Removed redundant tracking (now incorporated into NdisAllocatePacket, etc)
that was causing a bugcheck in the 'ping to us' case.  The memory was detected
tracked twice (correctly).

This condition was detected by mifritcher after I checked in.

reactos/drivers/net/tcpip/network

diff -u -r1.8 -r1.9
--- icmp.c	9 Jun 2004 18:11:38 -0000	1.8
+++ icmp.c	12 Jun 2004 04:46:18 -0000	1.9
@@ -91,7 +91,6 @@

         exFreePool(DataBuffer);
         return NULL;
     }

-    Track(NDIS_PACKET_TAG,NdisPacket);

 
     TI_DbgPrint(MAX_TRACE, ("NdisPacket at (0x%X).\n", NdisPacket));
 

@@ -103,7 +102,6 @@

         exFreePool(DataBuffer);
         return NULL;
     }

-    Track(NDIS_BUFFER_TAG,NdisBuffer);

     
     TI_DbgPrint(MAX_TRACE, ("NdisBuffer at (0x%X).\n", NdisBuffer));
 

reactos/drivers/net/tcpip/tcpip

diff -u -r1.2 -r1.3
--- memtrack.c	9 Jun 2004 18:11:38 -0000	1.2
+++ memtrack.c	12 Jun 2004 04:46:18 -0000	1.3
@@ -63,11 +63,10 @@

 	if( ThingInList->Thing == Thing ) {
 	    RemoveEntryList(Entry);
 	    

+	    KeReleaseSpinLock( &AllocatedObjectsLock, OldIrql );

 	    ShowTrackedThing( "Alloc", ThingInList, FileName, LineNo );

-	    

 	    ExFreePool( ThingInList );
 	    TrackDumpFL( FileName, LineNo );

-	    KeReleaseSpinLock( &AllocatedObjectsLock, OldIrql );

 	    DbgPrint("TRACK: SPECIFIED ALREADY ALLOCATED ITEM %x\n", Thing);
 	    KeBugCheck( 0 );
 	}