X-Git-Url: https://git.saurik.com/apple/xnu.git/blobdiff_plain/3903760236c30e3b5ace7a4eefac3a269d68957c..0a7de7458d150b5d4dffc935ba399be265ef0a1a:/bsd/netinet/ip_input.c diff --git a/bsd/netinet/ip_input.c b/bsd/netinet/ip_input.c index 35d7a9f20..35ee30666 100644 --- a/bsd/netinet/ip_input.c +++ b/bsd/netinet/ip_input.c @@ -1,5 +1,5 @@ /* - * Copyright (c) 2000-2016 Apple Inc. All rights reserved. + * Copyright (c) 2000-2018 Apple Inc. All rights reserved. * * @APPLE_OSREFERENCE_LICENSE_HEADER_START@ * @@ -66,7 +66,7 @@ * Version 2.0. */ -#define _IP_VHL +#define _IP_VHL #include #include @@ -137,24 +137,24 @@ #include #endif /* IPSEC */ -#define DBG_LAYER_BEG NETDBG_CODE(DBG_NETIP, 0) -#define DBG_LAYER_END NETDBG_CODE(DBG_NETIP, 2) -#define DBG_FNC_IP_INPUT NETDBG_CODE(DBG_NETIP, (2 << 8)) +#define DBG_LAYER_BEG NETDBG_CODE(DBG_NETIP, 0) +#define DBG_LAYER_END NETDBG_CODE(DBG_NETIP, 2) +#define DBG_FNC_IP_INPUT NETDBG_CODE(DBG_NETIP, (2 << 8)) #if IPSEC extern int ipsec_bypass; extern lck_mtx_t *sadb_mutex; -lck_grp_t *sadb_stat_mutex_grp; -lck_grp_attr_t *sadb_stat_mutex_grp_attr; -lck_attr_t *sadb_stat_mutex_attr; +lck_grp_t *sadb_stat_mutex_grp; +lck_grp_attr_t *sadb_stat_mutex_grp_attr; +lck_attr_t *sadb_stat_mutex_attr; decl_lck_mtx_data(, sadb_stat_mutex_data); -lck_mtx_t *sadb_stat_mutex = &sadb_stat_mutex_data; +lck_mtx_t *sadb_stat_mutex = &sadb_stat_mutex_data; #endif /* IPSEC */ MBUFQ_HEAD(fq_head); -static int frag_timeout_run; /* frag timer is scheduled to run */ +static int frag_timeout_run; /* frag timer is scheduled to run */ static void frag_timeout(void *); static void frag_sched_timeout(void); @@ -165,24 +165,24 @@ static void ip_input_second_pass(struct mbuf *, struct ifnet *, u_int32_t, int, int, struct ip_fw_in_args *, int); decl_lck_mtx_data(static, ipqlock); -static lck_attr_t *ipqlock_attr; -static lck_grp_t *ipqlock_grp; -static lck_grp_attr_t *ipqlock_grp_attr; +static lck_attr_t *ipqlock_attr; +static lck_grp_t *ipqlock_grp; +static lck_grp_attr_t *ipqlock_grp_attr; /* Packet reassembly stuff */ -#define IPREASS_NHASH_LOG2 6 -#define IPREASS_NHASH (1 << IPREASS_NHASH_LOG2) -#define IPREASS_HMASK (IPREASS_NHASH - 1) -#define IPREASS_HASH(x, y) \ +#define IPREASS_NHASH_LOG2 6 +#define IPREASS_NHASH (1 << IPREASS_NHASH_LOG2) +#define IPREASS_HMASK (IPREASS_NHASH - 1) +#define IPREASS_HASH(x, y) \ (((((x) & 0xF) | ((((x) >> 8) & 0xF) << 4)) ^ (y)) & IPREASS_HMASK) /* IP fragment reassembly queues (protected by ipqlock) */ static TAILQ_HEAD(ipqhead, ipq) ipq[IPREASS_NHASH]; /* ip reassembly queues */ -static int maxnipq; /* max packets in reass queues */ -static u_int32_t maxfragsperpacket; /* max frags/packet in reass queues */ -static u_int32_t nipq; /* # of packets in reass queues */ -static u_int32_t ipq_limit; /* ipq allocation limit */ -static u_int32_t ipq_count; /* current # of allocated ipq's */ +static int maxnipq; /* max packets in reass queues */ +static u_int32_t maxfragsperpacket; /* max frags/packet in reass queues */ +static u_int32_t nipq; /* # of packets in reass queues */ +static u_int32_t ipq_limit; /* ipq allocation limit */ +static u_int32_t ipq_count; /* current # of allocated ipq's */ static int sysctl_ipforwarding SYSCTL_HANDLER_ARGS; static int sysctl_maxnipq SYSCTL_HANDLER_ARGS; @@ -196,49 +196,54 @@ static int sysctl_ip_input_getperf SYSCTL_HANDLER_ARGS; int ipforwarding = 0; SYSCTL_PROC(_net_inet_ip, IPCTL_FORWARDING, forwarding, - CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED, &ipforwarding, 0, - sysctl_ipforwarding, "I", "Enable IP forwarding between interfaces"); + CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED, &ipforwarding, 0, + sysctl_ipforwarding, "I", "Enable IP forwarding between interfaces"); static int ipsendredirects = 1; /* XXX */ SYSCTL_INT(_net_inet_ip, IPCTL_SENDREDIRECTS, redirect, - CTLFLAG_RW | CTLFLAG_LOCKED, &ipsendredirects, 0, - "Enable sending IP redirects"); + CTLFLAG_RW | CTLFLAG_LOCKED, &ipsendredirects, 0, + "Enable sending IP redirects"); int ip_defttl = IPDEFTTL; SYSCTL_INT(_net_inet_ip, IPCTL_DEFTTL, ttl, CTLFLAG_RW | CTLFLAG_LOCKED, - &ip_defttl, 0, "Maximum TTL on IP packets"); + &ip_defttl, 0, "Maximum TTL on IP packets"); static int ip_dosourceroute = 0; SYSCTL_INT(_net_inet_ip, IPCTL_SOURCEROUTE, sourceroute, - CTLFLAG_RW | CTLFLAG_LOCKED, &ip_dosourceroute, 0, - "Enable forwarding source routed IP packets"); + CTLFLAG_RW | CTLFLAG_LOCKED, &ip_dosourceroute, 0, + "Enable forwarding source routed IP packets"); static int ip_acceptsourceroute = 0; SYSCTL_INT(_net_inet_ip, IPCTL_ACCEPTSOURCEROUTE, accept_sourceroute, - CTLFLAG_RW | CTLFLAG_LOCKED, &ip_acceptsourceroute, 0, - "Enable accepting source routed IP packets"); + CTLFLAG_RW | CTLFLAG_LOCKED, &ip_acceptsourceroute, 0, + "Enable accepting source routed IP packets"); static int ip_sendsourcequench = 0; SYSCTL_INT(_net_inet_ip, OID_AUTO, sendsourcequench, - CTLFLAG_RW | CTLFLAG_LOCKED, &ip_sendsourcequench, 0, - "Enable the transmission of source quench packets"); + CTLFLAG_RW | CTLFLAG_LOCKED, &ip_sendsourcequench, 0, + "Enable the transmission of source quench packets"); SYSCTL_PROC(_net_inet_ip, OID_AUTO, maxfragpackets, - CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED, &maxnipq, 0, sysctl_maxnipq, - "I", "Maximum number of IPv4 fragment reassembly queue entries"); + CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED, &maxnipq, 0, sysctl_maxnipq, + "I", "Maximum number of IPv4 fragment reassembly queue entries"); SYSCTL_UINT(_net_inet_ip, OID_AUTO, fragpackets, CTLFLAG_RD | CTLFLAG_LOCKED, - &nipq, 0, "Current number of IPv4 fragment reassembly queue entries"); + &nipq, 0, "Current number of IPv4 fragment reassembly queue entries"); SYSCTL_PROC(_net_inet_ip, OID_AUTO, maxfragsperpacket, - CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED, &maxfragsperpacket, 0, - sysctl_maxfragsperpacket, "I", - "Maximum number of IPv4 fragments allowed per packet"); + CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED, &maxfragsperpacket, 0, + sysctl_maxfragsperpacket, "I", + "Maximum number of IPv4 fragments allowed per packet"); static uint32_t ip_adj_clear_hwcksum = 0; SYSCTL_UINT(_net_inet_ip, OID_AUTO, adj_clear_hwcksum, - CTLFLAG_RW | CTLFLAG_LOCKED, &ip_adj_clear_hwcksum, 0, - "Invalidate hwcksum info when adjusting length"); + CTLFLAG_RW | CTLFLAG_LOCKED, &ip_adj_clear_hwcksum, 0, + "Invalidate hwcksum info when adjusting length"); + +static uint32_t ip_adj_partial_sum = 1; +SYSCTL_UINT(_net_inet_ip, OID_AUTO, adj_partial_sum, + CTLFLAG_RW | CTLFLAG_LOCKED, &ip_adj_partial_sum, 0, + "Perform partial sum adjustment of trailing bytes at IP layer"); /* * XXX - Setting ip_checkinterface mostly implements the receive side of @@ -255,33 +260,33 @@ SYSCTL_UINT(_net_inet_ip, OID_AUTO, adj_clear_hwcksum, */ static int ip_checkinterface = 0; SYSCTL_INT(_net_inet_ip, OID_AUTO, check_interface, CTLFLAG_RW | CTLFLAG_LOCKED, - &ip_checkinterface, 0, "Verify packet arrives on correct interface"); + &ip_checkinterface, 0, "Verify packet arrives on correct interface"); static int ip_chaining = 1; SYSCTL_INT(_net_inet_ip, OID_AUTO, rx_chaining, CTLFLAG_RW | CTLFLAG_LOCKED, - &ip_chaining, 1, "Do receive side ip address based chaining"); + &ip_chaining, 1, "Do receive side ip address based chaining"); static int ip_chainsz = 6; SYSCTL_INT(_net_inet_ip, OID_AUTO, rx_chainsz, CTLFLAG_RW | CTLFLAG_LOCKED, - &ip_chainsz, 1, "IP receive side max chaining"); + &ip_chainsz, 1, "IP receive side max chaining"); #if (DEBUG || DEVELOPMENT) static int ip_input_measure = 0; SYSCTL_PROC(_net_inet_ip, OID_AUTO, input_perf, - CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED, - &ip_input_measure, 0, sysctl_reset_ip_input_stats, "I", "Do time measurement"); + CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED, + &ip_input_measure, 0, sysctl_reset_ip_input_stats, "I", "Do time measurement"); static uint64_t ip_input_measure_bins = 0; SYSCTL_PROC(_net_inet_ip, OID_AUTO, input_perf_bins, - CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED, &ip_input_measure_bins, 0, - sysctl_ip_input_measure_bins, "I", - "bins for chaining performance data histogram"); + CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED, &ip_input_measure_bins, 0, + sysctl_ip_input_measure_bins, "I", + "bins for chaining performance data histogram"); static net_perf_t net_perf; SYSCTL_PROC(_net_inet_ip, OID_AUTO, input_perf_data, - CTLTYPE_STRUCT | CTLFLAG_RD | CTLFLAG_LOCKED, - 0, 0, sysctl_ip_input_getperf, "S,net_perf", - "IP input performance data (struct net_perf, net/net_perf.h)"); + CTLTYPE_STRUCT | CTLFLAG_RD | CTLFLAG_LOCKED, + 0, 0, sysctl_ip_input_getperf, "S,net_perf", + "IP input performance data (struct net_perf, net/net_perf.h)"); #endif /* (DEBUG || DEVELOPMENT) */ #if DIAGNOSTIC @@ -290,36 +295,36 @@ static int ipprintfs = 0; struct protosw *ip_protox[IPPROTO_MAX]; -static lck_grp_attr_t *in_ifaddr_rwlock_grp_attr; -static lck_grp_t *in_ifaddr_rwlock_grp; -static lck_attr_t *in_ifaddr_rwlock_attr; +static lck_grp_attr_t *in_ifaddr_rwlock_grp_attr; +static lck_grp_t *in_ifaddr_rwlock_grp; +static lck_attr_t *in_ifaddr_rwlock_attr; decl_lck_rw_data(, in_ifaddr_rwlock_data); -lck_rw_t *in_ifaddr_rwlock = &in_ifaddr_rwlock_data; +lck_rw_t *in_ifaddr_rwlock = &in_ifaddr_rwlock_data; /* Protected by in_ifaddr_rwlock */ -struct in_ifaddrhead in_ifaddrhead; /* first inet address */ -struct in_ifaddrhashhead *in_ifaddrhashtbl; /* inet addr hash table */ +struct in_ifaddrhead in_ifaddrhead; /* first inet address */ +struct in_ifaddrhashhead *in_ifaddrhashtbl; /* inet addr hash table */ -#define INADDR_NHASH 61 -static u_int32_t inaddr_nhash; /* hash table size */ -static u_int32_t inaddr_hashp; /* next largest prime */ +#define INADDR_NHASH 61 +static u_int32_t inaddr_nhash; /* hash table size */ +static u_int32_t inaddr_hashp; /* next largest prime */ static int ip_getstat SYSCTL_HANDLER_ARGS; struct ipstat ipstat; SYSCTL_PROC(_net_inet_ip, IPCTL_STATS, stats, - CTLTYPE_STRUCT | CTLFLAG_RD | CTLFLAG_LOCKED, - 0, 0, ip_getstat, "S,ipstat", - "IP statistics (struct ipstat, netinet/ip_var.h)"); + CTLTYPE_STRUCT | CTLFLAG_RD | CTLFLAG_LOCKED, + 0, 0, ip_getstat, "S,ipstat", + "IP statistics (struct ipstat, netinet/ip_var.h)"); #if IPCTL_DEFMTU SYSCTL_INT(_net_inet_ip, IPCTL_DEFMTU, mtu, CTLFLAG_RW | CTLFLAG_LOCKED, - &ip_mtu, 0, "Default MTU"); + &ip_mtu, 0, "Default MTU"); #endif /* IPCTL_DEFMTU */ #if IPSTEALTH -static int ipstealth = 0; +static int ipstealth = 0; SYSCTL_INT(_net_inet_ip, OID_AUTO, stealth, CTLFLAG_RW | CTLFLAG_LOCKED, - &ipstealth, 0, ""); + &ipstealth, 0, ""); #endif /* IPSTEALTH */ /* Firewall hooks */ @@ -335,20 +340,20 @@ ip_dn_io_t *ip_dn_io_ptr; #endif /* DUMMYNET */ SYSCTL_NODE(_net_inet_ip, OID_AUTO, linklocal, - CTLFLAG_RW | CTLFLAG_LOCKED, 0, "link local"); + CTLFLAG_RW | CTLFLAG_LOCKED, 0, "link local"); struct ip_linklocal_stat ip_linklocal_stat; SYSCTL_STRUCT(_net_inet_ip_linklocal, OID_AUTO, stat, - CTLFLAG_RD | CTLFLAG_LOCKED, &ip_linklocal_stat, ip_linklocal_stat, - "Number of link local packets with TTL less than 255"); + CTLFLAG_RD | CTLFLAG_LOCKED, &ip_linklocal_stat, ip_linklocal_stat, + "Number of link local packets with TTL less than 255"); SYSCTL_NODE(_net_inet_ip_linklocal, OID_AUTO, in, - CTLFLAG_RW | CTLFLAG_LOCKED, 0, "link local input"); + CTLFLAG_RW | CTLFLAG_LOCKED, 0, "link local input"); int ip_linklocal_in_allowbadttl = 1; SYSCTL_INT(_net_inet_ip_linklocal_in, OID_AUTO, allowbadttl, - CTLFLAG_RW | CTLFLAG_LOCKED, &ip_linklocal_in_allowbadttl, 0, - "Allow incoming link local packets with TTL less than 255"); + CTLFLAG_RW | CTLFLAG_LOCKED, &ip_linklocal_in_allowbadttl, 0, + "Allow incoming link local packets with TTL less than 255"); /* @@ -358,12 +363,12 @@ SYSCTL_INT(_net_inet_ip_linklocal_in, OID_AUTO, allowbadttl, * maintenance when the remote end is on a network that is not known * to us. */ -static int ip_nhops = 0; -static struct ip_srcrt { - struct in_addr dst; /* final destination */ - char nop; /* one NOP to align */ - char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ - struct in_addr route[MAX_IPOPTLEN / sizeof (struct in_addr)]; +static int ip_nhops = 0; +static struct ip_srcrt { + struct in_addr dst; /* final destination */ + char nop; /* one NOP to align */ + char srcopt[IPOPT_OFFSET + 1]; /* OPTVAL, OLEN and OFFSET */ + struct in_addr route[MAX_IPOPTLEN / sizeof(struct in_addr)]; } ip_srcrt; static void in_ifaddrhashtbl_init(void); @@ -386,7 +391,7 @@ static inline u_short ip_cksum(struct mbuf *, int); int ip_use_randomid = 1; SYSCTL_INT(_net_inet_ip, OID_AUTO, random_id, CTLFLAG_RW | CTLFLAG_LOCKED, - &ip_use_randomid, 0, "Randomize IP packets IDs"); + &ip_use_randomid, 0, "Randomize IP packets IDs"); /* * On platforms which require strict alignment (currently for anything but @@ -396,27 +401,27 @@ SYSCTL_INT(_net_inet_ip, OID_AUTO, random_id, CTLFLAG_RW | CTLFLAG_LOCKED, * it's needed later on. */ #if defined(__i386__) || defined(__x86_64__) -#define IP_HDR_ALIGNMENT_FIXUP(_m, _ifp, _action) do { } while (0) +#define IP_HDR_ALIGNMENT_FIXUP(_m, _ifp, _action) do { } while (0) #else /* !__i386__ && !__x86_64__ */ -#define IP_HDR_ALIGNMENT_FIXUP(_m, _ifp, _action) do { \ - if (!IP_HDR_ALIGNED_P(mtod(_m, caddr_t))) { \ - struct mbuf *_n; \ - struct ifnet *__ifp = (_ifp); \ - atomic_add_64(&(__ifp)->if_alignerrs, 1); \ - if (((_m)->m_flags & M_PKTHDR) && \ - (_m)->m_pkthdr.pkt_hdr != NULL) \ - (_m)->m_pkthdr.pkt_hdr = NULL; \ - _n = m_defrag_offset(_m, max_linkhdr, M_NOWAIT); \ - if (_n == NULL) { \ - atomic_add_32(&ipstat.ips_toosmall, 1); \ - m_freem(_m); \ - (_m) = NULL; \ - _action; \ - } else { \ - VERIFY(_n != (_m)); \ - (_m) = _n; \ - } \ - } \ +#define IP_HDR_ALIGNMENT_FIXUP(_m, _ifp, _action) do { \ + if (!IP_HDR_ALIGNED_P(mtod(_m, caddr_t))) { \ + struct mbuf *_n; \ + struct ifnet *__ifp = (_ifp); \ + atomic_add_64(&(__ifp)->if_alignerrs, 1); \ + if (((_m)->m_flags & M_PKTHDR) && \ + (_m)->m_pkthdr.pkt_hdr != NULL) \ + (_m)->m_pkthdr.pkt_hdr = NULL; \ + _n = m_defrag_offset(_m, max_linkhdr, M_NOWAIT); \ + if (_n == NULL) { \ + atomic_add_32(&ipstat.ips_toosmall, 1); \ + m_freem(_m); \ + (_m) = NULL; \ + _action; \ + } else { \ + VERIFY(_n != (_m)); \ + (_m) = _n; \ + } \ + } \ } while (0) #endif /* !__i386__ && !__x86_64__ */ @@ -439,9 +444,10 @@ ip_init_delayed(void) sin->sin_family = AF_INET; sin->sin_addr.s_addr = htonl(INADDR_LOOPBACK); error = in_control(NULL, SIOCSIFADDR, (caddr_t)&ifr, lo_ifp, kernproc); - if (error) + if (error) { printf("%s: failed to initialise lo0's address, error=%d\n", __func__, error); + } } /* @@ -457,19 +463,20 @@ ip_init(struct protosw *pp, struct domain *dp) int i; domain_proto_mtx_lock_assert_held(); - VERIFY((pp->pr_flags & (PR_INITIALIZED|PR_ATTACHED)) == PR_ATTACHED); + VERIFY((pp->pr_flags & (PR_INITIALIZED | PR_ATTACHED)) == PR_ATTACHED); /* ipq_alloc() uses mbufs for IP fragment queue structures */ - _CASSERT(sizeof (struct ipq) <= _MLEN); + _CASSERT(sizeof(struct ipq) <= _MLEN); /* * Some ioctls (e.g. SIOCAIFADDR) use ifaliasreq struct, which is * interchangeable with in_aliasreq; they must have the same size. */ - _CASSERT(sizeof (struct ifaliasreq) == sizeof (struct in_aliasreq)); + _CASSERT(sizeof(struct ifaliasreq) == sizeof(struct in_aliasreq)); - if (ip_initialized) + if (ip_initialized) { return; + } ip_initialized = 1; in_ifaddr_init(); @@ -494,8 +501,9 @@ ip_init(struct protosw *pp, struct domain *dp) } /* Initialize the entire ip_protox[] array to IPPROTO_RAW. */ - for (i = 0; i < IPPROTO_MAX; i++) + for (i = 0; i < IPPROTO_MAX; i++) { ip_protox[i] = pr; + } /* * Cycle through IP protocols and put them into the appropriate place * in ip_protox[], skipping protocols IPPROTO_{IP,RAW}. @@ -505,8 +513,9 @@ ip_init(struct protosw *pp, struct domain *dp) VERIFY(pr->pr_domain == dp); if (pr->pr_protocol != 0 && pr->pr_protocol != IPPROTO_RAW) { /* Be careful to only index valid IP protocols. */ - if (pr->pr_protocol < IPPROTO_MAX) + if (pr->pr_protocol < IPPROTO_MAX) { ip_protox[pr->pr_protocol] = pr; + } } } @@ -518,8 +527,9 @@ ip_init(struct protosw *pp, struct domain *dp) lck_mtx_lock(&ipqlock); /* Initialize IP reassembly queue. */ - for (i = 0; i < IPREASS_NHASH; i++) + for (i = 0; i < IPREASS_NHASH; i++) { TAILQ_INIT(&ipq[i]); + } maxnipq = nmbclusters / 32; maxfragsperpacket = 128; /* enough for 64k in 512 byte fragments */ @@ -553,19 +563,22 @@ in_ifaddrhashtbl_init(void) { int i, k, p; - if (in_ifaddrhashtbl != NULL) + if (in_ifaddrhashtbl != NULL) { return; + } PE_parse_boot_argn("inaddr_nhash", &inaddr_nhash, - sizeof (inaddr_nhash)); - if (inaddr_nhash == 0) + sizeof(inaddr_nhash)); + if (inaddr_nhash == 0) { inaddr_nhash = INADDR_NHASH; + } MALLOC(in_ifaddrhashtbl, struct in_ifaddrhashhead *, - inaddr_nhash * sizeof (*in_ifaddrhashtbl), + inaddr_nhash * sizeof(*in_ifaddrhashtbl), M_IFADDR, M_WAITOK | M_ZERO); - if (in_ifaddrhashtbl == NULL) + if (in_ifaddrhashtbl == NULL) { panic("in_ifaddrhashtbl_init allocation failed"); + } /* * Generate the next largest prime greater than inaddr_nhash. @@ -574,11 +587,13 @@ in_ifaddrhashtbl_init(void) for (;;) { p = 1; for (i = 3; i * i <= k; i += 2) { - if (k % i == 0) + if (k % i == 0) { p = 0; + } } - if (p == 1) + if (p == 1) { break; + } k += 2; } inaddr_hashp = k; @@ -592,10 +607,11 @@ inaddr_hashval(u_int32_t key) * the hash size, as documented in "Introduction to Algorithms" * (Cormen, Leiserson, Rivest). */ - if (inaddr_nhash > 1) - return ((key * inaddr_hashp) % inaddr_nhash); - else - return (0); + if (inaddr_nhash > 1) { + return (key * inaddr_hashp) % inaddr_nhash; + } else { + return 0; + } } void @@ -610,7 +626,7 @@ ip_proto_dispatch_in(struct mbuf *m, int hlen, u_int8_t proto, { struct ipfilter *filter; int seen = (inject_ipfref == NULL); - int changed_header = 0; + int changed_header = 0; struct ip *ip; void (*pr_input)(struct mbuf *, int len); @@ -618,8 +634,9 @@ ip_proto_dispatch_in(struct mbuf *m, int hlen, u_int8_t proto, ipf_ref(); TAILQ_FOREACH(filter, &ipv4_filters, ipf_link) { if (seen == 0) { - if ((struct ipfilter *)inject_ipfref == filter) + if ((struct ipfilter *)inject_ipfref == filter) { seen = 1; + } } else if (filter->ipf_filter.ipf_input) { errno_t result; @@ -633,8 +650,9 @@ ip_proto_dispatch_in(struct mbuf *m, int hlen, u_int8_t proto, m->m_pkthdr.rcvif, ipf_unref()); /* ipf_unref() already called */ - if (m == NULL) + if (m == NULL) { return; + } changed_header = 1; ip = mtod(m, struct ip *); @@ -644,8 +662,8 @@ ip_proto_dispatch_in(struct mbuf *m, int hlen, u_int8_t proto, ip->ip_sum = ip_cksum_hdr_in(m, hlen); } result = filter->ipf_filter.ipf_input( - filter->ipf_filter.cookie, (mbuf_t *)&m, - hlen, proto); + filter->ipf_filter.cookie, (mbuf_t *)&m, + hlen, proto); if (result == EJUSTRETURN) { ipf_unref(); return; @@ -661,7 +679,7 @@ ip_proto_dispatch_in(struct mbuf *m, int hlen, u_int8_t proto, } /* Perform IP header alignment fixup (post-filters), if needed */ - IP_HDR_ALIGNMENT_FIXUP(m, m->m_pkthdr.rcvif, return); + IP_HDR_ALIGNMENT_FIXUP(m, m->m_pkthdr.rcvif, return ); /* * If there isn't a specific lock for the protocol @@ -687,25 +705,25 @@ ip_proto_dispatch_in(struct mbuf *m, int hlen, u_int8_t proto, } struct pktchain_elm { - struct mbuf *pkte_head; - struct mbuf *pkte_tail; - struct in_addr pkte_saddr; - struct in_addr pkte_daddr; - uint16_t pkte_npkts; - uint16_t pkte_proto; - uint32_t pkte_nbytes; + struct mbuf *pkte_head; + struct mbuf *pkte_tail; + struct in_addr pkte_saddr; + struct in_addr pkte_daddr; + uint16_t pkte_npkts; + uint16_t pkte_proto; + uint32_t pkte_nbytes; }; typedef struct pktchain_elm pktchain_elm_t; /* Store upto PKTTBL_SZ unique flows on the stack */ -#define PKTTBL_SZ 7 +#define PKTTBL_SZ 7 static struct mbuf * ip_chain_insert(struct mbuf *packet, pktchain_elm_t *tbl) { - struct ip* ip; - int pkttbl_idx = 0; + struct ip* ip; + int pkttbl_idx = 0; ip = mtod(packet, struct ip*); @@ -721,16 +739,17 @@ ip_chain_insert(struct mbuf *packet, pktchain_elm_t *tbl) (ip->ip_src.s_addr == tbl[pkttbl_idx].pkte_saddr.s_addr) && (ip->ip_p == tbl[pkttbl_idx].pkte_proto)) { } else { - return (packet); + return packet; } } - if (tbl[pkttbl_idx].pkte_tail != NULL) + if (tbl[pkttbl_idx].pkte_tail != NULL) { mbuf_setnextpkt(tbl[pkttbl_idx].pkte_tail, packet); + } tbl[pkttbl_idx].pkte_tail = packet; tbl[pkttbl_idx].pkte_npkts += 1; tbl[pkttbl_idx].pkte_nbytes += packet->m_pkthdr.len; - return (NULL); + return NULL; } /* args is a dummy variable here for backward compatibility */ @@ -745,13 +764,16 @@ ip_input_second_pass_loop_tbl(pktchain_elm_t *tbl, struct ip_fw_in_args *args) ip_input_second_pass(m, m->m_pkthdr.rcvif, 0, tbl[i].pkte_npkts, tbl[i].pkte_nbytes, args, 0); - if (tbl[i].pkte_npkts > 2) + if (tbl[i].pkte_npkts > 2) { ipstat.ips_rxc_chainsz_gt2++; - if (tbl[i].pkte_npkts > 4) + } + if (tbl[i].pkte_npkts > 4) { ipstat.ips_rxc_chainsz_gt4++; + } #if (DEBUG || DEVELOPMENT) - if (ip_input_measure) + if (ip_input_measure) { net_perf_histogram(&net_perf, tbl[i].pkte_npkts); + } #endif /* (DEBUG || DEVELOPMENT) */ tbl[i].pkte_head = tbl[i].pkte_tail = NULL; tbl[i].pkte_npkts = 0; @@ -815,14 +837,16 @@ ip_input_dispatch_chain(struct mbuf *m) ip = mtod(tmp_mbuf, struct ip *); hlen = IP_VHL_HL(ip->ip_vhl) << 2; - while(tmp_mbuf) { + while (tmp_mbuf) { nxt_mbuf = mbuf_nextpkt(tmp_mbuf); mbuf_setnextpkt(tmp_mbuf, NULL); - if ((sw_lro) && (ip->ip_p == IPPROTO_TCP)) + if ((sw_lro) && (ip->ip_p == IPPROTO_TCP)) { tmp_mbuf = tcp_lro(tmp_mbuf, hlen); - if (tmp_mbuf) + } + if (tmp_mbuf) { ip_proto_dispatch_in(tmp_mbuf, hlen, ip->ip_p, 0); + } tmp_mbuf = nxt_mbuf; if (tmp_mbuf) { ip = mtod(tmp_mbuf, struct ip *); @@ -844,6 +868,76 @@ ip_input_setdst_chain(struct mbuf *m, uint32_t ifindex, struct in_ifaddr *ia) } } +static void +ip_input_adjust(struct mbuf *m, struct ip *ip, struct ifnet *inifp) +{ + boolean_t adjust = TRUE; + + ASSERT(m_pktlen(m) > ip->ip_len); + + /* + * Invalidate hardware checksum info if ip_adj_clear_hwcksum + * is set; useful to handle buggy drivers. Note that this + * should not be enabled by default, as we may get here due + * to link-layer padding. + */ + if (ip_adj_clear_hwcksum && + (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) && + !