HTTP/1.0 Web Server
I wrote this program around April, 2006 for inet4021 - Network Programming. The requirement was to create a web server adhering to RFC1945 (HTTP/1.0 specification) and achieving concurrency. I finished early, so I decided to add thread-safe logging (which, interestingly, led me to find out that my web server, running on my home connection on port 5555, got crawled by msnbot within a few hours of going online) and a chrooted environment (when spanwed as root, and then also suid’s to ‘nobody’).
The full project, including a README, makefile, and sample public_html directory is available here, including a bzipped tarball of all files. The source code is also provided below (reformatted from the tar’d copy for better 80-column compliance).
/* David R. Hedges * 2836226 * hedg0029 * 2006-04-18 * inet4021 */ #define _REENTRANT #include <errno.h> #include <stdio.h> #include <pthread.h> #include <stdlib.h> #include <string.h> #include <fcntl.h> #include <unistd.h> #include <ctype.h> #include <netinet/in.h> #include <sys/socket.h> #include <sys/types.h> #include <arpa/inet.h> #include <netdb.h> #define QUEUE_LENGTH 50 /* using a static-sized array for the 'queue' // instead of a linked list. this is the size // of that array (number of simultaneous // pending requests) */ #define NUM_THREADS 10 // defines how many worker threads to spawn static int reqBuf[QUEUE_LENGTH]; /* the 'queue' to hold sockfds of accepted clients */ static char addrBuf[QUEUE_LENGTH][16]; // put the IP of the client in here static int bufHead = 0; /* the 'front' of the queue (used in reqBuf array) */ static int bufTail = 0; /* the 'back' of the queue (new entries get added in bufTail++ */ static int log_fd; // file descriptor of the logfile static int threadCounter = 0; /* used for threads to be able to identify themselves during instantiation */ int debug = 0; /* variable used for determining whether or not to display debugging messages, set to 1 with -v flag */ int sockfd; /* globally visible sockfd for use in passing a socket fd around */ int done = 0; /* used for threads to figure out if execution is done */ char *errorpage = "<!DOCTYPE HTML PUBLIC \"-//IETF//DTD HTML 2.0//EN\">\n" "<html><head>\n<title>404 Not Found</title>\n</head><body>\n" "<h1>Not Found</h1>\n" "<p>The requested URL was not found on this server.</p>\n" "<hr>\n<address>p14serv/0.1</address>\n</body></html>"; char *serverName = "p14serv/0.1"; char *errorpage400 = "<!DOCTYPE HTML PUBLIC \"-//IETF//DTD HTML 2.0//EN\">\n" "<html><head>\n" "<title>400 Bad Request</title>\n" "</head><body>\n<h1>Bad Request</h1>\n" "<p>The method you used to request this page is not " "supported by this server. Please restrict your reqeuests " "to GET or HEAD request-methods.</p>\n<hr>\n" "<address>p14serv/0.1</address>\n</body></html>"; static pthread_mutex_t bufAccess = PTHREAD_MUTEX_INITIALIZER; static pthread_mutex_t logAccess = PTHREAD_MUTEX_INITIALIZER; static pthread_mutex_t doneAccess = PTHREAD_MUTEX_INITIALIZER; static pthread_cond_t bufAvail = PTHREAD_COND_INITIALIZER; //prototypes: int checkVerbose(int, char**); int checkPort(int, char**); int checkPath(int, char**); void *readerThread (void *arg); void *workerThread (void *arg); int main(int argc, char **argv) { int i; int ret_val; int port_number; char *errmsg; pthread_t reader; pthread_t workers[NUM_THREADS]; struct sockaddr_in serv_addr; char writeMsg[1024]; //create a new log file every time (open it before we chdir) //if ((log_fd = creat("httpserv_log", 0666)) < 0) { if ((log_fd = open("httpserv_log", O_WRONLY|O_CREAT|O_APPEND, 0666)) < 0) { errmsg = strerror(errno); printf("(%s:%d): Unable to create httpserv_log: %s.