#include <sys/param.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/sysctl.h>
#include <sys/malloc.h>
#include <sys/mbuf.h>
#include <sys/proc.h>
#include <sys/protosw.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/syslog.h>
#include <sys/mcache.h>
#include <kern/cpu_number.h>
#include <machine/endian.h>
#include <net/if.h>
#include <net/if_types.h>
#include <net/route.h>
#include <net/ntstat.h>
#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/ip.h>
#include <netinet/ip_icmp.h>
#include <netinet/in_var.h>
#include <netinet/icmp_var.h>
#include <netinet/in_pcb.h>
#include <netinet/ip_var.h>
#include <mach/sdt.h>
#if INET6
#include <netinet/ip6.h>
#include <netinet/icmp6.h>
#include <netinet6/nd6.h>
#include <netinet6/ip6_var.h>
#include <netinet6/in6_pcb.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>
#include <netinet/tcp_cc.h>
#include <kern/zalloc.h>
#if INET6
#include <netinet6/tcp6_var.h>
#endif
#include <netinet/tcpip.h>
#if TCPDEBUG
#include <netinet/tcp_debug.h>
u_char tcp_saveipgen[40];
struct tcphdr tcp_savetcp;
#endif
#if IPSEC
#include <netinet6/ipsec.h>
#if INET6
#include <netinet6/ipsec6.h>
#endif
#include <netkey/key.h>
#endif
#if CONFIG_MACF_NET || CONFIG_MACF_SOCKET
#include <security/mac_framework.h>
#endif
#include <sys/kdebug.h>
#include <netinet/lro_ext.h>
#define DBG_LAYER_BEG NETDBG_CODE(DBG_NETTCP, 0)
#define DBG_LAYER_END NETDBG_CODE(DBG_NETTCP, 2)
#define DBG_FNC_TCP_INPUT NETDBG_CODE(DBG_NETTCP, (3 << 8))
#define DBG_FNC_TCP_NEWCONN NETDBG_CODE(DBG_NETTCP, (7 << 8))
static int tcprexmtthresh = 2;
tcp_cc tcp_ccgen;
#if IPSEC
extern int ipsec_bypass;
#endif
extern int32_t total_sbmb_cnt;
struct tcpstat tcpstat;
static int log_in_vain = 0;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, log_in_vain, CTLFLAG_RW | CTLFLAG_LOCKED,
&log_in_vain, 0, "Log all incoming TCP connections");
static int blackhole = 0;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, blackhole, CTLFLAG_RW | CTLFLAG_LOCKED,
&blackhole, 0, "Do not send RST when dropping refused connections");
int tcp_delack_enabled = 3;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, delayed_ack, CTLFLAG_RW | CTLFLAG_LOCKED,
&tcp_delack_enabled, 0,
"Delay ACK to try and piggyback it onto a data packet");
int tcp_lq_overflow = 1;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, tcp_lq_overflow, CTLFLAG_RW | CTLFLAG_LOCKED,
&tcp_lq_overflow, 0,
"Listen Queue Overflow");
int tcp_recv_bg = 0;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, recvbg, CTLFLAG_RW | CTLFLAG_LOCKED,
&tcp_recv_bg, 0,
"Receive background");
#if TCP_DROP_SYNFIN
static int drop_synfin = 1;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, drop_synfin, CTLFLAG_RW | CTLFLAG_LOCKED,
&drop_synfin, 0, "Drop TCP packets with SYN+FIN set");
#endif
SYSCTL_NODE(_net_inet_tcp, OID_AUTO, reass, CTLFLAG_RW|CTLFLAG_LOCKED, 0,
"TCP Segment Reassembly Queue");
__private_extern__ int tcp_reass_maxseg = 0;
SYSCTL_INT(_net_inet_tcp_reass, OID_AUTO, maxsegments, CTLFLAG_RW | CTLFLAG_LOCKED,
&tcp_reass_maxseg, 0,
"Global maximum number of TCP Segments in Reassembly Queue");
__private_extern__ int tcp_reass_qsize = 0;
SYSCTL_INT(_net_inet_tcp_reass, OID_AUTO, cursegments, CTLFLAG_RD | CTLFLAG_LOCKED,
&tcp_reass_qsize, 0,
"Global number of TCP Segments currently in Reassembly Queue");
static int tcp_reass_overflows = 0;
SYSCTL_INT(_net_inet_tcp_reass, OID_AUTO, overflows, CTLFLAG_RD | CTLFLAG_LOCKED,
&tcp_reass_overflows, 0,
"Global number of TCP Segment Reassembly Queue Overflows");
__private_extern__ int slowlink_wsize = 8192;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, slowlink_wsize, CTLFLAG_RW | CTLFLAG_LOCKED,
&slowlink_wsize, 0, "Maximum advertised window size for slowlink");
int maxseg_unacked = 8;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, maxseg_unacked, CTLFLAG_RW | CTLFLAG_LOCKED,
&maxseg_unacked, 0, "Maximum number of outstanding segments left unacked");
int tcp_do_rfc3465 = 1;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, rfc3465, CTLFLAG_RW | CTLFLAG_LOCKED,
&tcp_do_rfc3465, 0, "");
int tcp_do_rfc3465_lim2 = 1;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, rfc3465_lim2, CTLFLAG_RW | CTLFLAG_LOCKED,
&tcp_do_rfc3465_lim2, 0, "Appropriate bytes counting w/ L=2*SMSS");
int rtt_samples_per_slot = 20;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, rtt_samples_per_slot, CTLFLAG_RW | CTLFLAG_LOCKED,
&rtt_samples_per_slot, 0, "Number of RTT samples stored for rtt history");
int tcp_allowed_iaj = ALLOWED_IAJ;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, recv_allowed_iaj, CTLFLAG_RW | CTLFLAG_LOCKED,
&tcp_allowed_iaj, 0, "Allowed inter-packet arrival jiter");
int tcp_acc_iaj_high_thresh = ACC_IAJ_HIGH_THRESH;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, acc_iaj_high_thresh, CTLFLAG_RW | CTLFLAG_LOCKED,
&tcp_acc_iaj_high_thresh, 0, "Used in calculating maximum accumulated IAJ");
u_int32_t tcp_do_autorcvbuf = 1;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, doautorcvbuf, CTLFLAG_RW | CTLFLAG_LOCKED,
&tcp_do_autorcvbuf, 0, "Enable automatic socket buffer tuning");
u_int32_t tcp_autorcvbuf_inc_shift = 3;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, autorcvbufincshift, CTLFLAG_RW | CTLFLAG_LOCKED,
&tcp_autorcvbuf_inc_shift, 0, "Shift for increment in receive socket buffer size");
u_int32_t tcp_autorcvbuf_max = 512 * 1024;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, autorcvbufmax, CTLFLAG_RW | CTLFLAG_LOCKED,
&tcp_autorcvbuf_max, 0, "Maximum receive socket buffer size");
int sw_lro = 1;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, lro, CTLFLAG_RW | CTLFLAG_LOCKED,
&sw_lro, 0, "Used to coalesce TCP packets");
int lrodebug = 0;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, lrodbg, CTLFLAG_RW | CTLFLAG_LOCKED,
&lrodebug, 0, "Used to debug SW LRO");
int lro_start = 3;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, lro_startcnt, CTLFLAG_RW | CTLFLAG_LOCKED,
&lro_start, 0, "Segments for starting LRO computed as power of 2");
extern int tcp_do_autosendbuf;
#if CONFIG_IFEF_NOWINDOWSCALE
int tcp_obey_ifef_nowindowscale = 0;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, obey_ifef_nowindowscale, CTLFLAG_RW | CTLFLAG_LOCKED,
&tcp_obey_ifef_nowindowscale, 0, "");
#endif
uint32_t tcp_rbuf_hiwat_shift = 1;
uint32_t tcp_rbuf_win_shift = 2;
extern int tcp_TCPTV_MIN;
extern int tcp_acc_iaj_high;
extern int tcp_acc_iaj_react_limit;
extern struct zone *tcp_reass_zone;
u_int32_t tcp_now;
struct timeval tcp_uptime;
lck_spin_t *tcp_uptime_lock;
struct inpcbhead tcb;
#define tcb6 tcb
struct inpcbinfo tcbinfo;
static void tcp_dooptions(struct tcpcb *, u_char *, int, struct tcphdr *,
struct tcpopt *, unsigned int);
static void tcp_pulloutofband(struct socket *,
struct tcphdr *, struct mbuf *, int);
static int tcp_reass(struct tcpcb *, struct tcphdr *, int *,
struct mbuf *);
static void tcp_xmit_timer(struct tcpcb *, int);
static inline unsigned int tcp_maxmtu(struct rtentry *);
static inline int tcp_stretch_ack_enable(struct tcpcb *tp);
#if TRAFFIC_MGT
static inline void update_iaj_state(struct tcpcb *tp, uint32_t tlen, int reset_size);
void compute_iaj(struct tcpcb *tp);
#endif
#if INET6
static inline unsigned int tcp_maxmtu6(struct rtentry *);
#endif
static void tcp_sbrcv_grow(struct tcpcb *tp, struct sockbuf *sb,
struct tcpopt *to, u_int32_t tlen);
void tcp_sbrcv_trim(struct tcpcb *tp, struct sockbuf *sb);
static void tcp_sbsnd_trim(struct sockbuf *sbsnd);
static inline void tcp_sbrcv_tstmp_check(struct tcpcb *tp);
static inline void tcp_sbrcv_reserve(struct tcpcb *tp, struct sockbuf *sb,
u_int32_t newsize, u_int32_t idealsize);
#define TCPTV_RCVNOTS_QUANTUM 100
#define TCP_RCVNOTS_BYTELEVEL 204800
#if INET6
#define ND6_HINT(tp) \
do { \
if ((tp) && (tp)->t_inpcb && \
((tp)->t_inpcb->inp_vflag & INP_IPV6) != 0 && \
(tp)->t_inpcb->in6p_route.ro_rt) \
nd6_nud_hint((tp)->t_inpcb->in6p_route.ro_rt, NULL, 0); \
} while (0)
#else
#define ND6_HINT(tp)
#endif
extern void add_to_time_wait(struct tcpcb *, uint32_t delay);
extern void postevent(struct socket *, struct sockbuf *, int);
extern void ipfwsyslog( int level, const char *format,...);
extern int ChkAddressOK( __uint32_t dstaddr, __uint32_t srcaddr );
extern int fw_verbose;
#if IPFIREWALL
#define log_in_vain_log( a ) { \
if ( (log_in_vain == 3 ) && (fw_verbose == 2)) { \
ipfwsyslog a ; \
} \
else log a ; \
}
#else
#define log_in_vain_log( a ) { log a; }
#endif
int tcp_rcvunackwin = TCPTV_UNACKWIN;
int tcp_maxrcvidle = TCPTV_MAXRCVIDLE;
int tcp_rcvsspktcnt = TCP_RCV_SS_PKTCOUNT;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, rcvsspktcnt, CTLFLAG_RW | CTLFLAG_LOCKED,
&tcp_rcvsspktcnt, 0, "packets to be seen before receiver stretches acks");
#define DELAY_ACK(tp, th) (CC_ALGO(tp)->delay_ack != NULL && CC_ALGO(tp)->delay_ack(tp, th))
static int tcp_dropdropablreq(struct socket *head);
static void tcp_newreno_partial_ack(struct tcpcb *tp, struct tcphdr *th);
static void update_base_rtt(struct tcpcb *tp, uint32_t rtt);
uint32_t get_base_rtt(struct tcpcb *tp);
void tcp_set_background_cc(struct socket *so);
void tcp_set_foreground_cc(struct socket *so);
static void tcp_set_new_cc(struct socket *so, uint16_t cc_index);
static void tcp_bwmeas_check(struct tcpcb *tp);
#if TRAFFIC_MGT
void
reset_acc_iaj(struct tcpcb *tp)
{
tp->acc_iaj = 0;
tp->iaj_rwintop = 0;
CLEAR_IAJ_STATE(tp);
}
static inline void
update_iaj_state(struct tcpcb *tp, uint32_t size, int rst_size)
{
if (rst_size > 0)
tp->iaj_size = 0;
if (tp->iaj_size == 0 || size >= tp->iaj_size) {
tp->iaj_size = size;
tp->iaj_rcv_ts = tcp_now;
tp->iaj_small_pkt = 0;
}
}
static inline int
isqrt(unsigned int val) {
unsigned int sqrt_cache[11] = {0, 1, 4, 9, 16, 25, 36, 49, 64, 81, 100};
unsigned int temp, g=0, b=0x8000, bshft=15;
if ( val <= 100) {
for (g = 0; g <= 10; ++g) {
if (sqrt_cache[g] > val) {
g--;
break;
} else if (sqrt_cache[g] == val) {
break;
}
}
} else {
do {
temp = (((g << 1) + b) << (bshft--));
if (val >= temp) {
g += b;
val -= temp;
}
b >>= 1;
} while ( b > 0 && val > 0);
}
return(g);
}
void
compute_iaj(struct tcpcb *tp)
{
#define MAX_ACC_IAJ (tcp_acc_iaj_high_thresh + tcp_acc_iaj_react_limit)
uint32_t allowed_iaj, acc_iaj = 0;
uint32_t cur_iaj = tcp_now - tp->iaj_rcv_ts;
uint32_t mean, temp;
int32_t cur_iaj_dev;
cur_iaj_dev = (cur_iaj - tp->avg_iaj);
allowed_iaj = tp->avg_iaj + tp->std_dev_iaj;
if (allowed_iaj < tcp_allowed_iaj)
allowed_iaj = tcp_allowed_iaj;
if (tp->iaj_pktcnt > IAJ_IGNORE_PKTCNT) {
if ( cur_iaj <= allowed_iaj ) {
if (tp->acc_iaj >= 2)
acc_iaj = tp->acc_iaj - 2;
else
acc_iaj = 0;
} else {
acc_iaj = tp->acc_iaj + (cur_iaj - allowed_iaj);
