Create UDP socket*/
clientSocket = socket(AF_INET, SOCK_DGRAM, 0);
portNum = atoi(argv[1]);
/*Configure settings in address struct*/
serverAddr.sin_family = AF_INET;
serverAddr.sin_port = htons(portNum);
serverAddr.sin_addr.s_addr = inet_addr("192.168.1.117");
// memset(serverAddr.sin_zero, '\0', sizeof serverAddr.sin_zero);
memset(&serverAddr,0,sizeof(serverAddr));
// binding
bind(clientSocket, (struct sockaddr *) &clientAddr, sizeof(clientAddr));
/*Initialize size variable to be used later on*/
addr_size = sizeof serverAddr;
while(1){
memset(buffer, 0, 1024);
buflen=65536;
recvfrom(clientSocket,buffer,buflen,0,(struct sockaddr *)&serverAddr, &addr_size);
//recvfrom(clientSocket,buffer,buflen,0,NULL,NULL);
printf("Received from server: %s\n",buffer);
/*
//********************************************************************
sprintf(buffer,"Hello Client");
buflen=strlen(buffer);
sendto(clientSocket,buffer,buflen,0,(struct sockaddr *)&serverAddr,addr_size);
printf("%s\n",buffer);
*/
}
return 0;
}
At a high level there are a few things that need to be addressed. Given the code in the post has a server that initiates communication with the client following line needs to be modified (and related code for these)
On the server side:
sendto(udpSocket,buffer,buflen,0,(struct sockaddr *)&serverStorage,addr_size);
Essentially what's happening here is that udpSocket is bound to serverAddr and we are trying to sendto serverStorage (2nd last argument). The trouble here is that the serverStorage is not bound to anything. Basically sendto has the source right but not the destination.
On the client side:
recvfrom(clientSocket,buffer,buflen,0,(struct sockaddr *)&serverAddr, &addr_size);
The clientSocket is not bound to anything so the system picks a random port where it will listen. In the recvfrom call, 2nd last argument captures the address details of the other side. Whereas in the code you have bound it to server's IP and port, it does not have any effect because it is filled in by the recvfrom call and returned.
To make things work here is what I'd suggest (you can add the details but have worked out the skeleton based on what you posted)
server_initiating_with_client.c
//...
int udpSocket, nBytes, buflen;
char buffer[1024];
int portNum;
struct sockaddr_in serverAddr, clientAddr;
struct sockaddr_storage serverStorage;
socklen_t addr_size, client_addr_size;
/*Create UDP socket*/
udpSocket = socket(AF_INET, SOCK_DGRAM, 0);
portNum=atoi(argv[1]);
/*Configure settings in address struct*/
memset(&serverAddr, 0, sizeof(serverAddr));
serverAddr.sin_family = AF_INET;
serverAddr.sin_port = htons(5000); // Making sure server has a unique port number
serverAddr.sin_addr.s_addr = htonl(INADDR_ANY);
/*Bind socket with address struct*/
int bStatus = bind(udpSocket, (struct sockaddr *) &serverAddr, sizeof(serverAddr));
if (bStatus < 0) {
printf("error binding on server's port number\n");
exit(1);
}
/*Initialize size variable to be used later on*/
addr_size = sizeof serverStorage;
/* Because server is initiating the communication here one needs to set the
* clientAddr and needs to know the port number of the client
*/
memset(&clientAddr, 0, sizeof(clientAddr));
clientAddr.sin_family = AF_INET;
clientAddr.sin_port = htons(portNum);
clientAddr.sin_addr.s_addr = htonl(INADDR_ANY);
while(1){
sprintf(buffer,"Hello Client");
buflen=strlen(buffer);
int bytesSent = sendto(udpSocket,buffer,buflen, 0,
(struct sockaddr *)&clientAddr,
sizeof(clientAddr));
if (bytesSent < 0) {
printf("Error sending/ communicating with client\n");
exit(1);
}
printf("Done sending %s\n", buffer);
sleep(1);
}
..
and on the client side something along these lines
//..
