/* * Copyright (c) 2002 - 2005 NetGroup, Politecnico di Torino (Italy) * Copyright (c) 2005 - 2008 CACE Technologies, Davis (California) * All rights reserved. * * Redistribution and use in source and binary forms, with or without * modification, are permitted provided that the following conditions * are met: * * 1. Redistributions of source code must retain the above copyright * notice, this list of conditions and the following disclaimer. * 2. Redistributions in binary form must reproduce the above copyright * notice, this list of conditions and the following disclaimer in the * documentation and/or other materials provided with the distribution. * 3. Neither the name of the Politecnico di Torino, CACE Technologies * nor the names of its contributors may be used to endorse or promote * products derived from this software without specific prior written * permission. * * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. * */ #ifdef HAVE_CONFIG_H #include "config.h" #endif #include <string.h> /* for strlen(), ... */ #include <stdlib.h> /* for malloc(), free(), ... */ #include <stdarg.h> /* for functions with variable number of arguments */ #include <errno.h> /* for the errno variable */ #include "pcap-int.h" #include "pcap-rpcap.h" #include "sockutils.h" /* * \file pcap-rpcap.c * * This file keeps all the new funtions that are needed for the RPCAP protocol. * Almost all the pcap functions need to be modified in order to become compatible * with the RPCAP protocol. However, you can find here only the ones that are completely new. * * This file keeps also the functions that are 'private', i.e. are needed by the RPCAP * protocol but are not exported to the user. * * \warning All the RPCAP functions that are allowed to return a buffer containing * the error description can return max PCAP_ERRBUF_SIZE characters. * However there is no guarantees that the string will be zero-terminated. * Best practice is to define the errbuf variable as a char of size 'PCAP_ERRBUF_SIZE+1' * and to insert manually a NULL character at the end of the buffer. This will * guarantee that no buffer overflows occur even if we use the printf() to show * the error on the screen. */ #define PCAP_STATS_STANDARD 0 /* Used by pcap_stats_remote to see if we want standard or extended statistics */ #define PCAP_STATS_EX 1 /* Used by pcap_stats_remote to see if we want standard or extended statistics */ /* Keeps a list of all the opened connections in the active mode. */ struct activehosts *activeHosts; /* * Private data for capturing on WinPcap devices. */ struct pcap_win { int nonblock; int rfmon_selfstart; /* a flag tells whether the monitor mode is set by itself */ int filtering_in_kernel; /* using kernel filter */ #ifdef HAVE_DAG_API int dag_fcs_bits; /* Number of checksum bits from link layer */ #endif }; /**************************************************** * * * Locally defined functions * * * ****************************************************/ static int rpcap_checkver(SOCKET sock, struct rpcap_header *header, char *errbuf); static struct pcap_stat *rpcap_stats_remote(pcap_t *p, struct pcap_stat *ps, int mode); static int pcap_pack_bpffilter(pcap_t *fp, char *sendbuf, int *sendbufidx, struct bpf_program *prog); static int pcap_createfilter_norpcappkt(pcap_t *fp, struct bpf_program *prog); static int pcap_updatefilter_remote(pcap_t *fp, struct bpf_program *prog); static int pcap_setfilter_remote(pcap_t *fp, struct bpf_program *prog); static int pcap_setsampling_remote(pcap_t *p); /**************************************************** * * * Function bodies * * * ****************************************************/ /* * \ingroup remote_pri_func * * \brief It traslates (i.e. de-serializes) a 'sockaddr_storage' structure from * the network byte order to the host byte order. * * It accepts a 'sockaddr_storage' structure as it is received from the network and it * converts it into the host byte order (by means of a set of ntoh() ). * The function will allocate the 'sockaddrout' variable according to the address family * in use. In case the address does not belong to the AF_INET nor AF_INET6 families, * 'sockaddrout' is not allocated and a NULL pointer is returned. * This usually happens because that address does not exist on the other host, so the * RPCAP daemon sent a 'sockaddr_storage' structure containing all 'zero' values. * * \param sockaddrin: a 'sockaddr_storage' pointer to the variable that has to be * de-serialized. * * \param sockaddrout: a 'sockaddr_storage' pointer to the variable that will contain * the de-serialized data. The structure returned can be either a 'sockaddr_in' or 'sockaddr_in6'. * This variable will be allocated automatically inside this function. * * \param errbuf: a pointer to a user-allocated buffer (of size PCAP_ERRBUF_SIZE) * that will contain the error message (in case there is one). * * \return '0' if everything is fine, '-1' if some errors occurred. Basically, the error * can be only the fact that the malloc() failed to allocate memory. * The error message is returned in the 'errbuf' variable, while the deserialized address * is returned into the 'sockaddrout' variable. * * \warning This function supports only AF_INET and AF_INET6 address families. * * \warning The sockaddrout (if not NULL) must be deallocated by the user. */ int rpcap_deseraddr(struct sockaddr_storage *sockaddrin, struct sockaddr_storage **sockaddrout, char *errbuf) { /* Warning: we support only AF_INET and AF_INET6 */ if (ntohs(sockaddrin->ss_family) == AF_INET) { struct sockaddr_in *sockaddr; sockaddr = (struct sockaddr_in *) sockaddrin; sockaddr->sin_family = ntohs(sockaddr->sin_family); sockaddr->sin_port = ntohs(sockaddr->sin_port); (*sockaddrout) = (struct sockaddr_storage *) malloc(sizeof(struct sockaddr_in)); if ((*sockaddrout) == NULL) { pcap_snprintf(errbuf, PCAP_ERRBUF_SIZE, "malloc() failed: %s", pcap_strerror(errno)); return -1; } memcpy(*sockaddrout, sockaddr, sizeof(struct sockaddr_in)); return 0; } if (ntohs(sockaddrin->ss_family) == AF_INET6) { struct sockaddr_in6 *sockaddr; sockaddr = (struct sockaddr_in6 *) sockaddrin; sockaddr->sin6_family = ntohs(sockaddr->sin6_family); sockaddr->sin6_port = ntohs(sockaddr->sin6_port); sockaddr->sin6_flowinfo = ntohl(sockaddr->sin6_flowinfo); sockaddr->sin6_scope_id = ntohl(sockaddr->sin6_scope_id); (*sockaddrout) = (struct sockaddr_storage *) malloc(sizeof(struct sockaddr_in6)); if ((*sockaddrout) == NULL) { pcap_snprintf(errbuf, PCAP_ERRBUF_SIZE, "malloc() failed: %s", pcap_strerror(errno)); return -1; } memcpy(*sockaddrout, sockaddr, sizeof(struct sockaddr_in6)); return 0; } /* It is neither AF_INET nor AF_INET6 */ *sockaddrout = NULL; return 0; } /* \ingroup remote_pri_func * * \brief It reads a packet from the network socket. This does not make use of * callback (hence the "nocb" string into its name). * * This function is called by the several pcap_next_ex() when they detect that * we have a remote capture and they are the client side. In that case, they need * to read packets from the socket. * * Parameters and return values are exactly the same of the pcap_next_ex(). * * \warning By choice, this function does not make use of semaphores. A smarter * implementation should put a semaphore into the data thread, and a signal will * be raised as soon as there is data into the socket buffer. * However this is complicated and it does not bring any advantages when reading * from the network, in which network delays can be much more important than * these optimizations. Therefore, we chose the following approach: * - the 'timeout' chosen by the user is split in two (half on the server side, * with the usual meaning, and half on the client side) * - this function checks for packets; if there are no packets, it waits for * timeout/2 and then it checks again. If packets are still missing, it returns, * otherwise it reads packets. */ static int pcap_read_nocb_remote(pcap_t *p, struct pcap_pkthdr **pkt_header, u_char **pkt_data) { struct rpcap_header *header; /* general header according to the RPCAP format */ struct rpcap_pkthdr *net_pkt_header; /* header of the packet */ char netbuf[RPCAP_NETBUF_SIZE]; /* size of the network buffer in which the packet is copied, just for UDP */ uint32 totread; /* number of bytes (of payload) currently read from the network (referred to the current pkt) */ int nread; int retval; /* generic return value */ /* Structures needed for the select() call */ fd_set rfds; /* set of socket descriptors we have to check */ struct timeval tv; /* maximum time the select() can block waiting for data */ struct pcap_md *md; /* structure used when doing a remote live capture */ md = (struct pcap_md *) ((u_char*)p->priv + sizeof(struct pcap_win)); /* * Define the read timeout, to be used in the select() * 'timeout', in pcap_t, is in milliseconds; we have to convert it into sec and microsec */ tv.tv_sec = p->opt.timeout / 1000; tv.tv_usec = (p->opt.timeout - tv.tv_sec * 1000) * 1000; /* Watch out sockdata to see if it has input */ FD_ZERO(&rfds); /* * 'fp->rmt_sockdata' has always to be set before calling the select(), * since it is cleared by the select() */ FD_SET(md->rmt_sockdata, &rfds); retval = select((int) md->rmt_sockdata + 1, &rfds, NULL, NULL, &tv); if (retval == -1) { sock_geterror("select(): ", p->errbuf, PCAP_ERRBUF_SIZE); return -1; } /* There is no data waiting, so return '0' */ if (retval == 0) return 0; /* * data is here; so, let's copy it into the user buffer. * I'm going to read a new packet; so I reset the number of bytes (payload only) read */ totread = 0; /* * We have to define 'header' as a pointer to a larger buffer, * because in case of UDP we have to read all the message within a single call */ header = (struct rpcap_header *) netbuf; net_pkt_header = (struct rpcap_pkthdr *) (netbuf + sizeof(struct rpcap_header)); if (md->rmt_flags & PCAP_OPENFLAG_DATATX_UDP) { /* Read the entire message from the network */ if (sock_recv(md->rmt_sockdata, netbuf, RPCAP_NETBUF_SIZE, SOCK_RECEIVEALL_NO, p->errbuf, PCAP_ERRBUF_SIZE) == -1) return -1; } else { if (sock_recv(md->rmt_sockdata, netbuf, sizeof(struct rpcap_header), SOCK_RECEIVEALL_YES, p->errbuf, PCAP_ERRBUF_SIZE) == -1) return -1; } /* Checks if the message is correct */ retval = rpcap_checkmsg(p->errbuf, md->rmt_sockdata, header, RPCAP_MSG_PACKET, 0); if (retval != RPCAP_MSG_PACKET) /* the message is not the one expected */ { switch (retval) { case -3: /* Unrecoverable network error */ return -1; /* Do nothing; just exit from here; the error code is already into the errbuf */ case -2: /* The other endpoint sent a message that is not allowed here */ case -1: /* The other endpoint has a version number that is not compatible with our */ return 0; /* Return 'no packets received' */ default: SOCK_ASSERT("Internal error", 1); return 0; /* Return 'no packets received' */ } } /* In case of TCP, read the remaining of the packet from the socket */ if (!(md->rmt_flags & PCAP_OPENFLAG_DATATX_UDP)) { /* Read the RPCAP packet header from the network */ nread = sock_recv(md->rmt_sockdata, (char *)net_pkt_header, sizeof(struct rpcap_pkthdr), SOCK_RECEIVEALL_YES, p->errbuf, PCAP_ERRBUF_SIZE); if (nread == -1) return -1; totread += nread; } if ((ntohl(net_pkt_header->caplen) + sizeof(struct pcap_pkthdr)) <= p->bufsize) { /* Initialize returned structures */ *pkt_header = (struct pcap_pkthdr *) p->buffer; *pkt_data = (u_char*)p->buffer + sizeof(struct pcap_pkthdr); (*pkt_header)->caplen = ntohl(net_pkt_header->caplen); (*pkt_header)->len = ntohl(net_pkt_header->len); (*pkt_header)->ts.tv_sec = ntohl(net_pkt_header->timestamp_sec); (*pkt_header)->ts.tv_usec = ntohl(net_pkt_header->timestamp_usec); /* * I don't update the counter of the packets dropped by the network since we're using TCP, * therefore no packets are dropped. Just update the number of packets received correctly */ md->TotCapt++; /* Copies the packet into the data buffer */ if (md->rmt_flags & PCAP_OPENFLAG_DATATX_UDP) { unsigned int npkt; /* * In case of UDP the packet has already been read, we have to copy it into 'buffer'. * Another option should be to declare 'netbuf' as 'static'. However this prevents * using several pcap instances within the same process (because the static buffer is shared among * all processes) */ memcpy(*pkt_data, netbuf + sizeof(struct rpcap_header) + sizeof(struct rpcap_pkthdr), (*pkt_header)->caplen); /* We're using UDP, so we need to update the counter of the packets dropped by the network */ npkt = ntohl(net_pkt_header->npkt); if (md->TotCapt != npkt) { md->TotNetDrops += (npkt - md->TotCapt); md->TotCapt = npkt; } } else { /* In case of TCP, read the remaining of the packet from the socket */ nread = sock_recv(md->rmt_sockdata, *pkt_data, (*pkt_header)->caplen, SOCK_RECEIVEALL_YES, p->errbuf, PCAP_ERRBUF_SIZE); if (nread == -1) return -1; totread += nread; /* Checks if all the data has been read; if not, discard the data in excess */ /* This check has to be done only on TCP connections */ if (totread != ntohl(header->plen)) sock_discard(md->rmt_sockdata, ntohl(header->plen) - totread, NULL, 0); } /* Packet read successfully */ return 1; } else { pcap_snprintf(p->errbuf, PCAP_ERRBUF_SIZE, "Received a packet that is larger than the internal buffer size."); return -1; } } /* \ingroup remote_pri_func * * \brief It reads a packet from the network socket. * * This function is called by the several pcap_read() when they detect that * we have a remote capture and they are the client side. In that case, they need * to read packets from the socket. * * This function relies on the pcap_read_nocb_remote to deliver packets. The * difference, here, is that as soon as a packet is read, it is delivered * to the application by means of a callback function. * * Parameters and return values are exactly the same of the pcap_read(). */ static int pcap_read_remote(pcap_t *p, int cnt, pcap_handler callback, u_char *user) { struct pcap_pkthdr *pkt_header; u_char *pkt_data; int n = 0; while ((n < cnt) || (cnt < 0)) { if (pcap_read_nocb_remote(p, &pkt_header, &pkt_data) == 1) { (*callback)(user, pkt_header, pkt_data); n++; } else return n; } return n; } /* \ingroup remote_pri_func * * \brief It sends a CLOSE command to the capture server. * * This function is called when the user wants to close a pcap_t adapter. * In case we're capturing from the network, it sends a command to the other * peer that says 'ok, let's stop capturing'. * This function is called automatically when the user calls the pcap_close(). * * Parameters and return values are exactly the same of the pcap_close(). * * \warning Since we're closing the connection, we do not check for errors. */ static void pcap_cleanup_remote(pcap_t *fp) { struct rpcap_header header; /* header of the RPCAP packet */ struct activehosts *temp; /* temp var needed to scan the host list chain, to detect if we're in active mode */ int active = 0; /* active mode or not? */ struct pcap_md *md; /* structure used when doing a remote live capture */ md = (struct pcap_md *) ((u_char*)fp->priv + sizeof(struct pcap_win)); /* detect if we're in active mode */ temp = activeHosts; while (temp) { if (temp->sockctrl == md->rmt_sockctrl) { active = 1; break; } temp = temp->next; } if (!active) { rpcap_createhdr(&header, RPCAP_MSG_CLOSE, 0, 0); /* I don't check for errors, since I'm going to close everything */ sock_send(md->rmt_sockctrl, (char *)&header, sizeof(struct rpcap_header), NULL, 0); } else { rpcap_createhdr(&header, RPCAP_MSG_ENDCAP_REQ, 0, 0); /* I don't check for errors, since I'm going to close everything */ sock_send(md->rmt_sockctrl, (char *)&header, sizeof(struct rpcap_header), NULL, 0); /* wait for the answer */ /* Don't check what we got, since the present libpcap does not uses this pcap_t anymore */ sock_recv(md->rmt_sockctrl, (char *)&header, sizeof(struct rpcap_header), SOCK_RECEIVEALL_YES, NULL, 0); if (ntohl(header.plen) != 0) sock_discard(md->rmt_sockctrl, ntohl(header.plen), NULL, 0); } if (md->rmt_sockdata) { sock_close(md->rmt_sockdata, NULL, 0); md->rmt_sockdata = 0; } if ((!active) && (md->rmt_sockctrl)) sock_close(md->rmt_sockctrl, NULL, 0); md->rmt_sockctrl = 0; if (md->currentfilter) { free(md->currentfilter); md->currentfilter = NULL; } /* To avoid inconsistencies in the number of sock_init() */ sock_cleanup(); } /* \ingroup remote_pri_func * * \brief It retrieves network statistics from the other peer. * * This function is just a void cointainer, since the work is done by the rpcap_stats_remote(). * See that funcion for more details. * * Parameters and return values are exactly the same of the pcap_stats(). */ static int pcap_stats_remote(pcap_t *p, struct pcap_stat *ps) { struct pcap_stat *retval; retval = rpcap_stats_remote(p, ps, PCAP_STATS_STANDARD); if (retval) return 0; else return -1; } #ifdef _WIN32 /* \ingroup remote_pri_func * * \brief It retrieves network statistics from the other peer. * * This function is just a void cointainer, since the work is done by the rpcap_stats_remote(). * See that funcion for more details. * * Parameters and return values are exactly the same of the pcap_stats_ex(). */ static struct pcap_stat *pcap_stats_ex_remote(pcap_t *p, int *pcap_stat_size) { *pcap_stat_size = sizeof (p->stat); /* PCAP_STATS_EX (third param) means 'extended pcap_stats()' */ return (rpcap_stats_remote(p, &(p->stat), PCAP_STATS_EX)); } #endif /* \ingroup remote_pri_func * * \brief It retrieves network statistics from the other peer. * * This function can be called in two modes: * - PCAP_STATS_STANDARD: if we want just standard statistics (i.e. the pcap_stats() ) * - PCAP_STATS_EX: if we want extended statistics (i.e. the pcap_stats_ex() ) * * This 'mode' parameter is needed because in the standard pcap_stats() the variable