[tulip] An interesting problem...

Tue, 03 Oct 2000 23:48:12 -0400

Hi all,

Ok...I've discovered something strange, but I'm not sure precisely where
to look to fix this problem. 

The problem: linksys LNE100TX card, using the ADMTek Centaur-P chipset,
with a Compaq proliant 6500 server...Dual PCI peerbus, but the kernel
I'm using supports that.

I figured out the chipset after comparing the output of lspci to the
pci_id_info structure in the tulip driver version 0.921.

lspci output = 

01:03.0 Ethernet controller: Bridgecom, Inc: Unknown device 0985 (rev
11)
        Subsystem: Bridgecom, Inc: Unknown device 0574
        Flags: bus master, medium devsel, latency 64, IRQ 10
        I/O ports at 5000
        Memory at c6ff0000 (32-bit, non-prefetchable)
        Capabilities: [c0] Power Management version 2

Again, according to tulip.c v0.921, a device # of 0985 should be a
ADMtek Centaur-P. 

I compiled the driver into the kernel and rebooted the machine...and
while it DOES work...it only works part way...The output of dmesg was a
little on the strange side...brace yourself.

eth0: ADMtek Comet rev 17 at 0xc8804000, 00:20:78:1C:A3:A2, IRQ 10.
eth0:  MII transceiver #1 config 3000 status 786d advertising 01e1.
tulip.c:v0.92l 8/19/2000  Written by Donald Becker <becker@scyld.com>
  http://www.scyld.com/network/tulip.html
eth1: ADMtek Comet rev 17 at 0xc8806000, 00:20:78:1C:A3:A2, IRQ 10.
eth1:  MII transceiver #1 config 1000 status 786d advertising 01e1.
tulip.c:v0.92l 8/19/2000  Written by Donald Becker <becker@scyld.com>
  http://www.scyld.com/network/tulip.html
eth2: ADMtek Comet rev 17 at 0xc8808000, 00:20:78:1C:A3:A2, IRQ 10.
eth2:  MII transceiver #1 config 1000 status 786d advertising 01e1.
tulip.c:v0.92l 8/19/2000  Written by Donald Becker <becker@scyld.com>
  http://www.scyld.com/network/tulip.html
eth3: ADMtek Comet rev 17 at 0xc880a000, 00:20:78:1C:A3:A2, IRQ 10.
eth3:  MII transceiver #1 config 1000 status 786d advertising 01e1.
tulip.c:v0.92l 8/19/2000  Written by Donald Becker <becker@scyld.com>
  http://www.scyld.com/network/tulip.html
eth4: ADMtek Comet rev 17 at 0xc880c000, 00:20:78:1C:A3:A2, IRQ 10.
eth4:  MII transceiver #1 config 1000 status 786d advertising 01e1.
tulip.c:v0.92l 8/19/2000  Written by Donald Becker <becker@scyld.com>
  http://www.scyld.com/network/tulip.html
eth5: ADMtek Comet rev 17 at 0xc880e000, 00:20:78:1C:A3:A2, IRQ 10.
eth5:  MII transceiver #1 config 1000 status 786d advertising 01e1.
tulip.c:v0.92l 8/19/2000  Written by Donald Becker <becker@scyld.com>
  http://www.scyld.com/network/tulip.html
eth6: ADMtek Comet rev 17 at 0xc8810000, 00:20:78:1C:A3:A2, IRQ 10.
eth6:  MII transceiver #1 config 1000 status 786d advertising 01e1.
tulip.c:v0.92l 8/19/2000  Written by Donald Becker <becker@scyld.com>
  http://www.scyld.com/network/tulip.html
eth7: ADMtek Comet rev 17 at 0xc8812000, 00:20:78:1C:A3:A2, IRQ 10.
eth7:  MII transceiver #1 config 1000 status 786d advertising 01e1.
tulip.c:v0.92l 8/19/2000  Written by Donald Becker <becker@scyld.com>
  http://www.scyld.com/network/tulip.html

Well...it appears it found the card...found 6 more than *I* expected it
to find. It guesses that it's a Comet card tho....

output of tulip-diag yields:

tulip-diag.c:v2.04 9/26/2000 Donald Becker (becker@scyld.com)
 http://www.scyld.com/diag/index.html
Index #1: Found a ADMtek AL985 Centaur-P adapter at 0x5000.
 Port selection is 100mbps-SYM/PCS 100baseTx scrambler, half-duplex.
 Transmit started, Receive started, half-duplex.
  The Rx process state is 'Waiting for packets'.
  The Tx process state is 'Idle'.
  The transmit threshold is 128.
 The Comet MAC registers are 1c782000 ffffa2a3 filter 8000000000000000.

The next thing is weird...

output of tulip-diag -aa:

tulip-diag.c:v1.19 10/2/99 Donald Becker (becker@cesdis.gsfc.nasa.gov)
Unable to find a recognized card in /proc/pci.
If there is a card in the machine, explicitly set the I/O port address
  using '-p <ioaddr> -t <chip_type_index>'
 Use '-t -1' to see the valid chip types.

output of tulip-diag -ee:

Index #1: Found a ADMtek AL985 Centaur-P adapter at 0x5000.
 Port selection is 100mbps-SYM/PCS 100baseTx scrambler, half-duplex.
 Transmit started, Receive started, half-duplex.
  The Rx process state is 'Waiting for packets'.
  The Tx process state is 'Idle'.
  The transmit threshold is 128.
 The Comet MAC registers are 1c782000 ffffa2a3 filter 8000000000000000.
EEPROM size is 8.
  Ethernet MAC Station Address 00:20:78:1c:a3:a2.
  Default connection type 'Autosense'.
  PCI IDs Vendor 1317 Device 0985  Subsystem 1317 0574
  PCI min_grant 255 max_latency 255.
  CSR18 power-up setting 0xa04c****.
EEPROM contents:
  0985 0002 0000 0000 2000 1c78 a2a3 0000
  0000 0400 0000 0000 0000 0000 0000 0100
  0985 1317 0574 1317 ffff 0000 0000 a04c
  0000 0000 0000 0000 0000 0000 0000 0000
  0000 0000 0000 0000 0000 0000 0000 0000
  0000 0000 0000 0000 0000 0000 0000 0000
  0000 0000 0000 0000 0000 0000 0000 0000
  0000 0000 0000 0000 0000 0000 0000 8337
 ID block CRC 0x71 (vs. 00).
  Full contents CRC 0x8337 (read as 0x8337).

Another thing...
When I set debug > 0, I get interesting messages about the linkpartner,
etc. which I assume are due to timeouts or something similiar? Lost
link? Irq problems? Any clues?

bp
shortwave@knight-hub.com