Re: [ixp1200] Debugging ETH ports of IXP1200 (fwd)

Magesh, Maybe you can try looking at the octalmac driver code and the Intel IXF440 Ethernet controller document until somebody has more useful advice on this. We did not run into problems like you described, either with the Intel firmware on the newer bridalveil cards or with Vera on the older evaluation board cards. We did not run Intel firmware on the older evaluation boards, so I cannot comment on how it would worked in that configuration. With both the configurations, I had the default card setting (autonegotiation or not, I dont know which one was the default), but it configured the port at 100 Mbps when connected to another PC or even with a switch. - abhijeet On Fri, 4 Oct 2002, Magesh Kannan wrote:

Hi,

Can someone from Intel on this list please help me on this? Please find enclosed a description of a problem in making the octal Fast Ethernet ports run at 100 Mbps when using the VERA runtime.

Today, I restored the flash to the original Intel supplied image, changed the jumper settings back to the factory settings, put the card back in the Evaluation system chassis and started diagnostics. The problem (of LEDs blinking on the ethernet ports and the link flapping) persists even with the Intel firmware. Is there a way to debug the ethernet ports on the evaluation board? In particular, is there a way to find out what speed the MAC has decided on after autonegotiation or to disable autonegotiation and fix the speed at 100 Mbps?

Thanks, Magesh

---------- Forwarded message ---------- Date: Wed, 2 Oct 2002 17:36:42 -0500 (CDT) From: Magesh Kannan Reply-To: ixp1200@CS.Princeton.EDU To: ixp1200@CS.Princeton.EDU Subject: [ixp1200] Debugging ETH ports of IXP1200

Hi,

This is a follow-up to my earlier emails about not being able to connect the Fast Ethernet ports on the IXP1200 to a PC using a straight-thru cable. Here is a recap.

If I connect any of the ports on the IXP1200 to an ethernet card on a Linux PC, the two LEDs on the IXP1200 port (which I think stand for Activity and Full-Duplex) and one of the eight LEDs in the middle (which I think is the Link LED) blink at regular intervals (even when no data is being transmitted) and no link seems to be established. If I run the diagnostic tool 'mii-diag' on the Linux side, this is the output I get.

================ Basic registers of MII PHY #1: 3100 7809 0181 b802 01e1 41e1 0003 0000. The autonegotiated capability is 01e0. The autonegotiated media type is 100baseTx-FD. Basic mode control register 0x3100: Auto-negotiation enabled. Basic mode status register 0x7809 ... 7809. Link status: not established. Your link partner advertised 41e1: 100baseTx-FD 100baseTx 10baseT-FD 10baseT. End of basic transceiver information.

Monitoring the MII transceiver status. 17:13:19.731 Baseline value of MII BMSR (basic mode status register) is 7809. 17:13:34.036 MII BMSR now 782d: Good link, NWay done, No Jabber (41e1). New link partner capability is 41e1 0003: 10/100 HD+FD switch. 17:13:34.326 MII BMSR now 7809: no link, NWay busy, No Jabber (0000). 17:13:41.236 MII BMSR now 782d: Good link, NWay done, No Jabber (41e1). New link partner capability is 41e1 0003: 10/100 HD+FD switch. 17:13:41.396 MII BMSR now 7809: no link, NWay busy, No Jabber (0000). 17:14:00.106 MII BMSR now 782d: Good link, NWay done, No Jabber (41e1). New link partner capability is 41e1 0003: 10/100 HD+FD switch. 17:14:00.136 MII BMSR now 7809: no link, NWay busy, No Jabber (0000). 17:14:09.846 MII BMSR now 7829: no link, NWay done, No Jabber (41e1). New link partner capability is 41e1 0003: 10/100 HD+FD switch. 17:14:09.856 MII BMSR now 782d: Good link, NWay done, No Jabber (41e1). 17:14:10.366 MII BMSR now 7809: no link, NWay busy, No Jabber (0000). 17:14:12.076 MII BMSR now 7829: no link, NWay done, No Jabber (41e1). New link partner capability is 41e1 0003: 10/100 HD+FD switch. 17:14:12.086 MII BMSR now 782d: Good link, NWay done, No Jabber (41e1). ================

The same behavior results with different NICs on the Linux side, one using eepro100 driver and another using the tulip driver on Linux 2.4.3-12.

As you can see from the output, the link seems to be flapping. I am sure that the cables that I use are fine (I tested them using a cable tester, connected them on other machines and even used a factory-made CAT5 cable). To cross-check this problem, I also connected one of the IXP1200 ports to a port on a Cisco Catalyst 3500 series switch and observed in the switch's console port that the concerned ethernet port was going up and down fairly regularly.

Has anyone seen this before? Given this behavior, is there a way to debug the ethernet ports on the IXP1200 board? In particular, is there a way to find out what speed the MAC has decided on and if possible, disable autonegotiation?

FYI, if I interconnect two ports of IXP1200 with each other with a crossover cable, the link is established fine, the yellow LED on the port is lit and link LED is also up.

Thanks for your help, Magesh