#include <sys/param.h>
#include <sys/systm.h>
#include <sys/callout.h>
#include <sys/kernel.h>
#include <sys/sysctl.h>
#include <sys/malloc.h>
#include <sys/mbuf.h>
#if INET6
#include <sys/domain.h>
#endif
#include <sys/proc.h>
#include <sys/kauth.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/protosw.h>
#include <sys/random.h>
#include <sys/syslog.h>
#include <kern/locks.h>
#include <net/route.h>
#include <net/if.h>
#define _IP_VHL
#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/ip.h>
#if INET6
#include <netinet/ip6.h>
#endif
#include <netinet/in_pcb.h>
#if INET6
#include <netinet6/in6_pcb.h>
#endif
#include <netinet/in_var.h>
#include <netinet/ip_var.h>
#if INET6
#include <netinet6/ip6_var.h>
#endif
#include <netinet/tcp.h>
#include <netinet/tcp_fsm.h>
#include <netinet/tcp_seq.h>
#include <netinet/tcp_timer.h>
#include <netinet/tcp_var.h>
#if INET6
#include <netinet6/tcp6_var.h>
#endif
#include <netinet/tcpip.h>
#if TCPDEBUG
#include <netinet/tcp_debug.h>
#endif
#include <netinet6/ip6protosw.h>
#if IPSEC
#include <netinet6/ipsec.h>
#if INET6
#include <netinet6/ipsec6.h>
#endif
#endif
#include <sys/md5.h>
#include <sys/kdebug.h>
#define DBG_FNC_TCP_CLOSE NETDBG_CODE(DBG_NETTCP, ((5 << 8) | 2))
extern int tcp_lq_overflow;
#if IPSEC
extern int ipsec_bypass;
extern lck_mtx_t *sadb_mutex;
#endif
int tcp_mssdflt = TCP_MSS;
SYSCTL_INT(_net_inet_tcp, TCPCTL_MSSDFLT, mssdflt, CTLFLAG_RW,
&tcp_mssdflt , 0, "Default TCP Maximum Segment Size");
#if INET6
int tcp_v6mssdflt = TCP6_MSS;
SYSCTL_INT(_net_inet_tcp, TCPCTL_V6MSSDFLT, v6mssdflt,
CTLFLAG_RW, &tcp_v6mssdflt , 0,
"Default TCP Maximum Segment Size for IPv6");
#endif
int tcp_minmss = TCP_MINMSS;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, minmss, CTLFLAG_RW,
&tcp_minmss , 0, "Minmum TCP Maximum Segment Size");
#ifdef FIX_WORKAROUND_FOR_3894301
__private_extern__ int tcp_minmssoverload = TCP_MINMSSOVERLOAD;
#else
__private_extern__ int tcp_minmssoverload = 0;
#endif
SYSCTL_INT(_net_inet_tcp, OID_AUTO, minmssoverload, CTLFLAG_RW,
&tcp_minmssoverload , 0, "Number of TCP Segments per Second allowed to"
"be under the MINMSS Size");
static int tcp_do_rfc1323 = 1;
SYSCTL_INT(_net_inet_tcp, TCPCTL_DO_RFC1323, rfc1323, CTLFLAG_RW,
&tcp_do_rfc1323 , 0, "Enable rfc1323 (high performance TCP) extensions");
static int tcp_do_rfc1644 = 0;
SYSCTL_INT(_net_inet_tcp, TCPCTL_DO_RFC1644, rfc1644, CTLFLAG_RW,
&tcp_do_rfc1644 , 0, "Enable rfc1644 (TTCP) extensions");
static int tcp_tcbhashsize = 0;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, tcbhashsize, CTLFLAG_RD,
&tcp_tcbhashsize, 0, "Size of TCP control-block hashtable");
static int do_tcpdrain = 0;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, do_tcpdrain, CTLFLAG_RW, &do_tcpdrain, 0,
"Enable tcp_drain routine for extra help when low on mbufs");
SYSCTL_INT(_net_inet_tcp, OID_AUTO, pcbcount, CTLFLAG_RD,
&tcbinfo.ipi_count, 0, "Number of active PCBs");
static int icmp_may_rst = 1;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, icmp_may_rst, CTLFLAG_RW, &icmp_may_rst, 0,
"Certain ICMP unreachable messages may abort connections in SYN_SENT");
static int tcp_strict_rfc1948 = 0;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, strict_rfc1948, CTLFLAG_RW,
&tcp_strict_rfc1948, 0, "Determines if RFC1948 is followed exactly");
static int tcp_isn_reseed_interval = 0;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, isn_reseed_interval, CTLFLAG_RW,
&tcp_isn_reseed_interval, 0, "Seconds between reseeding of ISN secret");
static void tcp_cleartaocache(void);
static void tcp_notify(struct inpcb *, int);
struct zone *sack_hole_zone;
#ifndef TCBHASHSIZE
