]> The Tcpdump Group git mirrors - libpcap/commitdiff
We shouldn't need to include <sys/file.h>, and, as I remember, it caused
authorguy <guy>
Wed, 6 Aug 2008 07:35:01 +0000 (07:35 +0000)
committerguy <guy>
Wed, 6 Aug 2008 07:35:01 +0000 (07:35 +0000)
problems on some platform.

fad-gifc.c [new file with mode: 0644]

diff --git a/fad-gifc.c b/fad-gifc.c
new file mode 100644 (file)
index 0000000..ef04bd9
--- /dev/null
@@ -0,0 +1,577 @@
+/* -*- Mode: c; tab-width: 8; indent-tabs-mode: 1; c-basic-offset: 8; -*- */
+/*
+ * Copyright (c) 1994, 1995, 1996, 1997, 1998
+ *     The Regents of the University of 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. All advertising materials mentioning features or use of this software
+ *    must display the following acknowledgement:
+ *     This product includes software developed by the Computer Systems
+ *     Engineering Group at Lawrence Berkeley Laboratory.
+ * 4. Neither the name of the University nor of the Laboratory may be used
+ *    to endorse or promote products derived from this software without
+ *    specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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.
+ */
+
+#ifndef lint
+static const char rcsid[] _U_ =
+    "@(#) $Header: /tcpdump/master/libpcap/fad-gifc.c,v 1.11.2.1 2008-08-06 07:35:01 guy Exp $ (LBL)";
+#endif
+
+#ifdef HAVE_CONFIG_H
+#include "config.h"
+#endif
+
+#include <sys/param.h>
+#include <sys/ioctl.h>
+#include <sys/socket.h>
+#ifdef HAVE_SYS_SOCKIO_H
+#include <sys/sockio.h>
+#endif
+#include <sys/time.h>                          /* concession to AIX */
+
+struct mbuf;           /* Squelch compiler warnings on some platforms for */
+struct rtentry;                /* declarations in <net/if.h> */
+#include <net/if.h>
+#include <netinet/in.h>
+
+#include <ctype.h>
+#include <errno.h>
+#include <memory.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+
+#include "pcap-int.h"
+
+#ifdef HAVE_OS_PROTO_H
+#include "os-proto.h"
+#endif
+
+/*
+ * This is fun.
+ *
+ * In older BSD systems, socket addresses were fixed-length, and
+ * "sizeof (struct sockaddr)" gave the size of the structure.
+ * All addresses fit within a "struct sockaddr".
+ *
+ * In newer BSD systems, the socket address is variable-length, and
+ * there's an "sa_len" field giving the length of the structure;
+ * this allows socket addresses to be longer than 2 bytes of family
+ * and 14 bytes of data.
+ *
+ * Some commercial UNIXes use the old BSD scheme, some use the RFC 2553
+ * variant of the old BSD scheme (with "struct sockaddr_storage" rather
+ * than "struct sockaddr"), and some use the new BSD scheme.
+ *
+ * Some versions of GNU libc use neither scheme, but has an "SA_LEN()"
+ * macro that determines the size based on the address family.  Other
+ * versions don't have "SA_LEN()" (as it was in drafts of RFC 2553
+ * but not in the final version).
+ *
+ * We assume that a UNIX that doesn't have "getifaddrs()" and doesn't have
+ * SIOCGLIFCONF, but has SIOCGIFCONF, uses "struct sockaddr" for the
+ * address in an entry returned by SIOCGIFCONF.
+ */
+#ifndef SA_LEN
+#ifdef HAVE_SOCKADDR_SA_LEN
+#define SA_LEN(addr)   ((addr)->sa_len)
+#else /* HAVE_SOCKADDR_SA_LEN */
+#define SA_LEN(addr)   (sizeof (struct sockaddr))
+#endif /* HAVE_SOCKADDR_SA_LEN */
+#endif /* SA_LEN */
+
+/*
+ * This is also fun.
+ *
+ * There is no ioctl that returns the amount of space required for all
+ * the data that SIOCGIFCONF could return, and if a buffer is supplied
+ * that's not large enough for all the data SIOCGIFCONF could return,
+ * on at least some platforms it just returns the data that'd fit with
+ * no indication that there wasn't enough room for all the data, much
+ * less an indication of how much more room is required.