that keeps the * statistics is allocated by the user. Unfortunately, this structure has been extended in order * to keep new stats. However, if the user has a smaller structure and it passes it to the pcap_stats, * thid function will try to fill in more data than the size of the structure, so that the application * goes in memory overflow. * So, we need to know it we have to copy just the standard fields, or the extended fields as well. * * In case we want to copy the extended fields as well, the problem of memory overflow does no * longer exist because the structure pcap_stat is no longer allocated by the program; * it is allocated by the library instead. * * \param p: the pcap_t structure related to the current instance. * * \param ps: a 'pcap_stat' structure, needed for compatibility with pcap_stat(), in which * the structure is allocated by the user. In case of pcap_stats_ex, this structure and the * function return value point to the same variable. * * \param mode: one of PCAP_STATS_STANDARD or PCAP_STATS_EX. * * \return The structure that keeps the statistics, or NULL in case of error. * The error string is placed in the pcap_t structure. */ static struct pcap_stat *rpcap_stats_remote(pcap_t *p, struct pcap_stat *ps, int mode) { struct rpcap_header header; /* header of the RPCAP packet */ struct rpcap_stats netstats; /* statistics sent on the network */ uint32 totread = 0; /* number of bytes of the payload read from the socket */ int nread; int retval; /* temp variable which stores functions return value */ struct pcap_md *md; /* structure used when doing a remote live capture */ md = (struct pcap_md *) ((u_char*)p->priv + sizeof(struct pcap_win)); /* * If the capture has still to start, we cannot ask statistics to the other peer, * so we return a fake number */ if (!md->rmt_capstarted) { if (mode == PCAP_STATS_STANDARD) { ps->ps_drop = 0; ps->ps_ifdrop = 0; ps->ps_recv = 0; } else { ps->ps_capt = 0; ps->ps_drop = 0; ps->ps_ifdrop = 0; ps->ps_netdrop = 0; ps->ps_recv = 0; ps->ps_sent = 0; } return ps; } rpcap_createhdr(&header, RPCAP_MSG_STATS_REQ, 0, 0); /* Send the PCAP_STATS command */ if (sock_send(md->rmt_sockctrl, (char *)&header, sizeof(struct rpcap_header), p->errbuf, PCAP_ERRBUF_SIZE)) goto error; /* Receive the RPCAP stats reply message */ if (sock_recv(md->rmt_sockctrl, (char *)&header, sizeof(struct rpcap_header), SOCK_RECEIVEALL_YES, p->errbuf, PCAP_ERRBUF_SIZE) == -1) goto error; /* Checks if the message is correct */ retval = rpcap_checkmsg(p->errbuf, md->rmt_sockctrl, &header, RPCAP_MSG_STATS_REPLY, RPCAP_MSG_ERROR, 0); if (retval != RPCAP_MSG_STATS_REPLY) /* the message is not the one expected */ { switch (retval) { case -3: /* Unrecoverable network error */ case -2: /* The other endpoint send a message that is not allowed here */ case -1: /* The other endpoint has a version number that is not compatible with our */ goto error; case RPCAP_MSG_ERROR: /* The other endpoint reported an error */ /* Update totread, since the rpcap_checkmsg() already purged the buffer */ totread = ntohl(header.plen); /* Do nothing; just exit; the error code is already into the errbuf */ goto error; default: pcap_snprintf(p->errbuf, PCAP_ERRBUF_SIZE, "Internal error"); goto error; } } nread = sock_recv(md->rmt_sockctrl, (char *)&netstats, sizeof(struct rpcap_stats), SOCK_RECEIVEALL_YES, p->errbuf, PCAP_ERRBUF_SIZE); if (nread == -1) goto error; totread += nread; if (mode == PCAP_STATS_STANDARD) { ps->ps_drop = ntohl(netstats.krnldrop); ps->ps_ifdrop = ntohl(netstats.ifdrop); ps->ps_recv = ntohl(netstats.ifrecv); } else { ps->ps_capt = md->TotCapt; ps->ps_drop = ntohl(netstats.krnldrop); ps->ps_ifdrop = ntohl(netstats.ifdrop); ps->ps_netdrop = md->TotNetDrops; ps->ps_recv = ntohl(netstats.ifrecv); ps->ps_sent = ntohl(netstats.svrcapt); } /* Checks if all the data has been read; if not, discard the data in excess */ if (totread != ntohl(header.plen)) { if (sock_discard(md->rmt_sockctrl, ntohl(header.plen) - totread, NULL, 0) == 1) goto error; } return ps; error: if (totread != ntohl(header.plen)) sock_discard(md->rmt_sockctrl, ntohl(header.plen) - totread, NULL, 0); return NULL; } /* \ingroup remote_pri_func * * \brief It opens a remote adapter by opening an RPCAP connection and so on. * * This function does basically the job of pcap_open_live() for a remote interface. * In other words, we have a pcap_read for win32, which reads packets from NPF, * another for LINUX, and so on. Now, we have a pcap_opensource_remote() as well. * The difference, here, is the capture thread does not start until the * pcap_startcapture_remote() is called. * * This is because, in remote capture, we cannot start capturing data as soon ad the * 'open adapter' command is sent. Suppose the remote adapter is already overloaded; * if we start a capture (which, by default, has a NULL filter) the new traffic can * saturate the network. * * Instead, we want to "open" the adapter, then send a "start capture" command only * when we're ready to start the capture. * This funtion does this job: it sends a "open adapter" command (according to the * RPCAP protocol), but it does not start the capture. * * Since the other libpcap functions do not share this way of life, we have to make * some dirty things in order to make everyting working. * * \param fp: A pointer to a pcap_t structure that has been previously created with * \ref pcap_create(). * \param source: see pcap_open(). * \param auth: see pcap_open(). * * \return 0 in case of success, -1 otherwise. In case of success, the pcap_t pointer in fp can be * used as a parameter to the following calls (pcap_compile() and so on). In case of * problems, fp->errbuf contains a text explanation of error. * * \warning In case we call the pcap_compile() and the capture is not started, the filter * will be saved into the pcap_t structure, and it will be sent to the other host later * (when the pcap_startcapture_remote() is called). */ int pcap_opensource_remote(pcap_t *fp, struct pcap_rmtauth *auth) { char host[PCAP_BUF_SIZE], ctrlport[PCAP_BUF_SIZE], iface[PCAP_BUF_SIZE]; char sendbuf[RPCAP_NETBUF_SIZE]; /* temporary buffer in which data to be sent is buffered */ int sendbufidx = 0; /* index which keeps the number of bytes currently buffered */ uint32 totread = 0; /* number of bytes of the payload read from the socket */ int nread; int retval; /* store the return value of the functions */ int active = 0; /* '1' if we're in active mode */ /* socket-related variables */ struct addrinfo hints; /* temp, needed to open a socket connection */ struct addrinfo *addrinfo; /* temp, needed to open a socket connection */ SOCKET sockctrl = 0; /* socket descriptor of the control connection */ /* RPCAP-related variables */ struct rpcap_header header; /* header of the RPCAP packet */ struct rpcap_openreply openreply; /* open reply message */ struct pcap_md *md; /* structure used when doing a remote live capture */ md = (struct pcap_md *) ((u_char*)fp->priv + sizeof(struct pcap_win)); /* * determine the type of the source (NULL, file, local, remote) * You must have a valid source string even if we're in active mode, because otherwise * the call to the following function will fail. */ if (pcap_parsesrcstr(fp->opt.device, &retval, host, ctrlport, iface, fp->errbuf) == -1) return -1; if (retval != PCAP_SRC_IFREMOTE) { pcap_snprintf(fp->errbuf, PCAP_ERRBUF_SIZE, "This function is able to open only remote interfaces"); return -1; } addrinfo = NULL; /* * Warning: this call can be the first one called by the user. * For this reason, we have to initialize the WinSock support. */ if (sock_init(fp->errbuf, PCAP_ERRBUF_SIZE) == -1) return -1; sockctrl = rpcap_remoteact_getsock(host, &active, fp->errbuf); if (sockctrl == INVALID_SOCKET) return -1; if (!active) { /* * We're not in active mode; let's try to open a new * control connection. */ memset(&hints, 0, sizeof(struct addrinfo)); hints.ai_family = PF_UNSPEC; hints.ai_socktype = SOCK_STREAM; if ((ctrlport == NULL) || (ctrlport[0] == 0)) { /* the user chose not to specify the port */ if (sock_initaddress(host, RPCAP_DEFAULT_NETPORT, &hints, &addrinfo, fp->errbuf, PCAP_ERRBUF_SIZE) == -1) return -1; } else { /* the user chose not to specify the port */ if (sock_initaddress(host, ctrlport, &hints, &addrinfo, fp->errbuf, PCAP_ERRBUF_SIZE) == -1) return -1; } if ((sockctrl = sock_open(addrinfo, SOCKOPEN_CLIENT, 0, fp->errbuf, PCAP_ERRBUF_SIZE)) == INVALID_SOCKET) goto error; freeaddrinfo(addrinfo); addrinfo = NULL; if (rpcap_sendauth(sockctrl, auth, fp->errbuf) == -1) goto error; } /* * Now it's time to start playing with the RPCAP protocol * RPCAP open command: create the request message */ if (sock_bufferize(NULL, sizeof(struct rpcap_header), NULL, &sendbufidx, RPCAP_NETBUF_SIZE, SOCKBUF_CHECKONLY, fp->errbuf, PCAP_ERRBUF_SIZE)) goto error; rpcap_createhdr((struct rpcap_header *) sendbuf, RPCAP_MSG_OPEN_REQ, 