\n", __FILE__, __LINE__, errmsg); //return -1; } if(argc == 1) { //no other command-line optios supplied, so set defaults port_number = 80; } else if(argc > 1) { //don't know whether flag is -v, port, or path if (strlen(argv[1]) >= 6 && strncmp(argv[1], "--help", 6) == 0) { printf("Usage is: ./httpserv [-v] [port] [path]\n"); return 0; } debug = checkVerbose(argc, argv); port_number = checkPort(argc, argv); ret_val = checkPath(argc, argv); if (ret_val == -1) { printf("Usage is: ./httpserv [-v] [port] [path]\n"); return -1; } if (debug == 0 && port_number == 0 && ret_val == 0) { printf("(%s:%d): Argument '%s' not understood.\n" "Usage is: ./httpserv [-v] [port] [path]\n", __FILE__, __LINE__, argv[1]); return 0; } if (port_number > 65535) { printf("Port number must be in the range [1-65535].\n"); return -1; } if (port_number == 0) { port_number = 80; } } //set up socket to listen; will be used by reader thread sockfd = socket(AF_INET, SOCK_STREAM, 0); if (sockfd < 0) { errmsg = strerror(errno); printf("(%s:%d): Unable to create initial socket: %s.\n", __FILE__, __LINE__, errmsg); return -1; } if (debug) { printf("socket() call completed\n"); } memset((char *) &serv_addr, '0', sizeof(serv_addr)); serv_addr.sin_family = AF_INET; serv_addr.sin_addr.s_addr = INADDR_ANY; serv_addr.sin_port = htons(port_number); if (bind(sockfd, (struct sockaddr *) &serv_addr, sizeof(serv_addr)) < 0) { errmsg = strerror(errno); printf("(%s:%d): Unable to bind socket: %s.\n", __FILE__, __LINE__, errmsg); return -1; } if (listen(sockfd,10) < 0) { errmsg = strerror(errno); printf("(%s:%d): Unable to listen() on socket: %s.\n", __FILE__, __LINE__, errmsg); return -1; } //if running as root, change to nobody now that we've created the socket and chrooted if (debug) { printf("(%s:%d#m): getuid() was %d, ", __FILE__, __LINE__, getuid()); } if (getuid() == 0) setuid(65534); if (debug) { printf("now it's %d.\n", getuid()); } //display our info: char hostname[128]; gethostname(hostname, 128); if (port_number == 80) { printf("\nWeb server running at:\thttp://%s/\n\n" "Press ctrl-c to terminate.\n", hostname); } else { printf("\nWeb server running at:\thttp://%s:%d/\n\n" "Press ctrl-c to terminate.\n", hostname, port_number); } //log startup time_t curtime = time(NULL); char timestamp[25]; strftime(timestamp, 25, "%Y-%m-%d %H:%M:%S%z", localtime(&curtime)); snprintf(writeMsg, 1024, "%s - %s started, listening on %s:%d\n", timestamp, serverName, hostname, port_number); //no other threads, so can log without risk of interruption: write(log_fd, writeMsg, strlen(writeMsg)); //spawn reader/listener thread if (pthread_create(&reader,NULL,readerThread,NULL) != 0) { errmsg = strerror(errno); printf("(%s:%d): Unable to spawn readerThread: %s.\n", __FILE__, __LINE__, errmsg); return -1; } for (i=0; i<NUM_THREADS; i++) { if(pthread_create(&workers[i],NULL,workerThread,NULL) != 0) { errmsg = strerror_r(errno, writeMsg, 1024); printf("(%s:%d#%d): Error creating workerThread(%d): %s.\n", __FILE__, __LINE__, i, i, errmsg); } else if(debug) { printf("(%s:%d#m): Created workerThread(%d).\n", __FILE__, __LINE__, i); } } //la de da, threads do things, now let's join them //join the reader thread if (pthread_join(reader,NULL) != 0) { errmsg = strerror_r(errno, writeMsg, 1024); printf("(%s:%d): Error re-joining reader/listener thread: %s.