Problem w/ Kingston TNE110TX. Analyzed in depth.

[Povl H. Pedersen]

This mail contais lots of debugging output and information. I am willing to 
help and test things, and have no problems rebooting, compiling new 
modules/kernels whatever. So if you need help, I am here. My hardware 
driver development experienc is an approximation to 0, but I am not afraid 
of doing printk statements or whatever to aid me help localize problems in 
drivers.


I have 2 Kingston TNE110TX connected to each other. One is in a PC (Win98SE 
/ Win2k dual boot), the other is in an Apple Network Server (non-MacOS, 
PowerPC machine). The machines are connected with a crossover cable, and 
are runing 100Mbit full-duplex.

I am using kernel 2.2.14 withthe 2.3 drivers, that is tulip.c 0.91x.

My problem is, that whenever the ANS loses its connection to the PC (like 
when the PC reboots), the Linux box will no longer talk to the PC. The 
heartbeat + 100Mbit LEDs are on, but the fullDuplex (FDX) LED is NOT! On 
the PC, the FDX is on. The two machines will not talk with each other no 
matter what I do to force speeds on the PC.

Loading/unloading the tulip module on the Linux does not help either. 
Rebooting the Linux when the PC is powered on works.

The Kingston card is detected as:
tulip.c:v0.91x 12/15/99 becker@cesdis.gsfc.nasa.gov
   The PCI BIOS has not enabled the device at 0/104!  Updating PCI command 
0004->0006.
eth1: Lite-On 82c168 PNIC rev 32 at 0xc8820000, 00:C0:F0:58:3C:D9, IRQ 23.
eth1:  MII transceiver #1 config 3100 status 7829 advertising 01e1.
eth1: Setting full-duplex based on MII #1 link partner capability of 01e1.

in dmesg when driver is loaded.

DIsconnecting the cable between the two machines leaves the 100Mbit LED on, 
reconnect, and the Link LED also goes on (no duplex).

On power up of the Linux box, even before RAM test is done (which happens 
early), the card defaults to to 3 working LEDs: LNK,100,FDX

The chips on the boards are labelled

The PNIC equivalent:
     Kingston
     KT98100
     1111053
     9946     <- So it is pretty recent
     FA733-000

Then there is also the MII interface or whatever, and there are 2 chips:
     MYSON
     MTD972F
     9949EKB363

And a much smaller
     MTD971  (probably support for the MTD972F)

Running with debug=15 and options=29 (forced MII 100Mbit full duplex) does 
not help either.

Here is more from dmesg, where I remove the card, unload the module, 
reloads it, unloads.
Only thing that is different first time is the MII status, 7829 instead of 
782d.

ulip.c:v0.91x 12/15/99 becker@cesdis.gsfc.nasa.gov
   The PCI BIOS has not enabled the device at 0/104!  Updating PCI command 
0004->0006.
eth1: Lite-On 82c168 PNIC rev 32 at 0xc8820000, 00:C0:F0:58:3C:D9, IRQ 23.
eth1:  MII transceiver #1 config 3100 status 7829 advertising 01e1.
eth1:  Advertising 0080 on PHY 1, previously advertising 01e1.
tulip.c:v0.91x 12/15/99 becker@cesdis.gsfc.nasa.gov
eth1: Lite-On 82c168 PNIC rev 32 at 0xc8822000, 00:C0:F0:58:3C:D9, IRQ 23.
eth1:  MII transceiver #1 config 3100 status 782d advertising 0080.
tulip.c:v0.91x 12/15/99 becker@cesdis.gsfc.nasa.gov
Found Lite-On 82c168 PNIC at PCI address 0xf3100000, IRQ 23.
eth1: Lite-On 82c168 PNIC rev 32 at 0xc8824000, 00:C0:F0:58:3C:D9, IRQ 23.
eth1:  MII transceiver #1 config 3100 status 782d advertising 0080.
eth1: tulip_open() irq 23.
eth1: PNIC PHY status is 201b87a, media MII 100baseTx.
eth1: Done tulip_open(), CSR0 00008000, CSR5 02520010 CSR6 810c2202.
eth1: interrupt  csr5=0x02670014 new csr5=0x02660010.
eth1: exiting interrupt, csr5=0x2660010.
eth1: interrupt  csr5=0x02670014 new csr5=0x02660010.
eth1: exiting interrupt, csr5=0x2660010.
eth1: interrupt  csr5=0x02670014 new csr5=0x02660010.
eth1: exiting interrupt, csr5=0x2660010.
eth1: interrupt  csr5=0x02670014 new csr5=0x02660010.
eth1: exiting interrupt, csr5=0x2660010.
eth1: interrupt  csr5=0x02670014 new csr5=0x02660010.
eth1: exiting interrupt, csr5=0x2660010.
eth1: Shutting down ethercard, status was 2560010.
eth1: interrupt  csr5=0x02670014 new csr5=0x02660010.
eth1: exiting interrupt, csr5=0x2660010.


---
Povl H. Pedersen

-------------------------------------------------------------------
To unsubscribe send a message body containing "unsubscribe"
to linux-tulip-bug-request@beowulf.org