Suspend to RAM regression retraced

[Jean-Marc Valin <Jean-Marc.Valin@USherbrooke.ca>]

Hi,

A while ago I reported a regression (http://lkml.org/lkml/2005/9/21/290)
in suspend to RAM that happened with kernel 2.6.12 on my Dell D600
laptop. After several months of testing (this is my work machine and the
bug takes time to reproduce), I have finally narrowed it down. It seems
like to problem was introduced between 2.6.12-rc5 and 2.6.12-rc6. 

Basically, what happens is that with 2.6.12-rc6, my machine *sometimes*
doesn't resume when I suspend it. This happens especially when it has
been running for a while. It almost always works when I just rebooted,
or if I just successfully resumed. So it behaves like "something" gets
randomly corrupted, at which point the machine still works, but will not
resume if I suspend it. Also, I've observed the same behaviour with and
without preemption enabled. 

Can someone have a look at what could cause the problem and fix it? I
can provide more information if needed. BTW, I'm not on the list so
please CC to me.

Thanks,

	Jean-Marc

P.S. Machine setup is:
Ubuntu 5.10 (but problem also observed on 5.04 and old Debian unstable)
Dell Latitude D600 (Bios rev. A14)
Pentium-M 1.6 GHz / 1 GB RAM
ATI Technologies, Inc. Radeon Mobility 9000 M9 (R250 Lf)