}
if (acc_iaj > MAX_ACC_IAJ)
acc_iaj = MAX_ACC_IAJ;
tp->acc_iaj = acc_iaj;
}
tp->avg_iaj = (((tp->avg_iaj << 4) - tp->avg_iaj) + cur_iaj) >> 4;
temp = tp->std_dev_iaj * tp->std_dev_iaj;
mean = (((temp << 4) - temp) + (cur_iaj_dev * cur_iaj_dev)) >> 4;
tp->std_dev_iaj = isqrt(mean);
DTRACE_TCP3(iaj, struct tcpcb *, tp, uint32_t, cur_iaj, uint32_t, allowed_iaj);
return;
}
#endif
static void
tcp_bwmeas_check(struct tcpcb *tp)
{
int32_t bw_meas_bytes;
uint32_t bw, bytes, elapsed_time;
bw_meas_bytes = tp->snd_una - tp->t_bwmeas->bw_start;
if ((tp->t_flagsext & TF_BWMEAS_INPROGRESS) != 0 &&
bw_meas_bytes >= (int32_t)(tp->t_bwmeas->bw_size)) {
bytes = bw_meas_bytes;
elapsed_time = tcp_now - tp->t_bwmeas->bw_ts;
if (elapsed_time > 0) {
bw = bytes / elapsed_time;
if ( bw > 0) {
if (tp->t_bwmeas->bw_sndbw > 0) {
tp->t_bwmeas->bw_sndbw =
(((tp->t_bwmeas->bw_sndbw << 3) - tp->t_bwmeas->bw_sndbw) + bw) >> 3;
} else {
tp->t_bwmeas->bw_sndbw = bw;
}
}
}
tp->t_flagsext &= ~(TF_BWMEAS_INPROGRESS);
}
}
static int
tcp_reass(tp, th, tlenp, m)
register struct tcpcb *tp;
register struct tcphdr *th;
int *tlenp;
struct mbuf *m;
{
struct tseg_qent *q;
struct tseg_qent *p = NULL;
struct tseg_qent *nq;
struct tseg_qent *te = NULL;
struct socket *so = tp->t_inpcb->inp_socket;
int flags;
int dowakeup = 0;
if (th == NULL)
goto present;
if ((tp->t_flags & TF_STRETCHACK) != 0)
tcp_reset_stretch_ack(tp);
tp->t_flagsext |= TF_RCVUNACK_WAITSS;
tp->rcv_waitforss = 0;
#if TRAFFIC_MGT
if (tp->acc_iaj > 0)
reset_acc_iaj(tp);
#endif
if (th->th_seq != tp->rcv_nxt &&
tcp_reass_qsize + 1 >= tcp_reass_maxseg) {
tcp_reass_overflows++;
tcpstat.tcps_rcvmemdrop++;
m_freem(m);
*tlenp = 0;
return (0);
}
te = (struct tseg_qent *) zalloc_noblock(tcp_reass_zone);
if (te == NULL) {
tcpstat.tcps_rcvmemdrop++;
m_freem(m);
return (0);
}
tcp_reass_qsize++;
LIST_FOREACH(q, &tp->t_segq, tqe_q) {
if (SEQ_GT(q->tqe_th->th_seq, th->th_seq))
break;
p = q;
}
if (p != NULL) {
register int i;
i = p->tqe_th->th_seq + p->tqe_len - th->th_seq;
if (i > 0) {
if (i >= *tlenp) {
tcpstat.tcps_rcvduppack++;
tcpstat.tcps_rcvdupbyte += *tlenp;
if (nstat_collect) {
nstat_route_rx(tp->t_inpcb->inp_route.ro_rt, 1, *tlenp, NSTAT_RX_FLAG_DUPLICATE);
locked_add_64(&tp->t_inpcb->inp_stat->rxpackets, 1);
locked_add_64(&tp->t_inpcb->inp_stat->rxbytes, *tlenp);
tp->t_stat.rxduplicatebytes += *tlenp;
}
m_freem(m);
zfree(tcp_reass_zone, te);
tcp_reass_qsize--;
goto present;
}
m_adj(m, i);
*tlenp -= i;
th->th_seq += i;
}
}
tcpstat.tcps_rcvoopack++;
tcpstat.tcps_rcvoobyte += *tlenp;
if (nstat_collect) {
nstat_route_rx(tp->t_inpcb->inp_route.ro_rt, 1, *tlenp, NSTAT_RX_FLAG_OUT_OF_ORDER);
locked_add_64(&tp->t_inpcb->inp_stat->rxpackets, 1);
locked_add_64(&tp->t_inpcb->inp_stat->rxbytes, *tlenp);
tp->t_stat.rxoutoforderbytes += *tlenp;
}
while (q) {
register int i = (th->th_seq + *tlenp) - q->tqe_th->th_seq;
if (i <= 0)
break;
if (i < q->tqe_len) {
q->tqe_th->th_seq += i;
q->tqe_len -= i;
m_adj(q->tqe_m, i);
break;
}
nq = LIST_NEXT(q, tqe_q);
LIST_REMOVE(q, tqe_q);
m_freem(q->tqe_m);
zfree(tcp_reass_zone, q);
tcp_reass_qsize--;
q = nq;
}
te->tqe_m = m;
te->tqe_th = th;
te->tqe_len = *tlenp;
if (p == NULL) {
LIST_INSERT_HEAD(&tp->t_segq, te, tqe_q);
} else {
LIST_INSERT_AFTER(p, te, tqe_q);
}
present:
if (!TCPS_HAVEESTABLISHED(tp->t_state))
return (0);
q = LIST_FIRST(&tp->t_segq);
if (!q || q->tqe_th->th_seq != tp->rcv_nxt) {
if (tp->t_flagsext & TF_LRO_OFFLOADED) {
tcp_lro_remove_state(tp->t_inpcb->inp_laddr,
tp->t_inpcb->inp_faddr,
th->th_dport, th->th_sport);
tp->t_flagsext &= ~TF_LRO_OFFLOADED;
}
return (0);
}
do {
tp->rcv_nxt += q->tqe_len;
flags = q->tqe_th->th_flags & TH_FIN;
nq = LIST_NEXT(q, tqe_q);
LIST_REMOVE(q, tqe_q);
if (so->so_state & SS_CANTRCVMORE)
m_freem(q->tqe_m);
else {
so_recv_data_stat(so, q->tqe_m, 0);
if (sbappendstream(&so->so_rcv, q->tqe_m))
dowakeup = 1;
if (tp->t_flagsext & TF_LRO_OFFLOADED) {
tcp_update_lro_seq(tp->rcv_nxt,
tp->t_inpcb->inp_laddr,
tp->t_inpcb->inp_faddr, th->th_dport, th->th_sport);
}
}
zfree(tcp_reass_zone, q);
tcp_reass_qsize--;
q = nq;
} while (q && q->tqe_th->th_seq == tp->rcv_nxt);
ND6_HINT(tp);
#if INET6
if ((tp->t_inpcb->inp_vflag & INP_IPV6) != 0) {
KERNEL_DEBUG(DBG_LAYER_BEG,
((tp->t_inpcb->inp_fport << 16) | tp->t_inpcb->inp_lport),
(((tp->t_inpcb->in6p_laddr.s6_addr16[0] & 0xffff) << 16) |
(tp->t_inpcb->in6p_faddr.s6_addr16[0] & 0xffff)),
0,0,0);
}
else
#endif
{
KERNEL_DEBUG(DBG_LAYER_BEG,
((tp->t_inpcb->inp_fport << 16) | tp->t_inpcb->inp_lport),
(((tp->t_inpcb->inp_laddr.s_addr & 0xffff) << 16) |
(tp->t_inpcb->inp_faddr.s_addr & 0xffff)),
0,0,0);
}
if (dowakeup)
sorwakeup(so);
return (flags);
}
static void
tcp_reduce_congestion_window(
struct tcpcb *tp)
{
if (CC_ALGO(tp)->pre_fr != NULL)
CC_ALGO(tp)->pre_fr(tp);
ENTER_FASTRECOVERY(tp);
tp->snd_recover = tp->snd_max;
tp->t_timer[TCPT_REXMT] = 0;
tp->t_rtttime = 0;
tp->ecn_flags |= TE_SENDCWR;
tp->snd_cwnd = tp->snd_ssthresh +
tp->t_maxseg * tcprexmtthresh;
}
#if INET6
int
tcp6_input(struct mbuf **mp, int *offp, int proto)
{
#pragma unused(proto)
register struct mbuf *m = *mp;
struct in6_ifaddr *ia6;
struct ifnet *ifp = ((m->m_flags & M_PKTHDR) && m->m_pkthdr.rcvif != NULL) ? m->m_pkthdr.rcvif: NULL;
IP6_EXTHDR_CHECK(m, *offp, sizeof(struct tcphdr), return IPPROTO_DONE);
MBUF_STRICT_DATA_ALIGNMENT_CHECK_32(m);
ia6 = ip6_getdstifaddr(m);
if (ia6 != NULL) {
IFA_LOCK_SPIN(&ia6->ia_ifa);
if (ia6->ia6_flags & IN6_IFF_ANYCAST) {
struct ip6_hdr *ip6;
IFA_UNLOCK(&ia6->ia_ifa);
IFA_REMREF(&ia6->ia_ifa);
ip6 = mtod(m, struct ip6_hdr *);
icmp6_error(m, ICMP6_DST_UNREACH,
ICMP6_DST_UNREACH_ADDR,
(caddr_t)&ip6->ip6_dst - (caddr_t)ip6);
if (ifp != NULL && ifp->if_tcp_stat != NULL)
atomic_add_64(&ifp->if_tcp_stat->icmp6unreach, 1);
return (IPPROTO_DONE);
}
IFA_UNLOCK(&ia6->ia_ifa);
IFA_REMREF(&ia6->ia_ifa);
}
tcp_input(m, *offp);
return (IPPROTO_DONE);
}
#endif
u_int8_t
tcp_cansbgrow(struct sockbuf *sb)
{
u_int32_t mblim = ((nmbclusters >> 1) << (MCLSHIFT - MSIZESHIFT));
u_int32_t sbspacelim = ((nmbclusters >> 4) << MCLSHIFT);
if ((total_sbmb_cnt < mblim) &&
(sb->sb_hiwat < sbspacelim)) {
return(1);
}
return(0);
}
void
tcp_sbrcv_reserve(struct tcpcb *tp,
struct sockbuf *sbrcv,
u_int32_t newsize,
u_int32_t idealsize) {
newsize = min(newsize, tcp_autorcvbuf_max);
newsize = min(newsize, TCP_MAXWIN << tp->rcv_scale);
if (newsize > sbrcv->sb_hiwat &&
(sbreserve(sbrcv, newsize) == 1)) {
sbrcv->sb_idealsize = min(max(sbrcv->sb_idealsize,
(idealsize != 0) ? idealsize : newsize),
tcp_autorcvbuf_max);
sbrcv->sb_idealsize = min(sbrcv->sb_idealsize,
TCP_MAXWIN << tp->rcv_scale);
}
}
static void
tcp_sbrcv_grow(struct tcpcb *tp, struct sockbuf *sbrcv,
struct tcpopt *to, u_int32_t pktlen) {
if (tcp_do_autorcvbuf == 0 ||
(sbrcv->sb_flags & SB_AUTOSIZE) == 0 ||
tcp_cansbgrow(sbrcv) == 0 ||
sbrcv->sb_hiwat >= tcp_autorcvbuf_max) {
goto out;
}
if (TSTMP_GT(tcp_now,
tp->rfbuf_ts + TCPTV_RCVBUFIDLE)) {
goto out;
}
if ((tp->t_flags & (TF_REQ_TSTMP | TF_RCVD_TSTMP)) !=
(TF_REQ_TSTMP | TF_RCVD_TSTMP)) {
if (TSTMP_GEQ(tcp_now,
tp->rfbuf_ts + TCPTV_RCVNOTS_QUANTUM)) {
if (tp->rfbuf_cnt >= TCP_RCVNOTS_BYTELEVEL) {
tcp_sbrcv_reserve(tp, sbrcv,
tcp_autorcvbuf_max, 0);
}
goto out;
} else {
tp->rfbuf_cnt += pktlen;
return;
}
} else if (to->to_tsecr != 0) {
if (TSTMP_GEQ(to->to_tsecr, tp->rfbuf_ts)) {
if ((tp->rfbuf_cnt > (sbrcv->sb_hiwat -
(sbrcv->sb_hiwat >> tcp_rbuf_hiwat_shift)) ||
(sbrcv->sb_hiwat - sbrcv->sb_cc) <
(tp->rfbuf_cnt >> tcp_rbuf_win_shift))) {
u_int32_t rcvbuf_inc;
rcvbuf_inc = tp->t_maxseg << tcp_autorcvbuf_inc_shift;
tcp_sbrcv_reserve(tp, sbrcv,
sbrcv->sb_hiwat + rcvbuf_inc,
(tp->rfbuf_cnt * 2));
}
goto out;
} else {
tp->rfbuf_cnt += pktlen;
return;
}
}
out:
tp->rfbuf_ts = 0;
tp->rfbuf_cnt = 0;
return;
}
void
tcp_sbrcv_trim(struct tcpcb *tp, struct sockbuf *sbrcv) {
if (tcp_do_autorcvbuf == 1 && sbrcv->sb_idealsize > 0 &&
sbrcv->sb_hiwat > sbrcv->sb_idealsize) {
int32_t trim;
u_int32_t diff = sbrcv->sb_hiwat - sbrcv->sb_idealsize;
u_int32_t advwin = tp->rcv_adv - tp->rcv_nxt;
u_int32_t leave = max(advwin, (sbrcv->sb_idealsize >> 4));
if (leave == 0)
leave = tp->t_maxseg << tcp_autorcvbuf_inc_shift;
trim = sbrcv->sb_hiwat - (sbrcv->sb_cc + leave);
trim = imin(trim, (int32_t)diff);
if (trim > 0)
sbreserve(sbrcv, (sbrcv->sb_hiwat - trim));
}
}
void
tcp_sbsnd_trim(struct sockbuf *sbsnd) {
if (tcp_do_autosendbuf == 1 &&
((sbsnd->sb_flags & (SB_AUTOSIZE | SB_TRIM)) ==
(SB_AUTOSIZE | SB_TRIM)) &&
(sbsnd->sb_idealsize > 0) &&
(sbsnd->sb_hiwat > sbsnd->sb_idealsize)) {
u_int32_t trim = 0;
if (sbsnd->sb_cc <= sbsnd->sb_idealsize) {
trim = sbsnd->sb_hiwat - sbsnd->sb_idealsize;
} else {
trim = sbsnd->sb_hiwat - sbsnd->sb_cc;
}
sbreserve(sbsnd, (sbsnd->sb_hiwat - trim));
}
if (sbsnd->sb_hiwat <= sbsnd->sb_idealsize)
sbsnd->sb_flags &= ~(SB_TRIM);
}
static inline void
tcp_sbrcv_tstmp_check(struct tcpcb *tp) {
struct socket *so = tp->t_inpcb->inp_socket;
u_int32_t newsize = 2 * tcp_recvspace;
struct sockbuf *sbrcv = &so->so_rcv;
if ((tp->t_flags & (TF_REQ_TSTMP | TF_RCVD_TSTMP)) !=
(TF_REQ_TSTMP | TF_RCVD_TSTMP) &&
(sbrcv->sb_flags & SB_AUTOSIZE) != 0) {
tcp_sbrcv_reserve(tp, sbrcv, newsize, 0);
}
}
static inline int
tcp_stretch_ack_enable(struct tcpcb *tp) {
if (tp->rcv_by_unackwin >= (maxseg_unacked * tp->t_maxseg) &&
TSTMP_GT(tp->rcv_unackwin + tcp_maxrcvidle, tcp_now) &&
(((tp->t_flagsext & TF_RCVUNACK_WAITSS) == 0) ||
(tp->rcv_waitforss >= tcp_rcvsspktcnt))) {
return(1);
}
return(0);
}
void
tcp_reset_stretch_ack(struct tcpcb *tp)
{
tp->t_flags &= ~(TF_STRETCHACK);
tp->rcv_by_unackwin = 0;
tp->rcv_unackwin = tcp_now + tcp_rcvunackwin;
}
void
tcp_input(m, off0)
struct mbuf *m;
int off0;
{
register struct tcphdr *th;
register struct ip *ip = NULL;
register struct ipovly *ipov;
register struct inpcb *inp;
u_char *optp = NULL;
int optlen = 0;
int len, tlen, off;
int drop_hdrlen;
register struct tcpcb *tp = 0;
register int thflags;
struct socket *so = 0;
int todrop, acked, ourfinisacked, needoutput = 0;
struct in_addr laddr;
#if INET6
struct in6_addr laddr6;
#endif
int dropsocket = 0;
int iss = 0, nosock = 0;
u_int32_t tiwin;
struct tcpopt to;
struct sockaddr_in *next_hop = NULL;
#if TCPDEBUG
short ostate = 0;
#endif
struct m_tag *fwd_tag;
u_char ip_ecn = IPTOS_ECN_NOTECT;
unsigned int ifscope, nocell = 0;
uint8_t isconnected, isdisconnected;
struct ifnet *ifp = ((m->m_flags & M_PKTHDR) && m->m_pkthdr.rcvif != NULL) ? m->m_pkthdr.rcvif: NULL;
int nlropkts = m->m_pkthdr.lro_npkts;
int mauxf_sw_lro_pkt = (m->m_pkthdr.aux_flags & MAUXF_SW_LRO_PKT) ? 1 : 0;
int turnoff_lro = 0;
#define TCP_INC_VAR(stat, npkts) do { \
if (mauxf_sw_lro_pkt) { \
