[tulip] PowerPC SBC with PMC ethernet - can't get it to work

[Donald Becker]

On Thu, 1 Feb 2001, Nicholas, John S wrote:

> We are having problems trying to get an Ethernet PMC card to work
> on a Synergy board (VME chassis).  We are adding a second Ethernet
> port the synergy board.  

>   0  12     1000       D         2        1      0     Symbios 885 SCSI
>   0  12.1   1000     701         2        2      0     Symbios 885 Ethernet

The 'yellowfin' driver is working correctly with the on-board Ethernet,
correct?


> We have tried both the "tulip" and "de4x5" drivers, neither worked right.
> With both, we get the LINK light on (GOOD) and the ACT light flickers.
...
> CPU: PowerPC in VME Backplane
>  Bus Idsel  Vendor  Device ID   Rev    Class Sub-Class Part Name
>   0  15     1011      26         2        6      4     DEC 21154 Bridge
>   1  18     1011      19        41        2      0     *** UNKNOWN ***

> Ethernet Card:
>   PMC/Fast Eathernet/21143
>   Board manufactured by Force Computers
> 
>   Placed on mezzanine board so IRQ would not conflict with VME 
>   Universe driver.  In this location, the board had an IRQ of 9
>   as reported by driver.

Are you seeing interrupts?  (Check the /proc/interrupt count.)

> Linux:  
>   Yellow dog  Version 2.2.15-a

What version of the Tulip driver?

> DE4X5:
>  This driver seemed to work on transmit.  
...
> TULIP:
>  With this driver, we first tried not setting any parameters.  Then we mostly
>  used media type of 12 (Serial 10baseT(no autoselect)). 
> 
> ---------------------------
> message log summary with following command
> insmod tulip debug=0x1ff 
> 
> eth1: Digital DS21143 Tulip rev 65 at 0xfe801000, 00:C0:8B:03:38:48, IRQ 9.

There should have been a version string before this.
If you have something before v0.92 you should upgrade.

> 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-FD (#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-FD (#5) described by a 21143 SYM PHY (4) block.

This EEPROM media table appears to be very similar to the example in the
design kit.  It might not be accurate for your board.  But we can't tell
that until the IRQ problem is resolved.

> eth1: tulip_open() irq 9.
...
> eth1: Done tulip_open(), CSR0 f8a08000, CSR5 f0320000 CSR6 b2422202.

Normal.

> eth1: 21143 negotiation status 000052ca, 10baseT.

This media selection was done by a software timer tick, but we should
have seen an interrupt.

> eth1: Tx hung, 7 vs. 0.
> eth1: 21140 transmit timed out, status f4678054, SIA 000052ca ffff0001 fffbffff cff0c000, resetting...

Bad.  The 21143 has raised an interrupt, but the interrupt handler
wasn't run.  The software timer tick is complaining about the Tx taking
too long.

> eth1: 21143 non-MII 100baseTx-FD transceiver control 08af/00a5.
> eth1:  Setting CSR15 to 08af0008/00a50008.
> eth1: Using media type 100baseTx-FD, CSR12 is ca.

This media switch is caused by the transmit timeout, not the regular
media detection code.  It wouldn't happen if the interrupts were getting
through.

Bottom line: figure out why interrupts are not working -- this isn't a
tulip driver bug.

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