\n", __FILE__, __LINE__, errmsg); return -1; } for (i=0; i<NUM_THREADS; i++) { if(pthread_join(workers[i],NULL) != 0) { errmsg = strerror_r(errno, writeMsg, 1024); printf("(%s:%d): Error re-joining worker thread %d: %s.\n", __FILE__, __LINE__, i, errmsg); return -1; } } if (close(log_fd) != 0) { perror("Error closing log file"); return -1; } return 0; } // usage is: ./httpserv [-v] [port] [directory] int checkVerbose(int argc, char **argv) { int i; for(i=1; i<argc; i++) { if(strlen(argv[i]) >= 2 && strncmp(argv[i], "-v", 2) == 0) { return 1; } } return 0; } int checkPort(int argc, char **argv) { int i, p; for(i=1; i<argc; i++) { p = atoi(argv[i]); if (p>0) { return p; } } return 0; } int checkPath(int argc, char **argv) { int i, ret_val; for(i=1; i<argc; i++) { #ifdef DEBUG printf("atoi(%s) = %d, strlen(%s) = %d, strncmp(%s, \"-v\", 2) = %d)\n", argv[i], atoi(argv[i]), argv[i], strlen(argv[i]), argv[i], strncmp(argv[i], "-v", 2)); #endif if((!atoi(argv[i])) && (strlen(argv[i]) >= 2) && strncmp(argv[i], "-v", 2) != 0) { //path declaration found, now check if it exists ret_val = chroot(argv[i]); #ifdef DEBUG printf("(%s:%d): chroot(%s) returns %d.\n", __FILE__, __LINE__, argv[i], ret_val); #endif if(ret_val < 0) { perror("Sad chrooting"); } if((ret_val = chdir(argv[i])) == 0) { #ifdef DEBUG printf("ret_val is %d\n", ret_val); printf("getuid = %d\n", getuid()); #endif return 1; } else { //path not found char *errmsg = strerror(errno); printf("(%s:%d): Specificed path '%s' not found: %s.\n", __FILE__, __LINE__, argv[i], errmsg); return -1; } } } return 0; } void *readerThread (void *arg) { struct sockaddr_in cli_addr; int tempsockfd; char *errmsg; char writeMsg[1024]; unsigned int clilen = sizeof(cli_addr); while (1) { tempsockfd = accept(sockfd, (struct sockaddr *) &cli_addr, &clilen); if (tempsockfd < 0) { errmsg = strerror_r(errno, writeMsg, 1024); snprintf(writeMsg, 1024, "[reader]: Error accepting connection '%s'\n", errmsg); pthread_mutex_lock(&logAccess); write(log_fd, writeMsg, strlen(writeMsg)); pthread_mutex_unlock(&logAccess); } else { if (debug) { printf("(%s:%d#r): Accepted connection %d from %s:%d\n", __FILE__, __LINE__, tempsockfd, inet_ntoa(cli_addr.sin_addr), ntohs(cli_addr.sin_port)); #ifdef DEBUG struct hostent *george; george = gethostbyaddr((void *) &cli_addr.sin_addr, clilen, AF_INET); if(george->h_name != NULL) { printf("hostname is '%s'\n", (char *) george->h_name); } else { printf("h_addr is null?\n"); } #endif } //put the temp sockfd into the shared buffer for a worker thread to pick up pthread_mutex_lock(&bufAccess); if (bufTail == QUEUE_LENGTH) bufTail = 0; //wrap around to beginning strncpy(addrBuf[bufTail], inet_ntoa(cli_addr.sin_addr), 16); reqBuf[bufTail++] = tempsockfd; pthread_mutex_unlock(&bufAccess); pthread_cond_signal(&bufAvail); if(debug) { printf("(%s:%d#r): Signaled bufAvail.\n", __FILE__, __LINE__); } } } if(debug) { printf("(%s:%d#r): Exiting reader thread.\n", __FILE__, __LINE__); } //"signal" (not exactly) the other threads that we're done pthread_mutex_lock(&doneAccess); done = 1; pthread_mutex_unlock(&doneAccess); //broadcast this, so all the threads wake up and finish //what they ahve to do, then exit pthread_cond_broadcast(&bufAvail); if (debug) { printf("(%s:%d#r): Returning from reader thread.