Red Hat Bugzilla – Bug 55742
spinlock code stuck
Last modified: 2007-04-18 12:38:02 EDT
From Bugzilla Helper:
User-Agent: Mozilla/4.0 (compatible; MSIE 5.5; Windows NT 5.0)
Description of problem:
Using kernel-2.4.2-2 and the Cisco-modified 2.4.2-7 (ALSA stuff added, can
get a list) an 'rpm -ba' of the kernel over NFS (options are
rw,v3,rsize=32768,wsize=32768,hard,udp,lock) makes the machine
unresponsive to network traffic (ping included). SysRq data tracked it
down to being a BKL spinlock issue. arjan, do you remember this IRC
conversation? You told me to test the 2.4.2-2 stock and the 2.4.9-6
errata. The errata doesn't break, but the stock does.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. over nfs with the options i used, mounting from a netapp f700 filer
2. rpm -ba the kernel
3. at the console, use sysrq to get eip
ALSA or ARLA ?
Either way, "fixed in errata" sounds like good advice....
Also, which network driver is this ?
ALSA, and eepro100. "Fixed in errata" is unfortunately not where we (Red Hat)
drop it. Again, because of Rational's lack of support beyond 2.4.2 currently,
Cisco cannot migrate to the errata. This is a good opportunity for them to use
developer hours, too. I'd like to facilitate the creation of a patch for them
to go on 2.4.2. I can provide the kernel SRPM they're currently using and
collaborate to gather any further diagnostic data.
I will escalate it through the proper channels if you like.