int clientSocket, portNum, nBytes, buflen;
char buffer[1024];
struct sockaddr_in serverAddr;
struct sockaddr_in clientAddr;
socklen_t addr_size;
/*Create UDP socket*/
clientSocket = socket(PF_INET, SOCK_DGRAM, 0);
portNum = atoi(argv[1]);
/*Configure settings in address struct*/
memset(&clientAddr, 0, sizeof(clientAddr));
clientAddr.sin_family = AF_INET;
clientAddr.sin_port = htons(portNum);
clientAddr.sin_addr.s_addr = htonl(INADDR_ANY);
memset(&serverAddr, 0, sizeof(serverAddr));
/*Bind on client side as well because you are specifically sending something over
* to the client. Usually the server will know where to talk back to the client
* but in your example because of the reversed semantics this will be needed
*/
bind(clientSocket, (struct sockaddr *) &clientAddr, sizeof(clientAddr));
/*Initialize size variable to be used later on*/
addr_size = sizeof serverAddr;
while(1){
memset(buffer, 0, 1024);
buflen=65536;
recvfrom(clientSocket,buffer,buflen,0,(struct sockaddr *)&serverAddr, &addr_size);
printf("Received from server: %s\n",buffer);
sleep(1);
}
return 0;
The code can be simplified if you reverse your comm from client to server, thereby, eliminating the need for a fixed port number on the client side.
just make a few changes in it. Assign a socket length predefined (socklen_t *) structure to typecast length
n = recvfrom(sockfd, (char *)buffer, MAXLINE,MSG_WAITALL, (struct sockaddr *) &servaddr, (socklen_t*)&len);
socklen_t* convert a length into socket length.
Make these changes at both client-side as well as a server-side
.
I'm writing an application that listens for UDP packets over a unix domain socket. Consider the following code block.
int sockfd;
struct sockaddr_un servaddr;
sockfd = socket(AF_LOCAL, SOCK_DGRAM, 0);
if(sockfd < 0)
{
perror("socket() failed");
}
unlink(port);
bzero(&servaddr, sizeof(servaddr));
servaddr.sun_family = AF_LOCAL;
strcpy(servaddr.sun_path, port);
if(bind(sockfd, (struct sockaddr *)&servaddr, sizeof(servaddr)) < 0)
{
perror("bind() failed");
close(sockfd);
}
int n;
struct sockaddr_un cliaddr;
socklen_t len = sizeof(cliaddr);
discovery_msgs client_message;
bzero(&client_message, sizeof(client_message));
// Wait for a message to be received
n = recvfrom(sock_fd, &client_message, sizeof(client_message), 0,
(struct sockaddr *) &cliaddr, &len);
// At this point n = 560, client_message is filled with the expected data
//len = 0 and cliaddr has no information about the client that sent the data
Now the type of client_message isn't really important, I'm receiving a UDP packet and client_message contains all of the data I expect. The problem begins when I look at cliaddr and len after calling recvfrom. cliaddr is not modified by recvfrom like it normally is with normal network TCP/UDP and len is set to 0 after the call(which means recvfrom wrote no data to &cliaddr). I need the information in cliaddr to be populated with the unix domain path so I can send a response.
What am I doing wrong?
The solution is binding the socket on the client side when using Unix domain sockets. Otherwise the transient pathname created for sending the UDP packet immediately disappears after sendto(), which explains why the client's address information is not available on the server side.
See Stevens Network Programming page 419 or see this for an example client implementation that solves this issue: libpix.org/unp/unixdgcli01_8c_source.html
#include "unp.h"
int
main(int argc, char **argv)
{
int sockfd;
struct sockaddr_un cliaddr, servaddr;
sockfd = Socket(AF_LOCAL, SOCK_DGRAM, 0);
bzero(&cliaddr, sizeof(cliaddr)); /* bind an address for us */
cliaddr.sun_family = AF_LOCAL;
strcpy(cliaddr.sun_path, tmpnam(NULL));
Bind(sockfd, (SA *) &cliaddr, sizeof(cliaddr));
bzero(&servaddr, sizeof(servaddr)); /* fill in server's address */
servaddr.sun_family = AF_LOCAL;
strcpy(servaddr.sun_path, UNIXDG_PATH);
dg_cli(stdin, sockfd, (SA *) &servaddr, sizeof(servaddr));
exit(0);
}
Note: unp.h defines Bind() which is simply bind() with some error checking(commonly used throughout Stevens Network Programming). In the same manner, (SA *) is the equivalent to (struct sockaddr *).
