I think that there may be a link between the kernel crashes I
reported earlier and some tunings with powertop. In powertop there
are some parameters that can be modified for decreasing the laptop
power consumption. Namely, I adjust parameters so that everything is
reported as "good" in the "tunable" powertop tab (this indeed
somehow decrease the power consumption reported by my laptop). These
parameters are "wireless power saving", "Enable* Audio codec power
management", "Enable SATA link power management", "VM writeback
timeout", and* "Runtime PM" for several PCI Devices.
I made some tests with the last SID kernel. If I do not tune
anything, the system remained totally stable for several days. Then
I tried some different tuning. It appeared that the kernel crashed
when returning from suspend to RAM immediately* after that the "VM
writeback timeout" was activated (i.e. switched on "good" in
powertop), and only this one. So the problem may comes from this. I
say "may", because it is difficult to be sure: the crash occurs
quite randomly, and is not 100% reproducible.
I hope this can help someone to identify the bug. Other people
experiencing similar crashes may try to look at these powertop
tunings for confirming/infirming this claim (and eventually for
solving their problem, at least temporarily).