This patch changes the #ifdef for knowing about the path to "if_packet.h"
to being the result of work done by configure. Rerunning autoconf to
regenerate configure from configure.in is required after applying this
patch.
There might be some appropriate use of #else here as it seems that
glibc ships with <netpacket/packet.h>. For example, if <netpacket/packet.h>
is present, is it also necessary to use <linux/if_packet.h> or
<netpacket/if_packet.h>?
The only properly documented location for this file that I can find
is that various man pages refer to <netpacket/packet.h> - and this
is what is now found in OpenSolaris. I suspect that checking for
that alone is not enough but I don't have any Linux boxes present
to test with :/
That aside, the change has merit if only because it replaces the
use of implied file presence through some other symbol with the
actual detection of the files being present using configure.
Summary of this patch: "yay linux!"
Darren
--- fad-getad.c.dist Mon Apr 5 10:54:05 2010
+++ fad-getad.c Sat May 8 22:11:18 2010
@@ -61,15 +61,19 @@
#endif
#ifdef AF_PACKET
-# ifdef __Lynx__
-/* LynxOS */
+#ifdef HAVE_NETPACKET_IF_PACKET_H
# include <netpacket/if_packet.h>
-# else
-/* Linux */
+#endif
+#ifdef HAVE_NETPACKET_PACKET_H
+# include <netpacket/packet.h>
+#endif
+#ifdef HAVE_LINUX_TYPES_H
# include <linux/types.h>
+#endif
+#ifdef HAVE_LINUX_IF_PACKET_H
# include <linux/if_packet.h>
-# endif
#endif
+#endif
/*
* This is fun.
--- configure.in.dist Sat May 8 22:08:18 2010
+++ configure.in Sat May 8 22:11:29 2010
@@ -29,6 +29,7 @@
AC_C___ATTRIBUTE__
AC_CHECK_HEADERS(sys/bitypes.h)
+AC_CHECK_HEADERS(linux/types.h linux/if_packet.h netpacket/if_packet.h
netpacket/packet.h)
AC_CHECK_TYPE([int8_t], ,
[AC_DEFINE([int8_t], [signed char],
-
This is the tcpdump-workers list.
Visit https://cod.sandelman.ca/ to unsubscribe.
