Date: Mon, 30 Aug 2004 21:10:47 -0500 From: Dan Nelson <dnelson@allantgroup.com> To: "Lucas (a.k.a T-Bird or bsdfan3)" <tbird-contact@cox.net> Cc: freebsd-questions@freebsd.org Subject: Re: Variable length packets? Message-ID: <20040831021047.GC33896@dan.emsphone.com> In-Reply-To: <000301c48efa$bfa00500$c022fc18@yourxu5v9frokn> References: <000301c48efa$bfa00500$c022fc18@yourxu5v9frokn>
index | next in thread | previous in thread | raw e-mail
In the last episode (Aug 30), Lucas (a.k.a T-Bird or bsdfan3) said:
> I am trying to implement a custom protocol that sends and receives
> variable-length packets on top of TCP/IPv4. The problem is that the
> length field of the packet is silently being mangled first becoming 0
> and then getting turned into a very large number (about 2-3 billion).
> The length field is a u_int32_t and I am using the byteorder
> routines. Source code snippets follow:
>
> --decl of struct packet_t--
> struct packet_t
> {
> u_int16_t num;
> u_int32_t len;
> char data[0];
> };
If these are different OSes, the structure may be packed differently.
There's almost certainly two bytes of padding between num and len to
ensure that len is 32-bit aligned, for example.
If you run ktrace on your client (or server), the kdump output will
include a hexdump of all data read or written, which might help you
determine what's going wrong.
--
Dan Nelson
dnelson@allantgroup.com
help
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20040831021047.GC33896>