In 2.1.127 and 2.1.126 using various pcap programs (eg tcpdump 3.4a6) on
ethernet, it seems that some timesamps reported are exactly 10ms earlier
than
they should be.
This does NOT happen in kernels as new as 2.1.122. I haven't tested
kernels between 2.1.122 and 2.1.126. This happens using two different
ethernet cards, a PCI DE500-BA at both 10 and 100MBps, and an old ISA
wd8013 card. It happens on busy and less busy networks, but is more
noticable on busy networks.
It seems to happen at a frequency of at least once per 1000 pkts.
eg:
14:51:14.708960 a.cs.waikato.ac.nz.nfs > b.cs.waikato.ac.nz.1487875662:
reply ok 28 lookup [|nfs] (DF)
14:51:14.699464 c.cs.waikato.ac.nz.6000 > d.cs.waikato.ac.nz.30040: P
928:960(32) ack 2609 win 31744 (DF)
14:51:14.710295 b.cs.waikato.ac.nz.1487875663 > a.cs.waikato.ac.nz.nfs:
116 lookup [|nfs]
Machine names were anonymised.
The method is this.
1. capture data:
tcpdump -c1000 >somefile
2. look for backwards going timestamps:
t.pl somefile
Here is the perl for t.pl
#!/usr/bin/perl
while( <> ) {
/\d+:\d+:([\d.]+)\s/;
if( $1 < $last ) {
print "$last_str";
print "$_";
$diff = $1-$last;
print "difference: ";
print +(int($diff*10000)/10000);
print "\n\n";
}
$last = $1;
$last_str = $_;
}
I'm not on this mailing list, I apologise if this is a known problem. To
reply to me use the address below.
Stephen
--
-----------------------------------------------------------------------
Stephen Donnelly (BCMS) email: sfd@cs.waikato.ac.nz
Waikato ATM Group Lab G1.31 phone +64 7 838 4466 x6728
Computer Science Department, University of Waikato, New Zealand
-----------------------------------------------------------------------
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/