stat += npkts; \
} else { \
stat++; \
} \
} while (0)
TCP_INC_VAR(tcpstat.tcps_rcvtotal, nlropkts);
if (!SLIST_EMPTY(&m->m_pkthdr.tags)) {
fwd_tag = m_tag_locate(m, KERNEL_MODULE_TAG_ID,
KERNEL_TAG_TYPE_IPFORWARD, NULL);
} else {
fwd_tag = NULL;
}
if (fwd_tag != NULL) {
struct ip_fwd_tag *ipfwd_tag = (struct ip_fwd_tag *)(fwd_tag+1);
next_hop = ipfwd_tag->next_hop;
m_tag_delete(m, fwd_tag);
}
#if INET6
struct ip6_hdr *ip6 = NULL;
int isipv6;
#endif
int rstreason;
struct proc *proc0=current_proc();
KERNEL_DEBUG(DBG_FNC_TCP_INPUT | DBG_FUNC_START,0,0,0,0,0);
#if INET6
isipv6 = (mtod(m, struct ip *)->ip_v == 6) ? 1 : 0;
#endif
bzero((char *)&to, sizeof(to));
#if INET6
if (isipv6) {
MBUF_STRICT_DATA_ALIGNMENT_CHECK_32(m);
ip6 = mtod(m, struct ip6_hdr *);
tlen = sizeof(*ip6) + ntohs(ip6->ip6_plen) - off0;
th = (struct tcphdr *)(void *)((caddr_t)ip6 + off0);
if ((apple_hwcksum_rx != 0) && (m->m_pkthdr.csum_flags & CSUM_DATA_VALID)) {
if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR)
th->th_sum = m->m_pkthdr.csum_data;
else {
if (in6_cksum(m, IPPROTO_TCP, off0, tlen)) {
th->th_sum = 0;
} else {
th->th_sum = 0xffff;
}
}
th->th_sum ^= 0xffff;
if (th->th_sum) {
tcpstat.tcps_rcvbadsum++;
if (ifp != NULL && ifp->if_tcp_stat != NULL)
atomic_add_64(&ifp->if_tcp_stat->badformat, 1);
goto dropnosock;
}
}
else {
if (in6_cksum(m, IPPROTO_TCP, off0, tlen)) {
tcpstat.tcps_rcvbadsum++;
if (ifp != NULL && ifp->if_tcp_stat != NULL)
atomic_add_64(&ifp->if_tcp_stat->badformat, 1);
goto dropnosock;
}
}
KERNEL_DEBUG(DBG_LAYER_BEG, ((th->th_dport << 16) | th->th_sport),
(((ip6->ip6_src.s6_addr16[0]) << 16) | (ip6->ip6_dst.s6_addr16[0])),
th->th_seq, th->th_ack, th->th_win);
if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
if (ifp != NULL && ifp->if_tcp_stat != NULL)
atomic_add_64(&ifp->if_tcp_stat->unspecv6, 1);
goto dropnosock;
}
DTRACE_TCP5(receive, sruct mbuf *, m, struct inpcb *, NULL,
struct ip6_hdr *, ip6, struct tcpcb *, NULL,
struct tcphdr *, th);
ip_ecn = (ntohl(ip6->ip6_flow) >> 20) & IPTOS_ECN_MASK;
} else
#endif
{
if (off0 > sizeof (struct ip)) {
ip_stripoptions(m, (struct mbuf *)0);
off0 = sizeof(struct ip);
if (m->m_pkthdr.csum_flags & CSUM_TCP_SUM16)
m->m_pkthdr.csum_flags = 0;
}
if (m->m_len < sizeof (struct tcpiphdr)) {
if ((m = m_pullup(m, sizeof (struct tcpiphdr))) == 0) {
tcpstat.tcps_rcvshort++;
return;
}
}
MBUF_STRICT_DATA_ALIGNMENT_CHECK_32(m);
ip = mtod(m, struct ip *);
ipov = (struct ipovly *)ip;
th = (struct tcphdr *)(void *)((caddr_t)ip + off0);
tlen = ip->ip_len;
if (m->m_pkthdr.aux_flags & MAUXF_SW_LRO_DID_CSUM) {
goto skip_checksum;
}
if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) {
if (m->m_pkthdr.csum_flags & CSUM_TCP_SUM16) {
u_short pseudo;
char b[9];
bcopy(ipov->ih_x1, b, sizeof (ipov->ih_x1));
bzero(ipov->ih_x1, sizeof (ipov->ih_x1));
ipov->ih_len = (u_short)tlen;
#if BYTE_ORDER != BIG_ENDIAN
HTONS(ipov->ih_len);
#endif
pseudo = in_cksum(m, sizeof (struct ip));
bcopy(b, ipov->ih_x1, sizeof (ipov->ih_x1));
th->th_sum = in_addword(pseudo, (m->m_pkthdr.csum_data & 0xFFFF));
} else {
if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR)
th->th_sum = m->m_pkthdr.csum_data;
else
th->th_sum = in_pseudo(ip->ip_src.s_addr,
ip->ip_dst.s_addr, htonl(m->m_pkthdr.csum_data +
ip->ip_len + IPPROTO_TCP));
}
th->th_sum ^= 0xffff;
} else {
char b[9];
bcopy(ipov->ih_x1, b, sizeof (ipov->ih_x1));
bzero(ipov->ih_x1, sizeof (ipov->ih_x1));
ipov->ih_len = (u_short)tlen;
#if BYTE_ORDER != BIG_ENDIAN
HTONS(ipov->ih_len);
#endif
len = sizeof (struct ip) + tlen;
th->th_sum = in_cksum(m, len);
bcopy(b, ipov->ih_x1, sizeof (ipov->ih_x1));
tcp_in_cksum_stats(len);
}
if (th->th_sum) {
tcpstat.tcps_rcvbadsum++;
if (ifp != NULL && ifp->if_tcp_stat != NULL)
atomic_add_64(&ifp->if_tcp_stat->badformat, 1);
if (lrodebug) printf("tcp_input: bad xsum len = %d, tlen = %d, flags = %x, csum_flags = %x.\n",len, tlen, m->m_flags, m->m_pkthdr.csum_flags);
goto dropnosock;
}
skip_checksum:
#if INET6
ip->ip_v = IPVERSION;
#endif
ip_ecn = (ip->ip_tos & IPTOS_ECN_MASK);
DTRACE_TCP5(receive, struct mbuf *, m, struct inpcb *, NULL,
struct ip *, ip, struct tcpcb *, NULL, struct tcphdr *, th);
KERNEL_DEBUG(DBG_LAYER_BEG, ((th->th_dport << 16) | th->th_sport),
(((ip->ip_src.s_addr & 0xffff) << 16) | (ip->ip_dst.s_addr & 0xffff)),
th->th_seq, th->th_ack, th->th_win);
}
off = th->th_off << 2;
if (off < sizeof (struct tcphdr) || off > tlen) {
tcpstat.tcps_rcvbadoff++;
if (ifp != NULL && ifp->if_tcp_stat != NULL)
atomic_add_64(&ifp->if_tcp_stat->badformat, 1);
goto dropnosock;
}
tlen -= off;
if (off > sizeof (struct tcphdr)) {
#if INET6
if (isipv6) {
IP6_EXTHDR_CHECK(m, off0, off, return);
ip6 = mtod(m, struct ip6_hdr *);
th = (struct tcphdr *)(void *)((caddr_t)ip6 + off0);
} else
#endif
{
if (m->m_len < sizeof(struct ip) + off) {
if ((m = m_pullup(m, sizeof (struct ip) + off)) == 0) {
tcpstat.tcps_rcvshort++;
return;
}
ip = mtod(m, struct ip *);
ipov = (struct ipovly *)ip;
th = (struct tcphdr *)(void *)((caddr_t)ip + off0);
}
}
optlen = off - sizeof (struct tcphdr);
optp = (u_char *)(th + 1);
if ((optlen == TCPOLEN_TSTAMP_APPA ||
(optlen > TCPOLEN_TSTAMP_APPA &&
optp[TCPOLEN_TSTAMP_APPA] == TCPOPT_EOL)) &&
*(u_int32_t *)(void *)optp == htonl(TCPOPT_TSTAMP_HDR) &&
(th->th_flags & TH_SYN) == 0) {
to.to_flags |= TOF_TS;
to.to_tsval = ntohl(*(u_int32_t *)(void *)(optp + 4));
to.to_tsecr = ntohl(*(u_int32_t *)(void *)(optp + 8));
optp = NULL;
}
}
thflags = th->th_flags;
#if TCP_DROP_SYNFIN
if (drop_synfin && (thflags & (TH_SYN|TH_FIN)) == (TH_SYN|TH_FIN)) {
if (ifp != NULL && ifp->if_tcp_stat != NULL)
atomic_add_64(&ifp->if_tcp_stat->synfin, 1);
goto dropnosock;
}
#endif
drop_hdrlen = off0 + off;
calculate_tcp_clock();
if ((m->m_flags & M_PKTHDR) && m->m_pkthdr.rcvif != NULL)
ifscope = m->m_pkthdr.rcvif->if_index;
else
ifscope = IFSCOPE_NONE;
#if BYTE_ORDER != BIG_ENDIAN
NTOHL(th->th_seq);
NTOHL(th->th_ack);
NTOHS(th->th_win);
NTOHS(th->th_urp);
#endif
findpcb:
isconnected = FALSE;
isdisconnected = FALSE;
#if IPFIREWALL_FORWARD
if (next_hop != NULL
#if INET6
&& isipv6 == 0
#endif
) {
inp = in_pcblookup_hash(&tcbinfo, ip->ip_src, th->th_sport,
ip->ip_dst, th->th_dport, 0, m->m_pkthdr.rcvif);
if (!inp) {
if (!next_hop->sin_port) {
inp = in_pcblookup_hash(&tcbinfo, ip->ip_src,
th->th_sport, next_hop->sin_addr,
th->th_dport, 1, m->m_pkthdr.rcvif);
} else {
inp = in_pcblookup_hash(&tcbinfo,
ip->ip_src, th->th_sport,
next_hop->sin_addr,
ntohs(next_hop->sin_port), 1,
m->m_pkthdr.rcvif);
}
}
} else
#endif
{
#if INET6
if (isipv6)
inp = in6_pcblookup_hash(&tcbinfo, &ip6->ip6_src, th->th_sport,
&ip6->ip6_dst, th->th_dport, 1,
m->m_pkthdr.rcvif);
else
#endif
inp = in_pcblookup_hash(&tcbinfo, ip->ip_src, th->th_sport,
ip->ip_dst, th->th_dport, 1, m->m_pkthdr.rcvif);
}
if (inp != NULL && (inp->inp_flags & INP_BOUND_IF))
ifscope = inp->inp_boundifp->if_index;
if (inp != NULL && (inp->inp_flags & INP_NO_IFT_CELLULAR))
nocell = 1;
#if IPSEC
if (ipsec_bypass == 0) {
#if INET6
if (isipv6) {
if (inp != NULL && ipsec6_in_reject_so(m, inp->inp_socket)) {
IPSEC_STAT_INCREMENT(ipsec6stat.in_polvio);
if (in_pcb_checkstate(inp, WNT_RELEASE, 0) == WNT_STOPUSING)
inp = NULL;
if (ifp != NULL && ifp->if_tcp_stat != NULL)
atomic_add_64(&ifp->if_tcp_stat->badformatipsec, 1);
goto dropnosock;
}
} else
#endif
if (inp != NULL && ipsec4_in_reject_so(m, inp->inp_socket)) {
IPSEC_STAT_INCREMENT(ipsecstat.in_polvio);
if (in_pcb_checkstate(inp, WNT_RELEASE, 0) == WNT_STOPUSING)
inp = NULL;
if (ifp != NULL && ifp->if_tcp_stat != NULL)
atomic_add_64(&ifp->if_tcp_stat->badformatipsec, 1);
goto dropnosock;
}
}
#endif
if (inp == NULL) {
if (log_in_vain) {
#if INET6
char dbuf[MAX_IPv6_STR_LEN], sbuf[MAX_IPv6_STR_LEN];
#else
char dbuf[MAX_IPv4_STR_LEN], sbuf[MAX_IPv4_STR_LEN];
#endif
#if INET6
if (isipv6) {
inet_ntop(AF_INET6, &ip6->ip6_dst, dbuf, sizeof(dbuf));
inet_ntop(AF_INET6, &ip6->ip6_src, sbuf, sizeof(sbuf));
} else
#endif
{
inet_ntop(AF_INET, &ip->ip_dst, dbuf, sizeof(dbuf));
inet_ntop(AF_INET, &ip->ip_src, sbuf, sizeof(sbuf));
}
switch (log_in_vain) {
case 1:
if(thflags & TH_SYN)
log(LOG_INFO,
"Connection attempt to TCP %s:%d from %s:%d\n",
dbuf, ntohs(th->th_dport),
sbuf,
ntohs(th->th_sport));
break;
case 2:
log(LOG_INFO,
"Connection attempt to TCP %s:%d from %s:%d flags:0x%x\n",
dbuf, ntohs(th->th_dport), sbuf,
ntohs(th->th_sport), thflags);
break;
case 3:
if ((thflags & TH_SYN) && !(thflags & TH_ACK) &&
!(m->m_flags & (M_BCAST | M_MCAST)) &&
#if INET6
((isipv6 && !IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, &ip6->ip6_src)) ||
(!isipv6 && ip->ip_dst.s_addr != ip->ip_src.s_addr))
#else
ip->ip_dst.s_addr != ip->ip_src.s_addr
#endif
)
log_in_vain_log((LOG_INFO,
"Stealth Mode connection attempt to TCP %s:%d from %s:%d\n",
dbuf, ntohs(th->th_dport),
sbuf,
ntohs(th->th_sport)));
break;
default:
break;
}
}
if (blackhole) {
if (m->m_pkthdr.rcvif && m->m_pkthdr.rcvif->if_type != IFT_LOOP)
switch (blackhole) {
case 1:
if (thflags & TH_SYN)
goto dropnosock;
break;
case 2:
goto dropnosock;
default:
goto dropnosock;
}
}
rstreason = BANDLIM_RST_CLOSEDPORT;
if (ifp != NULL && ifp->if_tcp_stat != NULL)
atomic_add_64(&ifp->if_tcp_stat->noconnnolist, 1);
goto dropwithresetnosock;
}
so = inp->inp_socket;
if (so == NULL) {
inp = NULL;
#if TEMPDEBUG
printf("tcp_input: no more socket for inp=%x. This shouldn't happen\n", inp);
#endif
goto dropnosock;
}
tcp_lock(so, 1, 0);
if (in_pcb_checkstate(inp, WNT_RELEASE, 1) == WNT_STOPUSING) {
tcp_unlock(so, 1, (void *)2);
inp = NULL; goto dropnosock;
}
tp = intotcpcb(inp);
if (tp == 0) {
rstreason = BANDLIM_RST_CLOSEDPORT;
if (ifp != NULL && ifp->if_tcp_stat != NULL)
atomic_add_64(&ifp->if_tcp_stat->noconnlist, 1);
goto dropwithreset;
}
if (tp->t_state == TCPS_CLOSED)
goto drop;
if ((thflags & TH_SYN) == 0)
tiwin = th->th_win << tp->snd_scale;
else
tiwin = th->th_win;
#if CONFIG_MACF_NET
if (mac_inpcb_check_deliver(inp, m, AF_INET, SOCK_STREAM))
goto drop;
#endif
if (tp->t_state == TCPS_LISTEN && (so->so_options & SO_ACCEPTCONN) == 0)
goto drop;
if (so->so_options & (SO_DEBUG|SO_ACCEPTCONN)) {
#if TCPDEBUG
if (so->so_options & SO_DEBUG) {
ostate = tp->t_state;
#if INET6
if (isipv6)
bcopy((char *)ip6, (char *)tcp_saveipgen,
sizeof(*ip6));
else
#endif
bcopy((char *)ip, (char *)tcp_saveipgen, sizeof(*ip));
tcp_savetcp = *th;
}
#endif
if (so->so_options & SO_ACCEPTCONN) {
register struct tcpcb *tp0 = tp;
struct socket *so2;
struct socket *oso;
struct sockaddr_storage from;
#if INET6
struct inpcb *oinp = sotoinpcb(so);
#endif
struct ifnet *head_ifscope;
unsigned int head_nocell, head_recvanyif;
head_ifscope = (inp->inp_flags & INP_BOUND_IF) ?
