May I ask the question again ?
I ended up crashing my system for good trying pm-utils so I guess that if someone knows how to avoid that, it is the right time to say it !
thanks

Christophe.

REF
basic-pm-debugging.txt - Debugging suspend and resume
http://www.kernel.org/doc/Documentation/power/basic-pm-debugging.txt

swsusp.txt - Goals, implementation, and usage of software suspend (ACPI S3)
http://www.kernel.org/doc/Documentation/power/swsusp.txt

In cooking, latest kernel upgrade (2.6.34 --> 2.6.36) broke pm-suspend on my laptop: it goes to sleep but on wakeup the LEDs flash and it tries to read the CD but the touchpad and keyboard are dead (no caps lock light response) and the screen remains black--i.e. system hang.

cat /proc/config.gz |gunzip -c|grep PM_
# CONFIG_PM_DEBUG is not set
CONFIG_PM_SLEEP_SMP=y
CONFIG_PM_SLEEP=y
CONFIG_PM_STD_PARTITION=""
# CONFIG_PM_RUNTIME is not set

So
basic-pm-debugging.txt - Debugging suspend and resume reference
can't be followed (/sys/power/pm_test is absent.)

Please set
CONFIG_PM_DEBUG=y
in next build.


Concerning older machines:

cat /proc/config.gz |gunzip -c|grep APM
# CONFIG_APM is not set

could be set and apmd included.

iirc, pm-utils 1.4 worked, so it's probably no a hal issue

Other possibilities are udev, and dbus...

but the kernel test would be most useful since I don't get any logged events from a hung live system.