[prev in list] [next in list] [prev in thread] [next in thread] 

List:       tcpdump-workers
Subject:    Re: [tcpdump-workers] [PATCH] print-sflow.c - actually print more
From:       Rick Jones <rick.jones2 () hp ! com>
Date:       2011-04-27 23:01:51
Message-ID: 1303945311.9358.756.camel () tardy
[Download RAW message or body]

On Wed, 2011-04-27 at 15:21 -0400, Michael Richardson wrote:
> Rick,  I've committed your pcap file and .out file.
> I edited the out file to remove the dates (-t option), and I suggest you
> want to generate one file for each -v level.
> 
> It's pretty important for me to have the .pcap and .out file.  You
> can run things directly as:
>     cd tests
>     ./TESTonce sflow_multiple_counter_30_pdus.pcap sflow_multiple_counter_30_pdus.out "-t -v"
> 
> The raw output goes into NEW/foo, and DIFF/foo should be zero lenght if
> things are okay.  If you like want is in NEW/foo, the cp NEW/foo.out
> foo.out.
> 

Cool.

I will try to be good about including updated .pcap and .out files with
subsequent fixes. And there likely will be subsequent fixes if I ever
get more time :) Looking at the code, there is a bug lurking if ever an
actual IPv6 agent id is used - right now it is ass-u-me-ing IPv4.

I still think that since all the routines being called are checking
lengths against their structure sizes some of the additional length
checks are redundant but I haven't worked it through completely and may
still be confused about some things there.  But, extra length checks
aren't nearly as bad as missing ones so I'm not going to sweat it too
much.

Also, I may be using the wrong masks and shifts for type and index in
counter samples and am just getting lucky.

rick

-
This is the tcpdump-workers list.
Visit https://cod.sandelman.ca/ to unsubscribe.
[prev in list] [next in list] [prev in thread] [next in thread]