#define TCBHASHSIZE 4096
#endif
#define ALIGNMENT 32
#define ALIGNM1 (ALIGNMENT - 1)
struct inp_tp {
union {
struct inpcb inp;
char align[(sizeof(struct inpcb) + ALIGNM1) & ~ALIGNM1];
} inp_tp_u;
struct tcpcb tcb;
};
#undef ALIGNMENT
#undef ALIGNM1
static struct tcpcb dummy_tcb;
extern struct inpcbhead time_wait_slots[];
extern int cur_tw_slot;
extern u_long *delack_bitmask;
extern u_long route_generation;
int get_inpcb_str_size()
{
return sizeof(struct inpcb);
}
int get_tcp_str_size()
{
return sizeof(struct tcpcb);
}
int tcp_freeq(struct tcpcb *tp);
void
tcp_init()
{
int hashsize = TCBHASHSIZE;
vm_size_t str_size;
int i;
struct inpcbinfo *pcbinfo;
tcp_ccgen = 1;
tcp_cleartaocache();
tcp_delacktime = TCPTV_DELACK;
tcp_keepinit = TCPTV_KEEP_INIT;
tcp_keepidle = TCPTV_KEEP_IDLE;
tcp_keepintvl = TCPTV_KEEPINTVL;
tcp_maxpersistidle = TCPTV_KEEP_IDLE;
tcp_msl = TCPTV_MSL;
read_random(&tcp_now, sizeof(tcp_now));
tcp_now = tcp_now & 0x7fffffff;
LIST_INIT(&tcb);
tcbinfo.listhead = &tcb;
pcbinfo = &tcbinfo;
if (!powerof2(hashsize)) {
printf("WARNING: TCB hash size not a power of 2\n");
hashsize = 512;
}
tcp_tcbhashsize = hashsize;
tcbinfo.hashsize = hashsize;
tcbinfo.hashbase = hashinit(hashsize, M_PCB, &tcbinfo.hashmask);
tcbinfo.porthashbase = hashinit(hashsize, M_PCB,
&tcbinfo.porthashmask);
str_size = (vm_size_t) sizeof(struct inp_tp);
tcbinfo.ipi_zone = (void *) zinit(str_size, 120000*str_size, 8192, "tcpcb");
sack_hole_zone = zinit(str_size, 120000*str_size, 8192, "sack_hole zone");
tcp_reass_maxseg = nmbclusters / 16;
#if INET6
#define TCP_MINPROTOHDR (sizeof(struct ip6_hdr) + sizeof(struct tcphdr))
#else
#define TCP_MINPROTOHDR (sizeof(struct tcpiphdr))
#endif
if (max_protohdr < TCP_MINPROTOHDR)
max_protohdr = TCP_MINPROTOHDR;
if (max_linkhdr + TCP_MINPROTOHDR > MHLEN)
panic("tcp_init");
#undef TCP_MINPROTOHDR
dummy_tcb.t_state = TCP_NSTATES;
dummy_tcb.t_flags = 0;
tcbinfo.dummy_cb = (caddr_t) &dummy_tcb;
pcbinfo->mtx_grp_attr = lck_grp_attr_alloc_init();
lck_grp_attr_setdefault(pcbinfo->mtx_grp_attr);
pcbinfo->mtx_grp = lck_grp_alloc_init("tcppcb", pcbinfo->mtx_grp_attr);
pcbinfo->mtx_attr = lck_attr_alloc_init();
lck_attr_setdefault(pcbinfo->mtx_attr);
if ((pcbinfo->mtx = lck_rw_alloc_init(pcbinfo->mtx_grp, pcbinfo->mtx_attr)) == NULL) {
printf("tcp_init: mutex not alloced!\n");
return;
}
in_pcb_nat_init(&tcbinfo, AF_INET, IPPROTO_TCP, SOCK_STREAM);
delack_bitmask = _MALLOC((4 * hashsize)/32, M_PCB, M_WAITOK);
if (delack_bitmask == 0)
panic("Delack Memory");
for (i=0; i < (tcbinfo.hashsize / 32); i++)
delack_bitmask[i] = 0;
for (i=0; i < N_TIME_WAIT_SLOTS; i++) {
LIST_INIT(&time_wait_slots[i]);
}
}
void
tcp_fillheaders(tp, ip_ptr, tcp_ptr)
struct tcpcb *tp;
void *ip_ptr;
void *tcp_ptr;
{
struct inpcb *inp = tp->t_inpcb;
struct tcphdr *tcp_hdr = (struct tcphdr *)tcp_ptr;
#if INET6
if ((inp->inp_vflag & INP_IPV6) != 0) {
struct ip6_hdr *ip6;
ip6 = (struct ip6_hdr *)ip_ptr;
ip6->ip6_flow = (ip6->ip6_flow & ~IPV6_FLOWINFO_MASK) |
(inp->in6p_flowinfo & IPV6_FLOWINFO_MASK);
ip6->ip6_vfc = (ip6->ip6_vfc & ~IPV6_VERSION_MASK) |
(IPV6_VERSION & IPV6_VERSION_MASK);
ip6->ip6_nxt = IPPROTO_TCP;
ip6->ip6_plen = sizeof(struct tcphdr);
ip6->ip6_src = inp->in6p_laddr;
ip6->ip6_dst = inp->in6p_faddr;
tcp_hdr->th_sum = 0;
} else
#endif
{
struct ip *ip = (struct ip *) ip_ptr;
ip->ip_vhl = IP_VHL_BORING;
ip->ip_tos = 0;
ip->ip_len = 0;
ip->ip_id = 0;
ip->ip_off = 0;
ip->ip_ttl = 0;
ip->ip_sum = 0;
ip->ip_p = IPPROTO_TCP;
ip->ip_src = inp->inp_laddr;
ip->ip_dst = inp->inp_faddr;
tcp_hdr->th_sum = in_pseudo(ip->ip_src.s_addr, ip->ip_dst.s_addr,
htons(sizeof(struct tcphdr) + IPPROTO_TCP));