+ *
+ * The only way to ensure that we got all the data is to pass a buffer
+ * large enough that the amount of space in the buffer *not* filled in
+ * is greater than the largest possible entry.
+ *
+ * We assume that's "sizeof(ifreq.ifr_name)" plus 255, under the assumption
+ * that no address is more than 255 bytes (on systems where the "sa_len"
+ * field in a "struct sockaddr" is 1 byte, e.g. newer BSDs, that's the
+ * case, and addresses are unlikely to be bigger than that in any case).
+ */
+#define MAX_SA_LEN     255
+
+#ifdef HAVE_PROC_NET_DEV
+/*
+ * Get from "/proc/net/dev" all interfaces listed there; if they're
+ * already in the list of interfaces we have, that won't add another
+ * instance, but if they're not, that'll add them.
+ *
+ * We don't bother getting any addresses for them; it appears you can't
+ * use SIOCGIFADDR on Linux to get IPv6 addresses for interfaces, and,
+ * although some other types of addresses can be fetched with SIOCGIFADDR,
+ * we don't bother with them for now.
+ *
+ * We also don't fail if we couldn't open "/proc/net/dev"; we just leave
+ * the list of interfaces as is.
+ */
+static int
+scan_proc_net_dev(pcap_if_t **devlistp, int fd, char *errbuf)
+{
+       FILE *proc_net_f;
+       char linebuf[512];
+       int linenum;
+       unsigned char *p;
+       char name[512]; /* XXX - pick a size */
+       char *q, *saveq;
+       struct ifreq ifrflags;
+       int ret = 0;
+
+       proc_net_f = fopen("/proc/net/dev", "r");
+       if (proc_net_f == NULL)
+               return (0);
+
+       for (linenum = 1;
+           fgets(linebuf, sizeof linebuf, proc_net_f) != NULL; linenum++) {
+               /*
+                * Skip the first two lines - they're headers.
+                */
+               if (linenum <= 2)
+                       continue;
+
+               p = &linebuf[0];
+
+               /*
+                * Skip leading white space.
+                */
+               while (*p != '\0' && isspace(*p))
+                       p++;
+               if (*p == '\0' || *p == '\n')
+                       continue;       /* blank line */
+
+               /*
+                * Get the interface name.
+                */
+               q = &name[0];
+               while (*p != '\0' && !isspace(*p)) {
+                       if (*p == ':') {
+                               /*
+                                * This could be the separator between a
+                                * name and an alias number, or it could be
+                                * the separator between a name with no
+                                * alias number and the next field.
+                                *
+                                * If there's a colon after digits, it
+                                * separates the name and the alias number,
+                                * otherwise it separates the name and the
+                                * next field.
+                                */
+                               saveq = q;
+                               while (isdigit(*p))
+                                       *q++ = *p++;
+                               if (*p != ':') {
+                                       /*
+                                        * That was the next field,
+                                        * not the alias number.
+                                        */
+                                       q = saveq;
+                               }
+                               break;
+                       } else
+                               *q++ = *p++;
+               }
+               *q = '\0';
+
+               /*
+                * Get the flags for this interface, and skip it if
+                * it's not up.
+                */
+               strncpy(ifrflags.ifr_name, name, sizeof(ifrflags.ifr_name));
+               if (ioctl(fd, SIOCGIFFLAGS, (char *)&ifrflags) < 0) {
+                       if (errno == ENXIO)
+                               continue;
+                       (void)snprintf(errbuf, PCAP_ERRBUF_SIZE,
+                           "SIOCGIFFLAGS: %.*s: %s",
+                           (int)sizeof(ifrflags.ifr_name),
+                           ifrflags.ifr_name,
+                           pcap_strerror(errno));
+                       ret = -1;
+                       break;
+               }
+               if (!(ifrflags.ifr_flags & IFF_UP))
+                       continue;
+
+               /*
+                * Add an entry for this interface, with no addresses.
+                */
+               if (pcap_add_if(devlistp, name, ifrflags.ifr_flags, NULL,
+                   errbuf) == -1) {
+                       /*
+                        * Failure.
+                        */
+                       ret = -1;
+                       break;
+               }
+       }
+       if (ret != -1) {
+               /*
+                * Well, we didn't fail for any other reason; did we
+                * fail due to an error reading the file?
