freebsd-skq/contrib/libpcap/pcap-int.h
cy 81062ad740 MFV r353141 (by phillip):
Update libpcap from 1.9.0 to 1.9.1.

MFC after:	2 weeks
2019-12-21 21:01:03 +00:00

534 lines
17 KiB
C

/*
* Copyright (c) 1994, 1995, 1996
* 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 pcap_int_h
#define pcap_int_h
#include <signal.h>
#include <pcap/pcap.h>
#include "varattrs.h"
#include "fmtutils.h"
/*
* Version string.
* Uses PACKAGE_VERSION from config.h.
*/
#define PCAP_VERSION_STRING "libpcap version " PACKAGE_VERSION
#ifdef __cplusplus
extern "C" {
#endif
#ifdef MSDOS
#include <fcntl.h>
#include <io.h>
#endif
/*
* Swap byte ordering of unsigned long long timestamp on a big endian
* machine.
*/
#define SWAPLL(ull) ((ull & 0xff00000000000000ULL) >> 56) | \
((ull & 0x00ff000000000000ULL) >> 40) | \
((ull & 0x0000ff0000000000ULL) >> 24) | \
((ull & 0x000000ff00000000ULL) >> 8) | \
((ull & 0x00000000ff000000ULL) << 8) | \
((ull & 0x0000000000ff0000ULL) << 24) | \
((ull & 0x000000000000ff00ULL) << 40) | \
((ull & 0x00000000000000ffULL) << 56)
/*
* Maximum snapshot length.
*
* Somewhat arbitrary, but chosen to be:
*
* 1) big enough for maximum-size Linux loopback packets (65549)
* and some USB packets captured with USBPcap:
*
* http://desowin.org/usbpcap/
*
* (> 131072, < 262144)
*
* and
*
* 2) small enough not to cause attempts to allocate huge amounts of
* memory; some applications might use the snapshot length in a
* savefile header to control the size of the buffer they allocate,
* so a size of, say, 2^31-1 might not work well. (libpcap uses it
* as a hint, but doesn't start out allocating a buffer bigger than
* 2 KiB, and grows the buffer as necessary, but not beyond the
* per-linktype maximum snapshot length. Other code might naively
* use it; we want to avoid writing a too-large snapshot length,
* in order not to cause that code problems.)
*
* We don't enforce this in pcap_set_snaplen(), but we use it internally.
*/
#define MAXIMUM_SNAPLEN 262144
struct pcap_opt {
char *device;
int timeout; /* timeout for buffering */
u_int buffer_size;
int promisc;
int rfmon; /* monitor mode */
int immediate; /* immediate mode - deliver packets as soon as they arrive */
int nonblock; /* non-blocking mode - don't wait for packets to be delivered, return "no packets available" */
int tstamp_type;
int tstamp_precision;
/*
* Platform-dependent options.
*/
#ifdef __linux__
int protocol; /* protocol to use when creating PF_PACKET socket */
#endif
#ifdef _WIN32
int nocapture_local;/* disable NPF loopback */
#endif
};
typedef int (*activate_op_t)(pcap_t *);
typedef int (*can_set_rfmon_op_t)(pcap_t *);
typedef int (*read_op_t)(pcap_t *, int cnt, pcap_handler, u_char *);
typedef int (*next_packet_op_t)(pcap_t *, struct pcap_pkthdr *, u_char **);
typedef int (*inject_op_t)(pcap_t *, const void *, size_t);
typedef void (*save_current_filter_op_t)(pcap_t *, const char *);
typedef int (*setfilter_op_t)(pcap_t *, struct bpf_program *);
typedef int (*setdirection_op_t)(pcap_t *, pcap_direction_t);
typedef int (*set_datalink_op_t)(pcap_t *, int);
typedef int (*getnonblock_op_t)(pcap_t *);
typedef int (*setnonblock_op_t)(pcap_t *, int);
typedef int (*stats_op_t)(pcap_t *, struct pcap_stat *);
#ifdef _WIN32
typedef struct pcap_stat *(*stats_ex_op_t)(pcap_t *, int *);
typedef int (*setbuff_op_t)(pcap_t *, int);
typedef int (*setmode_op_t)(pcap_t *, int);
typedef int (*setmintocopy_op_t)(pcap_t *, int);
typedef HANDLE (*getevent_op_t)(pcap_t *);
typedef int (*oid_get_request_op_t)(pcap_t *, bpf_u_int32, void *, size_t *);
typedef int (*oid_set_request_op_t)(pcap_t *, bpf_u_int32, const void *, size_t *);
typedef u_int (*sendqueue_transmit_op_t)(pcap_t *, pcap_send_queue *, int);
typedef int (*setuserbuffer_op_t)(pcap_t *, int);
typedef int (*live_dump_op_t)(pcap_t *, char *, int, int);
typedef int (*live_dump_ended_op_t)(pcap_t *, int);
typedef PAirpcapHandle (*get_airpcap_handle_op_t)(pcap_t *);
#endif
typedef void (*cleanup_op_t)(pcap_t *);
/*
* We put all the stuff used in the read code path at the beginning,
* to try to keep it together in the same cache line or lines.