\n", __FILE__, __LINE__); } return NULL; } void *workerThread (void *arg) { pthread_mutex_lock(&logAccess); int thread_id = threadCounter++; pthread_mutex_unlock(&logAccess); int ret_val, status_code; int sockfd, filefd, filesize; char *errmsg, *stringpos, *tok; char writemsg[1024], readmsg[5120]; char cli_addr[16]; char timestamp[25]; char method[5]; //will contain HEAD or GET (ideally) char filename[1024]; //will not contain leading / char content_type[1024]; char response_header[5120]; char user_agent[1024] = "-"; char request_line[1024]; if(debug) { printf("(%s:%d#%d): Spawning worker thread %d.\n", __FILE__, __LINE__, thread_id, thread_id); } while(1) { //hack to give another thread a chance to come and get the reqest sched_yield(); pthread_mutex_lock(&bufAccess); while (bufHead == bufTail && !done) {// bufEmpty() //sleep until there's something in the buffer pthread_cond_wait(&bufAvail, &bufAccess); } if(debug) { printf("(%s:%d#%d): Received bufAvail cond_signal.\n", __FILE__, __LINE__, thread_id); } if(bufHead == bufTail) { if(done) { pthread_mutex_unlock(&bufAccess); pthread_exit(NULL); } } //get our sockfd if (bufHead == QUEUE_LENGTH) bufHead = 0; strncpy(cli_addr, addrBuf[bufHead], 16); sockfd = reqBuf[bufHead++]; pthread_mutex_unlock(&bufAccess); if(debug) { printf("(%s:%d#%d): Picked up socket %d.\n", __FILE__, __LINE__, thread_id, sockfd); } memset(readmsg, '\0', 5120); stringpos = readmsg; while (strstr(readmsg, "\r\n\r\n") == NULL) { ret_val = read(sockfd, stringpos, 5120-(stringpos-readmsg)); if (ret_val <= 0) { errmsg = strerror_r(errno, writemsg, 1024); printf("(%s:%d#%d): No data read: %s.\n", __FILE__, __LINE__, thread_id, errmsg); close(sockfd); } stringpos += ret_val; } char tempString[5120]; strncpy(tempString, readmsg, 5120); time_t curtime = time(NULL); strftime(timestamp, 28, "%Y-%m-%d %H:%M:%S%z", localtime(&curtime)); tok = strtok(tempString, "\r\n"); strncpy(request_line, tok, 1024); if(debug && tok != NULL) { printf("(%s:%d#%d): [%s] Request is: \"%s\"\n", __FILE__, __LINE__, thread_id, cli_addr, tok); } #ifdef DEBUG snprintf(writemsg, 1024, "%s - [%s] - \"%s\"\n", timestamp, cli_addr, tok); pthread_mutex_lock(&logAccess); write(log_fd, writemsg, strlen(writemsg)); pthread_mutex_unlock(&logAccess); #endif //copy the request line to tempstring, and only parse that strncpy(tempString, tok, 1024); tok = strtok(tempString, " "); //copy GET or HEAD to method if (tok != NULL && (strlen(tok) >=3)) { strncpy(method, tok, 5); } else { snprintf(response_header, 5120, "HTTP/1.0 400 Bad Request\r\n" "Server: %s\r\n" "Allow: GET,HEAD\r\n" "Content-Length: %d\r\n" "Content-Type: %s\r\n\r\n", serverName, strlen(errorpage400), content_type); write(sockfd, response_header, strlen(response_header)); write(sockfd, errorpage400, strlen(errorpage400)); close(sockfd); snprintf(writemsg, 1024, "%s - [%s] - \"%s\" 400 %d \"-\" \"-\"\n", timestamp, cli_addr, request_line, strlen(errorpage400)); pthread_mutex_lock(&logAccess); write(log_fd, writemsg, strlen(writemsg)); pthread_mutex_unlock(&logAccess); if (debug) { //kill newline from log writemsg writemsg[strlen(writemsg)-1] = '\0'; printf("(%s:%d#%d): Wrote to logfile: '%s'\n", __FILE__, __LINE__, thread_id, writemsg); } continue; } //should be filename or full url tok = strtok(NULL, " "); if (strncasecmp(tok, "http://", 7) == 0) { stringpos = index(tok+7, '/'); //skips the http://, and finds a pointer to the next / if (stringpos == NULL) { strncpy(filename, "index.html", 