From Bugzilla Helper: User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.5) Gecko/20031007 Firebird/0.7 Description of problem: kernel OOPS on various PCI access - all appear to be caused by the airo driver. In testing, experienced OOPSs so far with: 1) starting X doing PCI scan 2) running lspci Version-Release number of selected component (if applicable): kernel-2.4.22-1.2129.nptl How reproducible: Always Steps to Reproduce: 1. boot Fedora with kernel specified 2. normal boot (including probe for airo), wait for X to start 3. OOPs caused by PCI scan when X starts 4. further runs of lspci causes OOPs Additional info: Will attach OOPs output in further post.
Created attachment 96494 [details] syslog output - /var/log/messages Attached syslog output (from /var/log/messages) shows boot and then OOPSs caused by X PCI probe, then lspci. OOPS looks like it may be triggered by proc_config_on_close in airo driver.
Possibly a dup of 110204? That also shows similar OOPS (scan_device etc.) on X startup after airo probed for PCI.
The problem appears to be triggered by kudzu - stopping the kudzu service running at boot fixed the issue for me. I am wondering if this is caused by the PCI unregister issue in the airo driver noted in this lkml post: http://marc.theaimsgroup.com/?l=linux-kernel&m=105948393807224&w=2 The kernel booted does not look to contain the PCI unregister fix noted and during kudzu probe I am presuming it may load the airo module to scan for devices and then unload it after scan. If this does not unregister correctly, subsequent access may cause the OOPS. If I get chance, I'll try to apply the driver fixes posted to lkml and re-test. Please shoot me down in flames if this is all complete rubbish of course... :-)
I haven't been able to reproduce the OOPS since moving to FC2 and 2.6 kernel. Assumed resolved.