+                */
+               if (ferror(proc_net_f)) {
+                       (void)snprintf(errbuf, PCAP_ERRBUF_SIZE,
+                           "Error reading /proc/net/dev: %s",
+                           pcap_strerror(errno));
+                       ret = -1;
+               }
+       }
+
+       (void)fclose(proc_net_f);
+       return (ret);
+}
+#endif /* HAVE_PROC_NET_DEV */
+
+/*
+ * Get a list of all interfaces that are up and that we can open.
+ * Returns -1 on error, 0 otherwise.
+ * The list, as returned through "alldevsp", may be null if no interfaces
+ * were up and could be opened.
+ *
+ * This is the implementation used on platforms that have SIOCGIFCONF but
+ * don't have any other mechanism for getting a list of interfaces.
+ *
+ * XXX - or platforms that have other, better mechanisms but for which
+ * we don't yet have code to use that mechanism; I think there's a better
+ * way on Linux, for example.
+ */
+int
+pcap_findalldevs(pcap_if_t **alldevsp, char *errbuf)
+{
+       pcap_if_t *devlist = NULL;
+       register int fd;
+       register struct ifreq *ifrp, *ifend, *ifnext;
+       int n;
+       struct ifconf ifc;
+       char *buf = NULL;
+       unsigned buf_size;
+#if defined (HAVE_SOLARIS) || defined (HAVE_HPUX10_20_OR_LATER)
+       char *p, *q;
+#endif
+       struct ifreq ifrflags, ifrnetmask, ifrbroadaddr, ifrdstaddr;
+       struct sockaddr *netmask, *broadaddr, *dstaddr;
+       size_t netmask_size, broadaddr_size, dstaddr_size;
+       int ret = 0;
+
+       /*
+        * Create a socket from which to fetch the list of interfaces.
+        */
+       fd = socket(AF_INET, SOCK_DGRAM, 0);
+       if (fd < 0) {
+               (void)snprintf(errbuf, PCAP_ERRBUF_SIZE,
+                   "socket: %s", pcap_strerror(errno));
+               return (-1);
+       }
+
+       /*
+        * Start with an 8K buffer, and keep growing the buffer until
+        * we have more than "sizeof(ifrp->ifr_name) + MAX_SA_LEN"
+        * bytes left over in the buffer or we fail to get the
+        * interface list for some reason other than EINVAL (which is
+        * presumed here to mean "buffer is too small").
+        */
+       buf_size = 8192;
+       for (;;) {
+               buf = malloc(buf_size);
+               if (buf == NULL) {
+                       (void)snprintf(errbuf, PCAP_ERRBUF_SIZE,
+                           "malloc: %s", pcap_strerror(errno));
+                       (void)close(fd);
+                       return (-1);
+               }
+
+               ifc.ifc_len = buf_size;
+               ifc.ifc_buf = buf;
+               memset(buf, 0, buf_size);
+               if (ioctl(fd, SIOCGIFCONF, (char *)&ifc) < 0
+                   && errno != EINVAL) {
+                       (void)snprintf(errbuf, PCAP_ERRBUF_SIZE,
+                           "SIOCGIFCONF: %s", pcap_strerror(errno));
+                       (void)close(fd);
+                       free(buf);
+                       return (-1);
+               }
+               if (ifc.ifc_len < buf_size &&
+                   (buf_size - ifc.ifc_len) > sizeof(ifrp->ifr_name) + MAX_SA_LEN)
+                       break;
+               free(buf);
+               buf_size *= 2;
+       }
+
+       ifrp = (struct ifreq *)buf;
+       ifend = (struct ifreq *)(buf + ifc.ifc_len);
+
+       for (; ifrp < ifend; ifrp = ifnext) {
+               /*
+                * XXX - what if this isn't an IPv4 address?  Can
+                * we still get the netmask, etc. with ioctls on
+                * an IPv4 socket?
+                *
+                * The answer is probably platform-dependent, and
+                * if the answer is "no" on more than one platform,
+                * the way you work around it is probably platform-
+                * dependent as well.
