> >Send patches if you find out how. The public part of the MacOS X source >code might be a place to look for clues (if you happen to find out what >they do to make multiword DMA 2 mode restore properly on all machines, I'm >sure BenH would appreciate a hint). > >The whole time spent for wakeup is in the kernel, reprobing the ADB bus >and restoring the IDE disk state after powering the disk back up. I can't >see how we can speed up the procedure.
I have a bunch of delays that could eventually be shortened once I'm sure the sleep process works on all boxes. But unless MacOS X, I have to support old crappy HW and MacOS X has the advantage of it's threaded driver model, allowing it to restore a bunch of things simultaneously. For now, I still have to figure out how to make sleep actually work on the Core99 machines (still losing RAM content for now). Ben.