*/
struct pcap {
/*
* Method to call to read packets on a live capture.
*/
read_op_t read_op;
/*
* Method to call to read the next packet from a savefile.
*/
next_packet_op_t next_packet_op;
#ifdef _WIN32
HANDLE handle;
#else
int fd;
#endif /* _WIN32 */
/*
* Read buffer.
*/
u_int bufsize;
void *buffer;
u_char *bp;
int cc;
sig_atomic_t break_loop; /* flag set to force break from packet-reading loop */
void *priv; /* private data for methods */
#ifdef ENABLE_REMOTE
struct pcap_samp rmt_samp; /* parameters related to the sampling process. */
#endif
int swapped;
FILE *rfile; /* null if live capture, non-null if savefile */
u_int fddipad;
struct pcap *next; /* list of open pcaps that need stuff cleared on close */
/*
* File version number; meaningful only for a savefile, but we
* keep it here so that apps that (mistakenly) ask for the
* version numbers will get the same zero values that they
* always did.
*/
int version_major;
int version_minor;
int snapshot;
int linktype; /* Network linktype */
int linktype_ext; /* Extended information stored in the linktype field of a file */
int tzoff; /* timezone offset */
int offset; /* offset for proper alignment */
int activated; /* true if the capture is really started */
int oldstyle; /* if we're opening with pcap_open_live() */
struct pcap_opt opt;
/*
* Place holder for pcap_next().
*/
u_char *pkt;
#ifdef _WIN32
struct pcap_stat stat; /* used for pcap_stats_ex() */
#endif
/* We're accepting only packets in this direction/these directions. */
pcap_direction_t direction;
/*
* Flags to affect BPF code generation.
*/
int bpf_codegen_flags;
#if !defined(_WIN32) && !defined(MSDOS)
int selectable_fd; /* FD on which select()/poll()/epoll_wait()/kevent()/etc. can be done */
/*
* In case there either is no selectable FD, or there is but
* it doesn't necessarily work (e.g., if it doesn't get notified
* if the packet capture timeout expires before the buffer
* fills up), this points to a timeout that should be used
* in select()/poll()/epoll_wait()/kevent() call. The pcap_t should
* be put into non-blocking mode, and, if the timeout expires on
* the call, an attempt should be made to read packets from all
* pcap_t's with a required timeout, and the code must be
* prepared not to see any packets from the attempt.
*/
struct timeval *required_select_timeout;
#endif
/*
* Placeholder for filter code if bpf not in kernel.
*/
struct bpf_program fcode;
char errbuf[PCAP_ERRBUF_SIZE + 1];
int dlt_count;
u_int *dlt_list;
int tstamp_type_count;
u_int *tstamp_type_list;
int tstamp_precision_count;
u_int *tstamp_precision_list;
struct pcap_pkthdr pcap_header; /* This is needed for the pcap_next_ex() to work */
/*
* More methods.
*/
activate_op_t activate_op;
can_set_rfmon_op_t can_set_rfmon_op;
inject_op_t inject_op;
save_current_filter_op_t save_current_filter_op;
setfilter_op_t setfilter_op;
setdirection_op_t setdirection_op;
set_datalink_op_t set_datalink_op;
getnonblock_op_t getnonblock_op;
setnonblock_op_t setnonblock_op;
stats_op_t stats_op;
/*
* Routine to use as callback for pcap_next()/pcap_next_ex().
*/
pcap_handler oneshot_callback;
#ifdef _WIN32
/*
* These are, at least currently, specific to the Win32 NPF
* driver.
