From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux 2.4.2-2 i686; en-US; rv:0.9.1) Gecko/20010607 Netscape6/6.1b1 Description of problem: Fairfax beta3 installed on a 6400 updated to 2.4.7-0.3 enterprise kernel, w/ 5GB ram and 2GB swap, test bios T42, and a perc3qc raid 1 container. While downloading several 1-10MB files with ncftp, kernel panics. May not happen everytime. Output of ksymoops attached. How reproducible: Sometimes Steps to Reproduce: 1. Fairfax beta3 installed on a 6400 updated to 2.4.7-0.3 enterprise kernel, w/ 5GB ram and 2GB swap, test bios T42, and a perc3qc raid 1 container. 2. Use ncftp to download several 1-10MB files. (I was using the kernel update rpms) 3. Look for kernel panic Actual Results: Kernel panic Expected Results: No kernel panic Additional info:
Created attachment 26497 [details] Output of ksymoops
Re-assigning to kernel
We (Red Hat) really need to fix this defect before next release.
Looks like the bcm5700 driver has a dereference NULL bug. I'm trying to find out who at Broadcom is the driver maintainer and have them look at it.
Broadcom has provided an updated driver, which will be in kernels 2.4.7-0.12 and later.
I noticed the mention of the broadcom driver and wanted to point out that the configuration exhibiting this error had the onboard intel nic configured and running on eth0 and the broadcom configured and running on the same subnet at eth1. I'm assuming that eth0 would have been the nic in use when the kernel panic occured.
>>EIP; f8a23966 <[bcm5700]MM_IndicateRxPackets+76/1b0> <===== in the oops trace suggests the broadcom driver was at least involved, and triggered me to classify this as a broadcom driver failure, especially as the driver had a known oops bug.
Closing as fixed.