Quoting Randall: "When we recognize a dup-ack we *will not* recognize it if for example if the rwnd changes even if new SACK information is reported in the sack blocks. This is due to the fact that in non-SACK you don't (on purpose) recognize ACK?s where the window changed (since you can?t really tell if its a plain window update or a dup-ack).. This means we occasionally miss out on stroking the dup-ack counter and getting out of recovery...." Another report of the same/similar problem: https://lists.freebsd.org/pipermail/freebsd-net/2015-September/043387.html This bug leads to us detecting loss later than we should have. And that leads to suboptimal loss recovery as a whole.
https://reviews.freebsd.org/D4225
A commit references this bug: Author: hiren Date: Tue Dec 8 21:21:48 UTC 2015 New revision: 292003 URL: https://svnweb.freebsd.org/changeset/base/292003 Log: One of the ways to detect loss is to count duplicate acks coming back from the other end till it reaches predetermined threshold which is 3 for us right now. Once that happens, we trigger fast-retransmit to do loss recovery. Main problem with the current implementation is that we don't honor SACK information well to detect whether an incoming ack is a dupack or not. RFC6675 has latest recommendations for that. According to it, dupack is a segment that arrives carrying a SACK block that identifies previously unknown information between snd_una and snd_max even if it carries new data, changes the advertised window, or moves the cumulative acknowledgment point. With the prevalence of Selective ACK (SACK) these days, improper handling can lead to delayed loss recovery. With the fix, new behavior looks like following: 0) th_ack < snd_una --> ignore Old acks are ignored. 1) th_ack == snd_una, !sack_changed --> ignore Acks with SACK enabled but without any new SACK info in them are ignored. 2) th_ack == snd_una, window == old_window --> increment Increment on a good dupack. 3) th_ack == snd_una, window != old_window, sack_changed --> increment When SACK enabled, it's okay to have advertized window changed if the ack has new SACK info. 4) th_ack > snd_una --> reset to 0 Reset to 0 when left edge moves. 5) th_ack > snd_una, sack_changed --> increment Increment if left edge moves but there is new SACK info. Here, sack_changed is the indicator that incoming ack has previously unknown SACK info in it. Note: This fix is not fully compliant to RFC6675. That may require a few changes to current implementation in order to keep per-sackhole dupack counter and change to the way we mark/handle sack holes. PR: 203663 Reviewed by: jtl MFC after: 3 weeks Sponsored by: Limelight Networks Differential Revision: https://reviews.freebsd.org/D4225 Changes: head/sys/netinet/tcp_input.c head/sys/netinet/tcp_sack.c head/sys/netinet/tcp_var.h