After installing Gentoo on my HP Omnibook xe3, I had major problems using the built-in EtherExpress Pro 100 interface. When running at 10 Mbps half duplex, the driver would lock up after a few seconds of network activity. A "wait_for_cmd_timeout" error would be logged to dmesg when this happened. Shutting down the interface and bringing it back up fixed things briefly. There were no problems running at 100 Mbps full duplex. I didn't get a chance to test other duplex/speed modes. I tried using the eepro100 drivers from the 2.4.18 and 2.4.19-pre7 source trees. The 2.4.18 version works fine at both 10 and 100 Mbps. 2.4.19-pre7 had the same lockups at 10 Mbps.
You may want to try gentoo-sources-2.4.19 (note gentoo-sources, not linux-sources). This is a 2.4.19-pre7-ac2 kernel, so we can see if Alan Cox fixed it. Sounds like Marcelo should be informed of this. I"ll send him an email.
I sent an email to Marcelo about this. I need to follow-up.
After installing and building from gentoo-sources-2.4.19 I still have the same problem. If I generate a significant amount of traffic, e.g. by running "emerge rsync" while connected at 10 Mbps half duplex, the connection hangs and dmesg displays several lines of eepro100: wait_for_cmd_done timeout! I can do more testing if needed.
I got the same on my Omnibook 6100 (got this during install, but it went away when I switched to the xfs-sources(2.4.18).... I discussed this with Joachim (migrating XFS to Gentoo) while testing his patch...But it's not only that...Going back to gentoo-sources gave me trouble with: a. the time-outs on my NIC (eepro100) back [kernel] eepro100: wait_for_cmd_done timeout! b. wacom-tablet stops working: [kernel] wacom_intuos_irq: received unknown report #1 c. My VGA-console gets f#^&-ed up (seems something wrong with the last line, after that diagonal output ! (although funny at first, very annoying quite quickly ;-) Is there more going on ?
The Intel-made driver is supposedly of inferior design-quality, but does not exibit this bug. I've installed it as a module, and the problem seems gone. See: http://support.intel.com/support/network/adapter/pro100/31351.htm#2
what is the status of this bug? Has this been tested in more recent gentoo kernels (preferrably gentoo-r9)
I'm assuming it works in 2.4.19-gentoo-r9, please re-open the bug if that is a bad assumption