Thanks for the quick reply.

> > I know this problem was discussed before on the list (msg:
> > http://www.tcpdump.org/lists/workers/2001/07/msg00001.html), and we
tried
> > that, but we still have the same problem.
>
> Tried what?
>
First we tried enabling the real time clock support for RH and rebuilt the
kernel and the libpcap from the source, but that only provided a partial
solution.  The timestamps became more accurate, but still seemed to be
taking "slices" of time and not down to the usec.

> Using an unmodified libpcap and tcpdump from tcpdump.org, as per my
> followup to the cited message:
>
> http://www.tcpdump.org/lists/workers/2001/07/msg00002.html
>
> If so, then, as per the aforementioned followup, the problem is probably
> in the Linux kernel's time-stamping code.  As such, I'd take this up
> with Red Hat, as it appears not to be a problem with tcpdump.org's
> libpcap code; the code we use to get the time stamp on Linux is
>
> if (ioctl(handle->fd, SIOCGSTAMP, &pcap_header.ts) == -1) {
> snprintf(handle->errbuf, sizeof(handle->errbuf),
> "ioctl: %s", pcap_strerror(errno));
> return -1;
> }
>
Cool.  That's what I was afraid of - might be more of a RH kernel issue.
Looks like it's time to poke around the kernel and see what's going on
there.

Thanks again for the help,
Scott

===============================================================
Scott Rose
Advanced Network Technologies Division
NIST

ph: 301-975-8439                       fax: 301-590-0932
http://www.nist.gov
===============================================================

-
This is the TCPDUMP workers list. It is archived at
http://www.tcpdump.org/lists/workers/index.html
To unsubscribe use mailto:[EMAIL PROTECTED]?body=unsubscribe

Reply via email to