Created attachment 22595 [details]
dmesg output from the affected machine

My test machine based on Athlon 64 X2 3800+ and a NVidia chipset suffers from hibernation regression where the second attempt to hibernate in a row hangs it right before the creation of the image.

Since the issue is 100% reproducible, I was able to bisect it and that turned up the following commit:

commit 3f4a782b5ce2698b1870b5a7b573cd721d4fce33
Author: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Date:   Fri Jul 3 11:25:16 2009 -0400

    [CPUFREQ] fix (utter) cpufreq_add_dev mess

    OK, I've tried to clean it up the best I could, but please test this with
    concurrent cpu hotplug and cpufreq add/remove in loops. I'm sure we will make
    other interesting findings.

    This is step one of fixing the overall locking dependency mess in cpufreq.

    Signed-off-by: Dave Jones <davej@redhat.com>

First-Bad-Commit : 3f4a782b5ce2698b1870b5a7b573cd721d4fce33