}
tcp_hdr->th_sport = inp->inp_lport;
tcp_hdr->th_dport = inp->inp_fport;
tcp_hdr->th_seq = 0;
tcp_hdr->th_ack = 0;
tcp_hdr->th_x2 = 0;
tcp_hdr->th_off = 5;
tcp_hdr->th_flags = 0;
tcp_hdr->th_win = 0;
tcp_hdr->th_urp = 0;
}
struct tcptemp *
tcp_maketemplate(tp)
struct tcpcb *tp;
{
struct mbuf *m;
struct tcptemp *n;
m = m_get(M_DONTWAIT, MT_HEADER);
if (m == NULL)
return (0);
m->m_len = sizeof(struct tcptemp);
n = mtod(m, struct tcptemp *);
tcp_fillheaders(tp, (void *)&n->tt_ipgen, (void *)&n->tt_t);
return (n);
}
void
tcp_respond(tp, ipgen, th, m, ack, seq, flags)
struct tcpcb *tp;
void *ipgen;
register struct tcphdr *th;
register struct mbuf *m;
tcp_seq ack, seq;
int flags;
{
register int tlen;
int win = 0;
struct route *ro = 0;
struct route sro;
struct ip *ip;
struct tcphdr *nth;
#if INET6
struct route_in6 *ro6 = 0;
struct route_in6 sro6;
struct ip6_hdr *ip6;
int isipv6;
#endif
int ipflags = 0;
#if INET6
isipv6 = IP_VHL_V(((struct ip *)ipgen)->ip_vhl) == 6;
ip6 = ipgen;
#endif
ip = ipgen;
if (tp) {
if (!(flags & TH_RST)) {
win = sbspace(&tp->t_inpcb->inp_socket->so_rcv);
if (win > (long)TCP_MAXWIN << tp->rcv_scale)
win = (long)TCP_MAXWIN << tp->rcv_scale;
}
#if INET6
if (isipv6)
ro6 = &tp->t_inpcb->in6p_route;
else
#endif
ro = &tp->t_inpcb->inp_route;
} else {
#if INET6
if (isipv6) {
ro6 = &sro6;
bzero(ro6, sizeof *ro6);
} else
#endif
{
ro = &sro;
bzero(ro, sizeof *ro);
}
}
if (m == 0) {
m = m_gethdr(M_DONTWAIT, MT_HEADER);
if (m == NULL)
return;
tlen = 0;
m->m_data += max_linkhdr;
#if INET6
if (isipv6) {
bcopy((caddr_t)ip6, mtod(m, caddr_t),
sizeof(struct ip6_hdr));
ip6 = mtod(m, struct ip6_hdr *);
nth = (struct tcphdr *)(ip6 + 1);
} else
#endif
{
bcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip));
ip = mtod(m, struct ip *);
nth = (struct tcphdr *)(ip + 1);
}
bcopy((caddr_t)th, (caddr_t)nth, sizeof(struct tcphdr));
flags = TH_ACK;
} else {
m_freem(m->m_next);
m->m_next = 0;
m->m_data = (caddr_t)ipgen;
tlen = 0;
#define xchg(a,b,type) { type t; t=a; a=b; b=t; }
#if INET6
if (isipv6) {
xchg(ip6->ip6_dst, ip6->ip6_src, struct in6_addr);
nth = (struct tcphdr *)(ip6 + 1);
} else
#endif
{
xchg(ip->ip_dst.s_addr, ip->ip_src.s_addr, n_long);
nth = (struct tcphdr *)(ip + 1);
}
if (th != nth) {
nth->th_sport = th->th_sport;
nth->th_dport = th->th_dport;
}
xchg(nth->th_dport, nth->th_sport, n_short);
#undef xchg
}
#if INET6
if (isipv6) {
ip6->ip6_plen = htons((u_short)(sizeof (struct tcphdr) +
tlen));
tlen += sizeof (struct ip6_hdr) + sizeof (struct tcphdr);
} else
#endif
{
tlen += sizeof (struct tcpiphdr);
ip->ip_len = tlen;
ip->ip_ttl = ip_defttl;
}
m->m_len = tlen;
m->m_pkthdr.len = tlen;
m->m_pkthdr.rcvif = 0;
nth->th_seq = htonl(seq);
nth->th_ack = htonl(ack);
nth->th_x2 = 0;
nth->th_off = sizeof (struct tcphdr) >> 2;
nth->th_flags = flags;
if (tp)
nth->th_win = htons((u_short) (win >> tp->rcv_scale));
else
nth->th_win = htons((u_short)win);
nth->th_urp = 0;
#if INET6
if (isipv6) {
nth->th_sum = 0;
nth->th_sum = in6_cksum(m, IPPROTO_TCP,
sizeof(struct ip6_hdr),
tlen - sizeof(struct ip6_hdr));
ip6->ip6_hlim = in6_selecthlim(tp ? tp->t_inpcb : NULL,
ro6 && ro6->ro_rt ?
ro6->ro_rt->rt_ifp :
NULL);
} else
#endif
{
nth->th_sum = in_pseudo(ip->ip_src.s_addr, ip->ip_dst.s_addr,
htons((u_short)(tlen - sizeof(struct ip) + ip->ip_p)));
m->m_pkthdr.csum_flags = CSUM_TCP;
m->m_pkthdr.csum_data = offsetof(struct tcphdr, th_sum);
}
#if TCPDEBUG
if (tp == NULL || (tp->t_inpcb->inp_socket->so_options & SO_DEBUG))
tcp_trace(TA_OUTPUT, 0, tp, mtod(m, void *), th, 0);
#endif
#if IPSEC
if (ipsec_bypass == 0 && ipsec_setsocket(m, tp ? tp->t_inpcb->inp_socket : NULL) != 0) {
