Zynx 21143 problems

Hello tulip members,

I believe this should be directed to tulip-bug, but mailman signed me up
for this list instead.

Anyway, my problem is this.  I am using Znyx ZX346Q 4-port cards with
Digital "21143-PD" chips on them.  I have tried a combination of Zynx's
own driver, Jeff Garzik's tulip drivers, and the driver available on
scyld.com with varying degrees of success.  Only the driver located at
scyld.com was able to push anything close to 100Mb and still be able
to use the MII registers.  (I believe there were carrier errors or
something to that effect on Jeff's version, and IIRC the Zynx card
didn't allow me to look at MII.)

I am for the most part happy with Donald's driver, with one fairly
troublesome exception.  When I physically disconnect the cable from the
NIC to the switch, and then reconnect it, link will not always be
re-established, and if it is, it still can take some time.

I should mention that I am connecting to a Cisco Catalyst 3500 XL, all
ports set to autonegotiation, and this behavior occurs on several cards
in different machines.  Also the output of dmesg during start up for one
of the ports is this (pardon the linewrapping):

eth1: Digital DS21143-xC Tulip rev 48 at 0xd08a5400, 00:C0:95:E0:BA:E9,
IRQ 21. 
eth1:  EEPROM default media type Autosense.
eth1:  Index #0 - Media 10baseT (#0) described by a 21142 Serial PHY (2)
block. 
eth1:  Index #1 - Media 10baseT-FDX (#4) described by a 21142 Serial PHY
(2) block. 
eth1:  Index #2 - Media 100baseTx (#3) described by a 21143 SYM PHY (4)
block. 
eth1:  Index #3 - Media 100baseTx-FDX (#5) described by a 21143 SYM
PHY(4) block.


The most common failure works something like this:

1. Remove cable.  mii-tool/diag report "no link"
2. Suddenly, mii-tool reports a link at 100-halfduplex.  Note, that I
   have not plugged the cable back in.
3. Insert cable.  Same status from mii-tool, but no connectivity.
4. Perhaps a minute passes (I haven't timed it, would that be helpful?),
   the following message may be printed:

eth1: Tx hung, 2439587 vs. 2439576.
eth1: Digital DS21143 Tulip transmit timed out, status f0260000, SIA
000010c6 ffff0001 fffaffbf 8ff50008, resetting...
eth1: transmit timed out, switching to 100baseTx media.

5. And the link is renegotiated.
   If that message is not printed, the link will remain down.

During any point of the lack of connectivity while the cable has been
reattached, I can do mii-tool -r eth1 and renegotiation occurs, and link
is reestablished.

I've done a fair amount of reading through the various sources in
search of a possible fix but I'm clearly not as well versed in the
nuances of NICs.  Can anyone offer me a clue?

Thanks for your help.

-- 
/* [email protected]               http://pheared.net/devel/ */
/* Network Security Engineer       http://pheared.net/~kevin */
/* Sabotage will set us free.   Throw a rock in the machine. */
/*   >++++++++++[<++++++++++>-]<.+++++.----.[-]++++++++++.   */
_______________________________________________
tulip mailing list, [email protected]
To change to digest mode or unsubscribe visit
http://www.scyld.com/mailman/listinfo/tulip