[PATCH] checksum problem on big endian systems

Hi,


A few days ago, Mike reported me a bug hapening on Mac OS X, and
probably with other big endian systems: some checksums were invalid,
sometimes ...

Here is a proposed patch (very short :) but that would be great if you
guys could test it on different systems/arch (eg sparc/solaris). I've
just tested on my OS X and an x86 linux to check it does not insert
a bug.

The problem was in libnet_in_cksum() (libnet_checksum.c) when
packet'size was odd, due to a bad casting and endianess problems.

I suspect this can also affect other places with such casting, but I
had no time to investigate any further.


	Frenchy pal Fred Raynal ;-)

--- libnet/src/libnet_checksum.c	Sat Jan  3 21:31:02 2004
+++ libnet-fr/src/libnet_checksum.c	Wed Feb 11 10:22:34 2004
@@ -52,7 +52,7 @@
     }
     if (len == 1)
     {
-        sum += *(u_int8_t *)addr;
+        sum += *(u_int16_t *)addr;
     }
 
     return (sum);