inp->inp_boundifp : NULL;
head_nocell = (inp->inp_flags & INP_NO_IFT_CELLULAR) ? 1 : 0;
head_recvanyif = (inp->inp_flags & INP_RECV_ANYIF);
if ((thflags & (TH_RST|TH_ACK|TH_SYN)) != TH_SYN) {
if (ifp != NULL && ifp->if_tcp_stat != NULL)
atomic_add_64(&ifp->if_tcp_stat->listbadsyn, 1);
if (thflags & TH_RST) {
goto drop;
}
if (thflags & TH_ACK) {
tp = NULL;
tcpstat.tcps_badsyn++;
rstreason = BANDLIM_RST_OPENPORT;
goto dropwithreset;
}
tcpstat.tcps_badsyn++;
goto drop;
}
KERNEL_DEBUG(DBG_FNC_TCP_NEWCONN | DBG_FUNC_START,0,0,0,0,0);
if (th->th_dport == th->th_sport) {
#if INET6
if (isipv6) {
if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
&ip6->ip6_src))
goto drop;
} else
#endif
if (ip->ip_dst.s_addr == ip->ip_src.s_addr)
goto drop;
}
if (m->m_flags & (M_BCAST|M_MCAST))
goto drop;
#if INET6
if (isipv6) {
if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
IN6_IS_ADDR_MULTICAST(&ip6->ip6_src))
goto drop;
} else
#endif
if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) ||
IN_MULTICAST(ntohl(ip->ip_src.s_addr)) ||
ip->ip_src.s_addr == htonl(INADDR_BROADCAST) ||
in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif))
goto drop;
#if INET6
if (isipv6 && !ip6_use_deprecated) {
struct in6_ifaddr *ia6;
ia6 = ip6_getdstifaddr(m);
if (ia6 != NULL) {
IFA_LOCK_SPIN(&ia6->ia_ifa);
if (ia6->ia6_flags & IN6_IFF_DEPRECATED) {
IFA_UNLOCK(&ia6->ia_ifa);
IFA_REMREF(&ia6->ia_ifa);
tp = NULL;
rstreason = BANDLIM_RST_OPENPORT;
if (ifp != NULL && ifp->if_tcp_stat != NULL)
atomic_add_64(&ifp->if_tcp_stat->deprecate6, 1);
goto dropwithreset;
}
IFA_UNLOCK(&ia6->ia_ifa);
IFA_REMREF(&ia6->ia_ifa);
}
}
#endif
if (so->so_filt) {
#if INET6
if (isipv6) {
struct sockaddr_in6 *sin6 = (struct sockaddr_in6*)&from;
sin6->sin6_len = sizeof(*sin6);
sin6->sin6_family = AF_INET6;
sin6->sin6_port = th->th_sport;
sin6->sin6_flowinfo = 0;
sin6->sin6_addr = ip6->ip6_src;
sin6->sin6_scope_id = 0;
}
else
#endif
{
struct sockaddr_in *sin = (struct sockaddr_in*)&from;
sin->sin_len = sizeof(*sin);
sin->sin_family = AF_INET;
sin->sin_port = th->th_sport;
sin->sin_addr = ip->ip_src;
}
so2 = sonewconn(so, 0, (struct sockaddr*)&from);
} else {
so2 = sonewconn(so, 0, NULL);
}
if (so2 == 0) {
tcpstat.tcps_listendrop++;
if (tcp_dropdropablreq(so)) {
if (so->so_filt)
so2 = sonewconn(so, 0, (struct sockaddr*)&from);
else
so2 = sonewconn(so, 0, NULL);
}
if (!so2)
goto drop;
}
inp = (struct inpcb *)so2->so_pcb;
tp = intotcpcb(inp);
oso = so;
tcp_unlock(so, 0, 0);
so = so2;
tcp_lock(so, 1, 0);
dropsocket++;
if (head_ifscope != NULL) {
inp->inp_flags |= INP_BOUND_IF;
inp->inp_boundifp = head_ifscope;
} else {
inp->inp_flags &= ~INP_BOUND_IF;
}
if (head_nocell) {
inp->inp_flags |= INP_NO_IFT_CELLULAR;
}
if (head_recvanyif)
inp->inp_flags |= INP_RECV_ANYIF;
else
inp->inp_flags &= ~INP_RECV_ANYIF;
#if INET6
if (isipv6)
inp->in6p_laddr = ip6->ip6_dst;
else {
inp->inp_vflag &= ~INP_IPV6;
inp->inp_vflag |= INP_IPV4;
#endif
inp->inp_laddr = ip->ip_dst;
#if INET6
}
#endif
inp->inp_lport = th->th_dport;
if (in_pcbinshash(inp, 0) != 0) {
#if INET6
if (isipv6)
inp->in6p_laddr = in6addr_any;
else
#endif
inp->inp_laddr.s_addr = INADDR_ANY;
inp->inp_lport = 0;
tcp_lock(oso, 0, 0);
tcp_unlock(oso, 1, 0);
goto drop;
}
#if INET6
if (isipv6) {
inp->inp_flags |=
oinp->inp_flags & INP_CONTROLOPTS;
if (oinp->in6p_outputopts)
inp->in6p_outputopts =
ip6_copypktopts(oinp->in6p_outputopts,
M_NOWAIT);
} else
#endif
inp->inp_options = ip_srcroute();
tcp_lock(oso, 0, 0);
#if IPSEC
if (sotoinpcb(oso)->inp_sp)
{
int error = 0;
if (inp->inp_sp != NULL)
error = ipsec_init_policy(so, &inp->inp_sp);
if (error != 0 || ipsec_copy_policy(sotoinpcb(oso)->inp_sp, inp->inp_sp))
printf("tcp_input: could not copy policy\n");
}
#endif
DTRACE_TCP4(state__change, void, NULL, struct inpcb *, inp,
struct tcpcb *, tp, int32_t, TCPS_LISTEN);
tp->t_state = TCPS_LISTEN;
tp->t_flags |= tp0->t_flags & (TF_NOPUSH|TF_NOOPT|TF_NODELAY);
tp->t_flagsext |= (tp0->t_flagsext & TF_RXTFINDROP);
tp->t_keepinit = tp0->t_keepinit;
tp->t_inpcb->inp_ip_ttl = tp0->t_inpcb->inp_ip_ttl;
if ((so->so_flags & SOF_NOTSENT_LOWAT) != 0)
tp->t_notsent_lowat = tp0->t_notsent_lowat;
tcp_unlock(oso, 1, 0);
tcp_set_max_rwinscale(tp, so);
KERNEL_DEBUG(DBG_FNC_TCP_NEWCONN | DBG_FUNC_END,0,0,0,0,0);
}
}
lck_mtx_assert(&((struct inpcb *)so->so_pcb)->inpcb_mtx, LCK_MTX_ASSERT_OWNED);
if (tp->t_state == TCPS_ESTABLISHED && tlen > 0) {
if (TSTMP_GT(tp->rcv_reset, tcp_now)) {
tp->rcv_pps++;
tp->rcv_byps += tlen + off;
if (tp->rcv_byps > tp->rcv_maxbyps)
tp->rcv_maxbyps = tp->rcv_byps;
if (tcp_minmss && tcp_minmssoverload && tp->rcv_pps > tcp_minmssoverload) {
if ((tp->rcv_byps / tp->rcv_pps) < tcp_minmss) {
char ipstrbuf[MAX_IPv6_STR_LEN];
printf("too many small tcp packets from "
"%s:%u, av. %ubyte/packet, "
"dropping connection\n",
#if INET6
isipv6 ?
inet_ntop(AF_INET6, &inp->in6p_faddr, ipstrbuf,
sizeof(ipstrbuf)) :
#endif
inet_ntop(AF_INET, &inp->inp_faddr, ipstrbuf,
sizeof(ipstrbuf)),
inp->inp_fport,
tp->rcv_byps / tp->rcv_pps);
tp = tcp_drop(tp, ECONNRESET);
goto drop;
}
}
} else {
tp->rcv_reset = tcp_now + TCP_RETRANSHZ;
tp->rcv_pps = 1;
tp->rcv_byps = tlen + off;
}
if ((tp->t_flagsext & TF_RCVUNACK_WAITSS) != 0) {
TCP_INC_VAR(tp->rcv_waitforss, nlropkts);
}
if (tcp_stretch_ack_enable(tp)) {
tp->t_flags |= TF_STRETCHACK;
tp->t_flagsext &= ~(TF_RCVUNACK_WAITSS);
tp->rcv_waitforss = 0;
} else {
tp->t_flags &= ~(TF_STRETCHACK);
}
if (TSTMP_GT(tp->rcv_unackwin, tcp_now)) {
tp->rcv_by_unackwin += (tlen + off);
} else {
tp->rcv_unackwin = tcp_now + tcp_rcvunackwin;
tp->rcv_by_unackwin = tlen + off;
}
}
if ((mauxf_sw_lro_pkt) && (nlropkts > 2)) {
tp->t_lropktlen += tlen;
}
if (ip_ecn == IPTOS_ECN_CE && tp->t_state == TCPS_ESTABLISHED &&
((tp->ecn_flags & (TE_ECN_ON)) == (TE_ECN_ON)) && tlen > 0 &&
SEQ_GEQ(th->th_seq, tp->last_ack_sent) &&
SEQ_LT(th->th_seq, tp->last_ack_sent + tp->rcv_wnd)) {
tp->ecn_flags |= TE_SENDECE;
}
if ((thflags & TH_CWR) == TH_CWR) {
tp->ecn_flags &= ~TE_SENDECE;
}
if (tp->t_state == TCPS_ESTABLISHED && (tp->t_flags & TF_STRETCHACK) != 0 &&
((ip_ecn == IPTOS_ECN_CE) || ((thflags & TH_CWR) == TH_CWR)))
tcp_reset_stretch_ack(tp);
if (sw_lro && (tp->t_flagsext & TF_LRO_OFFLOADED) &&
((tcp_now - tp->t_rcvtime) >= (TCP_IDLETIMEOUT(tp)))) {
turnoff_lro = 1;
}
tp->t_rcvtime = tcp_now;
if (TCPS_HAVEESTABLISHED(tp->t_state))
tp->t_timer[TCPT_KEEP] = OFFSET_FROM_START(tp, TCP_KEEPIDLE(tp));
if (tp->t_state != TCPS_LISTEN && optp)
tcp_dooptions(tp, optp, optlen, th, &to, ifscope);
if (tp->t_state == TCPS_SYN_SENT && (thflags & TH_SYN)) {
if (to.to_flags & TOF_SCALE) {
tp->t_flags |= TF_RCVD_SCALE;
tp->requested_s_scale = to.to_requested_s_scale;
tp->snd_wnd = th->th_win << tp->snd_scale;
tiwin = tp->snd_wnd;
}
if (to.to_flags & TOF_TS) {
tp->t_flags |= TF_RCVD_TSTMP;
tp->ts_recent = to.to_tsval;
tp->ts_recent_age = tcp_now;
}
if (to.to_flags & TOF_MSS)
tcp_mss(tp, to.to_mss, ifscope);
if (tp->sack_enable) {
if (!(to.to_flags & TOF_SACK))
tp->sack_enable = 0;
else
tp->t_flags |= TF_SACK_PERMIT;
}
}
#if TRAFFIC_MGT
if (tp->t_state == TCPS_ESTABLISHED &&
(thflags & (TH_SYN|TH_FIN|TH_RST|TH_URG|TH_ACK|TH_ECE|TH_PUSH)) == TH_ACK &&
((tp->t_flags & (TF_NEEDSYN|TF_NEEDFIN)) == 0) &&
((to.to_flags & TOF_TS) == 0 ||
TSTMP_GEQ(to.to_tsval, tp->ts_recent)) &&
th->th_seq == tp->rcv_nxt &&
LIST_EMPTY(&tp->t_segq)) {
int seg_size = tlen;
if (tp->iaj_pktcnt <= IAJ_IGNORE_PKTCNT) {
TCP_INC_VAR(tp->iaj_pktcnt, nlropkts);
}
if (m->m_pkthdr.aux_flags & MAUXF_SW_LRO_PKT) {
seg_size = m->m_pkthdr.lro_pktlen;
}
if ( tp->iaj_size == 0 || seg_size > tp->iaj_size ||
(seg_size == tp->iaj_size && tp->iaj_rcv_ts == 0)) {
update_iaj_state(tp, seg_size, 0);
} else {
if (seg_size == tp->iaj_size) {
compute_iaj(tp);
}
if (seg_size < tp->iaj_size) {
tp->iaj_small_pkt++;
if (tp->iaj_small_pkt > RESET_IAJ_SIZE_THRESH) {
update_iaj_state(tp, seg_size, 1);
} else {
CLEAR_IAJ_STATE(tp);
}
} else {
update_iaj_state(tp, seg_size, 0);
}
}
} else {
CLEAR_IAJ_STATE(tp);
}
#endif
if (tp->t_state == TCPS_ESTABLISHED &&
(thflags & (TH_SYN|TH_FIN|TH_RST|TH_URG|TH_ACK|TH_ECE)) == TH_ACK &&
((tp->t_flags & (TF_NEEDSYN|TF_NEEDFIN)) == 0) &&
((to.to_flags & TOF_TS) == 0 ||
TSTMP_GEQ(to.to_tsval, tp->ts_recent)) &&
th->th_seq == tp->rcv_nxt &&
tiwin && tiwin == tp->snd_wnd &&
tp->snd_nxt == tp->snd_max) {
if ((to.to_flags & TOF_TS) != 0 &&
SEQ_LEQ(th->th_seq, tp->last_ack_sent)) {
tp->ts_recent_age = tcp_now;
tp->ts_recent = to.to_tsval;
}
if (thflags & TH_FIN)
tp->t_flags |= TF_ACKNOW;
if (tlen == 0) {
if (SEQ_GT(th->th_ack, tp->snd_una) &&
SEQ_LEQ(th->th_ack, tp->snd_max) &&
tp->snd_cwnd >= tp->snd_ssthresh &&
(!IN_FASTRECOVERY(tp) &&
((!tp->sack_enable && tp->t_dupacks < tcprexmtthresh) ||
(tp->sack_enable && to.to_nsacks == 0 &&
TAILQ_EMPTY(&tp->snd_holes))))) {
++tcpstat.tcps_predack;
if (tp->t_rxtshift == 1 &&
TSTMP_LT(tcp_now, tp->t_badrxtwin)) {
++tcpstat.tcps_sndrexmitbad;
tp->snd_cwnd = tp->snd_cwnd_prev;
tp->snd_ssthresh =
tp->snd_ssthresh_prev;
tp->snd_recover = tp->snd_recover_prev;
if (tp->t_flags & TF_WASFRECOVERY)
ENTER_FASTRECOVERY(tp);
tp->snd_nxt = tp->snd_max;
tp->t_badrxtwin = 0;
tp->t_rxtshift = 0;
tp->rxt_start = 0;
tcp_bad_rexmt_fix_sndbuf(tp);
DTRACE_TCP5(cc, void, NULL, struct inpcb *, tp->t_inpcb,
struct tcpcb *, tp, struct tcphdr *, th,
int32_t, TCP_CC_BAD_REXMT_RECOVERY);
}
if (((to.to_flags & TOF_TS) != 0) && (to.to_tsecr != 0) &&
TSTMP_GEQ(tcp_now, to.to_tsecr)) {
tcp_xmit_timer(tp,
tcp_now - to.to_tsecr);
} else if (tp->t_rtttime &&
SEQ_GT(th->th_ack, tp->t_rtseq)) {
tcp_xmit_timer(tp, tcp_now - tp->t_rtttime);
}
acked = th->th_ack - tp->snd_una;
tcpstat.tcps_rcvackpack++;
tcpstat.tcps_rcvackbyte += acked;
if (CC_ALGO(tp)->inseq_ack_rcvd != NULL)
CC_ALGO(tp)->inseq_ack_rcvd(tp, th);
DTRACE_TCP5(cc, void, NULL, struct inpcb *, inp,
struct tcpcb *, tp, struct tcphdr *, th,
int32_t, TCP_CC_INSEQ_ACK_RCVD);
sbdrop(&so->so_snd, acked);
tcp_sbsnd_trim(&so->so_snd);
if (SEQ_GT(tp->snd_una, tp->snd_recover) &&
SEQ_LEQ(th->th_ack, tp->snd_recover))
tp->snd_recover = th->th_ack - 1;
