Author: jtl
Date: Fri Oct 7 10:47:32 2016
New Revision: 306802
URL: https://svnweb.freebsd.org/changeset/base/306802
Log:
MFC r296454:
Some cleanup in tcp_respond() in preparation for another change:
- Reorder variables by size
- Move initializer closer to where it is used
- Remove unneeded variable
MFC r296455:
As reported on the transport@ and current@ mailing lists, the FreeBSD TCP
stack is not compliant with RFC 7323, which requires that TCP stacks send
a timestamp option on all packets (except, optionally, RSTs) after the
session is established.
This patch adds that support. It also adds a TCP signature option to the
packet, if appropriate.
MFC r300764 (by jhb@):
Don't reuse the source mbuf in tcp_respond() if it is not writable.
Not all mbufs passed up from device drivers are M_WRITABLE(). In
particular, the Chelsio T4/T5 driver uses a feature called "buffer
packing" to receive multiple frames in a single receive buffer. The mbufs
to receive multiple frames in a single receive buffer. The mbufs for
these frames all share the same external storage so are treated as
read-only by the rest of the stack when multiple frames are in flight.
Previously tcp_respond() would blindly overwrite read-only mbufs when
INVARIANTS was disabled or panic with an assertion failure if INVARIANTS
was enabled. Note that the new case is a bit of a mix of the two other
cases in tcp_respond(). The TCP and IP headers must be copied explicitly
into the new mbuf instead of being inherited (similar to the m == NULL
case), but the addresses and ports must be swapped in the reply (similar
to the m != NULL case).
Modified:
stable/10/sys/netinet/tcp_subr.c
Directory Properties:
stable/10/ (props changed)
Modified: stable/10/sys/netinet/tcp_subr.c
==============================================================================
--- stable/10/sys/netinet/tcp_subr.c Fri Oct 7 06:29:24 2016
(r306801)
+++ stable/10/sys/netinet/tcp_subr.c Fri Oct 7 10:47:32 2016
(r306802)
@@ -556,16 +556,18 @@ void
tcp_respond(struct tcpcb *tp, void *ipgen, struct tcphdr *th, struct mbuf *m,
tcp_seq ack, tcp_seq seq, int flags)
{
- int tlen;
- int win = 0;
+ struct tcpopt to;
+ struct inpcb *inp;
struct ip *ip;
+ struct mbuf *optm;
struct tcphdr *nth;
+ u_char *optp;
#ifdef INET6
struct ip6_hdr *ip6;
int isipv6;
#endif /* INET6 */
- int ipflags = 0;
- struct inpcb *inp;
+ int optlen, tlen, win;
+ bool incl_opts;
KASSERT(tp != NULL || m != NULL, ("tcp_respond: tp and m both NULL"));
@@ -582,18 +584,21 @@ tcp_respond(struct tcpcb *tp, void *ipge
} else
inp = NULL;
+ incl_opts = false;
+ win = 0;
if (tp != NULL) {
if (!(flags & TH_RST)) {
win = sbspace(&inp->inp_socket->so_rcv);
if (win > (long)TCP_MAXWIN << tp->rcv_scale)
win = (long)TCP_MAXWIN << tp->rcv_scale;
}
+ if ((tp->t_flags & TF_NOOPT) == 0)
+ incl_opts = true;
}
if (m == NULL) {
m = m_gethdr(M_NOWAIT, MT_DATA);
if (m == NULL)
return;
- tlen = 0;
m->m_data += max_linkhdr;
#ifdef INET6
if (isipv6) {
@@ -610,17 +615,54 @@ tcp_respond(struct tcpcb *tp, void *ipge
}
bcopy((caddr_t)th, (caddr_t)nth, sizeof(struct tcphdr));
flags = TH_ACK;
+ } else if (!M_WRITABLE(m)) {
+ struct mbuf *n;
+
+ /* Can't reuse 'm', allocate a new mbuf. */
+ n = m_gethdr(M_NOWAIT, MT_DATA);
+ if (n == NULL) {
+ m_freem(m);
+ return;
+ }
+
+ if (!m_dup_pkthdr(n, m, M_NOWAIT)) {
+ m_freem(m);
+ m_freem(n);
+ return;
+ }
+
+ n->m_data += max_linkhdr;
+ /* m_len is set later */
+#define xchg(a,b,type) { type t; t=a; a=b; b=t; }
+#ifdef INET6
+ if (isipv6) {
+ bcopy((caddr_t)ip6, mtod(n, caddr_t),
+ sizeof(struct ip6_hdr));
+ ip6 = mtod(n, struct ip6_hdr *);
+ xchg(ip6->ip6_dst, ip6->ip6_src, struct in6_addr);
+ nth = (struct tcphdr *)(ip6 + 1);
+ } else
+#endif /* INET6 */
+ {
+ bcopy((caddr_t)ip, mtod(n, caddr_t), sizeof(struct ip));
+ ip = mtod(n, struct ip *);
+ xchg(ip->ip_dst.s_addr, ip->ip_src.s_addr, uint32_t);
+ nth = (struct tcphdr *)(ip + 1);
+ }
+ bcopy((caddr_t)th, (caddr_t)nth, sizeof(struct tcphdr));
+ xchg(nth->th_dport, nth->th_sport, uint16_t);
+ th = nth;
+ m_freem(m);
+ m = n;
} else {
/*
* reuse the mbuf.
- * XXX MRT We inherrit the FIB, which is lucky.
