Implement pseudo_AF_HDRCMPLT, which controls the state of the 'header

completion' flag.  If set, the interface output routine will assume that
the packet already has a valid link-level source address.  This defaults
to off (the address is overwritten)

PR:		kern/10680
Submitted by:	"Christopher N . Harrell" <cnh@mindspring.net>
Obtained from:	NetBSD

7 files changed, 58 insertions, 11 deletions

diff --git a/sys/net/bpf.c b/sys/net/bpf.c
index 6876ab2..5bd8ed5 100644
--- a/sys/net/bpf.c
+++ b/sys/net/bpf.c
@@ -579,6 +579,9 @@ bpfwrite(dev, uio, ioflag)
 	if (datlen > ifp->if_mtu)
 		return (EMSGSIZE);
 
+	if (d->bd_hdrcmplt)
+		dst.sa_family = pseudo_AF_HDRCMPLT;
+
 	s = splnet();
 #if BSD >= 199103
 	error = (*ifp->if_output)(ifp, m, &dst, (struct rtentry *)0);
@@ -626,6 +629,8 @@ reset_d(d)
  *  BIOCGSTATS		Get packet stats.
  *  BIOCIMMEDIATE	Set immediate mode.
  *  BIOCVERSION		Get filter language version.
+ *  BIOCGHDRCMPLT	Get "header already complete" flag
+ *  BIOCSHDRCMPLT	Set "header already complete" flag
  */
 /* ARGSUSED */
 static	int
@@ -822,6 +827,20 @@ bpfioctl(dev, cmd, addr, flags, p)
 			break;
 		}
 
+	/*
+	 * Get "header already complete" flag
+	 */
+	case BIOCGHDRCMPLT:
+		*(u_int *)addr = d->bd_hdrcmplt;
+		break;
+
+	/*
+	 * Set "header already complete" flag
+	 */
+	case BIOCSHDRCMPLT:
+		d->bd_hdrcmplt = *(u_int *)addr ? 1 : 0;
+		break;
+
 	case FIONBIO:		/* Non-blocking I/O */
 		break;
 
diff --git a/sys/net/bpf.h b/sys/net/bpf.h
index fa5e46c..64e1602 100644
--- a/sys/net/bpf.h
+++ b/sys/net/bpf.h
@@ -111,6 +111,8 @@ struct bpf_version {
 #define BIOCVERSION	_IOR('B',113, struct bpf_version)
 #define BIOCGRSIG	_IOR('B',114, u_int)
 #define BIOCSRSIG	_IOW('B',115, u_int)
+#define BIOCGHDRCMPLT	_IOR('B',116, u_int)
+#define BIOCSHDRCMPLT	_IOW('B',117, u_int)
 
 /*
  * Structure prepended to each packet.
diff --git a/sys/net/bpfdesc.h b/sys/net/bpfdesc.h
index d87cc40..1e638f7 100644
--- a/sys/net/bpfdesc.h
+++ b/sys/net/bpfdesc.h
@@ -76,6 +76,7 @@ struct bpf_d {
 	u_char		bd_promisc;	/* true if listening promiscuously */
 	u_char		bd_state;	/* idle, waiting, or timed out */
 	u_char		bd_immediate;	/* true to return on packet arrival */
+	int		bd_hdrcmplt;	/* false to fill in src lladdr automatically */
 	int		bd_async;	/* non-zero if packet reception should generate signal */
 	int		bd_sig;		/* signal to send upon packet reception */
 	struct sigio *	bd_sigio;	/* information for async I/O */
diff --git a/sys/net/if_ethersubr.c b/sys/net/if_ethersubr.c
index 489bcb2..9d59287 100644
--- a/sys/net/if_ethersubr.c
+++ b/sys/net/if_ethersubr.c
@@ -132,8 +132,8 @@ ether_output(ifp, m0, dst, rt0)
 	struct rtentry *rt0;
 {
 	short type;
-	int s, error = 0;
- 	u_char edst[6];
+	int s, error = 0, hdrcmplt = 0;
+ 	u_char esrc[6], edst[6];
 	register struct mbuf *m = m0;
 	register struct rtentry *rt;
 	register struct ether_header *eh;
@@ -326,6 +326,12 @@ ether_output(ifp, m0, dst, rt0)
 		} break;
 #endif /* LLC */
 