*/
stats_ex_op_t stats_ex_op;
setbuff_op_t setbuff_op;
setmode_op_t setmode_op;
setmintocopy_op_t setmintocopy_op;
getevent_op_t getevent_op;
oid_get_request_op_t oid_get_request_op;
oid_set_request_op_t oid_set_request_op;
sendqueue_transmit_op_t sendqueue_transmit_op;
setuserbuffer_op_t setuserbuffer_op;
live_dump_op_t live_dump_op;
live_dump_ended_op_t live_dump_ended_op;
get_airpcap_handle_op_t get_airpcap_handle_op;
#endif
cleanup_op_t cleanup_op;
};
/*
* BPF code generation flags.
*/
#define BPF_SPECIAL_VLAN_HANDLING 0x00000001 /* special VLAN handling for Linux */
/*
* This is a timeval as stored in a savefile.
* It has to use the same types everywhere, independent of the actual
* `struct timeval'; `struct timeval' has 32-bit tv_sec values on some
* platforms and 64-bit tv_sec values on other platforms, and writing
* out native `struct timeval' values would mean files could only be
* read on systems with the same tv_sec size as the system on which
* the file was written.
*/
struct pcap_timeval {
bpf_int32 tv_sec; /* seconds */
bpf_int32 tv_usec; /* microseconds */
};
/*
* This is a `pcap_pkthdr' as actually stored in a savefile.
*
* Do not change the format of this structure, in any way (this includes
* changes that only affect the length of fields in this structure),
* and do not make the time stamp anything other than seconds and
* microseconds (e.g., seconds and nanoseconds). Instead:
*
* introduce a new structure for the new format;
*
* send mail to "tcpdump-workers@lists.tcpdump.org", requesting
* a new magic number for your new capture file format, and, when
* you get the new magic number, put it in "savefile.c";
*
* use that magic number for save files with the changed record
* header;
*
* make the code in "savefile.c" capable of reading files with
* the old record header as well as files with the new record header
* (using the magic number to determine the header format).
*
* Then supply the changes by forking the branch at
*
* https://github.com/the-tcpdump-group/libpcap/issues
*
* and issuing a pull request, so that future versions of libpcap and
* programs that use it (such as tcpdump) will be able to read your new
* capture file format.
*/
struct pcap_sf_pkthdr {
struct pcap_timeval ts; /* time stamp */
bpf_u_int32 caplen; /* length of portion present */
bpf_u_int32 len; /* length this packet (off wire) */
};
/*
* How a `pcap_pkthdr' is actually stored in savefiles written
* by some patched versions of libpcap (e.g. the ones in Red
* Hat Linux 6.1 and 6.2).
*
* Do not change the format of this structure, in any way (this includes
* changes that only affect the length of fields in this structure).
* Instead, introduce a new structure, as per the above.
*/
struct pcap_sf_patched_pkthdr {
struct pcap_timeval ts; /* time stamp */
bpf_u_int32 caplen; /* length of portion present */
bpf_u_int32 len; /* length this packet (off wire) */
int index;
unsigned short protocol;
unsigned char pkt_type;
};
/*
* User data structure for the one-shot callback used for pcap_next()
* and pcap_next_ex().
*/
struct oneshot_userdata {
struct pcap_pkthdr *hdr;
const u_char **pkt;
pcap_t *pd;
};
#ifndef min
#define min(a, b) ((a) > (b) ? (b) : (a))
#endif
int pcap_offline_read(pcap_t *, int, pcap_handler, u_char *);
#include <stdarg.h>
#include "portability.h"
/*
* Does the packet count argument to a module's read routine say
* "supply packets until you run out of packets"?
*/
#define PACKET_COUNT_IS_UNLIMITED(count) ((count) <= 0)
/*
* Routines that most pcap implementations can use for non-blocking mode.
*/
#if !defined(_WIN32) && !defined(MSDOS)
int pcap_getnonblock_fd(pcap_t *);
int pcap_setnonblock_fd(pcap_t *p, int);
#endif
/*
* Internal interfaces for "pcap_create()".
*
* "pcap_create_interface()" is the routine to do a pcap_create on
* a regular network interface. There are multiple implementations
* of this, one for each platform type (Linux, BPF, DLPI, etc.),
* with the one used chosen by the configure script.
*
* "pcap_create_common()" allocates and fills in a pcap_t, for use
* by pcap_create routines.