+ * XXX MRT We inherit the FIB, which is lucky.
*/
m_freem(m->m_next);
m->m_next = NULL;
m->m_data = (caddr_t)ipgen;
/* m_len is set later */
- tlen = 0;
-#define xchg(a,b,type) { type t; t=a; a=b; b=t; }
#ifdef INET6
if (isipv6) {
xchg(ip6->ip6_dst, ip6->ip6_src, struct in6_addr);
@@ -643,12 +685,64 @@ tcp_respond(struct tcpcb *tp, void *ipge
xchg(nth->th_dport, nth->th_sport, uint16_t);
#undef xchg
}
+ tlen = 0;
+#ifdef INET6
+ if (isipv6)
+ tlen = sizeof (struct ip6_hdr) + sizeof (struct tcphdr);
+#endif
+#if defined(INET) && defined(INET6)
+ else
+#endif
+#ifdef INET
+ tlen = sizeof (struct tcpiphdr);
+#endif
+#ifdef INVARIANTS
+ m->m_len = 0;
+ KASSERT(M_TRAILINGSPACE(m) >= tlen,
+ ("Not enough trailing space for message (m=%p, need=%d, have=%ld)",
+ m, tlen, (long)M_TRAILINGSPACE(m)));
+#endif
+ m->m_len = tlen;
+ to.to_flags = 0;
+ if (incl_opts) {
+ /* Make sure we have room. */
+ if (M_TRAILINGSPACE(m) < TCP_MAXOLEN) {
+ m->m_next = m_get(M_NOWAIT, MT_DATA);
+ if (m->m_next) {
+ optp = mtod(m->m_next, u_char *);
+ optm = m->m_next;
+ } else
+ incl_opts = false;
+ } else {
+ optp = (u_char *) (nth + 1);
+ optm = m;
+ }
+ }
+ if (incl_opts) {
+ /* Timestamps. */
+ if (tp->t_flags & TF_RCVD_TSTMP) {
+ to.to_tsval = tcp_ts_getticks() + tp->ts_offset;
+ to.to_tsecr = tp->ts_recent;
+ to.to_flags |= TOF_TS;
+ }
+#ifdef TCP_SIGNATURE
+ /* TCP-MD5 (RFC2385). */
+ if (tp->t_flags & TF_SIGNATURE)
+ to.to_flags |= TOF_SIGNATURE;
+#endif
+
+ /* Add the options. */
+ tlen += optlen = tcp_addoptions(&to, optp);
+
+ /* Update m_len in the correct mbuf. */
+ optm->m_len += optlen;
+ } else
+ optlen = 0;
#ifdef INET6
if (isipv6) {
ip6->ip6_flow = 0;
ip6->ip6_vfc = IPV6_VERSION;
ip6->ip6_nxt = IPPROTO_TCP;
- tlen += sizeof (struct ip6_hdr) + sizeof (struct tcphdr);
ip6->ip6_plen = htons(tlen - sizeof(*ip6));
}
#endif
@@ -657,14 +751,12 @@ tcp_respond(struct tcpcb *tp, void *ipge
#endif
#ifdef INET
{
- tlen += sizeof (struct tcpiphdr);
ip->ip_len = htons(tlen);
ip->ip_ttl = V_ip_defttl;
if (V_path_mtu_discovery)
ip->ip_off |= htons(IP_DF);
}
#endif
- m->m_len = tlen;
m->m_pkthdr.len = tlen;
m->m_pkthdr.rcvif = NULL;
#ifdef MAC
@@ -686,7 +778,7 @@ tcp_respond(struct tcpcb *tp, void *ipge
nth->th_seq = htonl(seq);
nth->th_ack = htonl(ack);
nth->th_x2 = 0;
- nth->th_off = sizeof (struct tcphdr) >> 2;
+ nth->th_off = (sizeof (struct tcphdr) + optlen) >> 2;
nth->th_flags = flags;
if (tp != NULL)
nth->th_win = htons((u_short) (win >> tp->rcv_scale));
@@ -694,6 +786,13 @@ tcp_respond(struct tcpcb *tp, void *ipge
nth->th_win = htons((u_short)win);
nth->th_urp = 0;
+#ifdef TCP_SIGNATURE
+ if (to.to_flags & TOF_SIGNATURE) {
+ tcp_signature_compute(m, 0, 0, optlen, to.to_signature,
+ IPSEC_DIR_OUTBOUND);
+ }
+#endif
+
m->m_pkthdr.csum_data = offsetof(struct tcphdr, th_sum);
#ifdef INET6
if (isipv6) {
@@ -725,13 +824,13 @@ tcp_respond(struct tcpcb *tp, void *ipge
TCP_PROBE5(send, NULL, tp, mtod(m, const char *), tp, nth);
#ifdef INET6
if (isipv6)
- (void) ip6_output(m, NULL, NULL, ipflags, NULL, NULL, inp);
+ (void) ip6_output(m, NULL, NULL, 0, NULL, NULL, inp);
#endif /* INET6 */
#if defined(INET) && defined(INET6)
else
#endif
#ifdef INET
- (void) ip_output(m, NULL, NULL, ipflags, NULL, inp);
+ (void) ip_output(m, NULL, NULL, 0, NULL, inp);
#endif
}
_______________________________________________
svn-src-stable-10@freebsd.org mailing list
https://lists.freebsd.org/mailman/listinfo/svn-src-stable-10
To unsubscribe, send any mail to "svn-src-stable-10-unsubscr...@freebsd.org"