(inifp->if_flags & IFF_LOOPBACK) && + !(m->m_pkthdr.pkt_flags & PKTF_LOOP)) { + m->m_pkthdr.csum_flags &= ~CSUM_DATA_VALID; + m->m_pkthdr.csum_data = 0; + ipstat.ips_adj_hwcsum_clr++; + } + + /* + * If partial checksum information is available, subtract + * out the partial sum of postpended extraneous bytes, and + * update the checksum metadata accordingly. By doing it + * here, the upper layer transport only needs to adjust any + * prepended extraneous bytes (else it will do both.) + */ + if (ip_adj_partial_sum && + (m->m_pkthdr.csum_flags & (CSUM_DATA_VALID | CSUM_PARTIAL)) == + (CSUM_DATA_VALID | CSUM_PARTIAL)) { + m->m_pkthdr.csum_rx_val = m_adj_sum16(m, + m->m_pkthdr.csum_rx_start, m->m_pkthdr.csum_rx_start, + (ip->ip_len - m->m_pkthdr.csum_rx_start), + m->m_pkthdr.csum_rx_val); + } else if ((m->m_pkthdr.csum_flags & + (CSUM_DATA_VALID | CSUM_PARTIAL)) == + (CSUM_DATA_VALID | CSUM_PARTIAL)) { + /* + * If packet has partial checksum info and we decided not + * to subtract the partial sum of postpended extraneous + * bytes here (not the default case), leave that work to + * be handled by the other layers. For now, only TCP, UDP + * layers are capable of dealing with this. For all other + * protocols (including fragments), trim and ditch the + * partial sum as those layers might not implement partial + * checksumming (or adjustment) at all. + */ + if ((ip->ip_off & (IP_MF | IP_OFFMASK)) == 0 && + (ip->ip_p == IPPROTO_TCP || ip->ip_p == IPPROTO_UDP)) { + adjust = FALSE; + } else { + m->m_pkthdr.csum_flags &= ~CSUM_DATA_VALID; + m->m_pkthdr.csum_data = 0; + ipstat.ips_adj_hwcsum_clr++; + } + } + + if (adjust) { + ipstat.ips_adj++; + if (m->m_len == m->m_pkthdr.len) { + m->m_len = ip->ip_len; + m->m_pkthdr.len = ip->ip_len; + } else { + m_adj(m, ip->ip_len - m->m_pkthdr.len); + } + } +} + /* * First pass does all essential packet validation and places on a per flow * queue for doing operations that have same outcome for all packets of a flow. @@ -853,21 +947,21 @@ static ipinput_chain_ret_t ip_input_first_pass(struct mbuf *m, u_int32_t *div_info, struct ip_fw_in_args *args, int *ours, struct mbuf **modm) { - struct ip *ip; - struct ifnet *inifp; - unsigned int hlen; - int retval = IPINPUT_DOCHAIN; - int len = 0; - struct in_addr src_ip; + struct ip *ip; + struct ifnet *inifp; + unsigned int hlen; + int retval = IPINPUT_DOCHAIN; + int len = 0; + struct in_addr src_ip; #if IPFIREWALL - int i; + int i; #endif #if IPFIREWALL || DUMMYNET - struct m_tag *copy; - struct m_tag *p; - boolean_t delete = FALSE; - struct ip_fw_args args1; - boolean_t init = FALSE; + struct m_tag *copy; + struct m_tag *p; + boolean_t delete = FALSE; + struct ip_fw_args args1; + boolean_t init = FALSE; #endif ipfilter_t inject_filter_ref = NULL; @@ -899,8 +993,9 @@ ip_input_first_pass(struct mbuf *m, u_int32_t *div_info, /* * Don't bother searching for tag(s) if there's none. */ - if (SLIST_EMPTY(&m->m_pkthdr.tags)) + if (SLIST_EMPTY(&m->m_pkthdr.tags)) { goto ipfw_tags_done; + } /* Grab info from mtags prepended to the chain */ p = m_tag_first(m); @@ -910,7 +1005,7 @@ ip_input_first_pass(struct mbuf *m, u_int32_t *div_info, if (p->m_tag_type == KERNEL_TAG_TYPE_DUMMYNET) { struct dn_pkt_tag *dn_tag; - dn_tag = (struct dn_pkt_tag *)(p+1); + dn_tag = (struct dn_pkt_tag *)(p + 1); args->fwai_ipfw_rule = dn_tag->dn_ipfw_rule; args->fwai_pf_rule = dn_tag->dn_pf_rule; delete = TRUE; @@ -921,7 +1016,7 @@ ip_input_first_pass(struct mbuf *m, u_int32_t *div_info, if (p->m_tag_type == KERNEL_TAG_TYPE_DIVERT) { struct divert_tag *div_tag; - div_tag = (struct divert_tag *)(p+1); + div_tag = (struct divert_tag *)(p + 1); args->fwai_divert_rule = div_tag->cookie; delete = TRUE; } @@ -930,7 +1025,7 @@ ip_input_first_pass(struct mbuf *m, u_int32_t *div_info, if (p->m_tag_type == KERNEL_TAG_TYPE_IPFORWARD) { struct ip_fwd_tag *ipfwd_tag; - ipfwd_tag = (struct ip_fwd_tag *)(p+1); + ipfwd_tag = (struct ip_fwd_tag *)(p + 1); args->fwai_next_hop = ipfwd_tag->next_hop; delete = TRUE; } @@ -939,7 +1034,7 @@ ip_input_first_pass(struct mbuf *m, u_int32_t *div_info, copy = p; p = m_tag_next(m, p); m_tag_delete(m, copy); - } else { + } else { p = m_tag_next(m, p); } } else { @@ -948,8 +1043,9 @@ ip_input_first_pass(struct mbuf *m, u_int32_t *div_info, } #if DIAGNOSTIC - if (m == NULL || !(m->m_flags & M_PKTHDR)) + if (m == NULL || !(m->m_flags & M_PKTHDR)) { panic("ip_input no HDR"); + } #endif #if DUMMYNET @@ -959,11 +1055,13 @@ ip_input_first_pass(struct mbuf *m, u_int32_t *div_info, hlen = IP_VHL_HL(ip->ip_vhl) << 2; inject_filter_ref = ipf_get_inject_filter(m); #if IPFIREWALL - if (args->fwai_ipfw_rule) + if (args->fwai_ipfw_rule) { goto iphack; + } #endif /* IPFIREWALL */ - if (args->fwai_pf_rule) + if (args->fwai_pf_rule) { goto check_with_pf; + } } #endif /* DUMMYNET */ ipfw_tags_done: @@ -972,8 +1070,9 @@ ipfw_tags_done: /* * No need to process packet twice if we've already seen it. */ - if (!SLIST_EMPTY(&m->m_pkthdr.tags)) + if (!SLIST_EMPTY(&m->m_pkthdr.tags)) { inject_filter_ref = ipf_get_inject_filter(m); + } if (inject_filter_ref != NULL) { ip = mtod(m, struct ip *); hlen = IP_VHL_HL(ip->ip_vhl) << 2; @@ -985,21 +1084,21 @@ ipfw_tags_done: ip->ip_len = ntohs(ip->ip_len) - hlen; ip->ip_off = ntohs(ip->ip_off); ip_proto_dispatch_in(m, hlen, ip->ip_p, inject_filter_ref); - return (IPINPUT_DONE); + return IPINPUT_DONE; } - if (m->m_pkthdr.len < sizeof (struct ip)) { + if (m->m_pkthdr.len < sizeof(struct ip)) { OSAddAtomic(1, &ipstat.ips_total); OSAddAtomic(1, &ipstat.ips_tooshort); m_freem(m); - return (IPINPUT_FREED); + return IPINPUT_FREED; } - if (m->m_len < sizeof (struct ip) && - (m = m_pullup(m, sizeof (struct ip))) == NULL) { + if (m->m_len < sizeof(struct ip) && + (m = m_pullup(m, sizeof(struct ip))) == NULL) { OSAddAtomic(1, &ipstat.ips_total); OSAddAtomic(1, &ipstat.ips_toosmall); - return (IPINPUT_FREED); + return IPINPUT_FREED; } ip = mtod(m, struct ip *); @@ -1013,16 +1112,16 @@ ipfw_tags_done: OSAddAtomic(1, &ipstat.ips_badvers); KERNEL_DEBUG(DBG_LAYER_END, 0, 0, 0, 0, 0); m_freem(m); - return (IPINPUT_FREED); + return IPINPUT_FREED; } hlen = IP_VHL_HL(ip->ip_vhl) << 2; - if (hlen < sizeof (struct ip)) { + if (hlen < sizeof(struct ip)) { OSAddAtomic(1, &ipstat.ips_total); OSAddAtomic(1, &ipstat.ips_badhlen); KERNEL_DEBUG(DBG_LAYER_END, 0, 0, 0, 0, 0); m_freem(m); - return (IPINPUT_FREED); + return IPINPUT_FREED; } if (hlen > m->m_len) { @@ -1030,7 +1129,7 @@ ipfw_tags_done: OSAddAtomic(1, &ipstat.ips_total); OSAddAtomic(1, &ipstat.ips_badhlen); KERNEL_DEBUG(DBG_LAYER_END, 0, 0, 0, 0, 0); - return (IPINPUT_FREED); + return IPINPUT_FREED; } ip = mtod(m, struct ip *); *modm = m; @@ -1053,12 +1152,12 @@ ipfw_tags_done: * therefore we don't care so much about PKTF_IFINFO. */ if (!(inifp->if_flags & IFF_LOOPBACK) && - !(m->m_pkthdr.pkt_flags & PKTF_LOOP)) { + !(m->m_pkthdr.pkt_flags & PKTF_LOOP)) { OSAddAtomic(1, &ipstat.ips_total); OSAddAtomic(1, &ipstat.ips_badaddr); KERNEL_DEBUG(DBG_LAYER_END, 0, 0, 0, 0, 0); m_freem(m); - return (IPINPUT_FREED); + return IPINPUT_FREED; } } @@ -1073,7 +1172,7 @@ ipfw_tags_done: OSAddAtomic(1, &ipstat.ips_total); KERNEL_DEBUG(DBG_LAYER_END, 0, 0, 0, 0, 0); m_freem(m); - return (IPINPUT_FREED); + return IPINPUT_FREED; } } } @@ -1082,7 +1181,7 @@ ipfw_tags_done: OSAddAtomic(1, &ipstat.ips_total); KERNEL_DEBUG(DBG_LAYER_END, 0, 0, 0, 0, 0); m_freem(m); - return (IPINPUT_FREED); + return IPINPUT_FREED; } DTRACE_IP6(receive, struct mbuf *, m, struct inpcb *, NULL, @@ -1101,7 +1200,7 @@ ipfw_tags_done: OSAddAtomic(1, &ipstat.ips_badlen); KERNEL_DEBUG(DBG_LAYER_END, 0, 0, 0, 0, 0); m_freem(m); - return (IPINPUT_FREED); + return IPINPUT_FREED; } #if BYTE_ORDER != BIG_ENDIAN @@ -1119,31 +1218,11 @@ ipfw_tags_done: OSAddAtomic(1, &ipstat.ips_tooshort); KERNEL_DEBUG(DBG_LAYER_END, 0, 0, 0, 0, 0); m_freem(m); - return (IPINPUT_FREED); + return IPINPUT_FREED; } if (m->m_pkthdr.len > ip->ip_len) { - /* - * Invalidate hardware checksum info if ip_adj_clear_hwcksum - * is set; useful to handle buggy drivers. Note that this - * should not be enabled by default, as we may get here due - * to link-layer padding. - */ - if (ip_adj_clear_hwcksum && - (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) && - !(inifp->if_flags & IFF_LOOPBACK) && - !(m->m_pkthdr.pkt_flags & PKTF_LOOP)) { - m->m_pkthdr.csum_flags &= ~CSUM_DATA_VALID; - m->m_pkthdr.csum_data = 0; - ipstat.ips_adj_hwcsum_clr++; - } - - ipstat.ips_adj++; - if (m->m_len == m->m_pkthdr.len) { - m->m_len = ip->ip_len; - m->m_pkthdr.len = ip->ip_len; - } else - m_adj(m, ip->ip_len - m->m_pkthdr.len); + ip_input_adjust(m, ip, inifp); } /* for consistency */ @@ -1161,6 +1240,8 @@ check_with_pf: if (PF_IS_ENABLED) { int error; ip_input_cpout_args(args, &args1, &init); + ip = mtod(m, struct ip *); + src_ip = ip->ip_src; #if DUMMYNET error = pf_af_hook(inifp, NULL, &m, AF_INET, TRUE, &args1); @@ -1177,7 +1258,7 @@ check_with_pf: ip_input_update_nstat(inifp, src_ip, 1, len); KERNEL_DEBUG(DBG_LAYER_END, 0, 0, 0, 0, 0); OSAddAtomic(1, &ipstat.ips_total); - return (IPINPUT_FREED); + return IPINPUT_FREED; } ip = mtod(m, struct ip *); hlen = IP_VHL_HL(ip->ip_vhl) << 2; @@ -1209,9 +1290,9 @@ iphack: */ if (args->fwai_next_hop) { *ours = 1; - return (IPINPUT_DONTCHAIN); + return IPINPUT_DONTCHAIN; } -#endif /* IPFIREWALL_FORWARD */ +#endif /* IPFIREWALL_FORWARD */ ip_input_cpout_args(args, &args1, &init); args1.fwa_m = m; @@ -1219,12 +1300,13 @@ iphack: m = args1.fwa_m; if ((i & IP_FW_PORT_DENY_FLAG) || m == NULL) { /* drop */ - if (m) + if (m) { m_freem(m); + } ip_input_update_nstat(inifp, src_ip, 1, len); KERNEL_DEBUG(DBG_LAYER_END, 0, 0, 0, 0, 0); OSAddAtomic(1, &ipstat.ips_total); - return (IPINPUT_FREED); + return IPINPUT_FREED; } ip = mtod(m, struct ip *); /* just in case m changed */ *modm = m; @@ -1236,12 +1318,12 @@ iphack: #if DUMMYNET if (DUMMYNET_LOADED && (i & IP_FW_PORT_DYNT_FLAG) != 0) { /* Send packet to the appropriate pipe */ - ip_dn_io_ptr(m, i&0xffff, DN_TO_IP_IN, &args1, + ip_dn_io_ptr(m, i & 0xffff, DN_TO_IP_IN, &args1, DN_CLIENT_IPFW); ip_input_update_nstat(inifp, src_ip, 1, len); KERNEL_DEBUG(DBG_LAYER_END, 0, 0, 0, 0, 0); OSAddAtomic(1, &ipstat.ips_total); - return (IPINPUT_FREED); + return IPINPUT_FREED; } #endif /* DUMMYNET */ #if IPDIVERT @@ -1249,7 +1331,7 @@ iphack: /* Divert or tee packet */ *div_info = i; *ours = 1; - return (IPINPUT_DONTCHAIN); + return IPINPUT_DONTCHAIN; } #endif #if IPFIREWALL_FORWARD @@ -1265,7 +1347,7 @@ iphack: KERNEL_DEBUG(DBG_LAYER_END, 0, 0, 0, 0, 0); m_freem(m); OSAddAtomic(1, &ipstat.ips_total); - return (IPINPUT_FREED); + return IPINPUT_FREED; } #endif /* IPFIREWALL */ #if IPSEC | IPFIREWALL @@ -1277,17 +1359,17 @@ pass: * error was detected (causing an icmp message * to be sent and the original packet to be freed). */ - ip_nhops = 0; /* for source routed packets */ + ip_nhops = 0; /* for source routed packets */ #if IPFIREWALL - if (hlen > sizeof (struct ip) && + if (hlen > sizeof(struct ip) && ip_dooptions(m, 0, args->fwai_next_hop)) { #else /* !IPFIREWALL */ - if (hlen > sizeof (struct ip) && ip_dooptions(m, 0, NULL)) { + if (hlen > sizeof(struct ip) && ip_dooptions(m, 0, NULL)) { #endif /* !IPFIREWALL */ ip_input_update_nstat(inifp, src_ip, 1, len); KERNEL_DEBUG(DBG_LAYER_END, 0, 0, 0, 0, 0); OSAddAtomic(1, &ipstat.ips_total); - return (IPINPUT_FREED); + return IPINPUT_FREED; } /* @@ -1295,38 +1377,40 @@ pass: * if it is our fragment or someone else's plus the complexity of * divert and fw args makes it harder to do chaining. */ - if (ip->ip_off & ~(IP_DF | IP_RF)) - return (IPINPUT_DONTCHAIN); + if (ip->ip_off & ~(IP_DF | IP_RF)) { + return IPINPUT_DONTCHAIN; + } /* Allow DHCP/BootP responses through */ if ((inifp->if_eflags & IFEF_AUTOCONFIGURING) && - hlen == sizeof (struct ip) && ip->ip_p == IPPROTO_UDP) { + hlen == sizeof(struct ip) && ip->ip_p == IPPROTO_UDP) { struct udpiphdr *ui; - if (m->m_len < sizeof (struct udpiphdr) && - (m = m_pullup(m, sizeof (struct udpiphdr))) == NULL) { + if (m->m_len < sizeof(struct udpiphdr) && + (m = m_pullup(m, sizeof(struct udpiphdr))) == NULL) { OSAddAtomic(1, &udpstat.udps_hdrops); KERNEL_DEBUG(DBG_LAYER_END, 0, 0, 0, 0, 0); OSAddAtomic(1, &ipstat.ips_total); - return (IPINPUT_FREED); + return IPINPUT_FREED; } *modm = m; ui = mtod(m, struct udpiphdr *); if (ntohs(ui->ui_dport) == IPPORT_BOOTPC) { ip_setdstifaddr_info(m, inifp->if_index, NULL); - return (IPINPUT_DONTCHAIN); + return IPINPUT_DONTCHAIN; } } /* Avoid chaining raw sockets as ipsec checks occur later for them */ - if (ip_protox[ip->ip_p]->pr_flags & PR_LASTHDR) - return (IPINPUT_DONTCHAIN); + if (ip_protox[ip->ip_p]->pr_flags & PR_LASTHDR) { + return IPINPUT_DONTCHAIN; + } - return (retval); + return retval; #if !defined(__i386__) && !defined(__x86_64__) bad: m_freem(m); - return (IPINPUT_FREED); + return IPINPUT_FREED; #endif } @@ -1334,11 +1418,11 @@ static void ip_input_second_pass(struct mbuf *m, struct ifnet *inifp, u_int32_t div_info, int npkts_in_chain, int bytes_in_chain, struct ip_fw_in_args *args, int ours) { - unsigned int checkif; - struct mbuf *tmp_mbuf = NULL; - struct in_ifaddr *ia = NULL; - struct in_addr pkt_dst; - unsigned int hlen; + unsigned int checkif; + struct mbuf *tmp_mbuf = NULL; + struct in_ifaddr *ia = NULL; + struct in_addr pkt_dst; + unsigned int hlen; #if !IPFIREWALL #pragma unused (args) @@ -1367,8 +1451,9 @@ ip_input_second_pass(struct mbuf *m, struct ifnet *inifp, u_int32_t div_info, ip_input_update_nstat(inifp, ip->ip_src, npkts_in_chain, bytes_in_chain); - if (ours) + if (ours) { goto ours; + } /* * Check our list of addresses, to see if the packet is for us. @@ -1379,7 +1464,7 @@ ip_input_second_pass(struct mbuf *m, struct ifnet *inifp, u_int32_t div_info, tmp_mbuf = m; if (TAILQ_EMPTY(&in_ifaddrhead)) { while (tmp_mbuf) { - if (!(tmp_mbuf->m_flags & (M_MCAST|M_BCAST))) { + if (!(tmp_mbuf->m_flags & (M_MCAST | M_BCAST))) { ip_setdstifaddr_info(tmp_mbuf, inifp->if_index, NULL); } @@ -1418,7 +1503,7 @@ ip_input_second_pass(struct mbuf *m, struct ifnet *inifp, u_int32_t div_info, #if IPFIREWALL && (args->fwai_next_hop == NULL); #else /* !IPFIREWALL */ - ; + ; #endif /* !IPFIREWALL */ /* @@ -1539,8 +1624,9 @@ ours: #else m = ip_reass(m); #endif - if (m == NULL) + if (m == NULL) { return; + } ip = mtod(m, struct ip *); /* Get the header length of the reassembled packet */ hlen = IP_VHL_HL(ip->ip_vhl) << 2; @@ -1580,8 +1666,9 @@ ours: VERIFY(npkts_in_chain == 1); /* Clone packet if we're doing a 'tee' */ - if (div_info & IP_FW_PORT_TEE_FLAG) + if (div_info & IP_FW_PORT_TEE_FLAG) { clone = m_dup(m, M_DONTWAIT); + } /* Restore packet header fields to original values */ ip->ip_len += hlen; @@ -1631,12 +1718,13 @@ ours: VERIFY(npkts_in_chain == 1); fwd_tag = m_tag_create(KERNEL_MODULE_TAG_ID, - KERNEL_TAG_TYPE_IPFORWARD, sizeof (*ipfwd_tag), + KERNEL_TAG_TYPE_IPFORWARD, sizeof(*ipfwd_tag), M_NOWAIT, m); - if (fwd_tag == NULL) + if (fwd_tag == NULL) { goto bad; + } - ipfwd_tag = (struct ip_fwd_tag *)(fwd_tag+1); + ipfwd_tag = (struct ip_fwd_tag *)(fwd_tag + 1); ipfwd_tag->next_hop = args->fwai_next_hop; m_tag_prepend(m, fwd_tag); @@ -1651,7 +1739,6 @@ ours: ip->ip_src.s_addr, ip->ip_p, ip->ip_off, ip->ip_len); ip_input_dispatch_chain(m); - } #else /* !IPFIREWALL */ ip_input_dispatch_chain(m); @@ -1667,25 +1754,26 @@ bad: void ip_input_process_list(struct mbuf *packet_list) { - pktchain_elm_t pktchain_tbl[PKTTBL_SZ]; + pktchain_elm_t pktchain_tbl[PKTTBL_SZ]; - struct mbuf *packet = NULL; - struct mbuf *modm = NULL; /* modified mbuf */ - int retval = 0; - u_int32_t div_info = 0; - int ours = 0; + struct mbuf *packet = NULL; + struct mbuf *modm = NULL; /* modified mbuf */ + int retval = 0; + u_int32_t div_info = 0; + int ours = 0; #if (DEBUG || DEVELOPMENT) struct timeval start_tv; #endif /* (DEBUG || DEVELOPMENT) */ - int num_pkts = 0; + int num_pkts = 0; int chain = 0; struct ip_fw_in_args args; if (ip_chaining == 0) { struct mbuf *m = packet_list; #if (DEBUG || DEVELOPMENT) - if (ip_input_measure) + if (ip_input_measure) { net_perf_start_time(&net_perf, &start_tv); + } #endif /* (DEBUG || DEVELOPMENT) */ while (m) { @@ -1696,14 +1784,16 @@ ip_input_process_list(struct mbuf *packet_list) num_pkts++; } #if (DEBUG || DEVELOPMENT) - if (ip_input_measure) + if (ip_input_measure) { net_perf_measure_time(&net_perf, &start_tv, num_pkts); + } #endif /* (DEBUG || DEVELOPMENT) */ return; } #if (DEBUG || DEVELOPMENT) - if (ip_input_measure) + if (ip_input_measure) { net_perf_start_time(&net_perf, &start_tv); + } #endif /* (DEBUG || DEVELOPMENT) */ bzero(&pktchain_tbl, sizeof(pktchain_tbl)); @@ -1716,28 +1806,31 @@ restart_list_process: num_pkts++; modm = NULL; div_info = 0; - bzero(&args, sizeof (args)); + bzero(&args, sizeof(args)); retval = ip_input_first_pass(packet, &div_info, &args, &ours, &modm); if (retval == IPINPUT_DOCHAIN) { - if (modm) + if (modm) { packet = modm; + } packet = ip_chain_insert(packet, &pktchain_tbl[0]); if (packet == NULL) { ipstat.ips_rxc_chained++; chain++; - if (chain > ip_chainsz) + if (chain > ip_chainsz) { break; + } } else { ipstat.ips_rxc_collisions++; break; } } else if (retval == IPINPUT_DONTCHAIN) { /* in order to preserve order, exit from chaining */ - if (modm) + if (modm) { packet = modm; + } ipstat.ips_rxc_notchain++; break; } else { @@ -1746,8 +1839,9 @@ restart_list_process: } /* do second pass here for pktchain_tbl */ - if (chain) + if (chain) { ip_input_second_pass_loop_tbl(&pktchain_tbl[0], &args); + } if (packet) { /* @@ -1755,19 +1849,22 @@ restart_list_process: * ip_input_second_pass_loop_tbl(). */ #if (DEBUG || DEVELOPMENT) - if (ip_input_measure) + if (ip_input_measure) { net_perf_histogram(&net_perf, 1); + } #endif /* (DEBUG || DEVELOPMENT) */ ip_input_second_pass(packet, packet->m_pkthdr.rcvif, div_info, 1, packet->m_pkthdr.len, &args, ours); } - if (packet_list) + if (packet_list) { goto restart_list_process; + } #if (DEBUG || DEVELOPMENT) - if (ip_input_measure) + if (ip_input_measure) { net_perf_measure_time(&net_perf, &start_tv, num_pkts); + } #endif /* (DEBUG || DEVELOPMENT) */ } /* @@ -1784,11 +1881,11 @@ ip_input(struct mbuf *m) struct in_addr pkt_dst; #if IPFIREWALL int i; - u_int32_t div_info = 0; /* packet divert/tee info */ + u_int32_t div_info = 0; /* packet divert/tee info */ #endif #if IPFIREWALL || DUMMYNET struct ip_fw_args args; - struct m_tag *tag; + struct m_tag *tag; #endif ipfilter_t inject_filter_ref = NULL; struct ifnet *inifp; @@ -1806,13 +1903,14 @@ ip_input(struct mbuf *m) m->m_pkthdr.pkt_flags &= ~PKTF_FORWARDED; #if IPFIREWALL || DUMMYNET - bzero(&args, sizeof (struct ip_fw_args)); + bzero(&args, sizeof(struct ip_fw_args)); /* * Don't bother searching for tag(s) if there's none. */ - if (SLIST_EMPTY(&m->m_pkthdr.tags)) + if (SLIST_EMPTY(&m->m_pkthdr.tags)) { goto ipfw_tags_done; + } /* Grab info from mtags prepended to the chain */ #if DUMMYNET @@ -1820,7 +1918,7 @@ ip_input(struct mbuf *m) KERNEL_TAG_TYPE_DUMMYNET, NULL)) != NULL) { struct dn_pkt_tag *dn_tag; - dn_tag = (struct dn_pkt_tag *)(tag+1); + dn_tag = (struct dn_pkt_tag *)(tag + 1); args.fwa_ipfw_rule = dn_tag->dn_ipfw_rule; args.fwa_pf_rule = dn_tag->dn_pf_rule; @@ -1833,7 +1931,7 @@ ip_input(struct mbuf *m) KERNEL_TAG_TYPE_DIVERT, NULL)) != NULL) { struct divert_tag *div_tag; - div_tag = (struct divert_tag *)(tag+1); + div_tag = (struct divert_tag *)(tag + 1); args.fwa_divert_rule = div_tag->cookie; m_tag_delete(m, tag); @@ -1844,15 +1942,16 @@ ip_input(struct mbuf *m) KERNEL_TAG_TYPE_IPFORWARD, NULL)) != NULL) { struct ip_fwd_tag *ipfwd_tag; - ipfwd_tag = (struct ip_fwd_tag *)(tag+1); + ipfwd_tag = (struct ip_fwd_tag *)(tag + 1); args.fwa_next_hop = ipfwd_tag->next_hop; m_tag_delete(m, tag); } -#if DIAGNOSTIC - if (m == NULL || !(m->m_flags & M_PKTHDR)) +#if DIAGNOSTIC + if (m == NULL || !(m->m_flags & M_PKTHDR)) { panic("ip_input no HDR"); + } #endif #if DUMMYNET @@ -1862,11 +1961,13 @@ ip_input(struct mbuf *m) hlen = IP_VHL_HL(ip->ip_vhl) << 2; inject_filter_ref = ipf_get_inject_filter(m); #if IPFIREWALL - if (args.fwa_ipfw_rule) + if (args.fwa_ipfw_rule) { goto iphack; + } #endif /* IPFIREWALL */ - if (args.fwa_pf_rule) + if (args.fwa_pf_rule) { goto check_with_pf; + } } #endif /* DUMMYNET */ ipfw_tags_done: @@ -1875,8 +1976,9 @@ ipfw_tags_done: /* * No need to process packet twice if we've already seen it. */ - if (!SLIST_EMPTY(&m->m_pkthdr.tags)) + if (!SLIST_EMPTY(&m->m_pkthdr.tags)) { inject_filter_ref = ipf_get_inject_filter(m); + } if (inject_filter_ref != NULL) { ip = mtod(m, struct ip *); hlen = IP_VHL_HL(ip->ip_vhl) << 2; @@ -1892,11 +1994,12 @@ ipfw_tags_done: } OSAddAtomic(1, &ipstat.ips_total); - if (m->m_pkthdr.len < sizeof (struct ip)) + if (m->m_pkthdr.len < sizeof(struct ip)) { goto tooshort; + } - if (m->m_len < sizeof (struct ip) && - (m = m_pullup(m, sizeof (struct ip))) == NULL) { + if (m->m_len < sizeof(struct ip) && + (m = m_pullup(m, sizeof(struct ip))) == NULL) { OSAddAtomic(1, &ipstat.ips_toosmall); return; } @@ -1911,7 +2014,7 @@ ipfw_tags_done: } hlen = IP_VHL_HL(ip->ip_vhl) << 2; - if (hlen < sizeof (struct ip)) { /* minimum header length */ + if (hlen < sizeof(struct ip)) { /* minimum header length */ OSAddAtomic(1, &ipstat.ips_badhlen); goto bad; } @@ -1953,8 +2056,9 @@ ipfw_tags_done: if (ip->ip_ttl != MAXTTL) { OSAddAtomic(1, &ip_linklocal_stat.iplls_in_badttl); /* Silently drop link local traffic with bad TTL */ - if (!ip_linklocal_in_allowbadttl) + if (!ip_linklocal_in_allowbadttl) { goto bad; + } } } @@ -2014,27 +2118,7 @@ tooshort: goto bad; } if (m->m_pkthdr.len > ip->ip_len) { - /* - * Invalidate hardware checksum info if ip_adj_clear_hwcksum - * is set; useful to handle buggy drivers. Note that this - * should not be enabled by default, as we may get here due - * to link-layer padding. - */ - if (ip_adj_clear_hwcksum && - (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) && - !