Author: jhb
Date: Fri Jun 10 18:46:40 2011
New Revision: 222934
URL: http://svn.freebsd.org/changeset/base/222934
Log:
MFC 221209:
TCP reuses t_rxtshift to determine the backoff timer used for both the
persist state and the retransmit timer. However, the code that implements
"bad retransmit recovery" only checks t_rxtshift to see if an ACK has been
received in during the first retransmit timeout window. As a result, if
ticks has wrapped over to a negative value and a socket is in the persist
state, it can incorrectly treat an ACK from the remote peer as a
"bad retransmit recovery" and restore saved values such as snd_ssthresh and
snd_cwnd. However, if the socket has never had a retransmit timeout, then
these saved values will be zero, so snd_ssthresh and snd_cwnd will be set
to 0.
If the socket is in fast recovery (this can be caused by excessive
duplicate ACKs such as those fixed by 220794), then each ACK that arrives
triggers either NewReno or SACK partial ACK handling which clamps snd_cwnd
to be no larger than snd_ssthresh. In effect, the socket's send window
is permamently stuck at 0 even though the remote peer is advertising a
much larger window and pending data is only sent via TCP window probes
(so one byte every few seconds).
Fix this by adding a new TCP pcb flag (TF_PREVVALID) that indicates that
the various snd_*_prev fields in the pcb are valid and only perform
"bad retransmit recovery" if this flag is set in the pcb. The flag is set
on the first retransmit timeout that occurs and is cleared on subsequent
retransmit timeouts or when entering the persist state.
Modified:
stable/8/sys/netinet/tcp_input.c
stable/8/sys/netinet/tcp_output.c
stable/8/sys/netinet/tcp_timer.c
stable/8/sys/netinet/tcp_var.h
Directory Properties:
stable/8/sys/ (props changed)
stable/8/sys/amd64/include/xen/ (props changed)
stable/8/sys/cddl/contrib/opensolaris/ (props changed)
stable/8/sys/contrib/dev/acpica/ (props changed)
stable/8/sys/contrib/pf/ (props changed)
Modified: stable/8/sys/netinet/tcp_input.c
==============================================================================
--- stable/8/sys/netinet/tcp_input.c Fri Jun 10 17:06:25 2011
(r222933)
+++ stable/8/sys/netinet/tcp_input.c Fri Jun 10 18:46:40 2011
(r222934)
@@ -420,6 +420,7 @@ cc_cong_signal(struct tcpcb *tp, struct
if (tp->t_flags & TF_WASCRECOVERY)
ENTER_CONGRECOVERY(tp->t_flags);
tp->snd_nxt = tp->snd_max;
+ tp->t_flags &= ~TF_PREVVALID;
tp->t_badrxtwin = 0;
break;
}
@@ -1575,6 +1576,7 @@ tcp_do_segment(struct mbuf *m, struct tc
* "bad retransmit" recovery.
*/
if (tp->t_rxtshift == 1 &&
+ tp->t_flags & TF_PREVVALID &&
(int)(ticks - tp->t_badrxtwin) < 0) {
cc_cong_signal(tp, th, CC_RTO_ERR);
}
@@ -2522,7 +2524,8 @@ process_ACK:
* original cwnd and ssthresh, and proceed to transmit where
* we left off.
*/
- if (tp->t_rxtshift == 1 && (int)(ticks - tp->t_badrxtwin) < 0)
+ if (tp->t_rxtshift == 1 && tp->t_flags & TF_PREVVALID &&
+ (int)(ticks - tp->t_badrxtwin) < 0)
cc_cong_signal(tp, th, CC_RTO_ERR);
/*
Modified: stable/8/sys/netinet/tcp_output.c
==============================================================================
--- stable/8/sys/netinet/tcp_output.c Fri Jun 10 17:06:25 2011
(r222933)
+++ stable/8/sys/netinet/tcp_output.c Fri Jun 10 18:46:40 2011
(r222934)
@@ -1323,6 +1323,7 @@ tcp_setpersist(struct tcpcb *tp)
int t = ((tp->t_srtt >> 2) + tp->t_rttvar) >> 1;
int tt;
+ tp->t_flags &= ~TF_PREVVALID;
if (tcp_timer_active(tp, TT_REXMT))
panic("tcp_setpersist: retransmit pending");
/*
Modified: stable/8/sys/netinet/tcp_timer.c
==============================================================================
--- stable/8/sys/netinet/tcp_timer.c Fri Jun 10 17:06:25 2011
(r222933)
+++ stable/8/sys/netinet/tcp_timer.c Fri Jun 10 18:46:40 2011
(r222934)
@@ -506,7 +506,9 @@ tcp_timer_rexmt(void * xtp)
else
tp->t_flags &= ~TF_WASCRECOVERY;
tp->t_badrxtwin = ticks + (tp->t_srtt >> (TCP_RTT_SHIFT + 1));
- }
+ tp->t_flags |= TF_PREVVALID;
+ } else
+ tp->t_flags &= ~TF_PREVVALID;
TCPSTAT_INC(tcps_rexmttimeo);
if (tp->t_state == TCPS_SYN_SENT)
rexmt = TCP_REXMTVAL(tp) * tcp_syn_backoff[tp->t_rxtshift];
Modified: stable/8/sys/netinet/tcp_var.h
==============================================================================
--- stable/8/sys/netinet/tcp_var.h Fri Jun 10 17:06:25 2011
(r222933)
+++ stable/8/sys/netinet/tcp_var.h Fri Jun 10 18:46:40 2011
(r222934)
@@ -225,6 +225,7 @@ struct tcpcb {
#define TF_NEEDSYN 0x000400 /* send SYN (implicit state) */
#define TF_NEEDFIN 0x000800 /* send FIN (implicit state) */
#define TF_NOPUSH 0x001000 /* don't push */
+#define TF_PREVVALID 0x002000 /* saved values for bad rxmit
valid */
#define TF_MORETOCOME 0x010000 /* More data to be appended to
sock */
#define TF_LQ_OVERFLOW 0x020000 /* listen queue overflow */
#define TF_LASTIDLE 0x040000 /* connection was previously
idle */
_______________________________________________
svn-src-all@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "svn-src-all-unsubscr...@freebsd.org"
