diff options
author | Alberto Gonzalez Iniesta <agi@inittab.org> | 2016-12-27 18:25:47 +0100 |
---|---|---|
committer | Alberto Gonzalez Iniesta <agi@inittab.org> | 2016-12-27 18:25:47 +0100 |
commit | 3a2bbdb05ca6a6996e424c9fb225cb0d53804125 (patch) | |
tree | f29063da5bec4caf3853d49a22a09c8619eebd21 /src/openvpn/mroute.h | |
parent | d53dba59e78da865c4fe820386ff2f4f76925f3b (diff) |
New upstream version 2.4.0upstream/2.4.0
Diffstat (limited to 'src/openvpn/mroute.h')
-rw-r--r-- | src/openvpn/mroute.h | 226 |
1 files changed, 122 insertions, 104 deletions
diff --git a/src/openvpn/mroute.h b/src/openvpn/mroute.h index 8f7a064..0698348 100644 --- a/src/openvpn/mroute.h +++ b/src/openvpn/mroute.h @@ -5,7 +5,7 @@ * packet encryption, packet authentication, and * packet compression. * - * Copyright (C) 2002-2010 OpenVPN Technologies, Inc. <sales@openvpn.net> + * Copyright (C) 2002-2017 OpenVPN Technologies, Inc. <sales@openvpn.net> * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License version 2 @@ -76,49 +76,49 @@ #define MR_ARP 16 struct mroute_addr { - uint8_t len; /* length of address */ - uint8_t unused; - uint8_t type; /* MR_ADDR/MR_WITH flags */ - uint8_t netbits; /* number of bits in network part of address, - valid if MR_WITH_NETBITS is set */ - union { - uint8_t raw_addr[MR_MAX_ADDR_LEN]; /* actual address */ - uint8_t eth_addr[OPENVPN_ETH_ALEN]; - struct { - in_addr_t addr; /* _network order_ IPv4 address */ - in_port_t port; /* _network order_ TCP/UDP port */ - } v4; - struct { - struct in6_addr addr; - in_port_t port; /* _network order_ TCP/UDP port */ - } v6; - struct { - uint8_t prefix[12]; - in_addr_t addr; /* _network order_ IPv4 address */ - } v4mappedv6; - } + uint8_t len; /* length of address */ + uint8_t unused; + uint8_t type; /* MR_ADDR/MR_WITH flags */ + uint8_t netbits; /* number of bits in network part of address, + * valid if MR_WITH_NETBITS is set */ + union { + uint8_t raw_addr[MR_MAX_ADDR_LEN]; /* actual address */ + uint8_t eth_addr[OPENVPN_ETH_ALEN]; + struct { + in_addr_t addr; /* _network order_ IPv4 address */ + in_port_t port; /* _network order_ TCP/UDP port */ + } v4; + struct { + struct in6_addr addr; + in_port_t port; /* _network order_ TCP/UDP port */ + } v6; + struct { + uint8_t prefix[12]; + in_addr_t addr; /* _network order_ IPv4 address */ + } v4mappedv6; + } #ifndef HAVE_ANONYMOUS_UNION_SUPPORT /* Wrappers to support compilers that do not grok anonymous unions */ - mroute_union + mroute_union #define raw_addr mroute_union.raw_addr #define eth_addr mroute_union.eth_addr #define v4 mroute_union.v4 #define v6 mroute_union.v6 #define v4mappedv6 mroute_union.v4mappedv6 #endif - ; + ; }; /* Double-check that struct packing works as expected */ -static_assert (offsetof(struct mroute_addr, v4.port) == - offsetof(struct mroute_addr, v4) + 4, - "Unexpected struct packing of v4"); -static_assert (offsetof(struct mroute_addr, v6.port) == - offsetof(struct mroute_addr, v6) + 16, - "Unexpected struct packing of v6"); -static_assert (offsetof(struct mroute_addr, v4mappedv6.addr) == - offsetof(struct mroute_addr, v4mappedv6) + 12, - "Unexpected struct packing of v4mappedv6"); +static_assert(offsetof(struct mroute_addr, v4.port) == + offsetof(struct mroute_addr, v4) + 4, + "Unexpected struct packing of v4"); +static_assert(offsetof(struct mroute_addr, v6.port) == + offsetof(struct mroute_addr, v6) + 