(inifp->if_flags & IFF_LOOPBACK) && - !(m->m_pkthdr.pkt_flags & PKTF_LOOP)) { - m->m_pkthdr.csum_flags &= ~CSUM_DATA_VALID; - m->m_pkthdr.csum_data = 0; - ipstat.ips_adj_hwcsum_clr++; - } - - ipstat.ips_adj++; - if (m->m_len == m->m_pkthdr.len) { - m->m_len = ip->ip_len; - m->m_pkthdr.len = ip->ip_len; - } else - m_adj(m, ip->ip_len - m->m_pkthdr.len); + ip_input_adjust(m, ip, inifp); } /* for consistency */ @@ -2067,8 +2151,9 @@ check_with_pf: #endif /* PF */ #if IPSEC - if (ipsec_bypass == 0 && ipsec_gethist(m, NULL)) + if (ipsec_bypass == 0 && ipsec_gethist(m, NULL)) { goto pass; + } #endif #if IPFIREWALL @@ -2085,9 +2170,10 @@ iphack: * If we've been forwarded from the output side, then * skip the firewall a second time */ - if (args.fwa_next_hop) + if (args.fwa_next_hop) { goto ours; -#endif /* IPFIREWALL_FORWARD */ + } +#endif /* IPFIREWALL_FORWARD */ args.fwa_m = m; @@ -2095,8 +2181,9 @@ iphack: m = args.fwa_m; if ((i & IP_FW_PORT_DENY_FLAG) || m == NULL) { /* drop */ - if (m) + if (m) { m_freem(m); + } return; } ip = mtod(m, struct ip *); /* just in case m changed */ @@ -2107,7 +2194,7 @@ iphack: #if DUMMYNET if (DUMMYNET_LOADED && (i & IP_FW_PORT_DYNT_FLAG) != 0) { /* Send packet to the appropriate pipe */ - ip_dn_io_ptr(m, i&0xffff, DN_TO_IP_IN, &args, + ip_dn_io_ptr(m, i & 0xffff, DN_TO_IP_IN, &args, DN_CLIENT_IPFW); return; } @@ -2140,12 +2227,12 @@ pass: * error was detected (causing an icmp message * to be sent and the original packet to be freed). */ - ip_nhops = 0; /* for source routed packets */ + ip_nhops = 0; /* for source routed packets */ #if IPFIREWALL - if (hlen > sizeof (struct ip) && + if (hlen > sizeof(struct ip) && ip_dooptions(m, 0, args.fwa_next_hop)) { #else /* !IPFIREWALL */ - if (hlen > sizeof (struct ip) && ip_dooptions(m, 0, NULL)) { + if (hlen > sizeof(struct ip) && ip_dooptions(m, 0, NULL)) { #endif /* !IPFIREWALL */ return; } @@ -2156,7 +2243,7 @@ pass: * we receive might be for us (and let the upper layers deal * with it). */ - if (TAILQ_EMPTY(&in_ifaddrhead) && !(m->m_flags & (M_MCAST|M_BCAST))) { + if (TAILQ_EMPTY(&in_ifaddrhead) && !(m->m_flags & (M_MCAST | M_BCAST))) { ip_setdstifaddr_info(m, inifp->if_index, NULL); goto ours; } @@ -2192,7 +2279,7 @@ pass: #if IPFIREWALL && (args.fwa_next_hop == NULL); #else /* !IPFIREWALL */ - ; + ; #endif /* !IPFIREWALL */ /* @@ -2268,11 +2355,11 @@ pass: /* Allow DHCP/BootP responses through */ if ((inifp->if_eflags & IFEF_AUTOCONFIGURING) && - hlen == sizeof (struct ip) && ip->ip_p == IPPROTO_UDP) { + hlen == sizeof(struct ip) && ip->ip_p == IPPROTO_UDP) { struct udpiphdr *ui; - if (m->m_len < sizeof (struct udpiphdr) && - (m = m_pullup(m, sizeof (struct udpiphdr))) == NULL) { + if (m->m_len < sizeof(struct udpiphdr) && + (m = m_pullup(m, sizeof(struct udpiphdr))) == NULL) { OSAddAtomic(1, &udpstat.udps_hdrops); return; } @@ -2313,8 +2400,9 @@ ours: #else m = ip_reass(m); #endif - if (m == NULL) + if (m == NULL) { return; + } ip = mtod(m, struct ip *); /* Get the header length of the reassembled packet */ hlen = IP_VHL_HL(ip->ip_vhl) << 2; @@ -2352,8 +2440,9 @@ ours: struct mbuf *clone = NULL; /* Clone packet if we're doing a 'tee' */ - if (div_info & IP_FW_PORT_TEE_FLAG) + if (div_info & IP_FW_PORT_TEE_FLAG) { clone = m_dup(m, M_DONTWAIT); + } /* Restore packet header fields to original values */ ip->ip_len += hlen; @@ -2401,12 +2490,13 @@ ours: struct ip_fwd_tag *ipfwd_tag; fwd_tag = m_tag_create(KERNEL_MODULE_TAG_ID, - KERNEL_TAG_TYPE_IPFORWARD, sizeof (*ipfwd_tag), + KERNEL_TAG_TYPE_IPFORWARD, sizeof(*ipfwd_tag), M_NOWAIT, m); - if (fwd_tag == NULL) + if (fwd_tag == NULL) { goto bad; + } - ipfwd_tag = (struct ip_fwd_tag *)(fwd_tag+1); + ipfwd_tag = (struct ip_fwd_tag *)(fwd_tag + 1); ipfwd_tag->next_hop = args.fwa_next_hop; m_tag_prepend(m, fwd_tag); @@ -2422,8 +2512,9 @@ ours: if ((sw_lro) && (ip->ip_p == IPPROTO_TCP)) { m = tcp_lro(m, hlen); - if (m == NULL) + if (m == NULL) { return; + } } ip_proto_dispatch_in(m, hlen, ip->ip_p, 0); @@ -2431,8 +2522,9 @@ ours: #else /* !IPFIREWALL */ if ((sw_lro) && (ip->ip_p == IPPROTO_TCP)) { m = tcp_lro(m, hlen); - if (m == NULL) + if (m == NULL) { return; + } } ip_proto_dispatch_in(m, hlen, ip->ip_p, 0); #endif /* !IPFIREWALL */ @@ -2446,24 +2538,27 @@ bad: static void ipq_updateparams(void) { - lck_mtx_assert(&ipqlock, LCK_MTX_ASSERT_OWNED); + LCK_MTX_ASSERT(&ipqlock, LCK_MTX_ASSERT_OWNED); /* * -1 for unlimited allocation. */ - if (maxnipq < 0) + if (maxnipq < 0) { ipq_limit = 0; + } /* * Positive number for specific bound. */ - if (maxnipq > 0) + if (maxnipq > 0) { ipq_limit = maxnipq; + } /* * Zero specifies no further fragment queue allocation -- set the * bound very low, but rely on implementation elsewhere to actually * prevent allocation and reclaim current queues. */ - if (maxnipq == 0) + if (maxnipq == 0) { ipq_limit = 1; + } /* * Arm the purge timer if not already and if there's work to do */ @@ -2479,8 +2574,9 @@ sysctl_maxnipq SYSCTL_HANDLER_ARGS lck_mtx_lock(&ipqlock); i = maxnipq; error = sysctl_handle_int(oidp, &i, 0, req); - if (error || req->newptr == USER_ADDR_NULL) + if (error || req->newptr == USER_ADDR_NULL) { goto done; + } /* impose bounds */ if (i < -1 || i > (nmbclusters / 4)) { error = EINVAL; @@ -2490,7 +2586,7 @@ sysctl_maxnipq SYSCTL_HANDLER_ARGS ipq_updateparams(); done: lck_mtx_unlock(&ipqlock); - return (error); + return error; } static int @@ -2502,13 +2598,14 @@ sysctl_maxfragsperpacket SYSCTL_HANDLER_ARGS lck_mtx_lock(&ipqlock); i = maxfragsperpacket; error = sysctl_handle_int(oidp, &i, 0, req); - if (error || req->newptr == USER_ADDR_NULL) + if (error || req->newptr == USER_ADDR_NULL) { goto done; + } maxfragsperpacket = i; - ipq_updateparams(); /* see if we need to arm timer */ + ipq_updateparams(); /* see if we need to arm timer */ done: lck_mtx_unlock(&ipqlock); - return (error); + return error; } /* @@ -2519,7 +2616,7 @@ done: * When IPDIVERT enabled, keep additional state with each packet that * tells us if we need to divert or tee the packet we're building. * - * The IP header is *NOT* adjusted out of iplen. + * The IP header is *NOT* adjusted out of iplen (but in host byte order). */ static struct mbuf * #if IPDIVERT @@ -2544,7 +2641,7 @@ ip_reass(struct mbuf *m) uint16_t hash; struct fq_head dfq; - MBUFQ_INIT(&dfq); /* for deferred frees */ + MBUFQ_INIT(&dfq); /* for deferred frees */ /* If maxnipq or maxfragsperpacket is 0, never accept fragments. */ if (maxnipq == 0 || maxfragsperpacket == 0) { @@ -2553,10 +2650,10 @@ ip_reass(struct mbuf *m) m_freem(m); if (nipq > 0) { lck_mtx_lock(&ipqlock); - frag_sched_timeout(); /* purge stale fragments */ + frag_sched_timeout(); /* purge stale fragments */ lck_mtx_unlock(&ipqlock); } - return (NULL); + return NULL; } ip = mtod(m, struct ip *); @@ -2578,8 +2675,9 @@ ip_reass(struct mbuf *m) #if CONFIG_MACF_NET mac_ipq_label_compare(m, fp) && #endif - ip->ip_p == fp->ipq_p) + ip->ip_p == fp->ipq_p) { goto found; + } } fp = NULL; @@ -2617,35 +2715,52 @@ found: * * Perform 1's complement adjustment of octets that got included/ * excluded in the hardware-calculated checksum value. Ignore cases - * where the value includes or excludes the IP header span, as the - * sum for those octets would already be 0xffff and thus no-op. + * where the value includes the entire IPv4 header span, as the sum + * for those octets would already be 0 by the time we get here; IP + * has already performed its header checksum validation. Also take + * care of any trailing bytes and subtract out their partial sum. */ - if (ip->ip_p == IPPROTO_UDP && hlen == sizeof (struct ip) && + if (ip->ip_p == IPPROTO_UDP && hlen == sizeof(struct ip) && (m->m_pkthdr.csum_flags & (CSUM_DATA_VALID | CSUM_PARTIAL | CSUM_PSEUDO_HDR)) == (CSUM_DATA_VALID | CSUM_PARTIAL)) { - uint32_t start; + uint32_t start = m->m_pkthdr.csum_rx_start; + int32_t trailer = (m_pktlen(m) - ip->ip_len); + uint32_t swbytes = (uint32_t)trailer; - start = m->m_pkthdr.csum_rx_start; csum = m->m_pkthdr.csum_rx_val; - if (start != 0 && start != hlen) { + ASSERT(trailer >= 0); + if ((start != 0 && start != hlen) || trailer != 0) { #if BYTE_ORDER != BIG_ENDIAN if (start < hlen) { HTONS(ip->ip_len); HTONS(ip->ip_off); } -#endif +#endif /* BYTE_ORDER != BIG_ENDIAN */ /* callee folds in sum */ - csum = m_adj_sum16(m, start, hlen, csum); + csum = m_adj_sum16(m, start, hlen, + (ip->ip_len - hlen), csum); + if (hlen > start) { + swbytes += (hlen - start); + } else { + swbytes += (start - hlen); + } #if BYTE_ORDER != BIG_ENDIAN if (start < hlen) { NTOHS(ip->ip_off); NTOHS(ip->ip_len); } -#endif +#endif /* BYTE_ORDER != BIG_ENDIAN */ } csum_flags = m->m_pkthdr.csum_flags; + + if (swbytes != 0) { + udp_in_cksum_stats(swbytes); + } + if (trailer != 0) { + m_adj(m, -trailer); + } } else { csum = 0; csum_flags = 0; @@ -2700,8 +2815,9 @@ found: */ if (fp == NULL) { fp = ipq_alloc(M_DONTWAIT); - if (fp == NULL) + if (fp == NULL) { goto dropfrag; + } #if CONFIG_MACF_NET if (mac_ipq_label_init(fp, M_NOWAIT) != 0) { ipq_free(fp); @@ -2744,7 +2860,7 @@ found: *divinfo = 0; *divcookie = 0; #endif /* IPDIVERT */ - m = NULL; /* nothing to return */ + m = NULL; /* nothing to return */ goto done; } else { fp->ipq_nfrags++; @@ -2753,7 +2869,7 @@ found: #endif } -#define GETIP(m) ((struct ip *)((m)->m_pkthdr.pkt_hdr)) +#define GETIP(m) ((struct ip *)((m)->m_pkthdr.pkt_hdr)) /* * Handle ECN by comparing this segment with the first one; @@ -2763,20 +2879,25 @@ found: ecn = ip->ip_tos & IPTOS_ECN_MASK; ecn0 = GETIP(fp->ipq_frags)->ip_tos & IPTOS_ECN_MASK; if (ecn == IPTOS_ECN_CE) { - if (ecn0 == IPTOS_ECN_NOTECT) + if (ecn0 == IPTOS_ECN_NOTECT) { goto dropfrag; - if (ecn0 != IPTOS_ECN_CE) + } + if (ecn0 != IPTOS_ECN_CE) { GETIP(fp->ipq_frags)->ip_tos |= IPTOS_ECN_CE; + } } - if (ecn == IPTOS_ECN_NOTECT && ecn0 != IPTOS_ECN_NOTECT) + if (ecn == IPTOS_ECN_NOTECT && ecn0 != IPTOS_ECN_NOTECT) { goto dropfrag; + } /* * Find a segment which begins after this one does. */ - for (p = NULL, q = fp->ipq_frags; q; p = q, q = q->m_nextpkt) - if (GETIP(q)->ip_off > ip->ip_off) + for (p = NULL, q = fp->ipq_frags; q; p = q, q = q->m_nextpkt) { + if (GETIP(q)->ip_off > ip->ip_off) { break; + } + } /* * If there is a preceding segment, it may provide some of @@ -2790,8 +2911,9 @@ found: if (p) { i = GETIP(p)->ip_off + GETIP(p)->ip_len - ip->ip_off; if (i > 0) { - if (i >= ip->ip_len) + if (i >= ip->ip_len) { goto dropfrag; + } m_adj(m, i); fp->ipq_csum_flags = 0; ip->ip_off += i; @@ -2831,10 +2953,11 @@ found: * as that of the existing ones, accumulate checksum. Otherwise, * invalidate checksum offload info for the entire datagram. */ - if (csum_flags != 0 && csum_flags == fp->ipq_csum_flags) + if (csum_flags != 0 && csum_flags == fp->ipq_csum_flags) { fp->ipq_csum += csum; - else if (fp->ipq_csum_flags != 0) + } else if (fp->ipq_csum_flags != 0) { fp->ipq_csum_flags = 0; + } #if IPDIVERT /* @@ -2870,7 +2993,7 @@ found: ipstat.ips_fragdropped += fp->ipq_nfrags; frag_freef(head, fp); } - m = NULL; /* nothing to return */ + m = NULL; /* nothing to return */ goto done; } next += GETIP(q)->ip_len; @@ -2881,7 +3004,7 @@ found: ipstat.ips_fragdropped += fp->ipq_nfrags; frag_freef(head, fp); } - m = NULL; /* nothing to return */ + m = NULL; /* nothing to return */ goto done; } @@ -2894,7 +3017,7 @@ found: ipstat.ips_toolong++; ipstat.ips_fragdropped += fp->ipq_nfrags; frag_freef(head, fp); - m = NULL; /* nothing to return */ + m = NULL; /* nothing to return */ goto done; } @@ -2924,7 +3047,7 @@ found: ADDCARRY(csum); m->m_pkthdr.csum_rx_val = csum; - m->m_pkthdr.csum_rx_start = sizeof (struct ip); + m->m_pkthdr.csum_rx_start = sizeof(struct ip); m->m_pkthdr.csum_flags = fp->ipq_csum_flags; } else if ((m->m_pkthdr.rcvif->if_flags & IFF_LOOPBACK) || (m->m_pkthdr.pkt_flags & PKTF_LOOP)) { @@ -2961,25 +3084,27 @@ found: ip->ip_src = fp->ipq_src; ip->ip_dst = fp->ipq_dst; - fp->ipq_frags = NULL; /* return to caller as 'm' */ + fp->ipq_frags = NULL; /* return to caller as 'm' */ frag_freef(head, fp); fp = NULL; m->m_len += (IP_VHL_HL(ip->ip_vhl) << 2); m->m_data -= (IP_VHL_HL(ip->ip_vhl) << 2); /* some debugging cruft by sklower, below, will go away soon */ - if (m->m_flags & M_PKTHDR) /* XXX this should be done elsewhere */ + if (m->m_flags & M_PKTHDR) { /* XXX this should be done elsewhere */ m_fixhdr(m); + } ipstat.ips_reassembled++; /* arm the purge timer if not already and if there's work to do */ frag_sched_timeout(); lck_mtx_unlock(&ipqlock); /* perform deferred free (if needed) now that lock is dropped */ - if (!MBUFQ_EMPTY(&dfq)) + if (!MBUFQ_EMPTY(&dfq)) { MBUFQ_DRAIN(&dfq); + } VERIFY(MBUFQ_EMPTY(&dfq)); - return (m); + return m; done: VERIFY(m == NULL); @@ -2987,10 +3112,11 @@ done: frag_sched_timeout(); lck_mtx_unlock(&ipqlock); /* perform deferred free (if needed) */ - if (!MBUFQ_EMPTY(&dfq)) + if (!MBUFQ_EMPTY(&dfq)) { MBUFQ_DRAIN(&dfq); + } VERIFY(MBUFQ_EMPTY(&dfq)); - return (NULL); + return NULL; dropfrag: #if IPDIVERT @@ -2998,17 +3124,19 @@ dropfrag: *divcookie = 0; #endif /* IPDIVERT */ ipstat.ips_fragdropped++; - if (fp != NULL) + if (fp != NULL) { fp->ipq_nfrags--; + } /* arm the purge timer if not already and if there's work to do */ frag_sched_timeout(); lck_mtx_unlock(&ipqlock); m_freem(m); /* perform deferred free (if needed) */ - if (!MBUFQ_EMPTY(&dfq)) + if (!MBUFQ_EMPTY(&dfq)) { MBUFQ_DRAIN(&dfq); + } VERIFY(MBUFQ_EMPTY(&dfq)); - return (NULL); + return NULL; #undef GETIP } @@ -3019,7 +3147,7 @@ dropfrag: static void frag_freef(struct ipqhead *fhp, struct ipq *fp) { - lck_mtx_assert(&ipqlock, LCK_MTX_ASSERT_OWNED); + LCK_MTX_ASSERT(&ipqlock, LCK_MTX_ASSERT_OWNED); fp->ipq_nfrags = 0; if (fp->ipq_frags != NULL) { @@ -3050,7 +3178,7 @@ frag_timeout(void *arg) lck_mtx_lock(&ipqlock); for (i = 0; i < IPREASS_NHASH; i++) { - for (fp = TAILQ_FIRST(&ipq[i]); fp; ) { + for (fp = TAILQ_FIRST(&ipq[i]); fp;) { struct ipq *fpp; fpp = fp; @@ -3085,7 +3213,7 @@ frag_timeout(void *arg) static void frag_sched_timeout(void) { - lck_mtx_assert(&ipqlock, LCK_MTX_ASSERT_OWNED); + LCK_MTX_ASSERT(&ipqlock, LCK_MTX_ASSERT_OWNED); if (!frag_timeout_run && nipq > 0) { frag_timeout_run = 1; @@ -3123,18 +3251,19 @@ ipq_alloc(int how) * from nipq since the latter represents the elements already * in the reassembly queues. */ - if (ipq_limit > 0 && ipq_count > ipq_limit) - return (NULL); + if (ipq_limit > 0 && ipq_count > ipq_limit) { + return NULL; + } t = m_get(how, MT_FTABLE); if (t != NULL) { atomic_add_32(&ipq_count, 1); fp = mtod(t, struct ipq *); - bzero(fp, sizeof (*fp)); + bzero(fp, sizeof(*fp)); } else { fp = NULL; } - return (fp); + return fp; } static void @@ -3150,9 +3279,9 @@ ipq_free(struct ipq *fp) void ip_drain(void) { - frag_drain(); /* fragments */ - in_rtqdrain(); /* protocol cloned routes */ - in_arpdrain(NULL); /* cloned routes: ARP */ + frag_drain(); /* fragments */ + in_rtqdrain(); /* protocol cloned routes */ + in_arpdrain(NULL); /* cloned routes: ARP */ } /* @@ -3179,34 +3308,35 @@ ip_dooptions(struct mbuf *m, int pass, struct sockaddr_in *next_hop) struct in_addr *sin, dst; u_int32_t ntime; struct sockaddr_in ipaddr = { - sizeof (ipaddr), AF_INET, 0, { 0 }, { 0, } }; + sizeof(ipaddr), AF_INET, 0, { 0 }, { 0, } + }; /* Expect 32-bit aligned data pointer on strict-align platforms */ MBUF_STRICT_DATA_ALIGNMENT_CHECK_32(m); dst = ip->ip_dst; cp = (u_char *)(ip + 1); - cnt = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof (struct ip); + cnt = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof(struct ip); for (; cnt > 0; cnt -= optlen, cp += optlen) { opt = cp[IPOPT_OPTVAL]; - if (opt == IPOPT_EOL) + if (opt == IPOPT_EOL) { break; - if (opt == IPOPT_NOP) + } + if (opt == IPOPT_NOP) { optlen = 1; - else { - if (cnt < IPOPT_OLEN + sizeof (*cp)) { + } else { + if (cnt < IPOPT_OLEN + sizeof(*cp)) { code = &cp[IPOPT_OLEN] - (u_char *)ip; goto bad; } optlen = cp[IPOPT_OLEN]; - if (optlen < IPOPT_OLEN + sizeof (*cp) || + if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) { code = &cp[IPOPT_OLEN] - (u_char *)ip; goto bad; } } switch (opt) { - default: break; @@ -3221,7 +3351,7 @@ ip_dooptions(struct mbuf *m, int pass, struct sockaddr_in *next_hop) */ case IPOPT_LSRR: case IPOPT_SSRR: - if (optlen < IPOPT_OFFSET + sizeof (*cp)) { + if (optlen < IPOPT_OFFSET + sizeof(*cp)) { code = &cp[IPOPT_OLEN] - (u_char *)ip; goto bad; } @@ -3237,8 +3367,9 @@ ip_dooptions(struct mbuf *m, int pass, struct sockaddr_in *next_hop) code = ICMP_UNREACH_SRCFAIL; goto bad; } - if (!ip_dosourceroute) + if (!ip_dosourceroute) { goto nosourcerouting; + } /* * Loose routing, and not at next destination * yet; nothing to do except forward. @@ -3248,13 +3379,14 @@ ip_dooptions(struct mbuf *m, int pass, struct sockaddr_in *next_hop) IFA_REMREF(&ia->ia_ifa); ia = NULL; } - off--; /* 0 origin */ - if (off > optlen - (int)sizeof (struct in_addr)) { + off--; /* 0 origin */ + if (off > optlen - (int)sizeof(struct in_addr)) { /* * End of source route. Should be for us. */ - if (!ip_acceptsourceroute) + if (!ip_acceptsourceroute) { goto nosourcerouting; + } save_rte(cp, ip->ip_src); break; } @@ -3271,9 +3403,9 @@ nosourcerouting: "attempted source route from %s " "to %s\n", inet_ntop(AF_INET, &ip->ip_src, - buf, sizeof (buf)), + buf, sizeof(buf)), inet_ntop(AF_INET, &ip->ip_dst, - buf2, sizeof (buf2))); + buf2, sizeof(buf2))); type = ICMP_UNREACH; code = ICMP_UNREACH_SRCFAIL; goto bad; @@ -3284,7 +3416,7 @@ nosourcerouting: */ OSAddAtomic(1, &ipstat.ips_cantforward); m_freem(m); - return (1); + return 1; } } @@ -3292,12 +3424,12 @@ nosourcerouting: * locate outgoing interface */ (void) memcpy(&ipaddr.sin_addr, cp + off, - sizeof (ipaddr.sin_addr)); + sizeof(ipaddr.sin_addr)); if (opt == IPOPT_SSRR) { -#define INA struct in_ifaddr * +#define INA struct in_ifaddr * if ((ia = (INA)ifa_ifwithdstaddr( - SA(&ipaddr))) == NULL) { + SA(&ipaddr))) == NULL) { ia = (INA)ifa_ifwithnet(SA(&ipaddr)); } } else { @@ -3311,11 +3443,11 @@ nosourcerouting: ip->ip_dst = ipaddr.sin_addr; IFA_LOCK(&ia->ia_ifa); (void) memcpy(cp + off, &(IA_SIN(ia)->sin_addr), - sizeof (struct in_addr)); + sizeof(struct in_addr)); IFA_UNLOCK(&ia->ia_ifa); IFA_REMREF(&ia->ia_ifa); ia = NULL; - cp[IPOPT_OFFSET] += sizeof (struct in_addr); + cp[IPOPT_OFFSET] += sizeof(struct in_addr); /* * Let ip_intr's mcast routing check handle mcast pkts */ @@ -3323,7 +3455,7 @@ nosourcerouting: break; case IPOPT_RR: - if (optlen < IPOPT_OFFSET + sizeof (*cp)) { + if (optlen < IPOPT_OFFSET + sizeof(*cp)) { code = &cp[IPOPT_OFFSET] - (u_char *)ip; goto bad; } @@ -3334,11 +3466,12 @@ nosourcerouting: /* * If no space remains, ignore. */ - off--; /* 0 origin */ - if (off > optlen - (int)sizeof (struct in_addr)) + off--; /* 0 origin */ + if (off > optlen - (int)sizeof(struct in_addr)) { break; + } (void) memcpy(&ipaddr.sin_addr, &ip->ip_dst, - sizeof (ipaddr.sin_addr)); + sizeof(ipaddr.sin_addr)); /* * locate outgoing interface; if we're the destination, * use the incoming interface (should be same). @@ -3352,11 +3485,11 @@ nosourcerouting: } IFA_LOCK(&ia->ia_ifa); (void) memcpy(cp + off, &(IA_SIN(ia)->sin_addr), - sizeof (struct in_addr)); + sizeof(struct in_addr)); IFA_UNLOCK(&ia->ia_ifa); IFA_REMREF(&ia->ia_ifa); ia = NULL; - cp[IPOPT_OFFSET] += sizeof (struct in_addr); + cp[IPOPT_OFFSET] += sizeof(struct in_addr); break; case IPOPT_TS: @@ -3371,7 +3504,7 @@ nosourcerouting: goto bad; } if (ipt->ipt_ptr > - ipt->ipt_len - (int)sizeof (int32_t)) { + ipt->ipt_len - (int)sizeof(int32_t)) { if (++ipt->ipt_oflw == 0) { code = (u_char *)&ipt->ipt_ptr - (u_char *)ip; @@ -3381,13 +3514,12 @@ nosourcerouting: } sin = (struct in_addr *)(void *)(cp + ipt->ipt_ptr - 1); switch (ipt->ipt_flg) { - case IPOPT_TS_TSONLY: break; case IPOPT_TS_TSANDADDR: - if (ipt->ipt_ptr - 1 + sizeof (n_time) + - sizeof (struct in_addr) > ipt->ipt_len) { + if (ipt->ipt_ptr - 1 + sizeof(n_time) + + sizeof(struct in_addr) > ipt->ipt_len) { code = (u_char *)&ipt->ipt_ptr - (u_char *)ip; goto bad; @@ -3395,32 +3527,34 @@ nosourcerouting: ipaddr.sin_addr = dst; ia = (INA)ifaof_ifpforaddr(SA(&ipaddr), m->m_pkthdr.rcvif); - if (ia == NULL) + if (ia == NULL) { continue; + } IFA_LOCK(&ia->ia_ifa); (void) memcpy(sin, &IA_SIN(ia)->sin_addr, - sizeof (struct in_addr)); + sizeof(struct in_addr)); IFA_UNLOCK(&ia->ia_ifa); - ipt->ipt_ptr += sizeof (struct in_addr); + ipt->ipt_ptr += sizeof(struct in_addr); IFA_REMREF(&ia->ia_ifa); ia = NULL; break; case IPOPT_TS_PRESPEC: - if (ipt->ipt_ptr - 1 + sizeof (n_time) + - sizeof (struct in_addr) > ipt->ipt_len) { + if (ipt->ipt_ptr - 1 + sizeof(n_time) + + sizeof(struct in_addr) > ipt->ipt_len) { code = (u_char *)&ipt->ipt_ptr - (u_char *)ip; goto bad; } (void) memcpy(&ipaddr.sin_addr, sin, - sizeof (struct in_addr)); + sizeof(struct in_addr)); if ((ia = (struct in_ifaddr *)ifa_ifwithaddr( - SA(&ipaddr))) == NULL) + SA(&ipaddr))) == NULL) { continue; + } IFA_REMREF(&ia->ia_ifa); ia = NULL; - ipt->ipt_ptr += sizeof (struct in_addr); + ipt->ipt_ptr += sizeof(struct in_addr); break; default: @@ -3431,19 +3565,19 @@ nosourcerouting: } ntime = iptime(); (void) memcpy(cp + ipt->ipt_ptr - 1, &ntime, - sizeof (n_time)); - ipt->ipt_ptr += sizeof (n_time); + sizeof(n_time)); + ipt->ipt_ptr += sizeof(n_time); } } if (forward && ipforwarding) { ip_forward(m, 1, next_hop); - return (1); + return 1; } - return (0); + return 0; bad: icmp_error(m, type, code, 0, 0); OSAddAtomic(1, &ipstat.ips_badoptions); - return (1); + return 1; } /* @@ -3471,40 +3605,43 @@ ip_checkrouteralert(struct mbuf *m) found_ra = 0; cp = (u_char *)(ip + 1); - cnt = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof (struct ip); + cnt = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof(struct ip); for (; cnt > 0; cnt -= optlen, cp += optlen) { opt = cp[IPOPT_OPTVAL]; - if (opt == IPOPT_EOL) + if (opt == IPOPT_EOL) { break; - if (opt == IPOPT_NOP) + } + if (opt == IPOPT_NOP) { optlen = 1; - else { + } else { #ifdef DIAGNOSTIC - if (cnt < IPOPT_OLEN + sizeof (*cp)) + if (cnt < IPOPT_OLEN + sizeof(*cp)) { break; + } #endif optlen = cp[IPOPT_OLEN]; #ifdef DIAGNOSTIC - if (optlen < IPOPT_OLEN + sizeof (*cp) || optlen > cnt) + if (optlen < IPOPT_OLEN + sizeof(*cp) || optlen > cnt) { break; + } #endif } switch (opt) { case IPOPT_RA: #ifdef DIAGNOSTIC - if (optlen != IPOPT_OFFSET + sizeof (uint16_t) || - (*((uint16_t *)(void *)&cp[IPOPT_OFFSET]) != 0)) + if (optlen != IPOPT_OFFSET + sizeof(uint16_t) || + (*((uint16_t *)(void *)&cp[IPOPT_OFFSET]) != 0)) { break; - else + } else #endif - found_ra = 1; + found_ra = 1; break; default: break; } } - return (found_ra); + return found_ra; } /* @@ -3518,25 +3655,26 @@ ip_rtaddr(struct in_addr dst) struct ifaddr *rt_ifa; struct route ro; - bzero(&ro, sizeof (ro)); + bzero(&ro, sizeof(ro)); sin = SIN(&ro.ro_dst); sin->sin_family = AF_INET; - sin->sin_len = sizeof (*sin); + sin->sin_len = sizeof(*sin); sin->sin_addr = dst; rtalloc_ign(&ro, RTF_PRCLONING); if (ro.ro_rt == NULL) { ROUTE_RELEASE(&ro); - return (NULL); + return NULL; } RT_LOCK(ro.ro_rt); - if ((rt_ifa = ro.ro_rt->rt_ifa) != NULL) + if ((rt_ifa = ro.ro_rt->rt_ifa) != NULL) { IFA_ADDREF(rt_ifa); + } RT_UNLOCK(ro.ro_rt); ROUTE_RELEASE(&ro); - return ((struct in_ifaddr *)rt_ifa); + return (struct in_ifaddr *)rt_ifa; } /* @@ -3550,13 +3688,15 @@ save_rte(u_char *option, struct in_addr dst) olen = option[IPOPT_OLEN]; #if DIAGNOSTIC - if (ipprintfs) + if (ipprintfs) { printf("save_rte: olen %d\n", olen); + } #endif - if (olen > sizeof (ip_srcrt) - (1 + sizeof (dst))) + if (olen > sizeof(ip_srcrt) - (1 + sizeof(dst))) { return; + } bcopy(option, ip_srcrt.srcopt, olen); - ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof (struct in_addr); + ip_nhops = (olen - IPOPT_OFFSET - 1) / sizeof(struct in_addr); ip_srcrt.dst = dst; } @@ -3571,21 +3711,24 @@ ip_srcroute(void) struct in_addr *p, *q; struct mbuf *m; - if (ip_nhops == 0) - return (NULL); + if (ip_nhops == 0) { + return NULL; + } m = m_get(M_DONTWAIT, MT_HEADER); - if (m == NULL) - return (NULL); + if (m == NULL) { + return NULL; + } -#define OPTSIZ (sizeof (ip_srcrt.nop) + sizeof (ip_srcrt.srcopt)) +#define OPTSIZ (sizeof (ip_srcrt.nop) + sizeof (ip_srcrt.srcopt)) /* length is (nhops+1)*sizeof(addr) + sizeof(nop + srcrt header) */ - m->m_len = ip_nhops * sizeof (struct in_addr) + - sizeof (struct in_addr) + OPTSIZ; + m->m_len = ip_nhops * sizeof(struct in_addr) + + sizeof(struct in_addr) + OPTSIZ; #if DIAGNOSTIC - if (ipprintfs) + if (ipprintfs) { printf("ip_srcroute: nhops %d mlen %d", ip_nhops, m->m_len); + } #endif /* @@ -3594,9 +3737,10 @@ ip_srcroute(void) p = &ip_srcrt.route[ip_nhops - 1]; *(mtod(m, struct in_addr *)) = *p--; #if DIAGNOSTIC - if (ipprintfs) + if (ipprintfs) { printf(" hops %lx", (u_int32_t)ntohl(mtod(m, struct in_addr *)->s_addr)); + } #endif /* @@ -3604,10 +3748,10 @@ ip_srcroute(void) */ ip_srcrt.nop = IPOPT_NOP; ip_srcrt.srcopt[IPOPT_OFFSET] = IPOPT_MINOFF; - (void) memcpy(mtod(m, caddr_t) + sizeof (struct in_addr), + (void) memcpy(mtod(m, caddr_t) + sizeof(struct in_addr), &ip_srcrt.nop, OPTSIZ); q = (struct in_addr *)(void *)(mtod(m, caddr_t) + - sizeof (struct in_addr) + OPTSIZ); + sizeof(struct in_addr) + OPTSIZ); #undef OPTSIZ /* * Record return path as an IP source route, @@ -3615,8 +3759,9 @@ ip_srcroute(void) */ while (p >= ip_srcrt.route) { #if DIAGNOSTIC - if (ipprintfs) + if (ipprintfs) { printf(" %lx", (u_int32_t)ntohl(q->s_addr)); + } #endif *q++ = *p--; } @@ -3625,23 +3770,19 @@ ip_srcroute(void) */ *q = ip_srcrt.dst; #if DIAGNOSTIC - if (ipprintfs) + if (ipprintfs) { printf(" %lx\n", (u_int32_t)ntohl(q->s_addr)); + } #endif - return (m); + return m; } /* - * Strip out IP options, at higher - * level protocol in the kernel. - * Second argument is buffer to which options - * will be moved, and return value is their length. - * XXX should be deleted; last arg currently ignored. + * Strip out IP options, at higher level protocol in the kernel. */ void -ip_stripoptions(struct mbuf *m, struct mbuf *mopt) +ip_stripoptions(struct mbuf *m) { -#pragma unused(mopt) int i; struct ip *ip = mtod(m, struct ip *); caddr_t opts; @@ -3650,23 +3791,46 @@ ip_stripoptions(struct mbuf *m, struct mbuf *mopt) /* Expect 32-bit aligned data pointer on strict-align platforms */ MBUF_STRICT_DATA_ALIGNMENT_CHECK_32(m); - olen = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof (struct ip); + /* use bcopy() since it supports overlapping range */ + olen = (IP_VHL_HL(ip->ip_vhl) << 2) - sizeof(struct ip); opts = (caddr_t)(ip + 1); - i = m->m_len - (sizeof (struct ip) + olen); + i = m->m_len - (sizeof(struct ip) + olen); bcopy(opts + olen, opts, (unsigned)i); m->m_len -= olen; - if (m->m_flags & M_PKTHDR) + if (m->m_flags & M_PKTHDR) { m->m_pkthdr.len -= olen; - ip->ip_vhl = IP_MAKE_VHL(IPVERSION, sizeof (struct ip) >> 2); + } + ip->ip_vhl = IP_MAKE_VHL(IPVERSION, sizeof(struct ip) >> 2); + + /* + * We expect ip_{off,len} to be in host order by now, and + * that the original IP header length has been subtracted + * out from ip_len. Temporarily adjust ip_len for checksum + * recalculation, and restore it afterwards. + */ + ip->ip_len += sizeof(struct ip); + + /* recompute checksum now that IP header is smaller */ +#if BYTE_ORDER != BIG_ENDIAN + HTONS(ip->ip_len); + HTONS(ip->ip_off); +#endif /* BYTE_ORDER != BIG_ENDIAN */ + ip->ip_sum = in_cksum_hdr(ip); +#if BYTE_ORDER != BIG_ENDIAN + NTOHS(ip->ip_off); + NTOHS(ip->ip_len); +#endif /* BYTE_ORDER != BIG_ENDIAN */ + + ip->ip_len -= sizeof(struct ip); } u_char inetctlerrmap[PRC_NCMDS] = { - 0, 0, 0, 0, - 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, - ENETUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, - EMSGSIZE, EHOSTUNREACH, 0, 0, - 0, 0, 0, 0, - ENOPROTOOPT, ECONNREFUSED + 0, 0, 0, 0, + 0, EMSGSIZE, EHOSTDOWN, EHOSTUNREACH, + ENETUNREACH, EHOSTUNREACH, ECONNREFUSED, ECONNREFUSED, + EMSGSIZE, EHOSTUNREACH, 0, 0, + 0, 0, 0, 0, + ENOPROTOOPT, ECONNREFUSED }; static int @@ -3676,8 +3840,9 @@ sysctl_ipforwarding SYSCTL_HANDLER_ARGS int i, was_ipforwarding = ipforwarding; i = sysctl_handle_int(oidp, oidp->oid_arg1, oidp->oid_arg2, req); - if (i != 0 || req->newptr == USER_ADDR_NULL) - return (i); + if (i != 0 || req->newptr == USER_ADDR_NULL) { + return i; + } if (was_ipforwarding && !ipforwarding) { /* clean up IPv4 forwarding cached routes */ @@ -3688,14 +3853,14 @@ sysctl_ipforwarding SYSCTL_HANDLER_ARGS lck_mtx_lock(&ifp->if_cached_route_lock); ROUTE_RELEASE(&ifp->if_fwd_route); bzero(&ifp->if_fwd_route, - sizeof (ifp->if_fwd_route)); + sizeof(ifp->if_fwd_route)); lck_mtx_unlock(&ifp->if_cached_route_lock); } } ifnet_head_done(); } - return (0); + return 0; } /* @@ -3712,10 +3877,11 @@ ip_fwd_route_copyout(struct ifnet *ifp, struct route *dst) lck_mtx_convert_spin(&ifp->if_cached_route_lock); /* Minor sanity check */ - if (src->ro_rt != NULL && rt_key(src->ro_rt)->sa_family != AF_INET) + if (src->ro_rt != NULL && rt_key(src->ro_rt)->sa_family != AF_INET) { panic("%s: wrong or corrupted route: %p", __func__, src); + } - route_copyout(dst, src, sizeof (*dst)); + route_copyout(dst, src, sizeof(*dst)); lck_mtx_unlock(&ifp->if_cached_route_lock); } @@ -3729,11 +3895,13 @@ ip_fwd_route_copyin(struct ifnet *ifp, struct route *src) lck_mtx_convert_spin(&ifp->if_cached_route_lock); /* Minor sanity check */ - if (src->ro_rt != NULL && rt_key(src->ro_rt)->sa_family != AF_INET) + if (src->ro_rt != NULL && rt_key(src->ro_rt)->sa_family != AF_INET) { panic("%s: wrong or corrupted route: %p", __func__, src); + } - if (ifp->if_fwd_cacheok) - route_copyin(src, dst, sizeof (*src)); + if (ifp->if_fwd_cacheok) { + route_copyin(src, dst, sizeof(*src)); + } lck_mtx_unlock(&ifp->if_cached_route_lock); } @@ -3767,9 +3935,14 @@ ip_forward(struct mbuf *m, int srcrt, struct sockaddr_in *next_hop) n_long dest; struct in_addr pkt_dst; u_int32_t nextmtu = 0, len; - struct ip_out_args ipoa = { IFSCOPE_NONE, { 0 }, 0, 0, - SO_TC_UNSPEC, _NET_SERVICE_TYPE_UNSPEC }; + struct ip_out_args ipoa; struct ifnet *rcvifp = m->m_pkthdr.rcvif; + + bzero(&ipoa, sizeof(ipoa)); + ipoa.ipoa_boundif = IFSCOPE_NONE; + ipoa.ipoa_sotc = SO_TC_UNSPEC; + ipoa.ipoa_netsvctype = _NET_SERVICE_TYPE_UNSPEC; + #if IPSEC struct secpolicy *sp = NULL; int ipsecerror; @@ -3790,13 +3963,14 @@ ip_forward(struct mbuf *m, int srcrt, struct sockaddr_in *next_hop) #endif /* !IPFIREWALL */ #if DIAGNOSTIC - if (ipprintfs) + if (ipprintfs) { printf("forward: src %lx dst %lx ttl %x\n", (u_int32_t)ip->ip_src.s_addr, (u_int32_t)pkt_dst.s_addr, ip->ip_ttl); + } #endif - if (m->m_flags & (M_BCAST|M_MCAST) || !in_canforward(pkt_dst)) { + if (m->m_flags & (M_BCAST | M_MCAST) || !in_canforward(pkt_dst)) { OSAddAtomic(1, &ipstat.ips_cantforward); m_freem(m); return; @@ -3804,13 +3978,13 @@ ip_forward(struct mbuf *m, int srcrt, struct sockaddr_in *next_hop) #if IPSTEALTH if (!ipstealth) { #endif /* IPSTEALTH */ - if (ip->ip_ttl <= IPTTLDEC) { - icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, - dest, 0); - return; - } -#if IPSTEALTH + if (ip->ip_ttl <= IPTTLDEC) { + icmp_error(m, ICMP_TIMXCEED, ICMP_TIMXCEED_INTRANS, + dest, 0); + return; } +#if IPSTEALTH +} #endif /* IPSTEALTH */ #if PF @@ -3828,7 +4002,7 @@ ip_forward(struct mbuf *m, int srcrt, struct sockaddr_in *next_hop) ROUTE_RELEASE(&fwd_rt); sin->sin_family = AF_INET; - sin->sin_len = sizeof (*sin); + sin->sin_len = sizeof(*sin); sin->sin_addr = pkt_dst; rtalloc_scoped_ign(&fwd_rt, RTF_PRCLONING, ipoa.ipoa_boundif); @@ -3859,9 +4033,9 @@ ip_forward(struct mbuf *m, int srcrt, struct sockaddr_in *next_hop) #if IPSTEALTH if (!ipstealth) { #endif /* IPSTEALTH */ - ip->ip_ttl -= IPTTLDEC; + ip->ip_ttl -= IPTTLDEC; #if IPSTEALTH - } +} #endif /* IPSTEALTH */ /* @@ -3874,7 +4048,7 @@ ip_forward(struct mbuf *m, int srcrt, struct sockaddr_in *next_hop) */ RT_LOCK_SPIN(rt); if (rt->rt_ifp == m->m_pkthdr.rcvif && - !(rt->rt_flags & (RTF_DYNAMIC|RTF_MODIFIED)) && + !