m_freem(m);
return;
}
#endif
#if INET6
if (isipv6) {
(void)ip6_output(m, NULL, ro6, ipflags, NULL, NULL, 0);
if (ro6 == &sro6 && ro6->ro_rt) {
rtfree(ro6->ro_rt);
ro6->ro_rt = NULL;
}
} else
#endif
{
(void) ip_output_list(m, 0, NULL, ro, ipflags, NULL);
if (ro == &sro && ro->ro_rt) {
rtfree(ro->ro_rt);
ro->ro_rt = NULL;
}
}
}
struct tcpcb *
tcp_newtcpcb(inp)
struct inpcb *inp;
{
struct inp_tp *it;
register struct tcpcb *tp;
register struct socket *so = inp->inp_socket;
#if INET6
int isipv6 = (inp->inp_vflag & INP_IPV6) != 0;
#endif
if (so->cached_in_sock_layer == 0) {
it = (struct inp_tp *)inp;
tp = &it->tcb;
}
else
tp = (struct tcpcb *) inp->inp_saved_ppcb;
bzero((char *) tp, sizeof(struct tcpcb));
LIST_INIT(&tp->t_segq);
tp->t_maxseg = tp->t_maxopd =
#if INET6
isipv6 ? tcp_v6mssdflt :
#endif
tcp_mssdflt;
if (tcp_do_rfc1323)
tp->t_flags = (TF_REQ_SCALE|TF_REQ_TSTMP);
tp->sack_enable = tcp_do_sack;
TAILQ_INIT(&tp->snd_holes);
tp->t_inpcb = inp;
tp->t_srtt = TCPTV_SRTTBASE;
tp->t_rttvar = ((TCPTV_RTOBASE - TCPTV_SRTTBASE) << TCP_RTTVAR_SHIFT) / 4;
tp->t_rttmin = TCPTV_MIN;
tp->t_rxtcur = TCPTV_RTOBASE;
tp->snd_cwnd = TCP_MAXWIN << TCP_MAX_WINSHIFT;
tp->snd_ssthresh = TCP_MAXWIN << TCP_MAX_WINSHIFT;
tp->t_rcvtime = 0;
inp->inp_ip_ttl = ip_defttl;
inp->inp_ppcb = (caddr_t)tp;
return (tp);
}
struct tcpcb *
tcp_drop(tp, errno)
register struct tcpcb *tp;
int errno;
{
struct socket *so = tp->t_inpcb->inp_socket;
if (TCPS_HAVERCVDSYN(tp->t_state)) {
tp->t_state = TCPS_CLOSED;
(void) tcp_output(tp);
tcpstat.tcps_drops++;
} else
tcpstat.tcps_conndrops++;
if (errno == ETIMEDOUT && tp->t_softerror)
errno = tp->t_softerror;
so->so_error = errno;
return (tcp_close(tp));
}
struct tcpcb *
tcp_close(tp)
register struct tcpcb *tp;
{
struct inpcb *inp = tp->t_inpcb;
struct socket *so = inp->inp_socket;
#if INET6
int isipv6 = (inp->inp_vflag & INP_IPV6) != 0;
#endif
register struct rtentry *rt;
int dosavessthresh;
if ( inp->inp_ppcb == NULL)
return NULL;
{
int i;
for (i = 0; i < TCPT_NTIMERS; i++) {
tp->t_timer[i] = 0;
}
}
KERNEL_DEBUG(DBG_FNC_TCP_CLOSE | DBG_FUNC_START, tp,0,0,0,0);
switch (tp->t_state)
{
case TCPS_ESTABLISHED:
case TCPS_FIN_WAIT_1:
case TCPS_CLOSING:
case TCPS_CLOSE_WAIT:
case TCPS_LAST_ACK:
break;
}
if (tp->t_rttupdated >= 16) {
register u_long i = 0;
#if INET6
if (isipv6) {
struct sockaddr_in6 *sin6;
if ((rt = inp->in6p_route.ro_rt) == NULL)
goto no_valid_rt;
sin6 = (struct sockaddr_in6 *)rt_key(rt);
if (IN6_IS_ADDR_UNSPECIFIED(&sin6->sin6_addr))
goto no_valid_rt;
}
else
#endif
rt = inp->inp_route.ro_rt;
if (rt == NULL ||
((struct sockaddr_in *)rt_key(rt))->sin_addr.s_addr
== INADDR_ANY || rt->generation_id != route_generation) {
if (tp->t_state >= TCPS_CLOSE_WAIT)
tp->t_state = TCPS_CLOSING;
goto no_valid_rt;
}
if ((rt->rt_rmx.rmx_locks & RTV_RTT) == 0) {
i = tp->t_srtt *
(RTM_RTTUNIT / (PR_SLOWHZ * TCP_RTT_SCALE));
if (rt->rt_rmx.rmx_rtt && i)
rt->rt_rmx.rmx_rtt =
(rt->rt_rmx.rmx_rtt + i) / 2;
else
rt->rt_rmx.rmx_rtt = i;
tcpstat.tcps_cachedrtt++;
}
if ((rt->rt_rmx.rmx_locks & RTV_RTTVAR) == 0) {
i = tp->t_rttvar *
(RTM_RTTUNIT / (PR_SLOWHZ * TCP_RTTVAR_SCALE));
if (rt->rt_rmx.rmx_rttvar && i)
rt->rt_rmx.rmx_rttvar =
(rt->rt_rmx.rmx_rttvar + i) / 2;
else
rt->rt_rmx.rmx_rttvar = i;
tcpstat.tcps_cachedrttvar++;
}
i = tp->snd_ssthresh;
if (rt->rt_rmx.rmx_sendpipe != 0)
dosavessthresh = (i < rt->rt_rmx.rmx_sendpipe / 2);
else
dosavessthresh = (i < so->so_snd.sb_hiwat / 2);
if (((rt->rt_rmx.rmx_locks & RTV_SSTHRESH) == 0 &&
i != 0 && rt->rt_rmx.rmx_ssthresh != 0)
|| dosavessthresh) {
i = (i + tp->t_maxseg / 2) / tp->t_maxseg;
if (i < 2)
i = 2;
i *= (u_long)(tp->t_maxseg +
