From: Jan Klemkow <jan@openbsd.org>
Subject: Re: names for TSO and LRO
To: Alexander Bluhm <bluhm@openbsd.org>
Cc: tech@openbsd.org
Date: Thu, 17 Apr 2025 15:25:46 +0200

On Wed, Apr 16, 2025 at 11:59:43PM +0200, Jan Klemkow wrote:
> On Wed, Apr 16, 2025 at 11:39:21PM GMT, Alexander Bluhm wrote:
> > Now that we have software LRO, we need consistent naming with TSO.
> > One is called tcp_softlro and the other tcp_copper.  As claudio@
> > suggested the descriptive name "copper", I would like to have it
> > this way for LRO.  tcp_glue_enqueue() came to my mind.
> 
> In my opinion we should not make up new names for already established
> things.  LRO und TSO are established names [1].  If you google "tcp
> glue" you will very misleading results.  Looks like its a product name
> for some glue.  Also "tcp copper" won't get better search results.
> 
> Thus, I suggest to rename tcp_copper() to tcp_softtso() to get things
> consistent here.  Or, tcp_tso() and tcp_lro() to keep it shorter.
> 
> [1]: https://en.wikipedia.org/wiki/TCP_offload_engine

As explained in me message before.  I suggest this diff to
make this naming consistent.

ok?

bye,
jan

Index: netinet/ip_output.c
===================================================================
RCS file: /cvs/src/sys/netinet/ip_output.c,v
diff -u -p -r1.407 ip_output.c
--- netinet/ip_output.c	12 Mar 2025 01:44:27 -0000	1.407
+++ netinet/ip_output.c	17 Apr 2025 13:06:26 -0000
@@ -634,7 +634,7 @@ ip_output_ipsec_send(struct tdb *tdb, st
 	m->m_flags &= ~(M_MCAST | M_BCAST);
 
 	if (tso) {
-		error = tcp_chopper(m, &ml, encif, len);
+		error = tcp_softtso(m, &ml, encif, len);
 		if (error)
 			goto done;
 	} else {
Index: netinet/tcp_output.c
===================================================================
RCS file: /cvs/src/sys/netinet/tcp_output.c,v
diff -u -p -r1.153 tcp_output.c
--- netinet/tcp_output.c	17 Feb 2025 12:46:02 -0000	1.153
+++ netinet/tcp_output.c	17 Apr 2025 13:16:15 -0000
@@ -1199,7 +1199,7 @@ tcp_setpersist(struct tcpcb *tp)
 }
 
 int
-tcp_chopper(struct mbuf *m0, struct mbuf_list *ml, struct ifnet *ifp,
+tcp_softtso(struct mbuf *m0, struct mbuf_list *ml, struct ifnet *ifp,
     u_int mss)
 {
 	struct ip *ip = NULL;
@@ -1394,7 +1394,7 @@ tcp_if_output_tso(struct ifnet *ifp, str
 	}
 
 	/* as fallback do TSO in software */
-	if ((error = tcp_chopper(*mp, &ml, ifp, (*mp)->m_pkthdr.ph_mss)) ||
+	if ((error = tcp_softtso(*mp, &ml, ifp, (*mp)->m_pkthdr.ph_mss)) ||
 	    (error = if_output_ml(ifp, &ml, dst, rt)))
 		goto done;
 	tcpstat_inc(tcps_outswtso);
Index: netinet/tcp_var.h
===================================================================
RCS file: /cvs/src/sys/netinet/tcp_var.h,v
diff -u -p -r1.187 tcp_var.h
--- netinet/tcp_var.h	16 Apr 2025 17:17:06 -0000	1.187
+++ netinet/tcp_var.h	17 Apr 2025 13:07:22 -0000
@@ -731,7 +731,7 @@ struct tcpcb *
 	 tcp_newtcpcb(struct inpcb *, int);
 void	 tcp_notify(struct inpcb *, int);
 int	 tcp_output(struct tcpcb *);
-int	 tcp_chopper(struct mbuf *, struct mbuf_list *, struct ifnet *, u_int);
+int	 tcp_softtso(struct mbuf *, struct mbuf_list *, struct ifnet *, u_int);
 int	 tcp_if_output_tso(struct ifnet *, struct mbuf **, struct sockaddr *,
 	    struct rtentry *, uint32_t, u_int);
 void	 tcp_pulloutofband(struct socket *, u_int, struct mbuf *, int);
Index: netinet6/ip6_output.c
===================================================================
RCS file: /cvs/src/sys/netinet6/ip6_output.c,v
diff -u -p -r1.297 ip6_output.c
--- netinet6/ip6_output.c	2 Mar 2025 21:28:32 -0000	1.297
+++ netinet6/ip6_output.c	17 Apr 2025 13:09:10 -0000
@@ -2899,7 +2899,7 @@ ip6_output_ipsec_send(struct tdb *tdb, s
 	m->m_flags &= ~(M_BCAST | M_MCAST);
 
 	if (tso) {
-		error = tcp_chopper(m, &ml, encif, len);
+		error = tcp_softtso(m, &ml, encif, len);
 		if (error)
 			goto done;
 	} else {