(rt->rt_flags & (RTF_DYNAMIC | RTF_MODIFIED)) && satosin(rt_key(rt))->sin_addr.s_addr != INADDR_ANY && ipsendredirects && !srcrt && rt->rt_ifa != NULL) { struct in_ifaddr *ia = (struct in_ifaddr *)rt->rt_ifa; @@ -3884,10 +4058,11 @@ ip_forward(struct mbuf *m, int srcrt, struct sockaddr_in *next_hop) RT_CONVERT_LOCK(rt); IFA_LOCK_SPIN(&ia->ia_ifa); if ((src & ia->ia_subnetmask) == ia->ia_subnet) { - if (rt->rt_flags & RTF_GATEWAY) + if (rt->rt_flags & RTF_GATEWAY) { dest = satosin(rt->rt_gateway)->sin_addr.s_addr; - else + } else { dest = pkt_dst.s_addr; + } /* * Router requirements says to only send * host redirects. @@ -3895,9 +4070,10 @@ ip_forward(struct mbuf *m, int srcrt, struct sockaddr_in *next_hop) type = ICMP_REDIRECT; code = ICMP_REDIRECT_HOST; #if DIAGNOSTIC - if (ipprintfs) + if (ipprintfs) { printf("redirect (%d) to %lx\n", code, (u_int32_t)dest); + } #endif } IFA_UNLOCK(&ia->ia_ifa); @@ -3912,14 +4088,14 @@ ip_forward(struct mbuf *m, int srcrt, struct sockaddr_in *next_hop) tag = m_tag_create(KERNEL_MODULE_TAG_ID, KERNEL_TAG_TYPE_IPFORWARD, - sizeof (*ipfwd_tag), M_NOWAIT, m); + sizeof(*ipfwd_tag), M_NOWAIT, m); if (tag == NULL) { error = ENOBUFS; m_freem(m); goto done; } - ipfwd_tag = (struct ip_fwd_tag *)(tag+1); + ipfwd_tag = (struct ip_fwd_tag *)(tag + 1); ipfwd_tag->next_hop = next_hop; m_tag_prepend(m, tag); @@ -3963,15 +4139,16 @@ ip_forward(struct mbuf *m, int srcrt, struct sockaddr_in *next_hop) goto done; } } - if (mcopy == NULL) + if (mcopy == NULL) { goto done; + } switch (error) { - case 0: /* forwarded, but need redirect */ + case 0: /* forwarded, but need redirect */ /* type, code set above */ break; - case ENETUNREACH: /* shouldn't happen, checked above */ + case ENETUNREACH: /* shouldn't happen, checked above */ case EHOSTUNREACH: case ENETDOWN: case EHOSTDOWN: @@ -3988,13 +4165,15 @@ ip_forward(struct mbuf *m, int srcrt, struct sockaddr_in *next_hop) break; } else { RT_LOCK_SPIN(rt); - if (rt->rt_ifp != NULL) + if (rt->rt_ifp != NULL) { nextmtu = rt->rt_ifp->if_mtu; + } RT_UNLOCK(rt); } #ifdef IPSEC - if (ipsec_bypass) + if (ipsec_bypass) { break; + } /* * If the packet is routed over IPsec tunnel, tell the @@ -4005,8 +4184,9 @@ ip_forward(struct mbuf *m, int srcrt, struct sockaddr_in *next_hop) sp = ipsec4_getpolicybyaddr(mcopy, IPSEC_DIR_OUTBOUND, IP_FORWARDING, &ipsecerror); - if (sp == NULL) + if (sp == NULL) { break; + } /* * find the correct route for outer IPv4 @@ -4026,30 +4206,30 @@ ip_forward(struct mbuf *m, int srcrt, struct sockaddr_in *next_hop) ipsechdr = ipsec_hdrsiz(sp); ipm = mtod(mcopy, struct ip *); - bcopy(&sp->req->saidx, &saidx, sizeof (saidx)); + bcopy(&sp->req->saidx, &saidx, sizeof(saidx)); saidx.mode = sp->req->saidx.mode; saidx.reqid = sp->req->saidx.reqid; sin = SIN(&saidx.src); if (sin->sin_len == 0) { - sin->sin_len = sizeof (*sin); + sin->sin_len = sizeof(*sin); sin->sin_family = AF_INET; sin->sin_port = IPSEC_PORT_ANY; bcopy(&ipm->ip_src, &sin->sin_addr, - sizeof (sin->sin_addr)); + sizeof(sin->sin_addr)); } sin = SIN(&saidx.dst); if (sin->sin_len == 0) { - sin->sin_len = sizeof (*sin); + sin->sin_len = sizeof(*sin); sin->sin_family = AF_INET; sin->sin_port = IPSEC_PORT_ANY; bcopy(&ipm->ip_dst, &sin->sin_addr, - sizeof (sin->sin_addr)); + sizeof(sin->sin_addr)); } sav = key_allocsa_policy(&saidx); if (sav != NULL) { lck_mtx_lock(sadb_mutex); if (sav->sah != NULL) { - ro = &sav->sah->sa_route; + ro = (struct route *)&sav->sah->sa_route; if (ro->ro_rt != NULL) { RT_LOCK(ro->ro_rt); if (ro->ro_rt->rt_ifp != NULL) { @@ -4086,13 +4266,14 @@ ip_forward(struct mbuf *m, int srcrt, struct sockaddr_in *next_hop) } break; - case EACCES: /* ipfw denied packet */ + case EACCES: /* ipfw denied packet */ m_freem(mcopy); goto done; } - if (type == ICMP_UNREACH && code == ICMP_UNREACH_NEEDFRAG) + if (type == ICMP_UNREACH && code == ICMP_UNREACH_NEEDFRAG) { OSAddAtomic(1, &ipstat.ips_cantfrag); + } icmp_error(mcopy, type, code, dest, nextmtu); done: @@ -4108,7 +4289,7 @@ ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip, struct timeval tv; getmicrotime(&tv); - mp = sbcreatecontrol_mbuf((caddr_t)&tv, sizeof (tv), + mp = sbcreatecontrol_mbuf((caddr_t)&tv, sizeof(tv), SCM_TIMESTAMP, SOL_SOCKET, mp); if (*mp == NULL) { goto no_mbufs; @@ -4118,15 +4299,25 @@ ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip, uint64_t time; time = mach_absolute_time(); - mp = sbcreatecontrol_mbuf((caddr_t)&time, sizeof (time), + mp = sbcreatecontrol_mbuf((caddr_t)&time, sizeof(time), SCM_TIMESTAMP_MONOTONIC, SOL_SOCKET, mp); if (*mp == NULL) { goto no_mbufs; } } + if (inp->inp_socket->so_options & SO_TIMESTAMP_CONTINUOUS) { + uint64_t time; + + time = mach_continuous_time(); + mp = sbcreatecontrol_mbuf((caddr_t)&time, sizeof(time), + SCM_TIMESTAMP_CONTINUOUS, SOL_SOCKET, mp); + if (*mp == NULL) { + goto no_mbufs; + } + } if (inp->inp_flags & INP_RECVDSTADDR) { mp = sbcreatecontrol_mbuf((caddr_t)&ip->ip_dst, - sizeof (struct in_addr), IP_RECVDSTADDR, IPPROTO_IP, mp); + sizeof(struct in_addr), IP_RECVDSTADDR, IPPROTO_IP, mp); if (*mp == NULL) { goto no_mbufs; } @@ -4140,7 +4331,7 @@ ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip, /* options were tossed already */ if (inp->inp_flags & INP_RECVOPTS) { mp = sbcreatecontrol_mbuf((caddr_t)opts_deleted_above, - sizeof (struct in_addr), IP_RECVOPTS, IPPROTO_IP, mp); + sizeof(struct in_addr), IP_RECVOPTS, IPPROTO_IP, mp); if (*mp == NULL) { goto no_mbufs; } @@ -4148,7 +4339,7 @@ ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip, /* ip_srcroute doesn't do what we want here, need to fix */ if (inp->inp_flags & INP_RECVRETOPTS) { mp = sbcreatecontrol_mbuf((caddr_t)ip_srcroute(), - sizeof (struct in_addr), IP_RECVRETOPTS, IPPROTO_IP, mp); + sizeof(struct in_addr), IP_RECVRETOPTS, IPPROTO_IP, mp); if (*mp == NULL) { goto no_mbufs; } @@ -4163,7 +4354,7 @@ ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip, * Make sure to accomodate the largest possible * size of SA(if_lladdr)->sa_len. */ - _CASSERT(sizeof (sdlbuf) == (SOCK_MAXADDRLEN + 1)); + _CASSERT(sizeof(sdlbuf) == (SOCK_MAXADDRLEN + 1)); ifnet_head_lock_shared(); if ((ifp = m->m_pkthdr.rcvif) != NULL && @@ -4171,8 +4362,9 @@ ip_savecontrol(struct inpcb *inp, struct mbuf **mp, struct ip *ip, struct ifaddr *ifa = ifnet_addrs[ifp->if_index - 1]; struct sockaddr_dl *sdp; - if (!ifa || !ifa->ifa_addr) + if (!ifa || !ifa->ifa_addr) { goto makedummy; + } IFA_LOCK_SPIN(ifa); sdp = SDL(ifa->ifa_addr); @@ -4203,7 +4395,7 @@ makedummy: } if (inp->inp_flags & INP_RECVTTL) { mp = sbcreatecontrol_mbuf((caddr_t)&ip->ip_ttl, - sizeof (ip->ip_ttl), IP_RECVTTL, IPPROTO_IP, mp); + sizeof(ip->ip_ttl), IP_RECVTTL, IPPROTO_IP, mp); if (*mp == NULL) { goto no_mbufs; } @@ -4211,7 +4403,7 @@ makedummy: if (inp->inp_socket->so_flags & SOF_RECV_TRAFFIC_CLASS) { int tc = m_get_traffic_class(m); - mp = sbcreatecontrol_mbuf((caddr_t)&tc, sizeof (tc), + mp = sbcreatecontrol_mbuf((caddr_t)&tc, sizeof(tc), SO_TRAFFIC_CLASS, SOL_SOCKET, mp); if (*mp == NULL) { goto no_mbufs; @@ -4220,22 +4412,29 @@ makedummy: if (inp->inp_flags & INP_PKTINFO) { struct in_pktinfo pi; - bzero(&pi, sizeof (struct in_pktinfo)); - bcopy(&ip->ip_dst, &pi.ipi_addr, sizeof (struct in_addr)); + bzero(&pi, sizeof(struct in_pktinfo)); + bcopy(&ip->ip_dst, &pi.ipi_addr, sizeof(struct in_addr)); pi.ipi_ifindex = (m != NULL && m->m_pkthdr.rcvif != NULL) ? m->m_pkthdr.rcvif->if_index : 0; mp = sbcreatecontrol_mbuf((caddr_t)&pi, - sizeof (struct in_pktinfo), IP_RECVPKTINFO, IPPROTO_IP, mp); + sizeof(struct in_pktinfo), IP_RECVPKTINFO, IPPROTO_IP, mp); + if (*mp == NULL) { + goto no_mbufs; + } + } + if (inp->inp_flags & INP_RECVTOS) { + mp = sbcreatecontrol_mbuf((caddr_t)&ip->ip_tos, + sizeof(u_char), IP_RECVTOS, IPPROTO_IP, mp); if (*mp == NULL) { goto no_mbufs; } } - return (0); + return 0; no_mbufs: ipstat.ips_pktdropcntrl++; - return (ENOBUFS); + return ENOBUFS; } static inline u_short @@ -4259,20 +4458,22 @@ ip_cksum(struct mbuf *m, int hlen) m->m_pkthdr.csum_data = 0xffff; } - if (sum != 0) + if (sum != 0) { OSAddAtomic(1, &ipstat.ips_badsum); + } - return (sum); + return sum; } static int ip_getstat SYSCTL_HANDLER_ARGS { #pragma unused(oidp, arg1, arg2) - if (req->oldptr == USER_ADDR_NULL) - req->oldlen = (size_t)sizeof (struct ipstat); + if (req->oldptr == USER_ADDR_NULL) { + req->oldlen = (size_t)sizeof(struct ipstat); + } - return (SYSCTL_OUT(req, &ipstat, MIN(sizeof (ipstat), req->oldlen))); + return SYSCTL_OUT(req, &ipstat, MIN(sizeof(ipstat), req->oldlen)); } void @@ -4290,8 +4491,9 @@ ip_setsrcifaddr_info(struct mbuf *m, uint32_t src_idx, struct in_ifaddr *ia) m->m_pkthdr.src_ifindex = ia->ia_ifp->if_index; } else { m->m_pkthdr.src_ifindex = src_idx; - if (src_idx != 0) + if (src_idx != 0) { m->m_pkthdr.pkt_flags |= PKTF_IFAINFO; + } } } @@ -4310,8 +4512,9 @@ ip_setdstifaddr_info(struct mbuf *m, uint32_t dst_idx, struct in_ifaddr *ia) m->m_pkthdr.dst_ifindex = ia->ia_ifp->if_index; } else { m->m_pkthdr.dst_ifindex = dst_idx; - if (dst_idx != 0) + if (dst_idx != 0) { m->m_pkthdr.pkt_flags |= PKTF_IFAINFO; + } } } @@ -4320,16 +4523,19 @@ ip_getsrcifaddr_info(struct mbuf *m, uint32_t *src_idx, uint32_t *iaf) { VERIFY(m->m_flags & M_PKTHDR); - if (!(m->m_pkthdr.pkt_flags & PKTF_IFAINFO)) - return (-1); + if (!(m->m_pkthdr.pkt_flags & PKTF_IFAINFO)) { + return -1; + } - if (src_idx != NULL) + if (src_idx != NULL) { *src_idx = m->m_pkthdr.src_ifindex; + } - if (iaf != NULL) + if (iaf != NULL) { *iaf = 0; + } - return (0); + return 0; } int @@ -4337,16 +4543,19 @@ ip_getdstifaddr_info(struct mbuf *m, uint32_t *dst_idx, uint32_t *iaf) { VERIFY(m->m_flags & M_PKTHDR); - if (!(m->m_pkthdr.pkt_flags & PKTF_IFAINFO)) - return (-1); + if (!(m->m_pkthdr.pkt_flags & PKTF_IFAINFO)) { + return -1; + } - if (dst_idx != NULL) + if (dst_idx != NULL) { *dst_idx = m->m_pkthdr.dst_ifindex; + } - if (iaf != NULL) + if (iaf != NULL) { *iaf = 0; + } - return (0); + return 0; } /* @@ -4370,8 +4579,9 @@ gre_input(struct mbuf *m, int off) * If no matching tunnel that is up is found, we inject * the mbuf to raw ip socket to see if anyone picks it up. */ - if (m != NULL) + if (m != NULL) { rip_input(m, off); + } } /* @@ -4384,7 +4594,7 @@ ip_gre_register_input(gre_input_func_t fn) gre_input_func = fn; lck_mtx_unlock(inet_domain_mutex); - return (0); + return 0; } #if (DEBUG || DEVELOPMENT) @@ -4396,8 +4606,9 @@ sysctl_reset_ip_input_stats SYSCTL_HANDLER_ARGS i = ip_input_measure; error = sysctl_handle_int(oidp, &i, 0, req); - if (error || req->newptr == USER_ADDR_NULL) + if (error || req->newptr == USER_ADDR_NULL) { goto done; + } /* impose bounds */ if (i < 0 || i > 1) { error = EINVAL; @@ -4408,7 +4619,7 @@ sysctl_reset_ip_input_stats SYSCTL_HANDLER_ARGS } ip_input_measure = i; done: - return (error); + return error; } static int @@ -4420,8 +4631,9 @@ sysctl_ip_input_measure_bins SYSCTL_HANDLER_ARGS i = ip_input_measure_bins; error = sysctl_handle_quad(oidp, &i, 0, req); - if (error || req->newptr == USER_ADDR_NULL) + if (error || req->newptr == USER_ADDR_NULL) { goto done; + } /* validate data */ if (!net_perf_validate_bins(i)) { error = EINVAL; @@ -4429,16 +4641,17 @@ sysctl_ip_input_measure_bins SYSCTL_HANDLER_ARGS } ip_input_measure_bins = i; done: - return (error); + return error; } static int sysctl_ip_input_getperf SYSCTL_HANDLER_ARGS { #pragma unused(oidp, arg1, arg2) - if (req->oldptr == USER_ADDR_NULL) - req->oldlen = (size_t)sizeof (struct ipstat); + if (req->oldptr == USER_ADDR_NULL) { + req->oldlen = (size_t)sizeof(struct ipstat); + } - return (SYSCTL_OUT(req, &net_perf, MIN(sizeof (net_perf), req->oldlen))); + return SYSCTL_OUT(req, &net_perf, MIN(sizeof(net_perf), req->oldlen)); } #endif /* (DEBUG || DEVELOPMENT) */