0, (uint32) strlen(iface)); if (sock_bufferize(iface, (int) strlen(iface), sendbuf, &sendbufidx, RPCAP_NETBUF_SIZE, SOCKBUF_BUFFERIZE, fp->errbuf, PCAP_ERRBUF_SIZE)) goto error; if (sock_send(sockctrl, sendbuf, sendbufidx, fp->errbuf, PCAP_ERRBUF_SIZE)) goto error; /* Receive the RPCAP open reply message */ if (sock_recv(sockctrl, (char *)&header, sizeof(struct rpcap_header), SOCK_RECEIVEALL_YES, fp->errbuf, PCAP_ERRBUF_SIZE) == -1) goto error; /* Checks if the message is correct */ retval = rpcap_checkmsg(fp->errbuf, sockctrl, &header, RPCAP_MSG_OPEN_REPLY, RPCAP_MSG_ERROR, 0); if (retval != RPCAP_MSG_OPEN_REPLY) /* the message is not the one expected */ { switch (retval) { case -3: /* Unrecoverable network error */ case -2: /* The other endpoint send a message that is not allowed here */ case -1: /* The other endpoint has a version number that is not compatible with our */ goto error; case RPCAP_MSG_ERROR: /* The other endpoint reported an error */ /* Update totread, since the rpcap_checkmsg() already purged the buffer */ totread = ntohl(header.plen); /* Do nothing; just exit; the error code is already into the errbuf */ goto error; default: pcap_snprintf(fp->errbuf, PCAP_ERRBUF_SIZE, "Internal error"); goto error; } } nread = sock_recv(sockctrl, (char *)&openreply, sizeof(struct rpcap_openreply), SOCK_RECEIVEALL_YES, fp->errbuf, PCAP_ERRBUF_SIZE); if (nread == -1) goto error; totread += nread; /* Set proper fields into the pcap_t struct */ fp->linktype = ntohl(openreply.linktype); fp->tzoff = ntohl(openreply.tzoff); md->rmt_sockctrl = sockctrl; md->rmt_clientside = 1; /* This code is duplicated from the end of this function */ fp->read_op = pcap_read_remote; fp->setfilter_op = pcap_setfilter_remote; fp->getnonblock_op = NULL; /* This is not implemented in remote capture */ fp->setnonblock_op = NULL; /* This is not implemented in remote capture */ fp->stats_op = pcap_stats_remote; #ifdef _WIN32 fp->stats_ex_op = pcap_stats_ex_remote; #endif fp->cleanup_op = pcap_cleanup_remote; /* Checks if all the data has been read; if not, discard the data in excess */ if (totread != ntohl(header.plen)) { if (sock_discard(sockctrl, ntohl(header.plen) - totread, NULL, 0) == 1) goto error; } return 0; error: /* * When the connection has been established, we have to close it. So, at the * beginning of this function, if an error occur we return immediately with * a return NULL; when the connection is established, we have to come here * ('goto error;') in order to close everything properly. * * Checks if all the data has been read; if not, discard the data in excess */ if (totread != ntohl(header.plen)) sock_discard(sockctrl, ntohl(header.plen) - totread, NULL, 0); if (addrinfo) freeaddrinfo(addrinfo); if (!active) sock_close(sockctrl, NULL, 0); return -1; } /* \ingroup remote_pri_func * * \brief It starts a remote capture. * * This function is requires since the RPCAP protocol decouples the 'open' from the * 'start capture' functions. * This function takes all the parameters needed (which have been stored into the pcap_t structure) * and sends them to the server. * If everything is fine, it creates a new child thread that reads data from the network * and puts data it into the user buffer. * The pcap_read() will read data from the user buffer, as usual. * * The remote capture acts like a new "kernel", which puts packets directly into * the buffer pointed by pcap_t. * In fact, this function does not rely on a kernel that reads packets and put them * into the user buffer; it has to do that on its own. * * \param fp: the pcap_t descriptor of the device currently open. * * \return '0' if everything is fine, '-1' otherwise. The error message (if one) * is returned into the 'errbuf' field of the pcap_t structure. */ int pcap_startcapture_remote(pcap_t *fp) { char sendbuf[RPCAP_NETBUF_SIZE]; /* temporary buffer in which data to be sent is buffered */ int sendbufidx = 0; /* index which keeps the number of bytes currently buffered */ char portdata[PCAP_BUF_SIZE]; /* temp variable needed to keep the network port for the the data connection */ uint32 totread = 0; /* number of bytes of the payload read from the socket */ int nread; int retval; /* store the return value of the functions */ int active = 0; /* '1' if we're in active mode */ struct activehosts *temp; /* temp var needed to scan the host list chain, to detect if we're in active mode */ char host[INET6_ADDRSTRLEN + 1]; /* numeric name of the other host */ /* socket-related variables*/ struct addrinfo hints; /* temp, needed to open a socket connection */ struct addrinfo *addrinfo; /* temp, needed to open a socket connection */ SOCKET sockdata = 0; /* socket descriptor of the data connection */ struct sockaddr_storage saddr; /* temp, needed to retrieve the network data port chosen on the local machine */ socklen_t saddrlen; /* temp, needed to retrieve the network data port chosen on the local machine */ int ai_family; /* temp, keeps the address family used by the control connection */ /* RPCAP-related variables*/ struct rpcap_header header; /* header of the RPCAP packet */ struct rpcap_startcapreq *startcapreq; /* start capture request message */ struct rpcap_startcapreply startcapreply; /* start capture reply message */ /* Variables related to the buffer setting */ int res, itemp; int sockbufsize = 0; struct pcap_md *md; /* structure used when doing a remote live capture */ md = (struct pcap_md *) ((u_char*)fp->priv + sizeof(struct pcap_win)); /* * Let's check if sampling has been required. * If so, let's set it first */ if (pcap_setsampling_remote(fp) != 0) return -1; /* detect if we're in active mode */ temp = activeHosts; while (temp) { if (temp->sockctrl == md->rmt_sockctrl) { active = 1; break; } temp = temp->next; } addrinfo = NULL; /* * Gets the complete sockaddr structure used in the ctrl connection * This is needed to get the address family of the control socket * Tip: I cannot save the ai_family of the ctrl sock in the pcap_t struct, * since the ctrl socket can already be open in case of active mode; * so I would have to call getpeername() anyway */ saddrlen = sizeof(struct sockaddr_storage); if (getpeername(md->rmt_sockctrl, (struct sockaddr *) &saddr, &saddrlen) == -1) { sock_geterror("getsockname(): ", fp->errbuf, PCAP_ERRBUF_SIZE); goto error; } ai_family = ((struct sockaddr_storage *) &saddr)->ss_family; /* Get the numeric address of the remote host we are connected to */ if (getnameinfo((struct sockaddr *) &saddr, saddrlen, host, sizeof(host), NULL, 0, NI_NUMERICHOST)) { sock_geterror("getnameinfo(): ", fp->errbuf, PCAP_ERRBUF_SIZE); goto error; } /* * Data connection is opened by the server toward the client if: * - we're using TCP, and the user wants us to be in active mode * - we're using UDP */ if ((active) || (md->rmt_flags & PCAP_OPENFLAG_DATATX_UDP)) { /* * We have to create a new socket to receive packets * We have to do that immediately, since we have to tell the other * end which network port we picked up */ memset(&hints, 0, sizeof(struct addrinfo)); /* TEMP addrinfo is NULL in case of active */ hints.ai_family = ai_family; /* Use the same address family of the control socket */ hints.ai_socktype = (md->rmt_flags & PCAP_OPENFLAG_DATATX_UDP) ? SOCK_DGRAM : SOCK_STREAM; hints.ai_flags = AI_PASSIVE; /* Data connection is opened by the server toward the client */ /* Let's the server pick up a free network port for us */ if (sock_initaddress(NULL, "0", &hints, &addrinfo, fp->errbuf, PCAP_ERRBUF_SIZE) == -1) goto error; if ((sockdata = sock_open(addrinfo, SOCKOPEN_SERVER, 1 /* max 1 connection in queue */, fp->errbuf, PCAP_ERRBUF_SIZE)) == INVALID_SOCKET) goto error; /* addrinfo is no longer used */ freeaddrinfo(addrinfo); addrinfo = NULL; /* get the complete sockaddr structure used in the data connection */ saddrlen = sizeof(struct sockaddr_storage); if (getsockname(sockdata, (struct sockaddr *) &saddr, &saddrlen) == -1) { sock_geterror("getsockname(): ", fp->errbuf, PCAP_ERRBUF_SIZE); goto error; } /* Get the local port the system picked up */ if (getnameinfo((struct sockaddr *) &saddr, saddrlen, NULL, 0, portdata, sizeof(portdata), NI_NUMERICSERV)) { sock_geterror("getnameinfo(): ", fp->errbuf, PCAP_ERRBUF_SIZE); goto error; } } /* * Now it's time to start playing with the RPCAP protocol * RPCAP start capture command: create the request message */ if (sock_bufferize(NULL, sizeof(struct rpcap_header), NULL, &sendbufidx, RPCAP_NETBUF_SIZE, SOCKBUF_CHECKONLY, fp->errbuf, PCAP_ERRBUF_SIZE)) goto error; rpcap_createhdr((struct rpcap_header *) sendbuf, RPCAP_MSG_STARTCAP_REQ, 0, sizeof(struct rpcap_startcapreq) + sizeof(struct rpcap_filter) + fp->fcode.bf_len * sizeof(struct rpcap_filterbpf_insn)); /* Fill the structure needed to open an adapter remotely */ startcapreq = (struct rpcap_startcapreq *) &sendbuf[sendbufidx]; if (sock_bufferize(NULL, sizeof(struct