+                */
+               n = SA_LEN(&ifrp->ifr_addr) + sizeof(ifrp->ifr_name);
+               if (n < sizeof(*ifrp))
+                       ifnext = ifrp + 1;
+               else
+                       ifnext = (struct ifreq *)((char *)ifrp + n);
+
+               /*
+                * XXX - The 32-bit compatibility layer for Linux on IA-64
+                * is slightly broken. It correctly converts the structures
+                * to and from kernel land from 64 bit to 32 bit but 
+                * doesn't update ifc.ifc_len, leaving it larger than the 
+                * amount really used. This means we read off the end 
+                * of the buffer and encounter an interface with an 
+                * "empty" name. Since this is highly unlikely to ever 
+                * occur in a valid case we can just finish looking for 
+                * interfaces if we see an empty name.
+                */
+               if (!(*ifrp->ifr_name))
+                       break;
+
+               /*
+                * Skip entries that begin with "dummy".
+                * XXX - what are these?  Is this Linux-specific?
+                * Are there platforms on which we shouldn't do this?
+                */
+               if (strncmp(ifrp->ifr_name, "dummy", 5) == 0)
+                       continue;
+
+               /*
+                * Get the flags for this interface, and skip it if it's
+                * not up.
+                */
+               strncpy(ifrflags.ifr_name, ifrp->ifr_name,
+                   sizeof(ifrflags.ifr_name));
+               if (ioctl(fd, SIOCGIFFLAGS, (char *)&ifrflags) < 0) {
+                       if (errno == ENXIO)
+                               continue;
+                       (void)snprintf(errbuf, PCAP_ERRBUF_SIZE,
+                           "SIOCGIFFLAGS: %.*s: %s",
+                           (int)sizeof(ifrflags.ifr_name),
+                           ifrflags.ifr_name,
+                           pcap_strerror(errno));
+                       ret = -1;
+                       break;
+               }
+               if (!(ifrflags.ifr_flags & IFF_UP))
+                       continue;
+
+               /*
+                * Get the netmask for this address on this interface.
+                */
+               strncpy(ifrnetmask.ifr_name, ifrp->ifr_name,
+                   sizeof(ifrnetmask.ifr_name));
+               memcpy(&ifrnetmask.ifr_addr, &ifrp->ifr_addr,
+                   sizeof(ifrnetmask.ifr_addr));
+               if (ioctl(fd, SIOCGIFNETMASK, (char *)&ifrnetmask) < 0) {
+                       if (errno == EADDRNOTAVAIL) {
+                               /*
+                                * Not available.
+                                */
+                               netmask = NULL;
+                               netmask_size = 0;
+                       } else {
+                               (void)snprintf(errbuf, PCAP_ERRBUF_SIZE,
+                                   "SIOCGIFNETMASK: %.*s: %s",
+                                   (int)sizeof(ifrnetmask.ifr_name),
+                                   ifrnetmask.ifr_name,
+                                   pcap_strerror(errno));
+                               ret = -1;
+                               break;
+                       }
+               } else {
+                       netmask = &ifrnetmask.ifr_addr;
+                       netmask_size = SA_LEN(netmask);
+               }
+
+               /*
+                * Get the broadcast address for this address on this
+                * interface (if any).
+                */
+               if (ifrflags.ifr_flags & IFF_BROADCAST) {
+                       strncpy(ifrbroadaddr.ifr_name, ifrp->ifr_name,
+                           sizeof(ifrbroadaddr.ifr_name));
+                       memcpy(&ifrbroadaddr.ifr_addr, &ifrp->ifr_addr,
+                           sizeof(ifrbroadaddr.ifr_addr));
+                       if (ioctl(fd, SIOCGIFBRDADDR,
+                           (char *)&ifrbroadaddr) < 0) {
+                               if (errno == EADDRNOTAVAIL) {
+                                       /*
+                                        * Not available.
+                                        */
+                                       broadaddr = NULL;
+                                       broadaddr_size = 0;
+                               } else {
+                                       (void)snprintf(errbuf, PCAP_ERRBUF_SIZE,
+                                           "SIOCGIFBRDADDR: %.*s: %s",
+                                           (int)sizeof(ifrbroadaddr.ifr_name),
+                                           ifrbroadaddr.ifr_name,
+                                           pcap_strerror(errno));
+                                       ret = -1;
+                                       break;
+                               }
+                       } else {
+                               broadaddr = &ifrbroadaddr.ifr_broadaddr;
+                               broadaddr_size = SA_LEN(broadaddr);
+                       }
+               } else {
+                       /*
+                        * Not a broadcast interface, so no broadcast
+                        * address.