+	case pseudo_AF_HDRCMPLT:
+		hdrcmplt = 1;
+		eh = (struct ether_header *)dst->sa_data;
+		(void)memcpy(esrc, eh->ether_shost, sizeof (esrc));
+		/* FALLTHROUGH */
+
 	case AF_UNSPEC:
 		loop_copy = -1; /* if this is for us, don't do it */
 		eh = (struct ether_header *)dst->sa_data;
@@ -350,8 +356,12 @@ ether_output(ifp, m0, dst, rt0)
 	(void)memcpy(&eh->ether_type, &type,
 		sizeof(eh->ether_type));
  	(void)memcpy(eh->ether_dhost, edst, sizeof (edst));
- 	(void)memcpy(eh->ether_shost, ac->ac_enaddr,
-	    sizeof(eh->ether_shost));
+	if (hdrcmplt)
+		(void)memcpy(eh->ether_shost, esrc,
+			sizeof(eh->ether_shost));
+	else
+		(void)memcpy(eh->ether_shost, ac->ac_enaddr,
+			sizeof(eh->ether_shost));
 
 	/*
 	 * If a simplex interface, and the packet is being sent to our
diff --git a/sys/net/if_fddisubr.c b/sys/net/if_fddisubr.c
index 25409fd..eedc822 100644
--- a/sys/net/if_fddisubr.c
+++ b/sys/net/if_fddisubr.c
@@ -136,8 +136,8 @@ fddi_output(ifp, m0, dst, rt0)
 	struct rtentry *rt0;
 {
 	u_int16_t type;
-	int s, loop_copy = 0, error = 0;
- 	u_char edst[6];
+	int s, loop_copy = 0, error = 0, hdrcmplt = 0;
+ 	u_char esrc[6], edst[6];
 	register struct mbuf *m = m0;
 	register struct rtentry *rt;
 	register struct fddi_header *fh;
@@ -295,6 +295,15 @@ fddi_output(ifp, m0, dst, rt0)
 		} break;
 #endif /* LLC */	
 
+	case pseudo_AF_HDRCMPLT:
+	{
+		struct ether_header *eh;
+		hdrcmplt = 1;
+		eh = (struct ether_header *)dst->sa_data;
+ 		(void)memcpy((caddr_t)esrc, (caddr_t)eh->ether_shost, sizeof (esrc));
+		/* FALLTHROUGH */
+	}
+
 	case AF_UNSPEC:
 	{
 		struct ether_header *eh;
@@ -370,9 +379,12 @@ fddi_output(ifp, m0, dst, rt0)
 	fh->fddi_fc = FDDIFC_LLC_ASYNC|FDDIFC_LLC_PRIO4;
  	(void)memcpy((caddr_t)fh->fddi_dhost, (caddr_t)edst, sizeof (edst));
   queue_it:
- 	(void)memcpy((caddr_t)fh->fddi_shost, (caddr_t)ac->ac_enaddr,
-	    sizeof(fh->fddi_shost));
-
+	if (hdrcmplt)
+		(void)memcpy((caddr_t)fh->fddi_shost, (caddr_t)esrc,
+			sizeof(fh->fddi_shost));
+	else
+		(void)memcpy((caddr_t)fh->fddi_shost, (caddr_t)ac->ac_enaddr,
+			sizeof(fh->fddi_shost));
 	/*
 	 * If a simplex interface, and the packet is being sent to our
 	 * Ethernet address or a broadcast address, loopback a copy.
diff --git a/sys/sys/mount.h b/sys/sys/mount.h
index 1a68b17..b161718 100644
--- a/sys/sys/mount.h
+++ b/sys/sys/mount.h
@@ -355,7 +355,7 @@ struct vfsops {
 
 #include <net/radix.h>
 
-#define	AF_MAX		31	/* XXX */
+#define	AF_MAX		32	/* XXX */
 
 /*
  * Network address lookup element
diff --git a/sys/sys/socket.h b/sys/sys/socket.h
index a7b88d4..4303945 100644
--- a/sys/sys/socket.h
+++ b/sys/sys/socket.h
@@ -128,8 +128,11 @@ struct	linger {
 #define	AF_INET6	28		/* IPv6 */
 #define	AF_NATM		29		/* native ATM access */
 #define	AF_ATM		30		/* ATM */
+#define pseudo_AF_HDRCMPLT 31		/* Used by BPF to not rewrite headers
+					 * in interface output routine
+					 */
 
-#define	AF_MAX		31
+#define	AF_MAX		32
 
 /*
  * Structure used by kernel to store most