#if INET6
(isipv6 ? sizeof (struct ip6_hdr) +
sizeof (struct tcphdr) :
#endif
sizeof (struct tcpiphdr)
#if INET6
)
#endif
);
if (rt->rt_rmx.rmx_ssthresh)
rt->rt_rmx.rmx_ssthresh =
(rt->rt_rmx.rmx_ssthresh + i) / 2;
else
rt->rt_rmx.rmx_ssthresh = i;
tcpstat.tcps_cachedssthresh++;
}
}
rt = inp->inp_route.ro_rt;
if (rt) {
if ((tp->t_flags & TF_LQ_OVERFLOW) && tcp_lq_overflow &&
((rt->rt_rmx.rmx_locks & RTV_RTT) == 0)){
if (rt->rt_rmx.rmx_rtt == 0)
rt->rt_flags |= RTF_DELCLONE;
}
}
no_valid_rt:
(void) tcp_freeq(tp);
tcp_free_sackholes(tp);
#ifdef __APPLE__
if (so->cached_in_sock_layer)
inp->inp_saved_ppcb = (caddr_t) tp;
#endif
soisdisconnected(so);
#if INET6
if (INP_CHECK_SOCKAF(so, AF_INET6))
in6_pcbdetach(inp);
else
#endif
in_pcbdetach(inp);
tcpstat.tcps_closed++;
KERNEL_DEBUG(DBG_FNC_TCP_CLOSE | DBG_FUNC_END, tcpstat.tcps_closed,0,0,0,0);
return ((struct tcpcb *)0);
}
int
tcp_freeq(tp)
struct tcpcb *tp;
{
register struct tseg_qent *q;
int rv = 0;
while((q = LIST_FIRST(&tp->t_segq)) != NULL) {
LIST_REMOVE(q, tqe_q);
m_freem(q->tqe_m);
FREE(q, M_TSEGQ);
tcp_reass_qsize--;
rv = 1;
}
return (rv);
}
void
tcp_drain()
{
if (do_tcpdrain)
{
struct inpcb *inpb;
struct tcpcb *tcpb;
struct tseg_qent *te;
lck_rw_lock_exclusive(tcbinfo.mtx);
for (inpb = LIST_FIRST(tcbinfo.listhead); inpb;
inpb = LIST_NEXT(inpb, inp_list)) {
if ((tcpb = intotcpcb(inpb))) {
while ((te = LIST_FIRST(&tcpb->t_segq))
!= NULL) {
LIST_REMOVE(te, tqe_q);
m_freem(te->tqe_m);
FREE(te, M_TSEGQ);
tcp_reass_qsize--;
}
}
}
lck_rw_done(tcbinfo.mtx);
}
}
static void
tcp_notify(inp, error)
struct inpcb *inp;
int error;
{
struct tcpcb *tp;
if (inp == NULL || (inp->inp_state == INPCB_STATE_DEAD))
return;
tp = (struct tcpcb *)inp->inp_ppcb;
if (tp->t_state == TCPS_ESTABLISHED &&
(error == EHOSTUNREACH || error == ENETUNREACH ||
error == EHOSTDOWN)) {
return;
} else if (tp->t_state < TCPS_ESTABLISHED && tp->t_rxtshift > 3 &&
tp->t_softerror)
tcp_drop(tp, error);
else
tp->t_softerror = error;
#if 0
wakeup((caddr_t) &so->so_timeo);
sorwakeup(so);
sowwakeup(so);
#endif
}
static int
tcp_pcblist SYSCTL_HANDLER_ARGS
{
int error, i, n;
struct inpcb *inp, **inp_list;
inp_gen_t gencnt;
struct xinpgen xig;
lck_rw_lock_shared(tcbinfo.mtx);
if (req->oldptr == USER_ADDR_NULL) {
n = tcbinfo.ipi_count;
req->oldidx = 2 * (sizeof xig)
+ (n + n/8) * sizeof(struct xtcpcb);
lck_rw_done(tcbinfo.mtx);
return 0;
}
if (req->newptr != USER_ADDR_NULL) {
lck_rw_done(tcbinfo.mtx);
return EPERM;
}
gencnt = tcbinfo.ipi_gencnt;
n = tcbinfo.ipi_count;
bzero(&xig, sizeof(xig));
xig.xig_len = sizeof xig;
xig.xig_count = n;
xig.xig_gen = gencnt;
xig.xig_sogen = so_gencnt;
error = SYSCTL_OUT(req, &xig, sizeof xig);
if (error) {
lck_rw_done(tcbinfo.mtx);
return error;
}
if (n == 0) {
lck_rw_done(tcbinfo.mtx);
return 0;
}
inp_list = _MALLOC(n * sizeof *inp_list, M_TEMP, M_WAITOK);
if (inp_list == 0) {
lck_rw_done(tcbinfo.mtx);
return ENOMEM;
}
for (inp = LIST_FIRST(tcbinfo.listhead), i = 0; inp && i < n;
inp = LIST_NEXT(inp, inp_list)) {
#ifdef __APPLE__
if (inp->inp_gencnt <= gencnt && inp->inp_state != INPCB_STATE_DEAD)
#else
if (inp->inp_gencnt <= gencnt && !prison_xinpcb(req->p, inp))
#endif
inp_list[i++] = inp;
}
n = i;
error = 0;
for (i = 0; i < n; i++) {
inp = inp_list[i];
if (inp->inp_gencnt <= gencnt && inp->inp_state != INPCB_STATE_DEAD) {
struct xtcpcb xt;
caddr_t inp_ppcb;
bzero(&xt, sizeof(xt));
xt.xt_len = sizeof xt;
inpcb_to_compat(inp, &xt.xt_inp);
inp_ppcb = inp->inp_ppcb;
if (inp_ppcb != NULL) {
bcopy(inp_ppcb, &xt.xt_tp, sizeof xt.xt_tp);
}
else
bzero((char *) &xt.xt_tp, sizeof xt.xt_tp);
if (inp->inp_socket)