*/
pcap_t *pcap_create_interface(const char *, char *);
pcap_t *pcap_create_common(char *, size_t);
int pcap_do_addexit(pcap_t *);
void pcap_add_to_pcaps_to_close(pcap_t *);
void pcap_remove_from_pcaps_to_close(pcap_t *);
void pcap_cleanup_live_common(pcap_t *);
int pcap_check_activated(pcap_t *);
/*
* Internal interfaces for "pcap_findalldevs()".
*
* A pcap_if_list_t * is a reference to a list of devices.
*
* A get_if_flags_func is a platform-dependent function called to get
* additional interface flags.
*
* "pcap_platform_finddevs()" is the platform-dependent routine to
* find local network interfaces.
*
* "pcap_findalldevs_interfaces()" is a helper to find those interfaces
* using the "standard" mechanisms (SIOCGIFCONF, "getifaddrs()", etc.).
*
* "add_dev()" adds an entry to a pcap_if_list_t.
*
* "find_dev()" tries to find a device, by name, in a pcap_if_list_t.
*
* "find_or_add_dev()" checks whether a device is already in a pcap_if_list_t
* and, if not, adds an entry for it.
*/
struct pcap_if_list;
typedef struct pcap_if_list pcap_if_list_t;
typedef int (*get_if_flags_func)(const char *, bpf_u_int32 *, char *);
int pcap_platform_finddevs(pcap_if_list_t *, char *);
#if !defined(_WIN32) && !defined(MSDOS)
int pcap_findalldevs_interfaces(pcap_if_list_t *, char *,
int (*)(const char *), get_if_flags_func);
#endif
pcap_if_t *find_or_add_dev(pcap_if_list_t *, const char *, bpf_u_int32,
get_if_flags_func, const char *, char *);
pcap_if_t *find_dev(pcap_if_list_t *, const char *);
pcap_if_t *add_dev(pcap_if_list_t *, const char *, bpf_u_int32, const char *,
char *);
int add_addr_to_dev(pcap_if_t *, struct sockaddr *, size_t,
struct sockaddr *, size_t, struct sockaddr *, size_t,
struct sockaddr *dstaddr, size_t, char *errbuf);
#ifndef _WIN32
pcap_if_t *find_or_add_if(pcap_if_list_t *, const char *, bpf_u_int32,
get_if_flags_func, char *);
int add_addr_to_if(pcap_if_list_t *, const char *, bpf_u_int32,
get_if_flags_func,
struct sockaddr *, size_t, struct sockaddr *, size_t,
struct sockaddr *, size_t, struct sockaddr *, size_t, char *);
#endif
/*
* Internal interfaces for "pcap_open_offline()".
*
* "pcap_open_offline_common()" allocates and fills in a pcap_t, for use
* by pcap_open_offline routines.
*
* "pcap_adjust_snapshot()" adjusts the snapshot to be non-zero and
* fit within an int.
*
* "sf_cleanup()" closes the file handle associated with a pcap_t, if
* appropriate, and frees all data common to all modules for handling
* savefile types.
*/
pcap_t *pcap_open_offline_common(char *ebuf, size_t size);
bpf_u_int32 pcap_adjust_snapshot(bpf_u_int32 linktype, bpf_u_int32 snaplen);
void sf_cleanup(pcap_t *p);
/*
* Internal interfaces for doing user-mode filtering of packets and
* validating filter programs.
*/
/*
* Auxiliary data, for use when interpreting a filter intended for the
* Linux kernel when the kernel rejects the filter (requiring us to
* run it in userland). It contains VLAN tag information.
*/
struct bpf_aux_data {
u_short vlan_tag_present;
u_short vlan_tag;
};
/*
* Filtering routine that takes the auxiliary data as an additional
* argument.
*/
u_int bpf_filter_with_aux_data(const struct bpf_insn *,
const u_char *, u_int, u_int, const struct bpf_aux_data *);
/*
* Internal interfaces for both "pcap_create()" and routines that
* open savefiles.
*
* "pcap_oneshot()" is the standard one-shot callback for "pcap_next()"
* and "pcap_next_ex()".
*/
void pcap_oneshot(u_char *, const struct pcap_pkthdr *, const u_char *);
int install_bpf_program(pcap_t *, struct bpf_program *);
int pcap_strcasecmp(const char *, const char *);
#ifdef YYDEBUG
extern int pcap_debug;
#endif
#ifdef __cplusplus
}
#endif
#endif