Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 11 Oct 2017 06:28:46 +0000 (UTC)
From:      Sepherosa Ziehau <sephe@FreeBSD.org>
To:        src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-stable@freebsd.org, svn-src-stable-10@freebsd.org
Subject:   svn commit: r324520 - stable/10/sys/netinet
Message-ID:  <201710110628.v9B6SkLN018019@repo.freebsd.org>

next in thread | raw e-mail | index | archive | help
Author: sephe
Date: Wed Oct 11 06:28:46 2017
New Revision: 324520
URL: https://svnweb.freebsd.org/changeset/base/324520

Log:
  MFC 324050
  
      tcp: Don't "negotiate" MSS.
  
      _NO_ OSes actually "negotiate" MSS.
  
      RFC 879:
      "... This Maximum Segment Size (MSS) announcement (often mistakenly
      called a negotiation) ..."
  
      This negotiation behaviour was introduced 11 years ago by r159955
      without any explaination about why FreeBSD had to "negotiate" MSS:
  
          In syncache_respond() do not reply with a MSS that is larger than what
          the peer announced to us but make it at least tcp_minmss in size.
  
          Sponsored by:   TCP/IP Optimization Fundraise 2005
  
      The tcp_minmss behaviour is still kept.
  
      Syncookie fix was prodded by tuexen, who also helped to test this
      patch w/ packetdrill.
  
      Reviewed by:    tuexen, karels, bz (previous version)
      Sponsored by:   Microsoft
      Differential Revision:  https://reviews.freebsd.org/D12430

Modified:
  stable/10/sys/netinet/tcp_syncache.c
Directory Properties:
  stable/10/   (props changed)

Modified: stable/10/sys/netinet/tcp_syncache.c
==============================================================================
--- stable/10/sys/netinet/tcp_syncache.c	Wed Oct 11 06:08:01 2017	(r324519)
+++ stable/10/sys/netinet/tcp_syncache.c	Wed Oct 11 06:28:46 2017	(r324520)
@@ -1551,9 +1551,7 @@ syncache_respond(struct syncache *sc, const struct mbu
 	tlen = hlen + sizeof(struct tcphdr);
 
 	/* Determine MSS we advertize to other end of connection. */
-	mssopt = tcp_mssopt(&sc->sc_inc);
-	if (sc->sc_peer_mss)
-		mssopt = max( min(sc->sc_peer_mss, mssopt), V_tcp_minmss);
+	mssopt = max(tcp_mssopt(&sc->sc_inc), V_tcp_minmss);
 
 	/* XXX: Assume that the entire packet will fit in a header mbuf. */
 	KASSERT(max_linkhdr + tlen + TCP_MAXOLEN <= MHLEN,
@@ -1896,7 +1894,7 @@ syncookie_mac(struct in_conninfo *inc, tcp_seq irs, ui
 static tcp_seq
 syncookie_generate(struct syncache_head *sch, struct syncache *sc)
 {
-	u_int i, mss, secbit, wscale;
+	u_int i, secbit, wscale;
 	uint32_t iss, hash;
 	uint8_t *secbits;
 	union syncookie cookie;
@@ -1906,9 +1904,8 @@ syncookie_generate(struct syncache_head *sch, struct s
 	cookie.cookie = 0;
 
 	/* Map our computed MSS into the 3-bit index. */
-	mss = min(tcp_mssopt(&sc->sc_inc), max(sc->sc_peer_mss, V_tcp_minmss));
 	for (i = sizeof(tcp_sc_msstab) / sizeof(*tcp_sc_msstab) - 1;
-	     tcp_sc_msstab[i] > mss && i > 0;
+	     tcp_sc_msstab[i] > sc->sc_peer_mss && i > 0;
 	     i--)
 		;
 	cookie.flags.mss_idx = i;



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201710110628.v9B6SkLN018019>