1) The APM BIOS stinks with battstat. Unfortunately polling the battery (which battsat does once per second) via the APM bios seems to cause keyboard interrupts to get dropped, which is veryyyyyyyyyy annoying.
2) Whenever my linksys WPC11 PCMCIA card is plugged in and the AC power is plugged in, the machine refuses to sleep (rather, it almost always refuses to sleep). It does suspend if AC is unplugged and the PCMCIA card is plugged in, but this seems kind of dumb and it crashes about 10% of the time.
It starts up the fan like it's about to suspend, but the suspend never actually happens, and it makes a beeping noise. I've seen the same behavior with the orinoco_cs driver and the wlan-ng driver with RH9.
3) ACPI used to be very buggy and now very nearly works, but I get the exact same refusing to suspend behavior, but this time regardless of AC or PCMCIA cards plugged in or not. I've tried disabling PCMCIA altoghther with no luck.
To get around the #2 problem, I tried to have apmd unload the network driver before it suspends, but it appears it gets stuck even before it notifies to apmd.
I'm also a bit confused as to why suspend is doing anything at all. I looked at acpi-devel and read that the S3 state isn't supposed to work at all under 2.4. Is Fedora using some sort of unholy marriage of apm and acpi when acpi is enabled?
At this point, I'm not sure exactly what to blame. I assume that the APM BIOS is buggy, and ACPI is really my only hope.
Can anyone suggest a way to track this nonsense down?
Just for kicks I tried 2.6-test9 (kudos to whoever updated modutils to make this a reasonable endevor) which behaves differently, and will suspend and come back (though it appears that the uhci driver is totally trashed). Maybe there's some hope there.
Sorry for the sort of rambling post. I've been playing around with these issues for a few months now.
-- Brian Perkins
