[tulip] Question about RX-Drop errors and an appearant chipset lockup on the Phobos p430tx NIC.

Ben Greear greearb@candelatech.com
Sun Oct 20 15:19:00 2002


I have been running some tests generating 10Mbps of traffic on all 4 ports
of the Phobos p430TX 4-port NIC (02-00035 01, Rev G G2).  lspci shows this:

02:04.0 Ethernet controller: Digital Equipment Corporation DECchip 21142/43 (rev 41)
02:05.0 Ethernet controller: Digital Equipment Corporation DECchip 21142/43 (rev 41)
02:06.0 Ethernet controller: Digital Equipment Corporation DECchip 21142/43 (rev 41)
02:07.0 Ethernet controller: Digital Equipment Corporation DECchip 21142/43 (rev 41)

And the chipsets on the NIC are:  21143TD (x4) and 21152AB (x1)

I am using the stock driver in 2.4.20-pre9 currently.

I am seeing problems where after some time (an hour or more), one of the ports
will lock up.  It continiously gets rx-dropped errors, and it does not appear
to be sending anything either.  This problem can be made to go away by ifdown/ifup,
and sometimes just waiting a few minutes and re-starting traffic seems to help.

Also, even when the NIC is working normally, I get a few rx-dropped errors when
running at this speed.

Here is what mii-tool shows for the two ports.  The first one is probably working,
the second one is showing the rx-errors.  I have seen this problem on multiple cards
and multiple ports, but never more than one at a time.

[root@lf-demo1 root]# mii-tool -vv eth2
eth2: negotiated 100baseTx-FD, link ok
   registers for MII PHY 1:
     1000 782d 0013 78e1 05e1 41e1 0007 2001
     0000 ffff ffff ffff ffff ffff ffff ffff
     0084 4780 0000 00f4 0422 0000 0000 0000
     0000 0000 00c8 0000 ffff 0000 0c00 3878
   product info: vendor 00:04:de, model 14 rev 1
   basic mode:   autonegotiation enabled
   basic status: autonegotiation complete, link ok
   capabilities: 100baseTx-FD 100baseTx-HD 10baseT-FD 10baseT-HD
   advertising:  100baseTx-FD 100baseTx-HD 10baseT-FD 10baseT-HD flow-control
   link partner: 100baseTx-FD 100baseTx-HD 10baseT-FD 10baseT-HD


eth2: negotiated 100baseTx-FD, link ok
   registers for MII PHY 1:
     1000 782d 0013 78e1 05e1 41e1 0007 2001
     0000 ffff ffff ffff ffff ffff ffff ffff
     0084 4780 0000 00f4 0422 0000 0000 0000
     0000 0000 01c8 0000 ffff 0000 0c00 3888
   product info: vendor 00:04:de, model 14 rev 1
   basic mode:   autonegotiation enabled
   basic status: autonegotiation complete, link ok
   capabilities: 100baseTx-FD 100baseTx-HD 10baseT-FD 10baseT-HD
   advertising:  100baseTx-FD 100baseTx-HD 10baseT-FD 10baseT-HD flow-control
   link partner: 100baseTx-FD 100baseTx-HD 10baseT-FD 10baseT-HD



So, my questions are:

1)  Is there something I can tune to decrease the rx-dropped errors in the normal
     case?

2)  Any ideas for how I might go about debugging and fixing the problem of the
     total failure of a single port, as described above?

In the meantime, I'll work on trying Becker's driver to see if it makes the problem
go away...

Thanks,
Ben


-- 
Ben Greear <greearb@candelatech.com>       <Ben_Greear AT excite.com>
President of Candela Technologies Inc      http://www.candelatech.com
ScryMUD:  http://scry.wanfear.com     http://scry.wanfear.com/~greear