Ethereal-dev: Re: [Ethereal-dev] Van Jacobson dissector.
Thanks for the feed back. This is becoming more and more a learning
experience :)
It ought to be helpful in my second dissector.
> /* IP and TCP header types */
> typedef struct {
> #if BYTE_ORDER == LITTLE_ENDIAN
> guint8 ihl:4,
> version:4;
> #else
> guint8 version:4,
> ihl:4;
> #endif
You should stay away from using bitfields like this
in portable code. ANSI says that only unsigned ints can
hold bitfields. GCC accepts the above, but not all
compilers do (e.g., IBM's C compiler).
Look at the code in packet-ip.c to see what we did
to make this portable.
The VJ code comes from Linux. The linux ip structure considers bit field
order. I thought it was a good heurisitic
to consider bitfield endian'ness to be the same as byte order - I may have
been wrong. I have no problem with using the lo_nibble and hi_nibble approach.
Perhaps I can hold the world record of the fastest patch in the process :)
struct iphdr {
#if defined(__LITTLE_ENDIAN_BITFIELD)
__u8 ihl:4,
version:4;
#elif defined (__BIG_ENDIAN_BITFIELD)
__u8 version:4,
ihl:4;
#else
#error "Please fix <asm/byteorder.h>"
#endif
__u8 tos;
__u16 tot_len;
__u16 id;
__u16 frag_off;
__u8 ttl;
__u8 protocol;
__u16 check;
__u32 saddr;
__u32 daddr;
/*The options start here. */
};
> /* Copy header and form tvb */
> hdr_len = ((iphdr_type *)hdr_buf)->ihl * 4;
> hdr_len += ((tcphdr_type *)(hdr_buf + hdr_len))->doff * 4;
We just had an issue with this type of struct-casting
in tcp_graph.c. We stay away from this. Because hdr_buf is
being allocated from a GMemChunk, it might be aligned
properly for what you're doing. Any opinions on this?
I couldn't find tcp_graph.c ??
>
> /* Store the reconstructed header in frame data area */
> buf_hdr = g_mem_chunk_alloc(vj_header_memchunk);
I don't see a g_mem_chunk_free() call for the buf_hdr's that
are allocated.
I use g_mem_chunk_destroy in the initialization function for vj for that.
Thanks,
Irfan
--gilbert