I have two server codes:
the first server: send the client a char each time until the string is finished
int
main(int argc, char **argv)
{
int listenfd, connfd;
struct sockaddr_in servaddr;
char buff[MAXLINE];
time_t ticks;
char temp[1];
int i = 0;
listenfd = Socket(AF_INET, SOCK_STREAM, 0);
bzero(&servaddr, sizeof(servaddr));
servaddr.sin_family = AF_INET;
servaddr.sin_addr.s_addr = htonl(INADDR_ANY);
servaddr.sin_port = htons(9999); /* daytime server */
Bind(listenfd, (SA *) &servaddr, sizeof(servaddr));
Listen(listenfd, LISTENQ);
for ( ; ; ) {
connfd = Accept(listenfd, (SA *) NULL, NULL);
ticks = time(NULL);
snprintf(buff, sizeof(buff), "%.24s\r\n", ctime(&ticks));
for(i = 0; i < strlen(buff); i++)
{
temp[0] = buff[i];
Write(connfd, temp, strlen(temp));
}
Close(connfd);
}
}
the second server: send the client a string
int
main(int argc, char **argv)
{
int listenfd, connfd;
struct sockaddr_in servaddr;
char buff[MAXLINE];
time_t ticks;
char temp[1];
int i = 0;
listenfd = Socket(AF_INET, SOCK_STREAM, 0);
bzero(&servaddr, sizeof(servaddr));
servaddr.sin_family = AF_INET;
servaddr.sin_addr.s_addr = htonl(INADDR_ANY);
servaddr.sin_port = htons(9999); /* daytime server */
Bind(listenfd, (SA *) &servaddr, sizeof(servaddr));
Listen(listenfd, LISTENQ);
for ( ; ; ) {
connfd = Accept(listenfd, (SA *) NULL, NULL);
ticks = time(NULL);
snprintf(buff, sizeof(buff), "%.24s\r\n", ctime(&ticks));
Write(connfd, buff, strlen(buff));
Close(connfd);
}
}
the client:receive the chars sent by the server
int
main(int argc, char **argv)
{
int sockfd, n;
char recvline[MAXLINE + 1];
struct sockaddr_in servaddr;
int count = 0;
if (argc != 2)
err_quit("usage: a.out <IPaddress>");
if ( (sockfd = socket(AF_INET, SOCK_STREAM, 0)) < 0)
err_sys("socket error");
bzero(&servaddr, sizeof(servaddr));
servaddr.sin_family = AF_INET;
servaddr.sin_port = htons(9999); /* daytime server */
if (inet_pton(AF_INET, argv[1], &servaddr.sin_addr) <= 0)
err_quit("inet_pton error for %s", argv[1]);
if (connect(sockfd, (SA *) &servaddr, sizeof(servaddr)) < 0)
err_sys("connect error");
while ( (n = read(sockfd, recvline, MAXLINE)) > 0) {
recvline[n] = 0; /* null terminate */
count++;
if (fputs(recvline, stdout) == EOF)
err_sys("fputs error");
}
if (n < 0)
err_sys("read error");
printf("read time:%d\n", count);
exit(0);
}
the result is both of the output of variable count is 1. My question is why the first server's output is 1, I think the result should be strlen(buff) for the 1st server?
PS:I run the server and client on the same machine.
TCP is a stream protocol. As such the number of writes on one side will not cause the same amount of reads on the other side since the protocol doesn't preserve information about how the writes into the socket were made.
Usually, on the sender side there's a delay before a packet is sent in case you write more data to the socket so that more data can be stuffed into the same packet. One of the reasons for it is that a badly written server might flood the network with single byte packets.
On the receiver side, the protocol doesn't know why your data might have arrived as separate packets, it might have been split up because of the MTU, it might have been reassembled by some packet inspection software or appliance on the way, so whenever you read from your socket it will give you as much data as it can regardless of how it was sent to you.
On a local machine like in your setup it's likely that the client isn't even running while the server is writing, so even without buffering on the sender side it will not start reading until the server has written everything and therefore it will read everything in one go. Or not, you might be unlucky, your server gets preempted for long enough that the TCP implementation in your kernel thinks that there won't be any more data you'll be sending, send a single byte to the client, the client gets scheduled to run before the server runs again and the client will receive just one byte in the first read.