[tulip] Extremely stubborn 21143

[Donald Becker]

On Tue, 17 Oct 2000, Jon.Hinks wrote:

> Problem (if debug is set to 6) is that the ethernet card keeps on timing out
> and does not appear on the network. Cannot ping my box, and cannot ping
...
> I am using Mandrake 7.1, build 2.2.15-4mdk, and the latest version of the
> tulip driver with pci-scan. I have executed a `depmod -a` to make everything

What version are you using?
  I am certain that you aren't using the "latest version".
Please provide the detection message, which includes the version number.

> tulip-diag.c:v2.04 9/26/2000 Donald Becker (becker@scyld.com)
>  http://www.scyld.com/diag/index.html
> Index #1: Found a Digital DS21143 Tulip adapter at 0xf880.
>  Port selection is MII, half-duplex.

OK, the chip is in MII mode.

>  Transmit stopped, Receive stopped, half-duplex.
>   The Rx process state is 'Stopped'.
>   The Tx process state is 'Waiting for Tx to finish'.

Was the interface up or down when this was run?
   (See the 'ifconfig eth0' output.)

> EEPROM transceiver/media description for the Digital DS21143 Tulip chip.
> Leaf node at offset 30, default media type 0800 (Autosense).
>  3 transceiver description blocks:
>   Media 10baseT, block type 2, length 12.
>   Media 10baseT-Full Duplex, block type 2, length 12.

Hmmm, it claims that the on-chip 10Mbps serial transceiver is connected.

>   Media MII, block type 3, length 17.

And it also claims MII connected transceiver (?!)

>    MII interface PHY 0 (media type 11).
>    21143 MII initialization sequence is 2 words: 08ab 0001.
>    21143 MII reset sequence is 0 words:.
>     Media capabilities are 6000, advertising 01e1.
>     Full-duplex map 5000, Threshold map 0180.

This looks normal for a 10+100 MII transceiver.  The driver has no way to
know if the first two entries take priority over the MII transceiver.

>  MII PHY found at address 0, status 0x782d.
>  MII PHY #0 transceiver registers:
>    3400 782d ffff ffff 01e1 0081 0000 ffff

Several things are wrong here.
  - The transceiver isn't reporting its vendor or type: it is not a
    standards-conforming device.
  - The transceiver is at address #0, which is reserved for plug-in external
    transceivers.
  - The transceiver is isolated from the MII data bus
    (related to being at address #0)

The 'mii-diag' program would have reported all of these problems.
   http://www.scyld.com/diag/index.html

What type of hardware do you have?
It sounds like an early not-quite-right prototype board.

You can work around what I think that the problem is by turning on the
transceiver with
       mii-diag -r
or setting a NWay advertised capability, or setting any fixed speed.

Please send a report.

> The contents of this email are confidential to the intended recipient. It
> may not be disclosed to or used by anyone other than this addressee, nor may
> it be copied in any way. If received in error, please contact Berkeley

[[ The rest of the boilerplate omitted. ]]

These terms do not apply.  You have mailed to a published list.  You hold
the copyright to your writings, but by submitting questions to the list you
grant the list publisher (me) the right to publish in this list and related
forums.  "Related forums" may be collected archives of the list on any media,
and quotes from questions in FAQs.

(Long boilerplate appendages like this notice annoy me.  Automatically
attached copyright notices can never unilaterally extend your rights, so why
try to claim that they do?)


Donald Becker				becker@scyld.com
Scyld Computing Corporation		http://www.scyld.com
410 Severn Ave. Suite 210		Second Generation Beowulf Clusters
Annapolis MD 21403			410-990-9993