tp->snd_una = th->th_ack;
tp->snd_wl2 = th->th_ack;
tp->t_dupacks = 0;
m_freem(m);
ND6_HINT(tp);
if (tp->snd_una == tp->snd_max)
tp->t_timer[TCPT_REXMT] = 0;
else if (tp->t_timer[TCPT_PERSIST] == 0)
tp->t_timer[TCPT_REXMT] = OFFSET_FROM_START(tp, tp->t_rxtcur);
if ((tp->t_flagsext & TF_MEASURESNDBW) != 0 &&
tp->t_bwmeas != NULL)
tcp_bwmeas_check(tp);
sowwakeup(so);
if ((so->so_snd.sb_cc) || (tp->t_flags & TF_ACKNOW)) {
(void) tcp_output(tp);
}
tcp_check_timer_state(tp);
tcp_unlock(so, 1, 0);
KERNEL_DEBUG(DBG_FNC_TCP_INPUT | DBG_FUNC_END,0,0,0,0,0);
return;
}
} else if (th->th_ack == tp->snd_una &&
LIST_EMPTY(&tp->t_segq) &&
tlen <= tcp_sbspace(tp)) {
if (turnoff_lro) {
tcp_lro_remove_state(tp->t_inpcb->inp_laddr,
tp->t_inpcb->inp_faddr,
tp->t_inpcb->inp_lport,
tp->t_inpcb->inp_fport);
tp->t_flagsext &= ~TF_LRO_OFFLOADED;
tp->t_idleat = tp->rcv_nxt;
} else if (sw_lro && !mauxf_sw_lro_pkt && !isipv6 &&
(so->so_flags & SOF_USELRO) &&
(m->m_pkthdr.rcvif->if_type != IFT_CELLULAR) &&
(m->m_pkthdr.rcvif->if_type != IFT_LOOP) &&
((th->th_seq - tp->irs) >
(tp->t_maxseg << lro_start)) &&
((tp->t_idleat == 0) || ((th->th_seq -
tp->t_idleat) > (tp->t_maxseg << lro_start)))) {
tp->t_flagsext |= TF_LRO_OFFLOADED;
tcp_start_coalescing(ip, th, tlen);
tp->t_idleat = 0;
}
if (tp->sack_enable && tp->rcv_numsacks)
tcp_clean_sackreport(tp);
++tcpstat.tcps_preddat;
tp->rcv_nxt += tlen;
tp->snd_wl1 = th->th_seq;
tp->rcv_up = tp->rcv_nxt;
TCP_INC_VAR(tcpstat.tcps_rcvpack, nlropkts);
tcpstat.tcps_rcvbyte += tlen;
if (nstat_collect) {
if (m->m_pkthdr.aux_flags & MAUXF_SW_LRO_PKT) {
locked_add_64(&inp->inp_stat->rxpackets, m->m_pkthdr.lro_npkts);
}
else {
locked_add_64(&inp->inp_stat->rxpackets, 1);
}
locked_add_64(&inp->inp_stat->rxbytes, tlen);
}
ND6_HINT(tp);
tcp_sbrcv_grow(tp, &so->so_rcv, &to, tlen);
so_recv_data_stat(so, m, 0);
m_adj(m, drop_hdrlen);
if (sbappendstream(&so->so_rcv, m))
sorwakeup(so);
#if INET6
if (isipv6) {
KERNEL_DEBUG(DBG_LAYER_END, ((th->th_dport << 16) | th->th_sport),
(((ip6->ip6_src.s6_addr16[0]) << 16) | (ip6->ip6_dst.s6_addr16[0])),
th->th_seq, th->th_ack, th->th_win);
}
else
#endif
{
KERNEL_DEBUG(DBG_LAYER_END, ((th->th_dport << 16) | th->th_sport),
(((ip->ip_src.s_addr & 0xffff) << 16) | (ip->ip_dst.s_addr & 0xffff)),
th->th_seq, th->th_ack, th->th_win);
}
TCP_INC_VAR(tp->t_unacksegs, nlropkts);
if (DELAY_ACK(tp, th)) {
if ((tp->t_flags & TF_DELACK) == 0) {
tp->t_flags |= TF_DELACK;
tp->t_timer[TCPT_DELACK] = OFFSET_FROM_START(tp, tcp_delack);
}
} else {
tp->t_flags |= TF_ACKNOW;
tcp_output(tp);
}
tcp_check_timer_state(tp);
tcp_unlock(so, 1, 0);
KERNEL_DEBUG(DBG_FNC_TCP_INPUT | DBG_FUNC_END,0,0,0,0,0);
return;
}
}
lck_mtx_assert(&((struct inpcb *)so->so_pcb)->inpcb_mtx, LCK_MTX_ASSERT_OWNED);
{ int win;
win = tcp_sbspace(tp);
if (win < 0)
win = 0;
else {
if (tp->t_flags & TF_SLOWLINK && slowlink_wsize > 0)
win = min(win, slowlink_wsize);
}
tp->rcv_wnd = imax(win, (int)(tp->rcv_adv - tp->rcv_nxt));
}
switch (tp->t_state) {
case TCPS_LISTEN: {
register struct sockaddr_in *sin;
#if INET6
register struct sockaddr_in6 *sin6;
#endif
lck_mtx_assert(&((struct inpcb *)so->so_pcb)->inpcb_mtx, LCK_MTX_ASSERT_OWNED);
#if INET6
if (isipv6) {
MALLOC(sin6, struct sockaddr_in6 *, sizeof *sin6,
M_SONAME, M_NOWAIT);
if (sin6 == NULL)
goto drop;
bzero(sin6, sizeof(*sin6));
sin6->sin6_family = AF_INET6;
sin6->sin6_len = sizeof(*sin6);
sin6->sin6_addr = ip6->ip6_src;
sin6->sin6_port = th->th_sport;
laddr6 = inp->in6p_laddr;
if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr))
inp->in6p_laddr = ip6->ip6_dst;
if (in6_pcbconnect(inp, (struct sockaddr *)sin6,
proc0)) {
inp->in6p_laddr = laddr6;
FREE(sin6, M_SONAME);
goto drop;
}
FREE(sin6, M_SONAME);
} else
#endif
{
lck_mtx_assert(&((struct inpcb *)so->so_pcb)->inpcb_mtx, LCK_MTX_ASSERT_OWNED);
MALLOC(sin, struct sockaddr_in *, sizeof *sin, M_SONAME,
M_NOWAIT);
if (sin == NULL)
goto drop;
sin->sin_family = AF_INET;
sin->sin_len = sizeof(*sin);
sin->sin_addr = ip->ip_src;
sin->sin_port = th->th_sport;
bzero((caddr_t)sin->sin_zero, sizeof(sin->sin_zero));
laddr = inp->inp_laddr;
if (inp->inp_laddr.s_addr == INADDR_ANY)
inp->inp_laddr = ip->ip_dst;
if (in_pcbconnect(inp, (struct sockaddr *)sin, proc0, NULL)) {
inp->inp_laddr = laddr;
FREE(sin, M_SONAME);
goto drop;
}
FREE(sin, M_SONAME);
}
tcp_dooptions(tp, optp, optlen, th, &to, ifscope);
if (tp->sack_enable) {
if (!(to.to_flags & TOF_SACK))
tp->sack_enable = 0;
else
tp->t_flags |= TF_SACK_PERMIT;
}
if (iss)
tp->iss = iss;
else {
tp->iss = tcp_new_isn(tp);
}
tp->irs = th->th_seq;
tcp_sendseqinit(tp);
tcp_rcvseqinit(tp);
tp->snd_recover = tp->snd_una;
tp->snd_wnd = tiwin;
tp->t_flags |= TF_ACKNOW;
tp->t_unacksegs = 0;
DTRACE_TCP4(state__change, void, NULL, struct inpcb *, inp,
struct tcpcb *, tp, int32_t, TCPS_SYN_RECEIVED);
tp->t_state = TCPS_SYN_RECEIVED;
tp->t_timer[TCPT_KEEP] = OFFSET_FROM_START(tp,
tp->t_keepinit ? tp->t_keepinit : tcp_keepinit);
dropsocket = 0;
if (inp->inp_flowhash == 0)
inp->inp_flowhash = inp_calc_flowhash(inp);
so->so_flags &= ~(SOF_INCOMP_INPROGRESS);
tcpstat.tcps_accepts++;
if ((thflags & (TH_ECE | TH_CWR)) == (TH_ECE | TH_CWR)) {
tp->ecn_flags |= (TE_SETUPRECEIVED | TE_SENDIPECT);
}
#if CONFIG_IFEF_NOWINDOWSCALE
if (tcp_obey_ifef_nowindowscale && m->m_pkthdr.rcvif != NULL &&
(m->m_pkthdr.rcvif->if_eflags & IFEF_NOWINDOWSCALE)) {
tp->t_flags &= ~TF_REQ_SCALE;
}
#endif
goto trimthenstep6;
}
case TCPS_SYN_RECEIVED:
if ((thflags & TH_ACK) &&
(SEQ_LEQ(th->th_ack, tp->snd_una) ||
SEQ_GT(th->th_ack, tp->snd_max))) {
rstreason = BANDLIM_RST_OPENPORT;
if (ifp != NULL && ifp->if_tcp_stat != NULL)
atomic_add_64(&ifp->if_tcp_stat->ooopacket, 1);
goto dropwithreset;
}
break;
case TCPS_SYN_SENT:
if ((thflags & TH_ACK) &&
(SEQ_LEQ(th->th_ack, tp->iss) ||
SEQ_GT(th->th_ack, tp->snd_max))) {
rstreason = BANDLIM_UNLIMITED;
if (ifp != NULL && ifp->if_tcp_stat != NULL)
atomic_add_64(&ifp->if_tcp_stat->ooopacket, 1);
goto dropwithreset;
}
if (thflags & TH_RST) {
if ((thflags & TH_ACK) != 0) {
soevent(so,
(SO_FILT_HINT_LOCKED |
SO_FILT_HINT_CONNRESET));
tp = tcp_drop(tp, ECONNREFUSED);
postevent(so, 0, EV_RESET);
}
goto drop;
}
if ((thflags & TH_SYN) == 0)
goto drop;
tp->snd_wnd = th->th_win;
tp->irs = th->th_seq;
tcp_rcvseqinit(tp);
if (thflags & TH_ACK) {
tcpstat.tcps_connects++;
if ((thflags & (TH_ECE | TH_CWR)) == (TH_ECE)) {
tp->ecn_flags |= TE_SETUPRECEIVED;
}
else {
tp->ecn_flags &= ~TE_SENDIPECT;
}
#if CONFIG_MACF_NET && CONFIG_MACF_SOCKET
mac_socketpeer_label_associate_mbuf(m, so);
#endif
if ((tp->t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) ==
(TF_RCVD_SCALE|TF_REQ_SCALE)) {
tp->snd_scale = tp->requested_s_scale;
tp->rcv_scale = tp->request_r_scale;
}
tp->rcv_adv += min(tp->rcv_wnd, TCP_MAXWIN << tp->rcv_scale);
tp->snd_una++;
TCP_INC_VAR(tp->t_unacksegs, nlropkts);
if (DELAY_ACK(tp, th) && tlen != 0 ) {
if ((tp->t_flags & TF_DELACK) == 0) {
tp->t_flags |= TF_DELACK;
tp->t_timer[TCPT_DELACK] = OFFSET_FROM_START(tp, tcp_delack);
}
}
else {
tp->t_flags |= TF_ACKNOW;
}
tp->t_starttime = tcp_now;
tcp_sbrcv_tstmp_check(tp);
if (tp->t_flags & TF_NEEDFIN) {
DTRACE_TCP4(state__change, void, NULL, struct inpcb *, inp,
struct tcpcb *, tp, int32_t, TCPS_FIN_WAIT_1);
tp->t_state = TCPS_FIN_WAIT_1;
tp->t_flags &= ~TF_NEEDFIN;
thflags &= ~TH_SYN;
} else {
DTRACE_TCP4(state__change, void, NULL, struct inpcb *, inp,
struct tcpcb *, tp, int32_t, TCPS_ESTABLISHED);
tp->t_state = TCPS_ESTABLISHED;
tp->t_timer[TCPT_KEEP] = OFFSET_FROM_START(tp, TCP_KEEPIDLE(tp));
if (nstat_collect)
nstat_route_connect_success(tp->t_inpcb->inp_route.ro_rt);
}
isconnected = TRUE;
} else {
tp->t_flags |= TF_ACKNOW;
tp->t_timer[TCPT_REXMT] = 0;
DTRACE_TCP4(state__change, void, NULL, struct inpcb *, inp,
struct tcpcb *, tp, int32_t, TCPS_SYN_RECEIVED);
tp->t_state = TCPS_SYN_RECEIVED;
}
trimthenstep6:
th->th_seq++;
if (tlen > tp->rcv_wnd) {
todrop = tlen - tp->rcv_wnd;
m_adj(m, -todrop);
tlen = tp->rcv_wnd;
thflags &= ~TH_FIN;
tcpstat.tcps_rcvpackafterwin++;
tcpstat.tcps_rcvbyteafterwin += todrop;
}
tp->snd_wl1 = th->th_seq - 1;
tp->rcv_up = th->th_seq;
if (thflags & TH_ACK)
goto process_ACK;
goto step6;
case TCPS_LAST_ACK:
case TCPS_CLOSING:
case TCPS_TIME_WAIT:
break;
case TCPS_ESTABLISHED:
if (thflags & TH_SYN)
goto dropafterack;
break;
}
if (thflags & TH_RST) {
if ((SEQ_GEQ(th->th_seq, tp->last_ack_sent) &&
SEQ_LT(th->th_seq, tp->last_ack_sent + tp->rcv_wnd)) ||
(tp->rcv_wnd == 0 &&
((tp->last_ack_sent == th->th_seq) || ((tp->last_ack_sent -1) == th->th_seq)))) {
switch (tp->t_state) {
case TCPS_SYN_RECEIVED:
if (ifp != NULL && ifp->if_tcp_stat != NULL)
atomic_add_64(&ifp->if_tcp_stat->rstinsynrcv, 1);
so->so_error = ECONNREFUSED;
goto close;
case TCPS_ESTABLISHED:
if (tp->last_ack_sent != th->th_seq) {
tcpstat.tcps_badrst++;
goto drop;
}
case TCPS_FIN_WAIT_1:
case TCPS_CLOSE_WAIT:
case TCPS_FIN_WAIT_2:
so->so_error = ECONNRESET;
close:
postevent(so, 0, EV_RESET);
DTRACE_TCP4(state__change, void, NULL, struct inpcb *, inp,
struct tcpcb *, tp, int32_t, TCPS_CLOSED);
soevent(so,
(SO_FILT_HINT_LOCKED |
SO_FILT_HINT_CONNRESET));
tp->t_state = TCPS_CLOSED;
tcpstat.tcps_drops++;
tp = tcp_close(tp);
break;
case TCPS_CLOSING:
case TCPS_LAST_ACK:
tp = tcp_close(tp);
break;
case TCPS_TIME_WAIT:
break;
}
}
goto drop;
}
if ((to.to_flags & TOF_TS) != 0 && tp->ts_recent &&
TSTMP_LT(to.to_tsval, tp->ts_recent)) {
if ((int)(tcp_now - tp->ts_recent_age) > TCP_PAWS_IDLE) {
tp->ts_recent = 0;
} else {
tcpstat.tcps_rcvduppack++;
tcpstat.tcps_rcvdupbyte += tlen;
tcpstat.tcps_pawsdrop++;
if (nstat_collect) {
nstat_route_rx(tp->t_inpcb->inp_route.ro_rt, 1, tlen, NSTAT_RX_FLAG_DUPLICATE);
locked_add_64(&inp->inp_stat->rxpackets, 1);
locked_add_64(&inp->inp_stat->rxbytes, tlen);
tp->t_stat.rxduplicatebytes += tlen;
}
if (tlen)
goto dropafterack;
goto drop;
}
}
if (tp->t_state == TCPS_SYN_RECEIVED && SEQ_LT(th->th_seq, tp->irs)) {
rstreason = BANDLIM_RST_OPENPORT;
if (ifp != NULL && ifp->if_tcp_stat != NULL)
atomic_add_64(&ifp->if_tcp_stat->dospacket, 1);
goto dropwithreset;
}
