Whew -- big reorganization to allow ethernet sends of IP packets

NmapOps.h

@@ -122,6 +122,9 @@ class NmapOps {
   int TimeSinceStartMS(struct timeval *now=NULL); 
   struct in_addr v4source();
   const struct in_addr *v4sourceip();
+
+
+
   bool TCPScan(); /* Returns true if at least one chosen scan type is TCP */
   bool UDPScan(); /* Returns true if at least one chosen scan type is UDP */
 
@@ -138,6 +141,26 @@ class NmapOps {
 		             user). */
   int isr00t;
   int debugging;
+
+#define PACKET_SEND_NOPREF 1
+#define PACKET_SEND_ETH_WEAK 2
+#define PACKET_SEND_ETH_STRONG 4
+#define PACKET_SEND_ETH 6
+#define PACKET_SEND_IP_WEAK 8
+#define PACKET_SEND_IP_STRONG 16
+#define PACKET_SEND_IP 24
+
+  /* How should we send raw IP packets?  Nmap can generally use either
+     ethernet or raw ip sockets.  Which is better depends on platform
+     and goals.  A _STRONG preference means that Nmap should use the
+     preferred method whenever it is possible (obviously it isn't
+     always possible -- sending ethernet frames won't work over a PPP
+     connection).  This is useful when the other type doesn't work at
+     all.  A _WEAK preference means that Nmap may use the other type
+     where it is substantially more efficient to do so. For example,
+     Nmap will still do an ARP ping scan of a local network even when
+     the pref is SEND_IP_WEAK */
+  int sendpref;
   bool packetTrace() { return (debugging >= 3)? true : pTrace;  }
   bool versionTrace() { return packetTrace()? true : vTrace;  }
   // Note that packetTrace may turn on at high debug levels even if