From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6) Gecko/20040510 Description of problem: After upgrding from kernel-2.4.21-9.ELsmp to kernel-2.4.21-15.ELsmp tg3 either fails to negotiate speed and duplex with a Cisco Catalyst 4501R switch or the line drops just after negotion. The host hardware is a Dell 2650 with onboard nics. The driver/nic will work with other switches such as a Dell Power Connect. Log entries: Jun 9 03:33:11 phils kernel: tg3: eth0: Link is down. Jun 9 03:33:14 phils kernel: tg3: eth0: Link is up at 1000 Mbps, full duplex. Jun 9 03:33:14 phils kernel: tg3: eth0: Flow control is on for TX and on for RX . Jun 9 03:33:19 phils kernel: tg3: eth0: Link is down. Jun 9 03:33:22 phils kernel: tg3: eth0: Link is up at 1000 Mbps, full duplex. Jun 9 03:33:22 phils kernel: tg3: eth0: Flow control is on for TX and on for RX . Version-Release number of selected component (if applicable): kernel-2.4.21-15.ELsmp How reproducible: Always Steps to Reproduce: 1.Connect to Catalyst Switch 2. 3. Actual Results: Line can be observed being brought up and down continuosly Expected Results: Line should stay up Additional info:
I had same type symptoms. Catalyst switch set to 100/Full. Tried to use both mii-tool and ethtool to hard set eth0, but it always went to 100HD. Set switch to autonegotiate; set eth0 to autonegotiate, and it connected at 100FD. Odd. Some of my settings.... unit = Dell 2650 kernel = 2.4.21-15.0.3.ELsmp network: - class: NETWORK bus: PCI detached: 0 device: eth0 driver: tg3 desc: "Broadcom Corporation|NetXtreme BCM5701 Gigabit Ethernet" network.hwaddr: 00:0B:DB:A8:6E:29 vendorId: 14e4 deviceId: 1645 subVendorId: 1028 subDeviceId: 0121 pciType: 1 pcibus: 5 pcidev: 6 pcifn: 0 - class: NETWORK bus: PCI detached: 0 device: eth1 driver: tg3 desc: "Broadcom Corporation|NetXtreme BCM5701 Gigabit Ethernet" network.hwaddr: 00:0B:DB:A8:6E:2A vendorId: 14e4 deviceId: 1645 subVendorId: 1028 subDeviceId: 0121 pciType: 1 pcibus: 5 pcidev: 8 pcifn: 0 - log: **before autoneg** Aug 10 06:34:23 bes kernel: tg3: eth0: Link is down. Aug 10 06:34:26 bes kernel: tg3: eth0: Link is up at 100 Mbps, half duplex. Aug 10 06:34:26 bes kernel: tg3: eth0: Flow control is off for TX and off for RX. Aug 10 06:51:53 bes kernel: tg3: eth0: Link is down. **after autoneg** Aug 10 06:51:55 bes kernel: tg3: eth0: Link is up at 100 Mbps, full duplex. Aug 10 06:51:55 bes kernel: tg3: eth0: Flow control is off for TX and off for RX. --lwarden
That is expected behavior in comment #1. If the switch is not doing auto-negotiaton and the link has been forced to a certain speed, tg3 hardware has no way to determine the link settings. Without auto-negotiation there is no way for the switch to tell the tg3 PHY what duplex is in use, so half-duplex is used in such situations. If you tell ethtool to disable auto-negotiation it should do full-duplex for you if you ask it. You haven't indicated if you are doing that, are you?
This bug is filed against RHEL 3, which is in maintenance phase. During the maintenance phase, only security errata and select mission critical bug fixes will be released for enterprise products. Since this bug does not meet that criteria, it is now being closed. For more information of the RHEL errata support policy, please visit: http://www.redhat.com/security/updates/errata/ If you feel this bug is indeed mission critical, please contact your support representative. You may be asked to provide detailed information on how this bug is affecting you.