rpcap_startcapreq), NULL, &sendbufidx, RPCAP_NETBUF_SIZE, SOCKBUF_CHECKONLY, fp->errbuf, PCAP_ERRBUF_SIZE)) goto error; memset(startcapreq, 0, sizeof(struct rpcap_startcapreq)); /* By default, apply half the timeout on one side, half of the other */ fp->opt.timeout = fp->opt.timeout / 2; startcapreq->read_timeout = htonl(fp->opt.timeout); /* portdata on the openreq is meaningful only if we're in active mode */ if ((active) || (md->rmt_flags & PCAP_OPENFLAG_DATATX_UDP)) { sscanf(portdata, "%d", (int *)&(startcapreq->portdata)); /* cast to avoid a compiler warning */ startcapreq->portdata = htons(startcapreq->portdata); } startcapreq->snaplen = htonl(fp->snapshot); startcapreq->flags = 0; if (md->rmt_flags & PCAP_OPENFLAG_PROMISCUOUS) startcapreq->flags |= RPCAP_STARTCAPREQ_FLAG_PROMISC; if (md->rmt_flags & PCAP_OPENFLAG_DATATX_UDP) startcapreq->flags |= RPCAP_STARTCAPREQ_FLAG_DGRAM; if (active) startcapreq->flags |= RPCAP_STARTCAPREQ_FLAG_SERVEROPEN; startcapreq->flags = htons(startcapreq->flags); /* Pack the capture filter */ if (pcap_pack_bpffilter(fp, &sendbuf[sendbufidx], &sendbufidx, &fp->fcode)) goto error; if (sock_send(md->rmt_sockctrl, sendbuf, sendbufidx, fp->errbuf, PCAP_ERRBUF_SIZE)) goto error; /* Receive the RPCAP start capture reply message */ if (sock_recv(md->rmt_sockctrl, (char *)&header, sizeof(struct rpcap_header), SOCK_RECEIVEALL_YES, fp->errbuf, PCAP_ERRBUF_SIZE) == -1) goto error; /* Checks if the message is correct */ retval = rpcap_checkmsg(fp->errbuf, md->rmt_sockctrl, &header, RPCAP_MSG_STARTCAP_REPLY, RPCAP_MSG_ERROR, 0); if (retval != RPCAP_MSG_STARTCAP_REPLY) /* the message is not the one expected */ { switch (retval) { case -3: /* Unrecoverable network error */ case -2: /* The other endpoint send a message that is not allowed here */ case -1: /* The other endpoint has a version number that is not compatible with our */ goto error; case RPCAP_MSG_ERROR: /* The other endpoint reported an error */ /* Update totread, since the rpcap_checkmsg() already purged the buffer */ totread = ntohl(header.plen); /* Do nothing; just exit; the error code is already into the errbuf */ goto error; default: pcap_snprintf(fp->errbuf, PCAP_ERRBUF_SIZE, "Internal error"); goto error; } } nread = sock_recv(md->rmt_sockctrl, (char *)&startcapreply, sizeof(struct rpcap_startcapreply), SOCK_RECEIVEALL_YES, fp->errbuf, PCAP_ERRBUF_SIZE); if (nread == -1) goto error; totread += nread; /* * In case of UDP data stream, the connection is always opened by the daemon * So, this case is already covered by the code above. * Now, we have still to handle TCP connections, because: * - if we're in active mode, we have to wait for a remote connection * - if we're in passive more, we have to start a connection * * We have to do he job in two steps because in case we're opening a TCP connection, we have * to tell the port we're using to the remote side; in case we're accepting a TCP * connection, we have to wait this info from the remote side. */ if (!(md->rmt_flags & PCAP_OPENFLAG_DATATX_UDP)) { if (!active) { memset(&hints, 0, sizeof(struct addrinfo)); hints.ai_family = ai_family; /* Use the same address family of the control socket */ hints.ai_socktype = (md->rmt_flags & PCAP_OPENFLAG_DATATX_UDP) ? SOCK_DGRAM : SOCK_STREAM; pcap_snprintf(portdata, PCAP_BUF_SIZE, "%d", ntohs(startcapreply.portdata)); /* Let's the server pick up a free network port for us */ if (sock_initaddress(host, portdata, &hints, &addrinfo, fp->errbuf, PCAP_ERRBUF_SIZE) == -1) goto error; if ((sockdata = sock_open(addrinfo, SOCKOPEN_CLIENT, 0, fp->errbuf, PCAP_ERRBUF_SIZE)) == INVALID_SOCKET) goto error; /* addrinfo is no longer used */ freeaddrinfo(addrinfo); addrinfo = NULL; } else { SOCKET socktemp; /* We need another socket, since we're going to accept() a connection */ /* Connection creation */ saddrlen = sizeof(struct sockaddr_storage); socktemp = accept(sockdata, (struct sockaddr *) &saddr, &saddrlen); if (socktemp == -1) { sock_geterror("accept(): ", fp->errbuf, PCAP_ERRBUF_SIZE); goto error; } /* Now that I accepted the connection, the server socket is no longer needed */ sock_close(sockdata, fp->errbuf, PCAP_ERRBUF_SIZE); sockdata = socktemp; } } /* Let's save the socket of the data connection */ md->rmt_sockdata = sockdata; /* Allocates WinPcap/libpcap user buffer, which is a socket buffer in case of a remote capture */ /* It has the same size of the one used on the other side of the connection */ fp->bufsize = ntohl(startcapreply.bufsize); /* Let's get the actual size of the socket buffer */ itemp = sizeof(sockbufsize); res = getsockopt(sockdata, SOL_SOCKET, SO_RCVBUF, (char *)&sockbufsize, &itemp); if (res == -1) { sock_geterror("pcap_startcapture_remote()", fp->errbuf, PCAP_ERRBUF_SIZE); SOCK_ASSERT(fp->errbuf, 1); } /* * Warning: on some kernels (e.g. Linux), the size of the user buffer does not take * into account the pcap_header and such, and it is set equal to the snaplen. * In my view, this is wrong (the meaning of the bufsize became a bit strange). * So, here bufsize is the whole size of the user buffer. * In case the bufsize returned is too small, let's adjust it accordingly. */ if (fp->bufsize <= (u_int) fp->snapshot) fp->bufsize += sizeof(struct pcap_pkthdr); /* if the current socket buffer is smaller than the desired one */ if ((u_int) sockbufsize < fp->bufsize) { /* Loop until the buffer size is OK or the original socket buffer size is larger than this one */ while (1) { res = setsockopt(sockdata, SOL_SOCKET, SO_RCVBUF, (char *)&(fp->bufsize), sizeof(fp->bufsize)); if (res == 0) break; /* * If something goes wrong, half the buffer size (checking that it does not become smaller than * the current one) */ fp->bufsize /= 2; if ((u_int) sockbufsize >= fp->bufsize) { fp->bufsize = sockbufsize; break; } } } /* * Let's allocate the packet; this is required in order to put the packet somewhere when * extracting data from the socket * Since buffering has already been done in the socket buffer, here we need just a buffer, * whose size is equal to the pcap header plus the snapshot length */ fp->bufsize = fp->snapshot + sizeof(struct pcap_pkthdr); fp->buffer = (u_char *)malloc(fp->bufsize); if (fp->buffer == NULL) { pcap_snprintf(fp->errbuf, PCAP_ERRBUF_SIZE, "malloc: %s", pcap_strerror(errno)); goto error; } /* Checks if all the data has been read; if not, discard the data in excess */ if (totread != ntohl(header.plen)) { if (sock_discard(md->rmt_sockctrl, ntohl(header.plen) - totread, NULL, 0) == 1) goto error; } /* * In case the user does not want to capture RPCAP packets, let's update the filter * We have to update it here (instead of sending it into the 'StartCapture' message * because when we generate the 'start capture' we do not know (yet) all the ports * we're currently using. */ if (md->rmt_flags & PCAP_OPENFLAG_NOCAPTURE_RPCAP) { struct bpf_program fcode; if (pcap_createfilter_norpcappkt(fp, &fcode) == -1) goto error; /* We cannot use 'pcap_setfilter_remote' because formally the capture has not been started yet */ /* (the 'fp->rmt_capstarted' variable will be updated some lines below) */ if (pcap_updatefilter_remote(fp, &fcode) == -1) goto error; pcap_freecode(&fcode); } md->rmt_capstarted = 1; return 0; error: /* * When the connection has been established, we have to close it. So, at the * beginning of this function, if an error occur we return immediately with * a return NULL; when the connection is established, we have to come here * ('goto error;') in order to close everything properly. * * Checks if all the data has been read; if not, discard the data in excess */ if (totread != ntohl(header.plen)) sock_discard(md->rmt_sockctrl, ntohl(header.plen) - totread, NULL, 0); if ((sockdata) && (sockdata != -1)) /* we can be here because sockdata said 'error' */ sock_close(sockdata, NULL, 0); if (!active) sock_close(md->rmt_sockctrl, NULL, 0); /* * We do not have to call pcap_close() here, because this function is always called * by the user in case something bad happens */ // if (fp) // { // pcap_close(fp); // fp= NULL; // } return -1; } /* * \brief Takes a bpf program and sends it to the other host. * * This function can be called in two cases: * - the pcap_startcapture() is called (we have to send the filter along with * the 'start capture' command) * - we want to udpate the filter during a capture (i.e. the pcap_setfilter() * is called when the capture is still on) * * This function serializes the filter into the sending buffer ('sendbuf', passed * as a parameter) and return back. It does not send anything on the network. * * \param fp: the pcap_t descriptor of the device currently opened. * * \param sendbuf: the buffer on which the serialized data has to copied. * * \param sendbufidx: it is used to return the abounf of bytes copied into the buffer. * * \param prog: the bpf program we have to copy. * * \return '0' if everything is fine, '-1' otherwise. The