Description of Problem: Using Netgear FA-311 10/100 PCI ethernet card (NatSemi module) on a dual- processor PIII mobo, here are the initial bootup messages for eth0: Aug 24 13:15:56 localhost kernel: PCI: Found IRQ 11 for device 00:10.0 Aug 24 13:15:56 localhost kernel: eth0: NatSemi DP83815 at 0xf8917000, 00:02:e3:02:ef:f9, IRQ 11. Aug 24 13:15:56 localhost kernel: eth0: Transceiver status 0x7869 advertising 05e1. Aug 24 13:15:56 localhost kernel: eth0: Setting full-duplex based on negotiated link capability. When we get to run level 3, though, and try to bring up eth0, the kernel panics. It doesn't even write the panic message to the dmesg or boot.log or /var/log/messages, just dumps it to the screen and dies. We're going to attach a dump of the contents of the screen after this happens (and the machine locks up). Version-Release number of selected component (if applicable): How Reproducible: This happens every time if we choose the SMP option for the kernel and use a Netgear card. Removing a processor, changing network cards, or choosing the uni-processor kernel on boot, all prevent the crash. Get a Netgear PCI FA-311 10/100 card. Put it in a dual-processor motherboard. Install roswell and boot it up. Steps to Reproduce: 1. Install a Netgear PCI FA-311 10/100 card (natsemi chipset) in a dual- processor machine 2. Install roswell 3. Boot into networking config Actual Results: Expected Results: Additional Information: Going to copy the screen dump by hand, as it doesn't put it ot any file we can find, and attach to bug report after submission.
> The 2.4.7-2.99 kernel seems to fix the bug. We were able > to reproduce it every time on the earlier one, and it doesn't oops at > all now. thanks for testing!