d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Improve the TCP blackhole detection. The principle is to reduce the

Browse Source 
MSS in two steps and try each candidate two times. However, if two
candidates are the same (which is the case in TCP/IPv6), this candidate
was tested four times. This patch ensures that each candidate actually
reduced the MSS and is only tested 2 times. This reduces the time window
of missclassifying a temporary outage as an MTU issue.

Reviewed by:		jtl
MFC after:		1 week
Sponsored by:		Netflix, Inc.
Differential Revision:	https://reviews.freebsd.org/D24308
This commit is contained in:
Michael Tuexen 2020-04-14 16:35:05 +00:00
parent 9ed8321078
commit b89af8e16d
Notes: svn2git 2020-12-20 02:59:44 +00:00 
svn path=/head/; revision=359926
2 changed files with 39 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

50
sys/netinet/tcp_timer.c
View File

@ -723,16 +723,40 @@ tcp_timer_rexmt(void * xtp)
(V_tcp_pmtud_blackhole_detect == 3 && isipv6)) &&
((tp->t_state == TCPS_ESTABLISHED) ||
(tp->t_state == TCPS_FIN_WAIT_1))) {
/*
* Idea here is that at each stage of mtu probe (usually, 1448
* -> 1188 -> 524) should be given 2 chances to recover before
* further clamping down. 'tp->t_rxtshift % 2 == 0' should
* take care of that.
*/
if (tp->t_rxtshift == 1) {
/*
* We enter blackhole detection after the first
* unsuccessful timer based retransmission.
* Then we reduce up to two times the MSS, each
* candidate giving two tries of retransmissions.
* But we give a candidate only two tries, if it
* actually reduces the MSS.
*/
tp->t_blackhole_enter = 2;
tp->t_blackhole_exit = tp->t_blackhole_enter;
if (isipv6) {
#ifdef INET6
if (tp->t_maxseg > V_tcp_v6pmtud_blackhole_mss)
tp->t_blackhole_exit += 2;
if (tp->t_maxseg > V_tcp_v6mssdflt &&
V_tcp_v6pmtud_blackhole_mss > V_tcp_v6mssdflt)
tp->t_blackhole_exit += 2;
#endif
} else {
#ifdef INET
if (tp->t_maxseg > V_tcp_pmtud_blackhole_mss)
tp->t_blackhole_exit += 2;
if (tp->t_maxseg > V_tcp_mssdflt &&
V_tcp_pmtud_blackhole_mss > V_tcp_mssdflt)
tp->t_blackhole_exit += 2;
#endif
}
}
if (((tp->t_flags2 & (TF2_PLPMTU_PMTUD|TF2_PLPMTU_MAXSEGSNT)) ==
(TF2_PLPMTU_PMTUD|TF2_PLPMTU_MAXSEGSNT)) &&
(tp->t_rxtshift >= 2 && tp->t_rxtshift < 6 &&
tp->t_rxtshift % 2 == 0)) {
(tp->t_rxtshift >= tp->t_blackhole_enter &&
tp->t_rxtshift < tp->t_blackhole_exit &&
(tp->t_rxtshift - tp->t_blackhole_enter) % 2 == 0)) {
/*
* Enter Path MTU Black-hole Detection mechanism:
* - Disable Path MTU Discovery (IP "DF" bit).
@ -752,7 +776,8 @@ tcp_timer_rexmt(void * xtp)
*/
#ifdef INET6
if (isipv6 &&
tp->t_maxseg > V_tcp_v6pmtud_blackhole_mss) {
tp->t_maxseg > V_tcp_v6pmtud_blackhole_mss &&
V_tcp_v6pmtud_blackhole_mss > V_tcp_v6mssdflt) {
/* Use the sysctl tuneable blackhole MSS. */
tp->t_maxseg = V_tcp_v6pmtud_blackhole_mss;
TCPSTAT_INC(tcps_pmtud_blackhole_activated);
@ -771,7 +796,8 @@ tcp_timer_rexmt(void * xtp)
else
#endif
#ifdef INET
if (tp->t_maxseg > V_tcp_pmtud_blackhole_mss) {
if (tp->t_maxseg > V_tcp_pmtud_blackhole_mss &&
V_tcp_pmtud_blackhole_mss > V_tcp_mssdflt) {
/* Use the sysctl tuneable blackhole MSS. */
tp->t_maxseg = V_tcp_pmtud_blackhole_mss;
TCPSTAT_INC(tcps_pmtud_blackhole_activated);
@ -798,11 +824,9 @@ tcp_timer_rexmt(void * xtp)
* with a lowered MTU, maybe this isn't a blackhole and
* we restore the previous MSS and blackhole detection
* flags.
* The limit '6' is determined by giving each probe
* stage (1448, 1188, 524) 2 chances to recover.
*/
if ((tp->t_flags2 & TF2_PLPMTU_BLACKHOLE) &&
(tp->t_rxtshift >= 6)) {
(tp->t_rxtshift >= tp->t_blackhole_exit)) {
tp->t_flags2 |= TF2_PLPMTU_PMTUD;
tp->t_flags2 &= ~TF2_PLPMTU_BLACKHOLE;
tp->t_maxseg = tp->t_pmtud_saved_maxseg;

2
sys/netinet/tcp_var.h
View File

@ -169,6 +169,8 @@ struct tcpcb {
u_int t_starttime; /* time connection was established */
u_int t_pmtud_saved_maxseg; /* pre-blackhole MSS */
int t_blackhole_enter; /* when to enter blackhole detection */
int t_blackhole_exit; /* when to exit blackhole detection */
u_int t_rttmin; /* minimum rtt allowed */
u_int t_rttbest; /* best rtt we've seen */