error message (if one) * is returned into the 'errbuf' field of the pcap_t structure. */ static int pcap_pack_bpffilter(pcap_t *fp, char *sendbuf, int *sendbufidx, struct bpf_program *prog) { struct rpcap_filter *filter; struct rpcap_filterbpf_insn *insn; struct bpf_insn *bf_insn; struct bpf_program fake_prog; /* To be used just in case the user forgot to set a filter */ unsigned int i; if (prog->bf_len == 0) /* No filters have been specified; so, let's apply a "fake" filter */ { if (pcap_compile(fp, &fake_prog, NULL /* buffer */, 1, 0) == -1) return -1; prog = &fake_prog; } filter = (struct rpcap_filter *) sendbuf; if (sock_bufferize(NULL, sizeof(struct rpcap_filter), NULL, sendbufidx, RPCAP_NETBUF_SIZE, SOCKBUF_CHECKONLY, fp->errbuf, PCAP_ERRBUF_SIZE)) return -1; filter->filtertype = htons(RPCAP_UPDATEFILTER_BPF); filter->nitems = htonl((int32)prog->bf_len); if (sock_bufferize(NULL, prog->bf_len * sizeof(struct rpcap_filterbpf_insn), NULL, sendbufidx, RPCAP_NETBUF_SIZE, SOCKBUF_CHECKONLY, fp->errbuf, PCAP_ERRBUF_SIZE)) return -1; insn = (struct rpcap_filterbpf_insn *) (filter + 1); bf_insn = prog->bf_insns; for (i = 0; i < prog->bf_len; i++) { insn->code = htons(bf_insn->code); insn->jf = bf_insn->jf; insn->jt = bf_insn->jt; insn->k = htonl(bf_insn->k); insn++; bf_insn++; } return 0; } /* \ingroup remote_pri_func * * \brief Update a filter on a remote host. * * This function is called when the user wants to update a filter. * In case we're capturing from the network, it sends the filter to the other peer. * This function is *not* called automatically when the user calls the pcap_setfilter(). * There will be two cases: * - the capture is already on: in this case, pcap_setfilter() calls pcap_updatefilter_remote() * - the capture has not started yet: in this case, pcap_setfilter() stores the filter into * the pcap_t structure, and then the filter is sent with the pcap_startcap(). * * Parameters and return values are exactly the same of the pcap_setfilter(). * * \warning This function *does not* clear the packet currently into the buffers. Therefore, * the user has to expect to receive some packets that are related to the previous filter. * If you want to discard all the packets before applying a new filter, you have to close * the current capture session and start a new one. */ static int pcap_updatefilter_remote(pcap_t *fp, struct bpf_program *prog) { int retval; /* general variable used to keep the return value of other functions */ char sendbuf[RPCAP_NETBUF_SIZE];/* temporary buffer in which data to be sent is buffered */ int sendbufidx = 0; /* index which keeps the number of bytes currently buffered */ struct rpcap_header header; /* To keep the reply message */ struct pcap_md *md; /* structure used when doing a remote live capture */ md = (struct pcap_md *) ((u_char*)fp->priv + sizeof(struct pcap_win)); if (sock_bufferize(NULL, sizeof(struct rpcap_header), NULL, &sendbufidx, RPCAP_NETBUF_SIZE, SOCKBUF_CHECKONLY, fp->errbuf, PCAP_ERRBUF_SIZE)) return -1; rpcap_createhdr((struct rpcap_header *) sendbuf, RPCAP_MSG_UPDATEFILTER_REQ, 0, sizeof(struct rpcap_filter) + prog->bf_len * sizeof(struct rpcap_filterbpf_insn)); if (pcap_pack_bpffilter(fp, &sendbuf[sendbufidx], &sendbufidx, prog)) return -1; if (sock_send(md->rmt_sockctrl, sendbuf, sendbufidx, fp->errbuf, PCAP_ERRBUF_SIZE)) return -1; /* Waits for the answer */ if (sock_recv(md->rmt_sockctrl, (char *)&header, sizeof(struct rpcap_header), SOCK_RECEIVEALL_YES, fp->errbuf, PCAP_ERRBUF_SIZE) == -1) return -1; /* Checks if the message is correct */ retval = rpcap_checkmsg(fp->errbuf, md->rmt_sockctrl, &header, RPCAP_MSG_UPDATEFILTER_REPLY, 0); if (retval != RPCAP_MSG_UPDATEFILTER_REPLY) /* the message is not the one expected */ { switch (retval) { case -3: /* Unrecoverable network error */ case -2: /* The other endpoint sent a message that is not allowed here */ case -1: /* The other endpoint has a version number that is not compatible with our */ /* Do nothing; just exit from here; the error code is already into the errbuf */ return -1; default: SOCK_ASSERT("Internal error", 0); return -1; } } if (ntohl(header.plen) != 0) /* the message has an unexpected size */ { if (sock_discard(md->rmt_sockctrl, ntohl(header.plen), fp->errbuf, PCAP_ERRBUF_SIZE) == -1) return -1; } return 0; } /* * \ingroup remote_pri_func * * \brief Send a filter to a remote host. * * This function is called when the user wants to set a filter. * In case we're capturing from the network, it sends the filter to the other peer. * This function is called automatically when the user calls the pcap_setfilter(). * * Parameters and return values are exactly the same of the pcap_setfilter(). */ static int pcap_setfilter_remote(pcap_t *fp, struct bpf_program *prog) { struct pcap_md *md; /* structure used when doing a remote live capture */ md = (struct pcap_md *) ((u_char*)fp->priv + sizeof(struct pcap_win)); if (!md->rmt_capstarted) { /* copy filter into the pcap_t structure */ if (install_bpf_program(fp, prog) == -1) return -1; return 0; } /* we have to update a filter during run-time */ if (pcap_updatefilter_remote(fp, prog)) return -1; return 0; } /* * \ingroup remote_pri_func * * \brief Update the current filter in order not to capture rpcap packets. * * This function is called *only* when the user wants exclude RPCAP packets * related to the current session from the captured packets. * * \return '0' if everything is fine, '-1' otherwise. The error message (if one) * is returned into the 'errbuf' field of the pcap_t structure. */ static int pcap_createfilter_norpcappkt(pcap_t *fp, struct bpf_program *prog) { int RetVal = 0; struct pcap_md *md; /* structure used when doing a remote live capture */ md = (struct pcap_md *) ((u_char*)fp->priv + sizeof(struct pcap_win)); /* We do not want to capture our RPCAP traffic. So, let's update the filter */ if (md->rmt_flags & PCAP_OPENFLAG_NOCAPTURE_RPCAP) { struct sockaddr_storage saddr; /* temp, needed to retrieve the network data port chosen on the local machine */ socklen_t saddrlen; /* temp, needed to retrieve the network data port chosen on the local machine */ char myaddress[128]; char myctrlport[128]; char mydataport[128]; char peeraddress[128]; char peerctrlport[128]; char *newfilter; const int newstringsize = 1024; size_t currentfiltersize; /* Get the name/port of the other peer */ saddrlen = sizeof(struct sockaddr_storage); if (getpeername(md->rmt_sockctrl, (struct sockaddr *) &saddr, &saddrlen) == -1) { sock_geterror("getpeername(): ", fp->errbuf, PCAP_ERRBUF_SIZE); return -1; } if (getnameinfo((struct sockaddr *) &saddr, saddrlen, peeraddress, sizeof(peeraddress), peerctrlport, sizeof(peerctrlport), NI_NUMERICHOST | NI_NUMERICSERV)) { sock_geterror("getnameinfo(): ", fp->errbuf, PCAP_ERRBUF_SIZE); return -1; } /* We cannot check the data port, because this is available only in case of TCP sockets */ /* Get the name/port of the current host */ if (getsockname(md->rmt_sockctrl, (struct sockaddr *) &saddr, &saddrlen) == -1) { sock_geterror("getsockname(): ", fp->errbuf, PCAP_ERRBUF_SIZE); return -1; } /* Get the local port the system picked up */ if (getnameinfo((struct sockaddr *) &saddr, saddrlen, myaddress, sizeof(myaddress), myctrlport, sizeof(myctrlport), NI_NUMERICHOST | NI_NUMERICSERV)) { sock_geterror("getnameinfo(): ", fp->errbuf, PCAP_ERRBUF_SIZE); return -1; } /* Let's now check the data port */ if (getsockname(md->rmt_sockdata, (struct sockaddr *) &saddr, &saddrlen) == -1) { sock_geterror("getsockname(): ", fp->errbuf, PCAP_ERRBUF_SIZE); return -1; } /* Get the local port the system picked up */ if (getnameinfo((struct sockaddr *) &saddr, saddrlen, NULL, 0, mydataport, sizeof(mydataport), NI_NUMERICSERV)) { sock_geterror("getnameinfo(): ", fp->errbuf, PCAP_ERRBUF_SIZE); return -1; } currentfiltersize = strlen(md->currentfilter); newfilter = (char *)malloc(currentfiltersize + newstringsize + 1); if (currentfiltersize) { pcap_snprintf(newfilter, currentfiltersize + newstringsize, "(%s) and not (host %s and host %s and port %s and port %s) and not (host %s and host %s and port %s)", md->currentfilter, myaddress, peeraddress, myctrlport, peerctrlport, myaddress, peeraddress, mydataport); } else { pcap_snprintf(newfilter, currentfiltersize + newstringsize, "not (host %s and host %s and port %s and port %s) and not (host %s and host %s and port %s)", myaddress, peeraddress, myctrlport, peerctrlport, myaddress, peeraddress, mydataport); } newfilter[currentfiltersize + newstringsize] = 0; /* This is only an hack to make the pcap_compile() working properly */ md->rmt_clientside = 0; if (pcap_compile(fp, prog, newfilter, 1, 0) == -1) RetVal = -1; /* This is only an hack to make the pcap_compile() working properly */ md->rmt_clientside = 1; free(newfilter); } return RetVal; } /* * \ingroup remote_pri_func * * \brief Set sampling parameters in the remote