todrop = tp->rcv_nxt - th->th_seq;
if (todrop > 0) {
if (thflags & TH_SYN) {
thflags &= ~TH_SYN;
th->th_seq++;
if (th->th_urp > 1)
th->th_urp--;
else
thflags &= ~TH_URG;
todrop--;
}
if (todrop > tlen
|| (todrop == tlen && (thflags & TH_FIN) == 0)) {
thflags &= ~TH_FIN;
tp->t_flags |= TF_ACKNOW;
if (todrop == 1) {
soevent(so, SO_FILT_HINT_LOCKED |
SO_FILT_HINT_KEEPALIVE);
}
todrop = tlen;
tcpstat.tcps_rcvduppack++;
tcpstat.tcps_rcvdupbyte += todrop;
} else {
tcpstat.tcps_rcvpartduppack++;
tcpstat.tcps_rcvpartdupbyte += todrop;
}
if (nstat_collect) {
nstat_route_rx(tp->t_inpcb->inp_route.ro_rt, 1, todrop, NSTAT_RX_FLAG_DUPLICATE);
locked_add_64(&inp->inp_stat->rxpackets, 1);
locked_add_64(&inp->inp_stat->rxbytes, todrop);
tp->t_stat.rxduplicatebytes += todrop;
}
drop_hdrlen += todrop;
th->th_seq += todrop;
tlen -= todrop;
if (th->th_urp > todrop)
th->th_urp -= todrop;
else {
thflags &= ~TH_URG;
th->th_urp = 0;
}
}
if ((so->so_state & SS_NOFDREF) &&
tp->t_state > TCPS_CLOSE_WAIT && tlen) {
tp = tcp_close(tp);
tcpstat.tcps_rcvafterclose++;
rstreason = BANDLIM_UNLIMITED;
if (ifp != NULL && ifp->if_tcp_stat != NULL)
atomic_add_64(&ifp->if_tcp_stat->cleanup, 1);
goto dropwithreset;
}
todrop = (th->th_seq+tlen) - (tp->rcv_nxt+tp->rcv_wnd);
if (todrop > 0) {
tcpstat.tcps_rcvpackafterwin++;
if (todrop >= tlen) {
tcpstat.tcps_rcvbyteafterwin += tlen;
if (thflags & TH_SYN &&
tp->t_state == TCPS_TIME_WAIT &&
SEQ_GT(th->th_seq, tp->rcv_nxt)) {
iss = tcp_new_isn(tp);
tp = tcp_close(tp);
tcp_unlock(so, 1, 0);
goto findpcb;
}
if (tp->rcv_wnd == 0 && th->th_seq == tp->rcv_nxt) {
tp->t_flags |= TF_ACKNOW;
tcpstat.tcps_rcvwinprobe++;
} else
goto dropafterack;
} else
tcpstat.tcps_rcvbyteafterwin += todrop;
m_adj(m, -todrop);
tlen -= todrop;
thflags &= ~(TH_PUSH|TH_FIN);
}
if ((to.to_flags & TOF_TS) != 0 &&
SEQ_LEQ(th->th_seq, tp->last_ack_sent) &&
SEQ_LEQ(tp->last_ack_sent, th->th_seq + tlen +
((thflags & (TH_SYN|TH_FIN)) != 0))) {
tp->ts_recent_age = tcp_now;
tp->ts_recent = to.to_tsval;
}
if (thflags & TH_SYN) {
tp = tcp_drop(tp, ECONNRESET);
rstreason = BANDLIM_UNLIMITED;
postevent(so, 0, EV_RESET);
if (ifp != NULL && ifp->if_tcp_stat != NULL)
atomic_add_64(&ifp->if_tcp_stat->synwindow, 1);
goto dropwithreset;
}
if ((thflags & TH_ACK) == 0) {
if (tp->t_state == TCPS_SYN_RECEIVED ||
(tp->t_flags & TF_NEEDSYN))
goto step6;
else if (tp->t_flags & TF_ACKNOW)
goto dropafterack;
else
goto drop;
}
switch (tp->t_state) {
case TCPS_SYN_RECEIVED:
tcpstat.tcps_connects++;
if ((tp->t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) ==
(TF_RCVD_SCALE|TF_REQ_SCALE)) {
tp->snd_scale = tp->requested_s_scale;
tp->rcv_scale = tp->request_r_scale;
tp->snd_wnd = th->th_win << tp->snd_scale;
tiwin = tp->snd_wnd;
}
tp->t_starttime = tcp_now;
tcp_sbrcv_tstmp_check(tp);
if (tp->t_flags & TF_NEEDFIN) {
DTRACE_TCP4(state__change, void, NULL, struct inpcb *, inp,
struct tcpcb *, tp, int32_t, TCPS_FIN_WAIT_1);
tp->t_state = TCPS_FIN_WAIT_1;
tp->t_flags &= ~TF_NEEDFIN;
} else {
DTRACE_TCP4(state__change, void, NULL, struct inpcb *, inp,
struct tcpcb *, tp, int32_t, TCPS_ESTABLISHED);
tp->t_state = TCPS_ESTABLISHED;
tp->t_timer[TCPT_KEEP] = OFFSET_FROM_START(tp, TCP_KEEPIDLE(tp));
if (nstat_collect)
nstat_route_connect_success(tp->t_inpcb->inp_route.ro_rt);
}
if (tlen == 0 && (thflags & TH_FIN) == 0)
(void) tcp_reass(tp, (struct tcphdr *)0, &tlen,
(struct mbuf *)0);
tp->snd_wl1 = th->th_seq - 1;
isconnected = TRUE;
case TCPS_ESTABLISHED:
case TCPS_FIN_WAIT_1:
case TCPS_FIN_WAIT_2:
case TCPS_CLOSE_WAIT:
case TCPS_CLOSING:
case TCPS_LAST_ACK:
case TCPS_TIME_WAIT:
if (SEQ_GT(th->th_ack, tp->snd_max)) {
tcpstat.tcps_rcvacktoomuch++;
goto dropafterack;
}
if (tp->sack_enable &&
(to.to_nsacks > 0 || !TAILQ_EMPTY(&tp->snd_holes)))
tcp_sack_doack(tp, &to, th->th_ack);
if (SEQ_LEQ(th->th_ack, tp->snd_una)) {
if (tlen == 0 && tiwin == tp->snd_wnd) {
tcpstat.tcps_rcvdupack++;
if (tp->t_timer[TCPT_REXMT] == 0 ||
th->th_ack != tp->snd_una)
tp->t_dupacks = 0;
else if (++tp->t_dupacks > tcprexmtthresh ||
IN_FASTRECOVERY(tp)) {
if (tp->sack_enable && IN_FASTRECOVERY(tp)) {
int awnd;
awnd = (tp->snd_nxt - tp->snd_fack) +
tp->sackhint.sack_bytes_rexmit;
if (awnd < tp->snd_ssthresh) {
tp->snd_cwnd += tp->t_maxseg;
if (tp->snd_cwnd > tp->snd_ssthresh)
tp->snd_cwnd = tp->snd_ssthresh;
}
} else
tp->snd_cwnd += tp->t_maxseg;
DTRACE_TCP5(cc, void, NULL, struct inpcb *, inp,
struct tcpcb *, tp, struct tcphdr *, th,
int32_t, TCP_CC_IN_FASTRECOVERY);
(void) tcp_output(tp);
goto drop;
} else if (tp->t_dupacks == tcprexmtthresh) {
tcp_seq onxt = tp->snd_nxt;
if (tp->sack_enable) {
if (IN_FASTRECOVERY(tp)) {
tp->t_dupacks = 0;
break;
}
} else {
if (SEQ_LEQ(th->th_ack,
tp->snd_recover)) {
tp->t_dupacks = 0;
break;
}
}
if (CC_ALGO(tp)->pre_fr != NULL)
CC_ALGO(tp)->pre_fr(tp);
ENTER_FASTRECOVERY(tp);
tp->snd_recover = tp->snd_max;
tp->t_timer[TCPT_REXMT] = 0;
tp->t_rtttime = 0;
if ((tp->ecn_flags & TE_ECN_ON) == TE_ECN_ON) {
tp->ecn_flags |= TE_SENDCWR;
}
if (tp->sack_enable) {
tcpstat.tcps_sack_recovery_episode++;
tp->sack_newdata = tp->snd_nxt;
tp->snd_cwnd = tp->t_maxseg;
DTRACE_TCP5(cc, void, NULL, struct inpcb *, inp,
struct tcpcb *, tp, struct tcphdr *, th,
int32_t, TCP_CC_ENTER_FASTRECOVERY);
(void) tcp_output(tp);
goto drop;
}
tp->snd_nxt = th->th_ack;
tp->snd_cwnd = tp->t_maxseg;
(void) tcp_output(tp);
tp->snd_cwnd = tp->snd_ssthresh +
tp->t_maxseg * tp->t_dupacks;
if (SEQ_GT(onxt, tp->snd_nxt))
tp->snd_nxt = onxt;
DTRACE_TCP5(cc, void, NULL, struct inpcb *, inp,
struct tcpcb *, tp, struct tcphdr *, th,
int32_t, TCP_CC_ENTER_FASTRECOVERY);
goto drop;
}
} else
tp->t_dupacks = 0;
break;
}
if (IN_FASTRECOVERY(tp)) {
if (SEQ_LT(th->th_ack, tp->snd_recover)) {
if (tp->sack_enable)
tcp_sack_partialack(tp, th);
else
tcp_newreno_partial_ack(tp, th);
DTRACE_TCP5(cc, void, NULL, struct inpcb *, inp,
struct tcpcb *, tp, struct tcphdr *, th,
int32_t, TCP_CC_PARTIAL_ACK);
} else {
EXIT_FASTRECOVERY(tp);
if (CC_ALGO(tp)->post_fr != NULL)
CC_ALGO(tp)->post_fr(tp, th);
tp->t_dupacks = 0;
DTRACE_TCP5(cc, void, NULL, struct inpcb *, inp,
struct tcpcb *, tp, struct tcphdr *, th,
int32_t, TCP_CC_EXIT_FASTRECOVERY);
}
} else {
tp->t_dupacks = 0;
}
if (tp->t_flags & TF_NEEDSYN) {
tp->t_flags &= ~TF_NEEDSYN;
tp->snd_una++;
if ((tp->t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) ==
(TF_RCVD_SCALE|TF_REQ_SCALE)) {
tp->snd_scale = tp->requested_s_scale;
tp->rcv_scale = tp->request_r_scale;
}
}
process_ACK:
acked = th->th_ack - tp->snd_una;
tcpstat.tcps_rcvackpack++;
tcpstat.tcps_rcvackbyte += acked;
if (tp->t_rxtshift == 1 &&
TSTMP_LT(tcp_now, tp->t_badrxtwin)) {
++tcpstat.tcps_sndrexmitbad;
tp->snd_cwnd = tp->snd_cwnd_prev;
tp->snd_ssthresh = tp->snd_ssthresh_prev;
tp->snd_recover = tp->snd_recover_prev;
if (tp->t_flags & TF_WASFRECOVERY)
ENTER_FASTRECOVERY(tp);
tp->snd_nxt = tp->snd_max;
tp->t_badrxtwin = 0;
tp->t_rxtshift = 0;
tp->rxt_start = 0;
tcp_bad_rexmt_fix_sndbuf(tp);
DTRACE_TCP5(cc, void, NULL, struct inpcb *, inp,
struct tcpcb *, tp, struct tcphdr *, th,
int32_t, TCP_CC_BAD_REXMT_RECOVERY);
}
if (((to.to_flags & TOF_TS) != 0) && (to.to_tsecr != 0) &&
TSTMP_GEQ(tcp_now, to.to_tsecr)) {
tcp_xmit_timer(tp, tcp_now - to.to_tsecr);
} else if (tp->t_rtttime && SEQ_GT(th->th_ack, tp->t_rtseq)) {
tcp_xmit_timer(tp, tcp_now - tp->t_rtttime);
}
if (th->th_ack == tp->snd_max) {
tp->t_timer[TCPT_REXMT] = 0;
needoutput = 1;
} else if (tp->t_timer[TCPT_PERSIST] == 0)
tp->t_timer[TCPT_REXMT] = OFFSET_FROM_START(tp, tp->t_rxtcur);
if (acked == 0)
goto step6;
if ((thflags & TH_ECE) != 0 &&
((tp->ecn_flags & TE_ECN_ON) == TE_ECN_ON)) {
if (!tp->sack_enable && !IN_FASTRECOVERY(tp) &&
SEQ_GEQ(th->th_ack, tp->snd_recover)) {
tcp_reduce_congestion_window(tp);
DTRACE_TCP5(cc, void, NULL, struct inpcb *, inp,
struct tcpcb *, tp, struct tcphdr *, th,
int32_t, TCP_CC_ECN_RCVD);
}
}
if (!IN_FASTRECOVERY(tp)) {
if (CC_ALGO(tp)->ack_rcvd != NULL)
CC_ALGO(tp)->ack_rcvd(tp, th);
DTRACE_TCP5(cc, void, NULL, struct inpcb *, inp,
struct tcpcb *, tp, struct tcphdr *, th,
int32_t, TCP_CC_ACK_RCVD);
}
if (acked > so->so_snd.sb_cc) {
tp->snd_wnd -= so->so_snd.sb_cc;
sbdrop(&so->so_snd, (int)so->so_snd.sb_cc);
ourfinisacked = 1;
} else {
sbdrop(&so->so_snd, acked);
tcp_sbsnd_trim(&so->so_snd);
tp->snd_wnd -= acked;
ourfinisacked = 0;
}
if ( !IN_FASTRECOVERY(tp) &&
SEQ_GT(tp->snd_una, tp->snd_recover) &&
SEQ_LEQ(th->th_ack, tp->snd_recover))
tp->snd_recover = th->th_ack - 1;
if (IN_FASTRECOVERY(tp) &&
SEQ_GEQ(th->th_ack, tp->snd_recover))
EXIT_FASTRECOVERY(tp);
tp->snd_una = th->th_ack;
if (tp->sack_enable) {
if (SEQ_GT(tp->snd_una, tp->snd_recover))
tp->snd_recover = tp->snd_una;
}
if (SEQ_LT(tp->snd_nxt, tp->snd_una))
tp->snd_nxt = tp->snd_una;
if ((tp->t_flagsext & TF_MEASURESNDBW) != 0 &&
tp->t_bwmeas != NULL)
tcp_bwmeas_check(tp);
sowwakeup(so);
switch (tp->t_state) {
case TCPS_FIN_WAIT_1:
if (ourfinisacked) {
if (so->so_state & SS_CANTRCVMORE) {
add_to_time_wait(tp, tcp_maxidle);
isconnected = FALSE;
isdisconnected = TRUE;
}
DTRACE_TCP4(state__change, void, NULL, struct inpcb *, inp,
struct tcpcb *, tp, int32_t, TCPS_FIN_WAIT_2);
tp->t_state = TCPS_FIN_WAIT_2;
}
tp->t_flags |= TF_ACKNOW;
break;
case TCPS_CLOSING:
if (ourfinisacked) {
DTRACE_TCP4(state__change, void, NULL, struct inpcb *, inp,
struct tcpcb *, tp, int32_t, TCPS_TIME_WAIT);
tp->t_state = TCPS_TIME_WAIT;
tcp_canceltimers(tp);
if (tp->cc_recv != 0 &&
((int)(tcp_now - tp->t_starttime)) < tcp_msl)
add_to_time_wait(tp, tp->t_rxtcur * TCPTV_TWTRUNC);
else
add_to_time_wait(tp, 2 * tcp_msl);
isconnected = FALSE;
isdisconnected = TRUE;
}
tp->t_flags |= TF_ACKNOW;
break;
case TCPS_LAST_ACK:
if (ourfinisacked) {
tp = tcp_close(tp);
goto drop;
}
break;
case TCPS_TIME_WAIT:
add_to_time_wait(tp, 2 * tcp_msl);
goto dropafterack;
}
}
step6:
if ((thflags & TH_ACK) &&
(SEQ_LT(tp->snd_wl1, th->th_seq) ||
(tp->snd_wl1 == th->th_seq && (SEQ_LT(tp->snd_wl2, th->th_ack) ||
(tp->snd_wl2 == th->th_ack && tiwin > tp->snd_wnd))))) {
if (tlen == 0 &&
tp->snd_wl2 == th->th_ack && tiwin > tp->snd_wnd)
tcpstat.tcps_rcvwinupd++;