sotoxsocket(inp->inp_socket, &xt.xt_socket);
error = SYSCTL_OUT(req, &xt, sizeof xt);
}
}
if (!error) {
bzero(&xig, sizeof(xig));
xig.xig_len = sizeof xig;
xig.xig_gen = tcbinfo.ipi_gencnt;
xig.xig_sogen = so_gencnt;
xig.xig_count = tcbinfo.ipi_count;
error = SYSCTL_OUT(req, &xig, sizeof xig);
}
FREE(inp_list, M_TEMP);
lck_rw_done(tcbinfo.mtx);
return error;
}
SYSCTL_PROC(_net_inet_tcp, TCPCTL_PCBLIST, pcblist, CTLFLAG_RD, 0, 0,
tcp_pcblist, "S,xtcpcb", "List of active TCP connections");
#ifndef __APPLE__
static int
tcp_getcred(SYSCTL_HANDLER_ARGS)
{
struct sockaddr_in addrs[2];
struct inpcb *inp;
int error, s;
error = suser(req->p);
if (error)
return (error);
error = SYSCTL_IN(req, addrs, sizeof(addrs));
if (error)
return (error);
s = splnet();
inp = in_pcblookup_hash(&tcbinfo, addrs[1].sin_addr, addrs[1].sin_port,
addrs[0].sin_addr, addrs[0].sin_port, 0, NULL);
if (inp == NULL || inp->inp_socket == NULL) {
error = ENOENT;
goto out;
}
error = SYSCTL_OUT(req, inp->inp_socket->so_cred, sizeof(*(kauth_cred_t)0);
out:
splx(s);
return (error);
}
SYSCTL_PROC(_net_inet_tcp, OID_AUTO, getcred, CTLTYPE_OPAQUE|CTLFLAG_RW,
0, 0, tcp_getcred, "S,ucred", "Get the ucred of a TCP connection");
#if INET6
static int
tcp6_getcred(SYSCTL_HANDLER_ARGS)
{
struct sockaddr_in6 addrs[2];
struct inpcb *inp;
int error, s, mapped = 0;
error = suser(req->p);
if (error)
return (error);
error = SYSCTL_IN(req, addrs, sizeof(addrs));
if (error)
return (error);
if (IN6_IS_ADDR_V4MAPPED(&addrs[0].sin6_addr)) {
if (IN6_IS_ADDR_V4MAPPED(&addrs[1].sin6_addr))
mapped = 1;
else
return (EINVAL);
}
s = splnet();
if (mapped == 1)
inp = in_pcblookup_hash(&tcbinfo,
*(struct in_addr *)&addrs[1].sin6_addr.s6_addr[12],
addrs[1].sin6_port,
*(struct in_addr *)&addrs[0].sin6_addr.s6_addr[12],
addrs[0].sin6_port,
0, NULL);
else
inp = in6_pcblookup_hash(&tcbinfo, &addrs[1].sin6_addr,
addrs[1].sin6_port,
&addrs[0].sin6_addr, addrs[0].sin6_port,
0, NULL);
if (inp == NULL || inp->inp_socket == NULL) {
error = ENOENT;
goto out;
}
error = SYSCTL_OUT(req, inp->inp_socket->so_cred,
sizeof(*(kauth_cred_t)0);
out:
splx(s);
return (error);
}
SYSCTL_PROC(_net_inet6_tcp6, OID_AUTO, getcred, CTLTYPE_OPAQUE|CTLFLAG_RW,
0, 0,
tcp6_getcred, "S,ucred", "Get the ucred of a TCP6 connection");
#endif
#endif
void
tcp_ctlinput(cmd, sa, vip)
int cmd;
struct sockaddr *sa;
void *vip;
{
struct ip *ip = vip;
struct tcphdr *th;
struct in_addr faddr;
struct inpcb *inp;
struct tcpcb *tp;
void (*notify)(struct inpcb *, int) = tcp_notify;
tcp_seq icmp_seq;
faddr = ((struct sockaddr_in *)sa)->sin_addr;
if (sa->sa_family != AF_INET || faddr.s_addr == INADDR_ANY)
return;
if (cmd == PRC_QUENCH)
notify = tcp_quench;
else if (icmp_may_rst && (cmd == PRC_UNREACH_ADMIN_PROHIB ||
cmd == PRC_UNREACH_PORT) && ip)
notify = tcp_drop_syn_sent;
else if (cmd == PRC_MSGSIZE)
notify = tcp_mtudisc;
else if (PRC_IS_REDIRECT(cmd)) {
ip = 0;
notify = in_rtchange;
} else if (cmd == PRC_HOSTDEAD)
ip = 0;
else if ((unsigned)cmd > PRC_NCMDS || inetctlerrmap[cmd] == 0)
return;
if (ip) {
th = (struct tcphdr *)((caddr_t)ip
+ (IP_VHL_HL(ip->ip_vhl) << 2));
inp = in_pcblookup_hash(&tcbinfo, faddr, th->th_dport,
ip->ip_src, th->th_sport, 0, NULL);
if (inp != NULL && inp->inp_socket != NULL) {
tcp_lock(inp->inp_socket, 1, 0);
if (in_pcb_checkstate(inp, WNT_RELEASE, 1) == WNT_STOPUSING) {
tcp_unlock(inp->inp_socket, 1, 0);
return;
}
icmp_seq = htonl(th->th_seq);
tp = intotcpcb(inp);
if (SEQ_GEQ(icmp_seq, tp->snd_una) &&
SEQ_LT(icmp_seq, tp->snd_max))
(*notify)(inp, inetctlerrmap[cmd]);
tcp_unlock(inp->inp_socket, 1, 0);
}
} else
in_pcbnotifyall(&tcbinfo, faddr, inetctlerrmap[cmd], notify);
}
#if INET6