host. * * This function is called when the user wants to set activate sampling on the remote host. * * Sampling parameters are defined into the 'pcap_t' structure. * * \param p: the pcap_t descriptor of the device currently opened. * * \return '0' if everything is OK, '-1' is something goes wrong. The error message is returned * in the 'errbuf' member of the pcap_t structure. */ static int pcap_setsampling_remote(pcap_t *p) { int retval; /* general variable used to keep the return value of other functions */ char sendbuf[RPCAP_NETBUF_SIZE];/* temporary buffer in which data to be sent is buffered */ int sendbufidx = 0; /* index which keeps the number of bytes currently buffered */ struct rpcap_header header; /* To keep the reply message */ struct rpcap_sampling *sampling_pars; /* Structure that is needed to send sampling parameters to the remote host */ struct pcap_md *md; /* structure used when doing a remote live capture */ md = (struct pcap_md *) ((u_char*)p->priv + sizeof(struct pcap_win)); /* If no samping is requested, return 'ok' */ if (md->rmt_samp.method == PCAP_SAMP_NOSAMP) return 0; if (sock_bufferize(NULL, sizeof(struct rpcap_header), NULL, &sendbufidx, RPCAP_NETBUF_SIZE, SOCKBUF_CHECKONLY, p->errbuf, PCAP_ERRBUF_SIZE)) return -1; rpcap_createhdr((struct rpcap_header *) sendbuf, RPCAP_MSG_SETSAMPLING_REQ, 0, sizeof(struct rpcap_sampling)); /* Fill the structure needed to open an adapter remotely */ sampling_pars = (struct rpcap_sampling *) &sendbuf[sendbufidx]; if (sock_bufferize(NULL, sizeof(struct rpcap_sampling), NULL, &sendbufidx, RPCAP_NETBUF_SIZE, SOCKBUF_CHECKONLY, p->errbuf, PCAP_ERRBUF_SIZE)) return -1; memset(sampling_pars, 0, sizeof(struct rpcap_sampling)); sampling_pars->method = md->rmt_samp.method; sampling_pars->value = htonl(md->rmt_samp.value); if (sock_send(md->rmt_sockctrl, sendbuf, sendbufidx, p->errbuf, PCAP_ERRBUF_SIZE)) return -1; /* Waits for the answer */ if (sock_recv(md->rmt_sockctrl, (char *)&header, sizeof(struct rpcap_header), SOCK_RECEIVEALL_YES, p->errbuf, PCAP_ERRBUF_SIZE) == -1) return -1; /* Checks if the message is correct */ retval = rpcap_checkmsg(p->errbuf, md->rmt_sockctrl, &header, RPCAP_MSG_SETSAMPLING_REPLY, 0); if (retval != RPCAP_MSG_SETSAMPLING_REPLY) /* the message is not the one expected */ { switch (retval) { case -3: /* Unrecoverable network error */ case -2: /* The other endpoint sent a message that is not allowed here */ case -1: /* The other endpoint has a version number that is not compatible with our */ case RPCAP_MSG_ERROR: /* Do nothing; just exit from here; the error code is already into the errbuf */ return -1; default: SOCK_ASSERT("Internal error", 0); return -1; } } if (ntohl(header.plen) != 0) /* the message has an unexpected size */ { if (sock_discard(md->rmt_sockctrl, ntohl(header.plen), p->errbuf, PCAP_ERRBUF_SIZE) == -1) return -1; } return 0; } /********************************************************* * * * Miscellaneous functions * * * *********************************************************/ /* \ingroup remote_pri_func * \brief It sends a RPCAP error to the other peer. * * This function has to be called when the main program detects an error. This function * will send on the other peer the 'buffer' specified by the user. * This function *does not* request a RPCAP CLOSE connection. A CLOSE command must be sent * explicitly by the program, since we do not know it the error can be recovered in some * way or it is a non-recoverable one. * * \param sock: the socket we are currently using. * * \param error: an user-allocated (and '0' terminated) buffer that contains the error * description that has to be transmitted on the other peer. The error message cannot * be longer than PCAP_ERRBUF_SIZE. * * \param errcode: a integer which tells the other party the type of error we had; * currently is is not too much used. * * \param errbuf: a pointer to a user-allocated buffer (of size PCAP_ERRBUF_SIZE) * that will contain the error message (in case there is one). It could be network problem. * * \return '0' if everything is fine, '-1' if some errors occurred. The error message is returned * in the 'errbuf' variable. */ int rpcap_senderror(SOCKET sock, char *error, unsigned short errcode, char *errbuf) { char sendbuf[RPCAP_NETBUF_SIZE]; /* temporary buffer in which data to be sent is buffered */ int sendbufidx = 0; /* index which keeps the number of bytes currently buffered */ uint16 length; length = (uint16)strlen(error); if (length > PCAP_ERRBUF_SIZE) length = PCAP_ERRBUF_SIZE; rpcap_createhdr((struct rpcap_header *) sendbuf, RPCAP_MSG_ERROR, errcode, length); if (sock_bufferize(NULL, sizeof(struct rpcap_header), NULL, &sendbufidx, RPCAP_NETBUF_SIZE, SOCKBUF_CHECKONLY, errbuf, PCAP_ERRBUF_SIZE)) return -1; if (sock_bufferize(error, length, sendbuf, &sendbufidx, RPCAP_NETBUF_SIZE, SOCKBUF_BUFFERIZE, errbuf, PCAP_ERRBUF_SIZE)) return -1; if (sock_send(sock, sendbuf, sendbufidx, errbuf, PCAP_ERRBUF_SIZE)) return -1; return 0; } /* \ingroup remote_pri_func * \brief Sends the authentication message. * * It sends the authentication parameters on the control socket. * This function is required in order to open the connection with the other end party. * * \param sock: the socket we are currently using. * * \param auth: authentication parameters that have to be sent. * * \param errbuf: a pointer to a user-allocated buffer (of size PCAP_ERRBUF_SIZE) * that will contain the error message (in case there is one). It could be network problem * of the fact that the authorization failed. * * \return '0' if everything is fine, '-1' if some errors occurred. The error message is returned * in the 'errbuf' variable. * The error message could be also 'the authentication failed'. */ int rpcap_sendauth(SOCKET sock, struct pcap_rmtauth *auth, char *errbuf) { char sendbuf[RPCAP_NETBUF_SIZE]; /* temporary buffer in which data that has to be sent is buffered */ int sendbufidx = 0; /* index which keeps the number of bytes currently buffered */ uint16 length; /* length of the payload of this message */ struct rpcap_auth *rpauth; uint16 auth_type; struct rpcap_header header; int retval; /* temp variable which stores functions return value */ if (auth) { auth_type = auth->type; switch (auth->type) { case RPCAP_RMTAUTH_NULL: length = sizeof(struct rpcap_auth); break; case RPCAP_RMTAUTH_PWD: length = sizeof(struct rpcap_auth); if (auth->username) length += (uint16) strlen(auth->username); if (auth->password) length += (uint16) strlen(auth->password); break; default: pcap_snprintf(errbuf, PCAP_ERRBUF_SIZE, "Authentication type not recognized."); return -1; } } else { auth_type = RPCAP_RMTAUTH_NULL; length = sizeof(struct rpcap_auth); } if (sock_bufferize(NULL, sizeof(struct rpcap_header), NULL, &sendbufidx, RPCAP_NETBUF_SIZE, SOCKBUF_CHECKONLY, errbuf, PCAP_ERRBUF_SIZE)) return -1; rpcap_createhdr((struct rpcap_header *) sendbuf, RPCAP_MSG_AUTH_REQ, 0, length); rpauth = (struct rpcap_auth *) &sendbuf[sendbufidx]; if (sock_bufferize(NULL, sizeof(struct rpcap_auth), NULL, &sendbufidx, RPCAP_NETBUF_SIZE, SOCKBUF_CHECKONLY, errbuf, PCAP_ERRBUF_SIZE)) return -1; memset(rpauth, 0, sizeof(struct rpcap_auth)); rpauth->type = htons(auth_type); if (auth_type == RPCAP_RMTAUTH_PWD) { if (auth->username) rpauth->slen1 = (uint16) strlen(auth->username); else rpauth->slen1 = 0; if (sock_bufferize(auth->username, rpauth->slen1, sendbuf, &sendbufidx, RPCAP_NETBUF_SIZE, SOCKBUF_BUFFERIZE, errbuf, PCAP_ERRBUF_SIZE)) return -1; if (auth->password) rpauth->slen2 = (uint16) strlen(auth->password); else rpauth->slen2 = 0; if (sock_bufferize(auth->password, rpauth->slen2, sendbuf, &sendbufidx, RPCAP_NETBUF_SIZE, SOCKBUF_BUFFERIZE, errbuf, PCAP_ERRBUF_SIZE)) return -1; rpauth->slen1 = htons(rpauth->slen1); rpauth->slen2 = htons(rpauth->slen2); } if (sock_send(sock, sendbuf, sendbufidx, errbuf, PCAP_ERRBUF_SIZE)) return -1; if (sock_recv(sock, (char *)&header, sizeof(struct rpcap_header), SOCK_RECEIVEALL_YES, errbuf, PCAP_ERRBUF_SIZE) == -1) return -1; retval = rpcap_checkmsg(errbuf, sock, &header, RPCAP_MSG_AUTH_REPLY, RPCAP_MSG_ERROR, 0); if (retval != RPCAP_MSG_AUTH_REPLY) /* the message is not the one expected */ { switch (retval) { case -3: /* Unrecoverable network error */ case -2: /* The other endpoint sent a message that is not allowed here */ case -1: /* The other endpoint has a version number that is not compatible with our */ /* Do nothing; just exit from here; the error code is already into the errbuf */ return -1; case RPCAP_MSG_ERROR: return -1; default: SOCK_ASSERT("Internal error", 0); return -1; } } if (ntohl(header.plen)) { if (sock_discard(sock, ntohl(header.plen), errbuf, PCAP_ERRBUF_SIZE)) return -1; } return 0; } /* \ingroup remote_pri_func * \brief Creates a structure of type rpcap_header. * * This function is provided just because the creation of an rpcap header is quite a common * task. It accepts