tp->snd_wnd = tiwin;
tp->snd_wl1 = th->th_seq;
tp->snd_wl2 = th->th_ack;
if (tp->snd_wnd > tp->max_sndwnd)
tp->max_sndwnd = tp->snd_wnd;
needoutput = 1;
}
if ((thflags & TH_URG) && th->th_urp &&
TCPS_HAVERCVDFIN(tp->t_state) == 0) {
if (th->th_urp + so->so_rcv.sb_cc > sb_max) {
th->th_urp = 0;
thflags &= ~TH_URG;
goto dodata;
}
if (SEQ_GT(th->th_seq+th->th_urp, tp->rcv_up)) {
tp->rcv_up = th->th_seq + th->th_urp;
so->so_oobmark = so->so_rcv.sb_cc +
(tp->rcv_up - tp->rcv_nxt) - 1;
if (so->so_oobmark == 0) {
so->so_state |= SS_RCVATMARK;
postevent(so, 0, EV_OOB);
}
sohasoutofband(so);
tp->t_oobflags &= ~(TCPOOB_HAVEDATA | TCPOOB_HADDATA);
}
if (th->th_urp <= (u_int32_t)tlen
#if SO_OOBINLINE
&& (so->so_options & SO_OOBINLINE) == 0
#endif
)
tcp_pulloutofband(so, th, m,
drop_hdrlen);
} else {
if (SEQ_GT(tp->rcv_nxt, tp->rcv_up))
tp->rcv_up = tp->rcv_nxt;
}
dodata:
if (isconnected) {
soisconnected(so);
} else if (isdisconnected) {
soisdisconnected(so);
}
if (inp->inp_state == INPCB_STATE_DEAD) {
goto drop;
}
if ((tlen || (thflags & TH_FIN)) &&
TCPS_HAVERCVDFIN(tp->t_state) == 0) {
tcp_seq save_start = th->th_seq;
tcp_seq save_end = th->th_seq + tlen;
m_adj(m, drop_hdrlen);
if (th->th_seq == tp->rcv_nxt &&
LIST_EMPTY(&tp->t_segq) &&
TCPS_HAVEESTABLISHED(tp->t_state)) {
TCP_INC_VAR(tp->t_unacksegs, nlropkts);
if (DELAY_ACK(tp, th) &&
((tp->t_flags & TF_ACKNOW) == 0) ) {
if ((tp->t_flags & TF_DELACK) == 0) {
tp->t_flags |= TF_DELACK;
tp->t_timer[TCPT_DELACK] = OFFSET_FROM_START(tp, tcp_delack);
}
}
else {
tp->t_flags |= TF_ACKNOW;
}
tp->rcv_nxt += tlen;
thflags = th->th_flags & TH_FIN;
TCP_INC_VAR(tcpstat.tcps_rcvpack, nlropkts);
tcpstat.tcps_rcvbyte += tlen;
if (nstat_collect) {
if (m->m_pkthdr.aux_flags & MAUXF_SW_LRO_PKT) {
locked_add_64(&inp->inp_stat->rxpackets, m->m_pkthdr.lro_npkts);
} else {
locked_add_64(&inp->inp_stat->rxpackets, 1);
}
locked_add_64(&inp->inp_stat->rxbytes, tlen);
}
ND6_HINT(tp);
tcp_sbrcv_grow(tp, &so->so_rcv, &to, tlen);
so_recv_data_stat(so, m, drop_hdrlen);
if (sbappendstream(&so->so_rcv, m))
sorwakeup(so);
} else {
thflags = tcp_reass(tp, th, &tlen, m);
tp->t_flags |= TF_ACKNOW;
}
if (tlen > 0 && tp->sack_enable)
tcp_update_sack_list(tp, save_start, save_end);
if (tp->t_flags & TF_DELACK)
{
#if INET6
if (isipv6) {
KERNEL_DEBUG(DBG_LAYER_END, ((th->th_dport << 16) | th->th_sport),
(((ip6->ip6_src.s6_addr16[0]) << 16) | (ip6->ip6_dst.s6_addr16[0])),
th->th_seq, th->th_ack, th->th_win);
}
else
#endif
{
KERNEL_DEBUG(DBG_LAYER_END, ((th->th_dport << 16) | th->th_sport),
(((ip->ip_src.s_addr & 0xffff) << 16) | (ip->ip_dst.s_addr & 0xffff)),
th->th_seq, th->th_ack, th->th_win);
}
}
} else {
m_freem(m);
thflags &= ~TH_FIN;
}
if (thflags & TH_FIN) {
if (TCPS_HAVERCVDFIN(tp->t_state) == 0) {
socantrcvmore(so);
postevent(so, 0, EV_FIN);
TCP_INC_VAR(tp->t_unacksegs, nlropkts);
if (DELAY_ACK(tp, th) && (tp->t_flags & TF_NEEDSYN)) {
if ((tp->t_flags & TF_DELACK) == 0) {
tp->t_flags |= TF_DELACK;
tp->t_timer[TCPT_DELACK] = OFFSET_FROM_START(tp, tcp_delack);
}
}
else {
tp->t_flags |= TF_ACKNOW;
}
tp->rcv_nxt++;
}
switch (tp->t_state) {
case TCPS_SYN_RECEIVED:
tp->t_starttime = tcp_now;
case TCPS_ESTABLISHED:
DTRACE_TCP4(state__change, void, NULL, struct inpcb *, inp,
struct tcpcb *, tp, int32_t, TCPS_CLOSE_WAIT);
tp->t_state = TCPS_CLOSE_WAIT;
break;
case TCPS_FIN_WAIT_1:
DTRACE_TCP4(state__change, void, NULL, struct inpcb *, inp,
struct tcpcb *, tp, int32_t, TCPS_CLOSING);
tp->t_state = TCPS_CLOSING;
break;
case TCPS_FIN_WAIT_2:
DTRACE_TCP4(state__change, void, NULL, struct inpcb *, inp,
struct tcpcb *, tp, int32_t, TCPS_TIME_WAIT);
tp->t_state = TCPS_TIME_WAIT;
tcp_canceltimers(tp);
if (tp->cc_recv != 0 &&
((int)(tcp_now - tp->t_starttime)) < tcp_msl) {
add_to_time_wait(tp, tp->t_rxtcur * TCPTV_TWTRUNC);
tp->t_flags |= TF_ACKNOW;
tp->t_unacksegs = 0;
}
else
add_to_time_wait(tp, 2 * tcp_msl);
soisdisconnected(so);
break;
case TCPS_TIME_WAIT:
add_to_time_wait(tp, 2 * tcp_msl);
break;
}
}
#if TCPDEBUG
if (so->so_options & SO_DEBUG)
tcp_trace(TA_INPUT, ostate, tp, (void *)tcp_saveipgen,
&tcp_savetcp, 0);
#endif
if (needoutput || (tp->t_flags & TF_ACKNOW)) {
(void) tcp_output(tp);
}
tcp_check_timer_state(tp);
tcp_unlock(so, 1, 0);
KERNEL_DEBUG(DBG_FNC_TCP_INPUT | DBG_FUNC_END,0,0,0,0,0);
return;
dropafterack:
if (tp->t_state == TCPS_SYN_RECEIVED && (thflags & TH_ACK) &&
(SEQ_GT(tp->snd_una, th->th_ack) ||
SEQ_GT(th->th_ack, tp->snd_max)) ) {
rstreason = BANDLIM_RST_OPENPORT;
if (ifp != NULL && ifp->if_tcp_stat != NULL)
atomic_add_64(&ifp->if_tcp_stat->dospacket, 1);
goto dropwithreset;
}
#if TCPDEBUG
if (so->so_options & SO_DEBUG)
tcp_trace(TA_DROP, ostate, tp, (void *)tcp_saveipgen,
&tcp_savetcp, 0);
#endif
m_freem(m);
tp->t_flags |= TF_ACKNOW;
(void) tcp_output(tp);
tcp_unlock(so, 1, 0);
KERNEL_DEBUG(DBG_FNC_TCP_INPUT | DBG_FUNC_END,0,0,0,0,0);
return;
dropwithresetnosock:
nosock = 1;
dropwithreset:
if ((thflags & TH_RST) || m->m_flags & (M_BCAST|M_MCAST))
goto drop;
#if INET6
if (isipv6) {
if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
IN6_IS_ADDR_MULTICAST(&ip6->ip6_src))
goto drop;
} else
#endif
if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) ||
IN_MULTICAST(ntohl(ip->ip_src.s_addr)) ||
ip->ip_src.s_addr == htonl(INADDR_BROADCAST) ||
in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif))
goto drop;
#if ICMP_BANDLIM
if (badport_bandlim(rstreason) < 0)
goto drop;
#endif
#if TCPDEBUG
if (tp == 0 || (tp->t_inpcb->inp_socket->so_options & SO_DEBUG))
tcp_trace(TA_DROP, ostate, tp, (void *)tcp_saveipgen,
&tcp_savetcp, 0);
#endif
if (thflags & TH_ACK)
tcp_respond(tp, mtod(m, void *), th, m, (tcp_seq)0, th->th_ack,
TH_RST, ifscope, nocell);
else {
if (thflags & TH_SYN)
tlen++;
tcp_respond(tp, mtod(m, void *), th, m, th->th_seq+tlen,
(tcp_seq)0, TH_RST|TH_ACK, ifscope, nocell);
}
if (dropsocket) {
(void) soabort(so);
tcp_unlock(so, 1, 0);
}
else if ((inp != NULL) && (nosock == 0)) {
tcp_unlock(so, 1, 0);
}
KERNEL_DEBUG(DBG_FNC_TCP_INPUT | DBG_FUNC_END,0,0,0,0,0);
return;
dropnosock:
nosock = 1;
drop:
#if TCPDEBUG
if (tp == 0 || (tp->t_inpcb->inp_socket->so_options & SO_DEBUG))
tcp_trace(TA_DROP, ostate, tp, (void *)tcp_saveipgen,
&tcp_savetcp, 0);
#endif
m_freem(m);
if (dropsocket) {
(void) soabort(so);
tcp_unlock(so, 1, 0);
}
else if (nosock == 0) {
tcp_unlock(so, 1, 0);
}
KERNEL_DEBUG(DBG_FNC_TCP_INPUT | DBG_FUNC_END,0,0,0,0,0);
return;
}
static void
tcp_dooptions(tp, cp, cnt, th, to, input_ifscope)
struct tcpcb *tp;
u_char *cp;
int cnt;
struct tcphdr *th;
struct tcpopt *to;
unsigned int input_ifscope;
{
u_short mss = 0;
int opt, optlen;
for (; cnt > 0; cnt -= optlen, cp += optlen) {
opt = cp[0];
if (opt == TCPOPT_EOL)
break;
if (opt == TCPOPT_NOP)
optlen = 1;
else {
if (cnt < 2)
break;
optlen = cp[1];
if (optlen < 2 || optlen > cnt)
break;
}
switch (opt) {
default:
continue;
case TCPOPT_MAXSEG:
if (optlen != TCPOLEN_MAXSEG)
continue;
if (!(th->th_flags & TH_SYN))
continue;
bcopy((char *) cp + 2, (char *) &mss, sizeof(mss));
#if BYTE_ORDER != BIG_ENDIAN
NTOHS(mss);
#endif
break;
case TCPOPT_WINDOW:
if (optlen != TCPOLEN_WINDOW)
continue;
if (!(th->th_flags & TH_SYN))
continue;
tp->t_flags |= TF_RCVD_SCALE;
tp->requested_s_scale = min(cp[2], TCP_MAX_WINSHIFT);
break;
case TCPOPT_TIMESTAMP:
if (optlen != TCPOLEN_TIMESTAMP)
continue;
to->to_flags |= TOF_TS;
bcopy((char *)cp + 2,
(char *)&to->to_tsval, sizeof(to->to_tsval));
#if BYTE_ORDER != BIG_ENDIAN
NTOHL(to->to_tsval);
#endif
bcopy((char *)cp + 6,
(char *)&to->to_tsecr, sizeof(to->to_tsecr));
#if BYTE_ORDER != BIG_ENDIAN
NTOHL(to->to_tsecr);
#endif
if (th->th_flags & TH_SYN) {
tp->t_flags |= TF_RCVD_TSTMP;
tp->ts_recent = to->to_tsval;
tp->ts_recent_age = tcp_now;
}
break;
case TCPOPT_SACK_PERMITTED:
if (!tcp_do_sack ||
optlen != TCPOLEN_SACK_PERMITTED)
continue;
if (th->th_flags & TH_SYN)
to->to_flags |= TOF_SACK;
break;
case TCPOPT_SACK:
if (optlen <= 2 || (optlen - 2) % TCPOLEN_SACK != 0)
continue;
to->to_nsacks = (optlen - 2) / TCPOLEN_SACK;
to->to_sacks = cp + 2;
tcpstat.tcps_sack_rcv_blocks++;
break;
}
}
if (th->th_flags & TH_SYN)
tcp_mss(tp, mss, input_ifscope);
}
static void
tcp_pulloutofband(so, th, m, off)
struct socket *so;
struct tcphdr *th;
register struct mbuf *m;
int off;
{
int cnt = off + th->th_urp - 1;
while (cnt >= 0) {
if (m->m_len > cnt) {
char *cp = mtod(m, caddr_t) + cnt;
struct tcpcb *tp = sototcpcb(so);
tp->t_iobc = *cp;
tp->t_oobflags |= TCPOOB_HAVEDATA;
bcopy(cp+1, cp, (unsigned)(m->m_len - cnt - 1));
m->m_len--;
if (m->m_flags & M_PKTHDR)
m->m_pkthdr.len--;
return;
}
cnt -= m->m_len;
m = m->m_next;
if (m == 0)
break;
}
panic("tcp_pulloutofband");
}
uint32_t
get_base_rtt(struct tcpcb *tp)
{
uint32_t base_rtt = 0, i;
for (i = 0; i < N_RTT_BASE; ++i) {
if (tp->rtt_hist[i] != 0 &&
(base_rtt == 0 || tp->rtt_hist[i] < base_rtt))
base_rtt = tp->rtt_hist[i];
}
return base_rtt;
}
void
update_base_rtt(struct tcpcb *tp, uint32_t rtt)
{
if (++tp->rtt_count >= rtt_samples_per_slot) {
int i=0;
for (i = (N_RTT_BASE-1); i > 0; --i) {
tp->rtt_hist[i] = tp->rtt_hist[i-1];
}
tp->rtt_hist[0] = rtt;
tp->rtt_count = 0;
} else {
tp->rtt_hist[0] = min(tp->rtt_hist[0], rtt);
}
}
static void
tcp_xmit_timer(tp, rtt)
register struct tcpcb *tp;
int rtt;
{
register int delta;
tcpstat.tcps_rttupdated++;
tp->t_rttupdated++;
if (rtt > 0) {
tp->t_rttcur = rtt;
update_base_rtt(tp, rtt);
}
if (tp->t_srtt != 0) {
delta = (rtt << TCP_DELTA_SHIFT)
- (tp->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT));
if ((tp->t_srtt += delta) <= 0)
tp->t_srtt = 1;
if (delta < 0)
delta = -delta;
delta -= tp->t_rttvar >> (TCP_RTTVAR_SHIFT - TCP_DELTA_SHIFT);
if ((tp->t_rttvar += delta) <= 0)
tp->t_rttvar = 1;
if (tp->t_rttbest == 0 ||
tp->t_rttbest > (tp->t_srtt + tp->t_rttvar))
tp->t_rttbest = tp->t_srtt + tp->t_rttvar;
} else {
tp->t_srtt = rtt << TCP_RTT_SHIFT;
tp->t_rttvar = rtt << (TCP_RTTVAR_SHIFT - 1);
}
nstat_route_rtt(tp->t_inpcb->inp_route.ro_rt, tp->t_srtt, tp->t_rttvar);
tp->t_rtttime = 0;
tp->t_rxtshift = 0;
tp->rxt_start = 0;
TCPT_RANGESET(tp->t_rxtcur, TCP_REXMTVAL(tp),