16, + "Unexpected struct packing of v6"); +static_assert(offsetof(struct mroute_addr, v4mappedv6.addr) == + offsetof(struct mroute_addr, v4mappedv6) + 12, + "Unexpected struct packing of v4mappedv6"); /* * Number of bits in an address. Should be raised for IPv6. @@ -129,122 +129,140 @@ static_assert (offsetof(struct mroute_addr, v4mappedv6.addr) == * Used to help maintain CIDR routing table. */ struct mroute_helper { - unsigned int cache_generation; /* incremented when route added */ - int ageable_ttl_secs; /* host route cache entry time-to-live*/ - int n_net_len; /* length of net_len array */ - uint8_t net_len[MR_HELPER_NET_LEN]; /* CIDR netlengths in descending order */ - int net_len_refcount[MR_HELPER_NET_LEN]; /* refcount of each netlength */ + unsigned int cache_generation; /* incremented when route added */ + int ageable_ttl_secs; /* host route cache entry time-to-live*/ + int n_net_len; /* length of net_len array */ + uint8_t net_len[MR_HELPER_NET_LEN]; /* CIDR netlengths in descending order */ + int net_len_refcount[MR_HELPER_NET_LEN]; /* refcount of each netlength */ }; struct openvpn_sockaddr; -bool mroute_extract_openvpn_sockaddr (struct mroute_addr *addr, - const struct openvpn_sockaddr *osaddr, - bool use_port); +bool mroute_extract_openvpn_sockaddr(struct mroute_addr *addr, + const struct openvpn_sockaddr *osaddr, + bool use_port); -bool mroute_learnable_address (const struct mroute_addr *addr); +bool mroute_learnable_address(const struct mroute_addr *addr); -uint32_t mroute_addr_hash_function (const void *key, uint32_t iv); -bool mroute_addr_compare_function (const void *key1, const void *key2); +uint32_t mroute_addr_hash_function(const void *key, uint32_t iv); -void mroute_addr_init (struct mroute_addr *addr); +bool mroute_addr_compare_function(const void *key1, const void *key2); -const char *mroute_addr_print (const struct mroute_addr *ma, - struct gc_arena *gc); +void mroute_addr_init(struct mroute_addr *addr); + +const char *mroute_addr_print(const struct mroute_addr *ma, + struct gc_arena *gc); #define MAPF_SUBNET (1<<0) #define MAPF_IA_EMPTY_IF_UNDEF (1<<1) #define MAPF_SHOW_ARP (1<<2) -const char *mroute_addr_print_ex (const struct mroute_addr *ma, - const unsigned int flags, - struct gc_arena *gc); +const char *mroute_addr_print_ex(const struct mroute_addr *ma, + const unsigned int flags, + struct gc_arena *gc); + +void mroute_addr_mask_host_bits(struct mroute_addr *ma); + +struct mroute_helper *mroute_helper_init(int ageable_ttl_secs); + +void mroute_helper_free(struct mroute_helper *mh); -void mroute_addr_mask_host_bits (struct mroute_addr *ma); +void mroute_helper_add_iroute46(struct mroute_helper *mh, int netbits); -struct mroute_helper *mroute_helper_init (int ageable_ttl_secs); -void mroute_helper_free (struct mroute_helper *mh); -void mroute_helper_add_iroute46 (struct mroute_helper *mh, int netbits); -void mroute_helper_del_iroute46 (struct mroute_helper *mh, int netbits); +void mroute_helper_del_iroute46(struct mroute_helper *mh, int netbits); /* * Given a raw packet in buf, return the src and dest * addresses of the packet. */ static inline unsigned int -mroute_extract_addr_from_packet (struct mroute_addr *src, - struct mroute_addr *dest, - struct mroute_addr *esrc, - struct mroute_addr *edest, - const struct buffer *buf, - int tunnel_type) +mroute_extract_addr_from_packet(struct mroute_addr *src, + struct