all the values that appears into an rpcap_header, and it puts them in * place using the proper hton() calls. * * \param header: a pointer to a user-allocated buffer which will contain the serialized * header, ready to be sent on the network. * * \param type: a value (in the host by order) which will be placed into the header.type * field and that represents the type of the current message. * * \param value: a value (in the host by order) which will be placed into the header.value * field and that has a message-dependent meaning. * * \param length: a value (in the host by order) which will be placed into the header.length * field and that represents the payload length of the message. * * \return Nothing. The serialized header is returned into the 'header' variable. */ void rpcap_createhdr(struct rpcap_header *header, uint8 type, uint16 value, uint32 length) { memset(header, 0, sizeof(struct rpcap_header)); header->ver = RPCAP_VERSION; header->type = type; header->value = htons(value); header->plen = htonl(length); } /* ingroup remote_pri_func * \brief Checks if the header of the received message is correct. * * This function is a way to easily check if the message received, in a certain * state of the RPCAP protocol Finite State Machine, is valid. This function accepts, * as a parameter, the list of message types that are allowed in a certain situation, * and it returns the one which occurs. * * \param errbuf: a pointer to a user-allocated buffer (of size PCAP_ERRBUF_SIZE) * that will contain the error message (in case there is one). It could be either problem * occurred inside this function (e.g. a network problem in case it tries to send an * error on the other peer and the send() call fails), an error message which has been * sent to us from the other party, or a version error (the message receive has a version * number that is incompatible with our). * * \param sock: the socket that has to be used to receive data. This function can * read data from socket in case the version contained into the message is not compatible * with our. In that case, all the message is purged from the socket, so that the following * recv() calls will return a new message. * * \param header: a pointer to and 'rpcap_header' structure that keeps the data received from * the network (still in network byte order) and that has to be checked. * * \param first: this function has a variable number of parameters. From this point on, * all the messages that are valid in this context must be passed as parameters. * The message type list must be terminated with a '0' value, the null message type, * which means 'no more types to check'. The RPCAP protocol does not define anything with * message type equal to zero, so there is no ambiguity in using this value as a list terminator. * * \return The message type of the message that has been detected. In case of errors (e.g. the * header contains a type that is not listed among the allowed types), this function will * return the following codes: * - (-1) if the version is incompatible. * - (-2) if the code is not among the one listed into the parameters list * - (-3) if a network error (connection reset, ...) * - RPCAP_MSG_ERROR if the message is an error message (it follow that the RPCAP_MSG_ERROR * could not be present in the allowed message-types list, because this function checks * for errors anyway) * * In case either the version is incompatible or nothing matches (i.e. it returns '-1' or '-2'), * it discards the message body (i.e. it reads the remaining part of the message from the * network and it discards it) so that the application is ready to receive a new message. */ int rpcap_checkmsg(char *errbuf, SOCKET sock, struct rpcap_header *header, uint8 first, ...) { va_list ap; uint8 type; int32 len; va_start(ap, first); /* Check if the present version of the protocol can handle this message */ if (rpcap_checkver(sock, header, errbuf)) { SOCK_ASSERT(errbuf, 1); va_end(ap); return -1; } type = first; while (type != 0) { /* * The message matches with one of the types listed * There is no need of conversions since both values are uint8 * * Check if the other side reported an error. * If yes, it retrieves it and it returns it back to the caller */ if (header->type == RPCAP_MSG_ERROR) { len = ntohl(header->plen); if (len >= PCAP_ERRBUF_SIZE) { if (sock_recv(sock, errbuf, PCAP_ERRBUF_SIZE - 1, SOCK_RECEIVEALL_YES, errbuf, PCAP_ERRBUF_SIZE)) return -3; sock_discard(sock, len - (PCAP_ERRBUF_SIZE - 1), NULL, 0); /* Put '\0' at the end of the string */ errbuf[PCAP_ERRBUF_SIZE - 1] = 0; } else { if (sock_recv(sock, errbuf, len, SOCK_RECEIVEALL_YES, errbuf, PCAP_ERRBUF_SIZE) == -1) return -3; /* Put '\0' at the end of the string */ errbuf[len] = 0; } va_end(ap); return header->type; } if (header->type == type) { va_end(ap); return header->type; } /* get next argument */ type = va_arg(ap, int); } /* we already have an error, so please discard this one */ sock_discard(sock, ntohl(header->plen), NULL, 0); pcap_snprintf(errbuf, PCAP_ERRBUF_SIZE, "The other endpoint sent a message that is not allowed here."); SOCK_ASSERT(errbuf, 1); va_end(ap); return -2; } /* \ingroup remote_pri_func * \brief Checks if the version contained into the message is compatible with * the one handled by this implementation. * * Right now, this function does not have any sophisticated task: if the versions * are different, it returns -1 and it discards the message. * It is expected that in the future this message will become more complex. * * \param sock: the socket that has to be used to receive data. This function can * read data from socket in case the version contained into the message is not compatible * with our. In that case, all the message is purged from the socket, so that the following * recv() calls will return a new (clean) message. * * \param header: a pointer to and 'rpcap_header' structure that keeps the data received from * the network (still in network byte order) and that has to be checked. * * \param errbuf: a pointer to a user-allocated buffer (of size PCAP_ERRBUF_SIZE) * that will contain the error message (in case there is one). The error message is * "incompatible version". * * \return '0' if everything is fine, '-1' if some errors occurred. The error message is returned * in the 'errbuf' variable. */ static int rpcap_checkver(SOCKET sock, struct rpcap_header *header, char *errbuf) { /* * This is a sample function. * * In the real world, you have to check at the type code, * and decide accordingly. */ if (header->ver != RPCAP_VERSION) { pcap_snprintf(errbuf, PCAP_ERRBUF_SIZE, "Incompatible version number: message discarded."); /* we already have an error, so please discard this one */ sock_discard(sock, ntohl(header->plen), NULL, 0); return -1; } return 0; } /* \ingroup remote_pri_func * * \brief It returns the socket currently used for this active connection * (active mode only) and provides an indication of whether this connection * is in active mode or not. * * This function is just for internal use; it returns the socket ID of the * active connection currently opened. * * \param host: a string that keeps the host name of the host for which we * want to get the socket ID for that active connection. * * \param isactive: a pointer to an int that is set to 1 if there's an * active connection to that host and 0 otherwise. * * \param errbuf: a pointer to a user-allocated buffer (of size * PCAP_ERRBUF_SIZE) that will contain the error message (in case * there is one). * * \return the socket identifier if everything is fine, '0' if this host * is not in the active host list. An indication of whether this host * is in the active host list is returned into the isactive variable. * It returns 'INVALID_SOCKET' in case of error. The error message is * returned into the errbuf variable. */ SOCKET rpcap_remoteact_getsock(const char *host, int *isactive, char *errbuf) { struct activehosts *temp; /* temp var needed to scan the host list chain */ struct addrinfo hints, *addrinfo, *ai_next; /* temp var needed to translate between hostname to its address */ int retval; /* retrieve the network address corresponding to 'host' */ addrinfo = NULL; memset(&hints, 0, sizeof(struct addrinfo)); hints.ai_family = PF_UNSPEC; hints.ai_socktype = SOCK_STREAM; retval = getaddrinfo(host, "0", &hints, &addrinfo); if (retval != 0) { pcap_snprintf(errbuf, PCAP_ERRBUF_SIZE, "getaddrinfo() %s", gai_strerror(retval)); *isactive = 0; return INVALID_SOCKET; } temp = activeHosts; while (temp) { ai_next = addrinfo; while (ai_next) { if (sock_cmpaddr(&temp->host, (struct sockaddr_storage *) ai_next->ai_addr) == 0) { *isactive = 1; return (temp->sockctrl); } ai_next = ai_next->ai_next; } temp = temp->next; } if (addrinfo) freeaddrinfo(addrinfo); /* * The host for which you want to get the socket ID does not have an * active connection. */ *isactive = 0; return 0; }