void
tcp6_ctlinput(cmd, sa, d)
int cmd;
struct sockaddr *sa;
void *d;
{
struct tcphdr th;
void (*notify)(struct inpcb *, int) = tcp_notify;
struct ip6_hdr *ip6;
struct mbuf *m;
struct ip6ctlparam *ip6cp = NULL;
const struct sockaddr_in6 *sa6_src = NULL;
int off;
struct tcp_portonly {
u_int16_t th_sport;
u_int16_t th_dport;
} *thp;
if (sa->sa_family != AF_INET6 ||
sa->sa_len != sizeof(struct sockaddr_in6))
return;
if (cmd == PRC_QUENCH)
notify = tcp_quench;
else if (cmd == PRC_MSGSIZE)
notify = tcp_mtudisc;
else if (!PRC_IS_REDIRECT(cmd) &&
((unsigned)cmd > PRC_NCMDS || inet6ctlerrmap[cmd] == 0))
return;
if (d != NULL) {
ip6cp = (struct ip6ctlparam *)d;
m = ip6cp->ip6c_m;
ip6 = ip6cp->ip6c_ip6;
off = ip6cp->ip6c_off;
sa6_src = ip6cp->ip6c_src;
} else {
m = NULL;
ip6 = NULL;
off = 0;
sa6_src = &sa6_any;
}
if (ip6) {
if (m->m_pkthdr.len < off + sizeof(*thp))
return;
bzero(&th, sizeof(th));
m_copydata(m, off, sizeof(*thp), (caddr_t)&th);
in6_pcbnotify(&tcbinfo, sa, th.th_dport,
(struct sockaddr *)ip6cp->ip6c_src,
th.th_sport, cmd, notify);
} else
in6_pcbnotify(&tcbinfo, sa, 0, (struct sockaddr *)sa6_src,
0, cmd, notify);
}
#endif
#define ISN_BYTES_PER_SECOND 1048576
u_char isn_secret[32];
int isn_last_reseed;
MD5_CTX isn_ctx;
tcp_seq
tcp_new_isn(tp)
struct tcpcb *tp;
{
u_int32_t md5_buffer[4];
tcp_seq new_isn;
struct timeval timenow;
if (((tp->t_state == TCPS_LISTEN) || (tp->t_state == TCPS_TIME_WAIT))
&& tcp_strict_rfc1948 == 0)
#ifdef __APPLE__
return random();
#else
return arc4random();
#endif
getmicrotime(&timenow);
if ((isn_last_reseed == 0) ||
((tcp_strict_rfc1948 == 0) && (tcp_isn_reseed_interval > 0) &&
(((u_int)isn_last_reseed + (u_int)tcp_isn_reseed_interval*hz)
< (u_int)timenow.tv_sec))) {
#ifdef __APPLE__
read_random(&isn_secret, sizeof(isn_secret));
#else
read_random_unlimited(&isn_secret, sizeof(isn_secret));
#endif
isn_last_reseed = timenow.tv_sec;
}
MD5Init(&isn_ctx);
MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->inp_fport, sizeof(u_short));
MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->inp_lport, sizeof(u_short));
#if INET6
if ((tp->t_inpcb->inp_vflag & INP_IPV6) != 0) {
MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->in6p_faddr,
sizeof(struct in6_addr));
MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->in6p_laddr,
sizeof(struct in6_addr));
} else
#endif
{
MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->inp_faddr,
sizeof(struct in_addr));
MD5Update(&isn_ctx, (u_char *) &tp->t_inpcb->inp_laddr,
sizeof(struct in_addr));
}
MD5Update(&isn_ctx, (u_char *) &isn_secret, sizeof(isn_secret));
MD5Final((u_char *) &md5_buffer, &isn_ctx);
new_isn = (tcp_seq) md5_buffer[0];
new_isn += timenow.tv_sec * (ISN_BYTES_PER_SECOND / hz);
return new_isn;
}
void
tcp_quench(
struct inpcb *inp,
__unused int errno
)
{
struct tcpcb *tp = intotcpcb(inp);
if (tp)
tp->snd_cwnd = tp->t_maxseg;
}
void
tcp_drop_syn_sent(inp, errno)
struct inpcb *inp;
int errno;
{
struct tcpcb *tp = intotcpcb(inp);
if (tp && tp->t_state == TCPS_SYN_SENT)
tcp_drop(tp, errno);
}
void
tcp_mtudisc(
struct inpcb *inp,
__unused int errno
)
{
struct tcpcb *tp = intotcpcb(inp);
struct rtentry *rt;
struct rmxp_tao *taop;
struct socket *so = inp->inp_socket;
int offered;
int mss;
#if INET6
int isipv6 = (tp->t_inpcb->inp_vflag & INP_IPV6) != 0;
#endif
if (tp) {
#if INET6
if (isipv6)
rt = tcp_rtlookup6(inp);
else
#endif
rt = tcp_rtlookup(inp);
if (!rt || !rt->rt_rmx.rmx_mtu) {
tp->t_maxopd = tp->t_maxseg =
#if INET6
isipv6 ? tcp_v6mssdflt :
#endif
tcp_mssdflt;
return;
}
taop = rmx_taop(rt->rt_rmx);
offered = taop->tao_mssopt;
mss = rt->rt_rmx.rmx_mtu -
#if INET6
(isipv6 ?
