Realtek RT8139 (onboard) - failed to receive packet in loopback mode

Mike Tancsa mike at
Thu Jun 17 20:53:38 GMT 2004

At 04:04 PM 17/06/2004, Bill Paul wrote:



>I don't know what your problem is, and you have not provided enough
>data to figure it out. (Just how do you know there are many FCS errors

Looking at the switch port statistics.  The Nortels/Bays have the following 
info per port (this is not the port as it was already cleared..)

Packets:                      41910550    Packets: 
Multicasts:                         15    Multicasts: 
Broadcasts:                     113405    Broadcasts: 
Total Octets:               2935490261    Total 
Octets:                142434347
Lost Packets:                        0    Lost 
Packets:                        0
Errors:                          0    Collisions:                          0
Frame Errors:                        0    Single 
Collisions:                   0
Undersized Packets:                  0    Multiple 
Collisions:                 0
Oversized Packets:                   0    Excessive 
Collisions:                0
Packets 64 bytes:             22042474    Packets 64 
bytes:             13026195
         65-127 bytes           7534680            65-127 
bytes          15818483
         128-255 bytes           723547            128-255 
bytes          4809496
         256-511 bytes           551054            256-511 
bytes           412113
         512-1023 bytes         3792979            512-1023 
bytes          248253
         1024-1518 bytes        7265825            1024-1518 
bytes       23308940
Filtered Packets:                    0    Deferred 
Packets:                    0
Flooded Packets:                     1    Late 
Collisions:                     0

The FCS errors were steadily increasing.... And yes, we tried new cables 
and new ports.

>  How did you meature them?

Simple ftp http gets... It was quite evident soon as I logged in via ssh 
that something was not right.   The ftp from a host on a connected network 
is slow and bursty.

>Did you test performance too?

Yes, it was crap... About the same sort of performance as one would expect 
if there were a duplex mismatch, yet we would see the same performance 
where it was auto-auto, 100full-100full 100half-100half... Barely a Mb/s on 
a directly connected host off the switch.

>so, how did you do that? What switch do you have?)

Noticed first on a Bay Centillion 100 and repeated on a Baystack 350T.

The units are scheduled to go back to the supplier as we swapped them out 
for a slightly more expensive MB with onboard fxps.  I can dig one of them 
out if you want more information before I send it back.


More information about the freebsd-current mailing list