max(tp->t_rttmin, rtt + 2), TCPTV_REXMTMAX,
TCP_ADD_REXMTSLOP(tp));
tp->t_softerror = 0;
}
static inline unsigned int
tcp_maxmtu(struct rtentry *rt)
{
unsigned int maxmtu;
RT_LOCK_ASSERT_HELD(rt);
if (rt->rt_rmx.rmx_mtu == 0)
maxmtu = rt->rt_ifp->if_mtu;
else
maxmtu = MIN(rt->rt_rmx.rmx_mtu, rt->rt_ifp->if_mtu);
return (maxmtu);
}
#if INET6
static inline unsigned int
tcp_maxmtu6(struct rtentry *rt)
{
unsigned int maxmtu;
struct nd_ifinfo *ndi;
RT_LOCK_ASSERT_HELD(rt);
lck_rw_lock_shared(nd_if_rwlock);
if ((ndi = ND_IFINFO(rt->rt_ifp)) != NULL && !ndi->initialized)
ndi = NULL;
if (ndi != NULL)
lck_mtx_lock(&ndi->lock);
if (rt->rt_rmx.rmx_mtu == 0)
maxmtu = IN6_LINKMTU(rt->rt_ifp);
else
maxmtu = MIN(rt->rt_rmx.rmx_mtu, IN6_LINKMTU(rt->rt_ifp));
if (ndi != NULL)
lck_mtx_unlock(&ndi->lock);
lck_rw_done(nd_if_rwlock);
return (maxmtu);
}
#endif
void
tcp_mss(tp, offer, input_ifscope)
struct tcpcb *tp;
int offer;
unsigned int input_ifscope;
{
register struct rtentry *rt;
struct ifnet *ifp;
register int rtt, mss;
u_int32_t bufsize;
struct inpcb *inp;
struct socket *so;
struct rmxp_tao *taop;
int origoffer = offer;
u_int32_t sb_max_corrected;
int isnetlocal = 0;
#if INET6
int isipv6;
int min_protoh;
#endif
inp = tp->t_inpcb;
#if INET6
isipv6 = ((inp->inp_vflag & INP_IPV6) != 0) ? 1 : 0;
min_protoh = isipv6 ? sizeof (struct ip6_hdr) + sizeof (struct tcphdr)
: sizeof (struct tcpiphdr);
#else
#define min_protoh (sizeof (struct tcpiphdr))
#endif
#if INET6
if (isipv6) {
rt = tcp_rtlookup6(inp, input_ifscope);
if (rt != NULL &&
(IN6_IS_ADDR_LOOPBACK(&inp->in6p_faddr) ||
IN6_IS_ADDR_LINKLOCAL(&inp->in6p_faddr) ||
rt->rt_gateway->sa_family == AF_LINK ||
in6_localaddr(&inp->in6p_faddr))) {
tp->t_flags |= TF_LOCAL;
}
}
else
#endif
{
rt = tcp_rtlookup(inp, input_ifscope);
if (rt != NULL &&
(rt->rt_gateway->sa_family == AF_LINK ||
rt->rt_ifp->if_flags & IFF_LOOPBACK ||
in_localaddr(inp->inp_faddr))) {
tp->t_flags |= TF_LOCAL;
}
}
isnetlocal = (tp->t_flags & TF_LOCAL);
if (rt == NULL) {
tp->t_maxopd = tp->t_maxseg =
#if INET6
isipv6 ? tcp_v6mssdflt :
#endif
tcp_mssdflt;
return;
}
ifp = rt->rt_ifp;
if (ifp->if_type == IFT_PPP && slowlink_wsize > 0 &&
ifp->if_baudrate > 9600 && ifp->if_baudrate <= 128000) {
tp->t_flags |= TF_SLOWLINK;
}
so = inp->inp_socket;
taop = rmx_taop(rt->rt_rmx);
if (offer == -1)
offer = taop->tao_mssopt;
if (offer == 0)
offer =
#if INET6
isipv6 ? tcp_v6mssdflt :
#endif
tcp_mssdflt;
else {
offer = max(offer, tcp_minmss);
offer = max(offer, 64);
}
taop->tao_mssopt = offer;
if (tp->t_srtt == 0 && (rtt = rt->rt_rmx.rmx_rtt) != 0) {
tcp_getrt_rtt(tp, rt);
} else {
tp->t_rttmin = isnetlocal ? tcp_TCPTV_MIN : TCPTV_REXMTMIN;
}
#if INET6
mss = (isipv6 ? tcp_maxmtu6(rt) : tcp_maxmtu(rt));
#else
mss = tcp_maxmtu(rt);
#endif
mss -= min_protoh;
if (rt->rt_rmx.rmx_mtu == 0) {
#if INET6
if (isipv6) {
if (!isnetlocal)
mss = min(mss, tcp_v6mssdflt);
} else
#endif
if (!isnetlocal)
mss = min(mss, tcp_mssdflt);
}
mss = min(mss, offer);
tp->t_maxopd = mss;
if ((tp->t_flags & (TF_REQ_TSTMP|TF_NOOPT)) == TF_REQ_TSTMP &&
(origoffer == -1 ||
(tp->t_flags & TF_RCVD_TSTMP) == TF_RCVD_TSTMP))
mss -= TCPOLEN_TSTAMP_APPA;
tp->t_maxseg = mss;
sb_max_corrected = (sb_max * (u_int64_t)MCLBYTES) / (MSIZE + MCLBYTES);
#if RTV_SPIPE
bufsize = rt->rt_rmx.rmx_sendpipe;
if (bufsize < so->so_snd.sb_hiwat)
#endif
bufsize = so->so_snd.sb_hiwat;
if (bufsize < mss)
mss = bufsize;
else {
bufsize = (((bufsize + (u_int64_t)mss - 1) / (u_int64_t)mss) * (u_int64_t)mss);
if (bufsize > sb_max_corrected)
bufsize = sb_max_corrected;
(void)sbreserve(&so->so_snd, bufsize);
}
tp->t_maxseg = mss;
#if RTV_RPIPE
bufsize = rt->rt_rmx.rmx_recvpipe;
if (bufsize < so->so_rcv.sb_hiwat)
#endif
bufsize = so->so_rcv.sb_hiwat;
if (bufsize > mss) {
bufsize = (((bufsize + (u_int64_t)mss - 1) / (u_int64_t)mss) * (u_int64_t)mss);
if (bufsize > sb_max_corrected)
bufsize = sb_max_corrected;
(void)sbreserve(&so->so_rcv, bufsize);
}
set_tcp_stream_priority(so);
if (rt->rt_rmx.rmx_ssthresh) {
tp->snd_ssthresh = max(2 * mss, rt->rt_rmx.rmx_ssthresh);
tcpstat.tcps_usedssthresh++;
} else {
tp->snd_ssthresh = TCP_MAXWIN << TCP_MAX_WINSHIFT;
}
if (CC_ALGO(tp)->cwnd_init != NULL)
CC_ALGO(tp)->cwnd_init(tp);
DTRACE_TCP5(cc, void, NULL, struct inpcb *, tp->t_inpcb, struct tcpcb *, tp,
struct tcphdr *, NULL, int32_t, TCP_CC_CWND_INIT);
RT_UNLOCK(rt);
}
int
tcp_mssopt(tp)
struct tcpcb *tp;
{
struct rtentry *rt;
int mss;
#if INET6
int isipv6;
int min_protoh;
#endif
#if INET6
isipv6 = ((tp->t_inpcb->inp_vflag & INP_IPV6) != 0) ? 1 : 0;
min_protoh = isipv6 ? sizeof (struct ip6_hdr) + sizeof (struct tcphdr)
: sizeof (struct tcpiphdr);
#else
#define min_protoh (sizeof (struct tcpiphdr))
#endif
#if INET6
if (isipv6)
rt = tcp_rtlookup6(tp->t_inpcb, IFSCOPE_NONE);
else
#endif
rt = tcp_rtlookup(tp->t_inpcb, IFSCOPE_NONE);
if (rt == NULL) {
return (
#if INET6
isipv6 ? tcp_v6mssdflt :
#endif
tcp_mssdflt);
}
if (rt->rt_ifp->if_type == IFT_PPP && slowlink_wsize > 0 &&
rt->rt_ifp->if_baudrate > 9600 && rt->rt_ifp->if_baudrate <= 128000) {
tp->t_flags |= TF_SLOWLINK;
}
#if INET6
mss = (isipv6 ? tcp_maxmtu6(rt) : tcp_maxmtu(rt));
#else
mss = tcp_maxmtu(rt);
#endif
RT_UNLOCK(rt);
return (mss - min_protoh);
}
static void
tcp_newreno_partial_ack(tp, th)
struct tcpcb *tp;
struct tcphdr *th;
{
tcp_seq onxt = tp->snd_nxt;
u_int32_t ocwnd = tp->snd_cwnd;
tp->t_timer[TCPT_REXMT] = 0;
tp->t_rtttime = 0;
tp->snd_nxt = th->th_ack;
tp->snd_cwnd = tp->t_maxseg + (th->th_ack - tp->snd_una);
tp->t_flags |= TF_ACKNOW;
(void) tcp_output(tp);
tp->snd_cwnd = ocwnd;
if (SEQ_GT(onxt, tp->snd_nxt))
tp->snd_nxt = onxt;
if (tp->snd_cwnd > th->th_ack - tp->snd_una)
tp->snd_cwnd -= th->th_ack - tp->snd_una;
else
tp->snd_cwnd = 0;
tp->snd_cwnd += tp->t_maxseg;
}
static int
tcp_dropdropablreq(struct socket *head)
{
struct socket *so, *sonext;
unsigned int i, j, qlen;
static int rnd;
static struct timeval old_runtime;
static unsigned int cur_cnt, old_cnt;
struct timeval tv;
struct inpcb *inp = NULL;
struct tcpcb *tp;
if ((head->so_options & SO_ACCEPTCONN) == 0)
return 0;
so = TAILQ_FIRST(&head->so_incomp);
if (!so)
return 0;
microtime(&tv);
if ((i = (tv.tv_sec - old_runtime.tv_sec)) != 0) {
old_runtime = tv;
old_cnt = cur_cnt / i;
cur_cnt = 0;
}
qlen = head->so_incqlen;
if (++cur_cnt > qlen || old_cnt > qlen) {
rnd = (314159 * rnd + 66329) & 0xffff;
j = ((qlen + 1) * rnd) >> 16;
while (j-- && so)
so = TAILQ_NEXT(so, so_list);
}
while (so) {
inp = (struct inpcb *)so->so_pcb;
sonext = TAILQ_NEXT(so, so_list);
if (in_pcb_checkstate(inp, WNT_ACQUIRE, 0) != WNT_STOPUSING) {
if (lck_mtx_try_lock(&inp->inpcb_mtx)) {
so->so_usecount++;
if ((so->so_usecount == 2) &&
(so->so_state & SS_INCOMP) != 0 &&
(so->so_flags & SOF_INCOMP_INPROGRESS) == 0)
break;
else {
in_pcb_checkstate(inp, WNT_RELEASE, 1);
tcp_unlock(so, 1, 0);
}
}
else {
in_pcb_checkstate(inp, WNT_RELEASE, 1);
}
}
so = sonext;
}
if (!so)
return 0;
if (in_pcb_checkstate(inp, WNT_RELEASE, 1) == WNT_STOPUSING) {
tcp_unlock(so, 1, 0);
return 0;
}
if (so->so_usecount != 2 || !(so->so_state & SS_INCOMP)) {
tcp_unlock(so, 1, 0);
return 0;
}
TAILQ_REMOVE(&head->so_incomp, so, so_list);
tcp_unlock(head, 0, 0);
lck_mtx_assert(&inp->inpcb_mtx, LCK_MTX_ASSERT_OWNED);
tp = sototcpcb(so);
so->so_flags |= SOF_OVERFLOW;
so->so_head = NULL;
tcp_close(tp);
tp->t_unacksegs = 0;
if (inp->inp_wantcnt > 0 && inp->inp_wantcnt != WNT_STOPUSING) {
so->so_usecount--;
tcp_unlock(so, 1, 0);
} else {
tcp_unlock(so, 0, 0);
lck_rw_lock_exclusive(tcbinfo.mtx);
tcp_lock(so, 0, 0);
so->so_usecount--;
if (so->so_usecount != 1 ||
(inp->inp_wantcnt > 0 && inp->inp_wantcnt != WNT_STOPUSING)) {
tcp_unlock(so, 1, 0);
} else {
so->so_usecount--;
in_pcbdispose(inp);
}
lck_rw_done(tcbinfo.mtx);
}
tcpstat.tcps_drops++;
tcp_lock(head, 0, 0);
head->so_incqlen--;
head->so_qlen--;
return(1);
}
void
tcp_set_background_cc(struct socket *so)
{
tcp_set_new_cc(so, TCP_CC_ALGO_BACKGROUND_INDEX);
}
void
tcp_set_foreground_cc(struct socket *so)
{
tcp_set_new_cc(so, TCP_CC_ALGO_NEWRENO_INDEX);
}
static void
tcp_set_new_cc(struct socket *so, uint16_t cc_index)
{
struct inpcb *inp = sotoinpcb(so);
struct tcpcb *tp = intotcpcb(inp);
uint16_t old_cc_index = 0;
if (tp->tcp_cc_index != cc_index) {
old_cc_index = tp->tcp_cc_index;
if (CC_ALGO(tp)->cleanup != NULL)
CC_ALGO(tp)->cleanup(tp);
tp->tcp_cc_index = cc_index;
if (tp->snd_nxt > tp->iss) {
if (CC_ALGO(tp)->switch_to != NULL)
CC_ALGO(tp)->switch_to(tp, old_cc_index);
} else {
if (CC_ALGO(tp)->init != NULL)
CC_ALGO(tp)->init(tp);
}
DTRACE_TCP5(cc, void, NULL, struct inpcb *, inp,
struct tcpcb *, tp, struct tcphdr *, NULL,
int32_t, TCP_CC_CHANGE_ALGO);
}
}
void
tcp_set_recv_bg(struct socket *so)
{
if (!IS_TCP_RECV_BG(so))
so->so_traffic_mgt_flags |= TRAFFIC_MGT_TCP_RECVBG;
}
void
tcp_clear_recv_bg(struct socket *so)
{
if (IS_TCP_RECV_BG(so))
so->so_traffic_mgt_flags &= ~(TRAFFIC_MGT_TCP_RECVBG);
}
void
inp_fc_unthrottle_tcp(struct inpcb *inp)
{
struct tcpcb *tp = inp->inp_ppcb;
if (CC_ALGO(tp)->pre_fr != NULL)
CC_ALGO(tp)->pre_fr(tp);
tp->snd_cwnd = tp->snd_ssthresh;
tp->t_bytes_acked = 0;
tp->t_rxtshift = 0;
tcp_output(tp);
}
static int
tcp_getstat SYSCTL_HANDLER_ARGS
{
#pragma unused(oidp, arg1, arg2)
int error;
if (req->oldptr == 0) {
req->oldlen= (size_t)sizeof(struct tcpstat);
}
error = SYSCTL_OUT(req, &tcpstat, MIN(sizeof (tcpstat), req->oldlen));
return (error);
}
SYSCTL_PROC(_net_inet_tcp, TCPCTL_STATS, stats, CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0,
tcp_getstat, "S,tcpstat", "TCP statistics (struct tcpstat, netinet/tcp_var.h)");
static int
sysctl_rexmtthresh SYSCTL_HANDLER_ARGS
{
#pragma unused(arg1, arg2)
int error, val = tcprexmtthresh;
error = sysctl_handle_int(oidp, &val, 0, req);
if (error || !req->newptr)
return (error);
if (val < 2 || val > 3)
return (EINVAL);
tcprexmtthresh = val;
return (0);
}
SYSCTL_PROC(_net_inet_tcp, OID_AUTO, rexmt_thresh, CTLTYPE_INT|CTLFLAG_RW | CTLFLAG_LOCKED,
&tcprexmtthresh, 0, &sysctl_rexmtthresh, "I", "Duplicate ACK Threshold for Fast Retransmit");