sizeof(struct ip6_hdr) + sizeof(struct tcphdr) :
#endif
sizeof(struct tcpiphdr)
#if INET6
)
#endif
;
if (offered)
mss = min(mss, offered);
if (tp->t_maxopd <= mss)
return;
tp->t_maxopd = mss;
if ((tp->t_flags & (TF_REQ_TSTMP|TF_NOOPT)) == TF_REQ_TSTMP &&
(tp->t_flags & TF_RCVD_TSTMP) == TF_RCVD_TSTMP)
mss -= TCPOLEN_TSTAMP_APPA;
if (so->so_snd.sb_hiwat < mss)
mss = so->so_snd.sb_hiwat;
tp->t_maxseg = mss;
tcpstat.tcps_mturesent++;
tp->t_rtttime = 0;
tp->snd_nxt = tp->snd_una;
tcp_output(tp);
}
}
struct rtentry *
tcp_rtlookup(inp)
struct inpcb *inp;
{
struct route *ro;
struct rtentry *rt;
ro = &inp->inp_route;
if (ro == NULL)
return (NULL);
rt = ro->ro_rt;
if (rt == NULL || !(rt->rt_flags & RTF_UP) || rt->generation_id != route_generation) {
if (inp->inp_faddr.s_addr != INADDR_ANY) {
ro->ro_dst.sa_family = AF_INET;
ro->ro_dst.sa_len = sizeof(struct sockaddr_in);
((struct sockaddr_in *) &ro->ro_dst)->sin_addr =
inp->inp_faddr;
rtalloc(ro);
rt = ro->ro_rt;
}
}
return rt;
}
#if INET6
struct rtentry *
tcp_rtlookup6(inp)
struct inpcb *inp;
{
struct route_in6 *ro6;
struct rtentry *rt;
ro6 = &inp->in6p_route;
rt = ro6->ro_rt;
if (rt == NULL || !(rt->rt_flags & RTF_UP)) {
if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) {
struct sockaddr_in6 *dst6;
dst6 = (struct sockaddr_in6 *)&ro6->ro_dst;
dst6->sin6_family = AF_INET6;
dst6->sin6_len = sizeof(*dst6);
dst6->sin6_addr = inp->in6p_faddr;
rtalloc((struct route *)ro6);
rt = ro6->ro_rt;
}
}
return rt;
}
#endif
#if IPSEC
size_t
ipsec_hdrsiz_tcp(tp)
struct tcpcb *tp;
{
struct inpcb *inp;
struct mbuf *m;
size_t hdrsiz;
struct ip *ip;
#if INET6
struct ip6_hdr *ip6 = NULL;
#endif
struct tcphdr *th;
if ((tp == NULL) || ((inp = tp->t_inpcb) == NULL))
return 0;
MGETHDR(m, M_DONTWAIT, MT_DATA);
if (!m)
return 0;
lck_mtx_lock(sadb_mutex);
#if INET6
if ((inp->inp_vflag & INP_IPV6) != 0) {
ip6 = mtod(m, struct ip6_hdr *);
th = (struct tcphdr *)(ip6 + 1);
m->m_pkthdr.len = m->m_len =
sizeof(struct ip6_hdr) + sizeof(struct tcphdr);
tcp_fillheaders(tp, ip6, th);
hdrsiz = ipsec6_hdrsiz(m, IPSEC_DIR_OUTBOUND, inp);
} else
#endif
{
ip = mtod(m, struct ip *);
th = (struct tcphdr *)(ip + 1);
m->m_pkthdr.len = m->m_len = sizeof(struct tcpiphdr);
tcp_fillheaders(tp, ip, th);
hdrsiz = ipsec4_hdrsiz(m, IPSEC_DIR_OUTBOUND, inp);
}
lck_mtx_unlock(sadb_mutex);
m_free(m);
return hdrsiz;
}
#endif
struct rmxp_tao *
tcp_gettaocache(inp)
struct inpcb *inp;
{
struct rtentry *rt;
#if INET6
if ((inp->inp_vflag & INP_IPV6) != 0)
rt = tcp_rtlookup6(inp);
else
#endif
rt = tcp_rtlookup(inp);
if (rt == NULL ||
(rt->rt_flags & (RTF_UP|RTF_HOST)) != (RTF_UP|RTF_HOST))
return NULL;
return rmx_taop(rt->rt_rmx);
}
static void
tcp_cleartaocache()
{
}
int
tcp_lock(so, refcount, lr)
struct socket *so;
int refcount;
int lr;
{
int lr_saved;
#ifdef __ppc__
if (lr == 0) {
__asm__ volatile("mflr %0" : "=r" (lr_saved));
}
else lr_saved = lr;
#endif
if (so->so_pcb) {
lck_mtx_lock(((struct inpcb *)so->so_pcb)->inpcb_mtx);
}
else {
panic("tcp_lock: so=%x NO PCB! lr=%x\n", so, lr_saved);
lck_mtx_lock(so->so_proto->pr_domain->dom_mtx);
}
if (so->so_usecount < 0)
panic("tcp_lock: so=%x so_pcb=%x lr=%x ref=%x\n",
so, so->so_pcb, lr_saved, so->so_usecount);
if (refcount)
so->so_usecount++;
so->reserved3 = (void *)lr_saved;
return (0);
}
int
tcp_unlock(so, refcount, lr)
struct socket *so;
int refcount;
int lr;
{
int lr_saved;
#ifdef __ppc__
if (lr == 0) {
__asm__ volatile("mflr %0" : "=r" (lr_saved));
}
else lr_saved = lr;
#endif
#ifdef MORE_TCPLOCK_DEBUG
printf("tcp_unlock: so=%x sopcb=%x lock=%x ref=%x lr=%x\n",
so, so->so_pcb, ((struct inpcb *)so->so_pcb)->inpcb_mtx, so->so_usecount, lr_saved);
#endif
if (refcount)
so->so_usecount--;
if (so->so_usecount < 0)
panic("tcp_unlock: so=%x usecount=%x\n", so, so->so_usecount);
if (so->so_pcb == NULL) {
panic("tcp_unlock: so=%x NO PCB usecount=%x lr=%x\n", so, so->so_usecount, lr_saved);
lck_mtx_unlock(so->so_proto->pr_domain->dom_mtx);
}
else {
lck_mtx_assert(((struct inpcb *)so->so_pcb)->inpcb_mtx, LCK_MTX_ASSERT_OWNED);
lck_mtx_unlock(((struct inpcb *)so->so_pcb)->inpcb_mtx);
}
so->reserved4 = (void *)lr_saved;
return (0);
}
lck_mtx_t *
tcp_getlock(so, locktype)
struct socket *so;
int locktype;
{
struct inpcb *inp = sotoinpcb(so);
if (so->so_pcb) {
if (so->so_usecount < 0)
panic("tcp_getlock: so=%x usecount=%x\n", so, so->so_usecount);
return(inp->inpcb_mtx);
}
else {
panic("tcp_getlock: so=%x NULL so_pcb\n", so);
return (so->so_proto->pr_domain->dom_mtx);
}
}