+                        */
+                       broadaddr = NULL;
+                       broadaddr_size = 0;
+               }
+
+               /*
+                * Get the destination address for this address on this
+                * interface (if any).
+                */
+               if (ifrflags.ifr_flags & IFF_POINTOPOINT) {
+                       strncpy(ifrdstaddr.ifr_name, ifrp->ifr_name,
+                           sizeof(ifrdstaddr.ifr_name));
+                       memcpy(&ifrdstaddr.ifr_addr, &ifrp->ifr_addr,
+                           sizeof(ifrdstaddr.ifr_addr));
+                       if (ioctl(fd, SIOCGIFDSTADDR,
+                           (char *)&ifrdstaddr) < 0) {
+                               if (errno == EADDRNOTAVAIL) {
+                                       /*
+                                        * Not available.
+                                        */
+                                       dstaddr = NULL;
+                                       dstaddr_size = 0;
+                               } else {
+                                       (void)snprintf(errbuf, PCAP_ERRBUF_SIZE,
+                                           "SIOCGIFDSTADDR: %.*s: %s",
+                                           (int)sizeof(ifrdstaddr.ifr_name),
+                                           ifrdstaddr.ifr_name,
+                                           pcap_strerror(errno));
+                                       ret = -1;
+                                       break;
+                               }
+                       } else {
+                               dstaddr = &ifrdstaddr.ifr_dstaddr;
+                               dstaddr_size = SA_LEN(dstaddr);
+                       }
+               } else {
+                       /*
+                        * Not a point-to-point interface, so no destination
+                        * address.
+                        */
+                       dstaddr = NULL;
+                       dstaddr_size = 0;
+               }
+
+#if defined (HAVE_SOLARIS) || defined (HAVE_HPUX10_20_OR_LATER)
+               /*
+                * If this entry has a colon followed by a number at
+                * the end, it's a logical interface.  Those are just
+                * the way you assign multiple IP addresses to a real
+                * interface, so an entry for a logical interface should
+                * be treated like the entry for the real interface;
+                * we do that by stripping off the ":" and the number.
+                */
+               p = strchr(ifrp->ifr_name, ':');
+               if (p != NULL) {
+                       /*
+                        * We have a ":"; is it followed by a number?
+                        */
+                       q = p + 1;
+                       while (isdigit((unsigned char)*q))
+                               q++;
+                       if (*q == '\0') {
+                               /*
+                                * All digits after the ":" until the end.
+                                * Strip off the ":" and everything after
+                                * it.
+                                */
+                               *p = '\0';
+                       }
+               }
+#endif
+
+               /*
+                * Add information for this address to the list.
+                */
+               if (add_addr_to_iflist(&devlist, ifrp->ifr_name,
+                   ifrflags.ifr_flags, &ifrp->ifr_addr,
+                   SA_LEN(&ifrp->ifr_addr), netmask, netmask_size,
+                   broadaddr, broadaddr_size, dstaddr, dstaddr_size,
+                   errbuf) < 0) {
+                       ret = -1;
+                       break;
+               }
+       }
+       free(buf);
+
+#ifdef HAVE_PROC_NET_DEV
+       if (ret != -1) {
+               /*
+                * We haven't had any errors yet; now read "/proc/net/dev",
+                * and add to the list of interfaces all interfaces listed
+                * there that we don't already have, because, on Linux,
+                * SIOCGIFCONF reports only interfaces with IPv4 addresses,
+                * so you need to read "/proc/net/dev" to get the names of
+                * the rest of the interfaces.
+                */
+               ret = scan_proc_net_dev(&devlist, fd, errbuf);
+       }
+#endif
+       (void)close(fd);
+
+       if (ret != -1) {
+               /*
+                * We haven't had any errors yet; do any platform-specific
+                * operations to add devices.
+                */
+               if (pcap_platform_finddevs(&devlist, errbuf) < 0)
+                       ret = -1;
+       }
+
+       if (ret == -1) {
+               /*
+                * We had an error; free the list we've been constructing.
+                */
+               if (devlist != NULL) {
+                       pcap_freealldevs(devlist);
+                       devlist = NULL;
+               }
+       }
+
+       *alldevsp = devlist;
+       return (ret);
+}