mroute_addr *dest, + struct mroute_addr *esrc, + struct mroute_addr *edest, + const struct buffer *buf, + int tunnel_type) { - unsigned int mroute_extract_addr_ipv4 (struct mroute_addr *src, - struct mroute_addr *dest, - const struct buffer *buf); - - unsigned int mroute_extract_addr_ether (struct mroute_addr *src, - struct mroute_addr *dest, - struct mroute_addr *esrc, - struct mroute_addr *edest, - const struct buffer *buf); - unsigned int ret = 0; - verify_align_4 (buf); - if (tunnel_type == DEV_TYPE_TUN) - ret = mroute_extract_addr_ipv4 (src, dest, buf); - else if (tunnel_type == DEV_TYPE_TAP) - ret = mroute_extract_addr_ether (src, dest, esrc, edest, buf); - return ret; + unsigned int mroute_extract_addr_ipv4(struct mroute_addr *src, + struct mroute_addr *dest, + const struct buffer *buf); + + unsigned int mroute_extract_addr_ether(struct mroute_addr *src, + struct mroute_addr *dest, + struct mroute_addr *esrc, + struct mroute_addr *edest, + const struct buffer *buf); + + unsigned int ret = 0; + verify_align_4(buf); + if (tunnel_type == DEV_TYPE_TUN) + { + ret = mroute_extract_addr_ipv4(src, dest, buf); + } + else if (tunnel_type == DEV_TYPE_TAP) + { + ret = mroute_extract_addr_ether(src, dest, esrc, edest, buf); + } + return ret; } static inline bool -mroute_addr_equal (const struct mroute_addr *a1, const struct mroute_addr *a2) +mroute_addr_equal(const struct mroute_addr *a1, const struct mroute_addr *a2) { - if (a1->type != a2->type) - return false; - if (a1->netbits != a2->netbits) - return false; - if (a1->len != a2->len) - return false; - return memcmp (a1->raw_addr, a2->raw_addr, a1->len) == 0; + if (a1->type != a2->type) + { + return false; + } + if (a1->netbits != a2->netbits) + { + return false; + } + if (a1->len != a2->len) + { + return false; + } + return memcmp(a1->raw_addr, a2->raw_addr, a1->len) == 0; } static inline const uint8_t * -mroute_addr_hash_ptr (const struct mroute_addr *a) +mroute_addr_hash_ptr(const struct mroute_addr *a) { - /* NOTE: depends on ordering of struct mroute_addr */ - return (uint8_t *) &a->type; + /* NOTE: depends on ordering of struct mroute_addr */ + return (uint8_t *) &a->type; } static inline uint32_t -mroute_addr_hash_len (const struct mroute_addr *a) +mroute_addr_hash_len(const struct mroute_addr *a) { - return (uint32_t) a->len + 2; + return (uint32_t) a->len + 2; } static inline void -mroute_extract_in_addr_t (struct mroute_addr *dest, const in_addr_t src) +mroute_extract_in_addr_t(struct mroute_addr *dest, const in_addr_t src) { - dest->type = MR_ADDR_IPV4; - dest->netbits = 0; - dest->len = 4; - dest->v4.addr = htonl (src); + dest->type = MR_ADDR_IPV4; + dest->netbits = 0; + dest->len = 4; + dest->v4.addr = htonl(src); } static inline in_addr_t -in_addr_t_from_mroute_addr (const struct mroute_addr *addr) +in_addr_t_from_mroute_addr(const struct mroute_addr *addr) { - if ((addr->type & MR_ADDR_MASK) == MR_ADDR_IPV4 && addr->netbits == 0 && addr->len == 4) { - return ntohl(addr->v4.addr); - } else { - return 0; - } + if ((addr->type & MR_ADDR_MASK) == MR_ADDR_IPV4 && addr->netbits == 0 && addr->len == 4) + { + return ntohl(addr->v4.addr); + } + else + { + return 0; + } } static inline void -mroute_addr_reset (struct mroute_addr *ma) +mroute_addr_reset(struct mroute_addr *ma) { - ma->len = 0; - ma->type = MR_ADDR_NONE; + ma->len = 0; + ma->type = MR_ADDR_NONE; } #endif /* P2MP_SERVER */ |