Re: [RFC PATCH 0/2]: hw timestamp support

[Aaron Turner <synfinatic () gmail com>]

On Wed, May 26, 2010 at 9:03 AM, Mcmillan, Scott A
<scott.a.mcmillan () intel com> wrote:
> Guy,
>
> Both the 'raw' and 'nic' timestamps are in the form of seconds since the Unix epoch, plus fractions of a second.  
> Please see my response to Darren for more info on the difference between these two timestamp sources.
>
> The only reason I can think of to use 'raw' is for debugging.
>
> Regarding whether one would always want to use hw timestamps if they are available... I think in general, one would 
> want to use hw timestamps.  The reason I didn't make it the default is that limitations of current hw can lead to 
> potentially confusing results.  For example, the Intel 82580 nic only hw timestamps RX general purpose packets, 
> meaning TX general purpose packets are still timestamped in sw.  Hence, timestamps may not be monotonic in a capture 
> stream (although RX timestamps will be monotonic and TX timestamps will be monotonic).
>
> If the consensus is to use hw timestamps by default if available, I have no issues modifying the patch accordingly.  
> Similarly, if the 'raw' timesource is not seen as potentially useful, I can change that too.

I'd prefer hw timestamps *not* be the default as tools like tcpreplay
which utilize the timestamps during processing would have problems.
Perhaps pcap-ng could encode some kind of delta metadata between the
hw & sw timestamps so that it could be handled properly, but I can't
think of a way off hand of determining the correct timestamp values
post-capture time.

Thanks,
Aaron

-- 
Aaron Turner
http://synfin.net/         Twitter: @synfinatic
http://tcpreplay.synfin.net/ - Pcap editing and replay tools for Unix & Windows
Those who would give up essential Liberty, to purchase a little temporary
Safety, deserve neither Liberty nor Safety.
    -- Benjamin Franklin
"carpe diem quam minimum credula postero"
-
This is the tcpdump-workers list.
Visit https://cod.sandelman.ca/ to unsubscribe.