11); } else { strncpy(filename, stringpos+1, 1024); } } else { if (tok[0] == '/') { strncpy(filename, tok+1, 1024); } else { strncpy(filename, tok, 1024); } } if (debug) { printf("(%s:%d#%d): Filename is '%s'\n", __FILE__, __LINE__, thread_id, filename); } if ((filefd = open(filename, O_RDONLY)) <= 0) { errmsg = strerror_r(errno, writemsg, 1024); if (debug) { printf("(%s:%d#%d): Error opening file '%s'\n", __FILE__, __LINE__, thread_id, filename); } status_code = 404; } else { filesize = lseek(filefd, 0, SEEK_END); lseek(filefd, 0, SEEK_SET); status_code = 200; } //get user-agent: strncpy(tempString, readmsg, 5120); tok = strtok(tempString, "\r\n"); while (strncasecmp(tok, "user-agent", 10) != 0 && tok != NULL) { //printf("tok is '%s'\n", tok); tok = strtok(NULL, "\r\n"); } if (strncasecmp(tok, "user-agent", 10) == 0) { //user-agent found strncpy(user_agent, index(tok, ' ')+1, 1024); } if (status_code == 200) { //determine file type for use in content-type if (strncasecmp(rindex(filename, '.'), ".htm", 4) == 0) { strncpy(content_type, "text/html", 10); if(debug) { printf("(%s:%d#%d): content-type is text/html\n", __FILE__, __LINE__, thread_id); } } else if (strncasecmp(rindex(filename, '.'), ".jpg", 4) == 0) { strncpy(content_type, "image/jpeg", 11); } else if (strncasecmp(rindex(filename, '.'), ".gif", 4) == 0) { strncpy(content_type, "image/gif", 10); } else { strncpy(content_type, "text/plain", 11); } } else { strncpy(content_type, "text/html", 10); } if(status_code == 404) { snprintf(response_header, 5120, "HTTP/1.0 404 Not Found\r\n" "Server: %s\r\n" "Content-Length: %d\r\n" "Content-Type: %s\r\n\r\n", serverName, strlen(errorpage), content_type); write(sockfd, response_header, strlen(response_header)); if (strncmp(method, "GET", 4) == 0) { write(sockfd, errorpage, strlen(errorpage)); } filesize = strlen(errorpage); } else if(status_code == 200) { snprintf(response_header, 5120, "HTTP/1.0 200 OK\r\n" "Server: %s\r\n" "Content-Length: %d\r\n" "Content-Type: %s\r\n\r\n", serverName, filesize, content_type); write(sockfd, response_header, strlen(response_header)); if (strncmp(method, "GET", 4) == 0) { ssize_t readbytes, writebytes; while ((readbytes = read(filefd, writemsg, 1024)) > 0) { //currently assuming that we'll only get <= 0 when EOF arrives, not an error stringpos = writemsg; while ((writebytes = write(sockfd, stringpos, readbytes)) < readbytes) { if (debug) { printf("(%s:%d#%d): Writebytes was %d.\n", __FILE__, __LINE__, thread_id, writebytes); } //we didn't get to write the whole buffer to the new file, so we need to write the rest readbytes -= writebytes; stringpos += writebytes; }// done writing the buffer to the socket if (debug) { printf("(%s:%d#%d): Wrote %d bytes to sockfd.\n", __FILE__, __LINE__, thread_id, writebytes); } } } close(filefd); } //log the request curtime = time(NULL); strftime(timestamp, 28, "%Y-%m-%d %H:%M:%S%z", localtime(&curtime)); snprintf(writemsg, 1024, "%s - [%s] - \"%s\" %d %d \"-\" \"%s\"\n", timestamp, cli_addr, request_line, status_code, filesize, user_agent); pthread_mutex_lock(&logAccess); write(log_fd, writemsg, strlen(writemsg)); pthread_mutex_unlock(&logAccess); if (debug) { //kill the newline in the write msg writemsg[strlen(writemsg)-1] = '\0'; printf("(%s:%d#%d): Wrote to logfile: '%s'\n", __FILE__, __LINE__, thread_id, writemsg); } close(sockfd); } return NULL; }
Download this code: 20060419-1746-drh-httpserv/httpserv.c
