I'm using FC on a Dell 5150 laptop, with a nvidia GeForce 5200Go and the NVidia closed-source drivers. Power suspend _almost_ works with this setup, except that on resume, the screen remains powered off. However - if I use Matthew Garrett's vbetool (available from http://www.srcf.ucam.org/~mjg59/vbetool/) then I can get things working again. After a resume, a 'vbetool post' reinitialises the screen and powers it back up, and I'm a happy little bunny again. Please consider shipping this tool, and maybe hooking it into the ACPI power control setup. It makes things work, and I can't give higher praise than that!
Hm, this seems like something better fixed in the drivers and subsystems itself. Running this randomly looks like a bad idea.
Yes, in an ideal world the drivers would do all this, and do it right. Unfortunately, the nvidia driver is necessary, broken, and not fixable (because of it's license). The vbetool makes the laptop far more useful than it would otherwise be (lack of suspend is a serious nuisance).
For a lot of hardware, there is inadequate documentation to allow for reinitialisation from scratch. For laptops, it's also necessary to reinitialise the backlight controller. In the general case, we have no idea whatsoever how to do that. vbetool takes the success rate in ACPI suspend/resume from ~20% to ~80%, which isn't too bad.
What are the consequences of running vbetool on a device that hasn't been tested to work with it before?
Some machines with extremely broken BIOSes will hang. The incidence rate seems to be very low, though - I've had two reports of this out of several thousand tests (it's installed and run by default on the current Ubuntu testing distribution). I wouldn't recommend it be run by default unless ACPI S3 is being used.
Included in the pm-utils package as of today.