On Wed, 22 Sep 2004, Oliver Neukum wrote: > Remember, mmapped pages are not written to swap.
It's not clear that they need to be. They will just remain in system memory. Or wherever the system memory goes during a power-off suspend. Likewise for iomapped pages remaining in device memory. > Affected are all > partitions mounted rw. And it is not limited to disks. There are also > nfs, nbd, i-scsi ... Again, these things shouldn't matter. If there are dirty buffers, they can remain dirty across the suspend. Whether that's a good idea is a separate question -- but the PM core shouldn't be responsible for flushing buffers. That's the block subsystem's job. And all the flushing should be completed _before_ the suspend code starts changing any devices' PM state. > I am sure the swsusp people will really appreciate the added > necessity to first compute the set of devices that need to run. Left unsaid is how a suspended device can be woken up automatically when it is needed. Presumably that would be handled by whoever put it to sleep in the first place... And if you manually suspend a drive that contains a swap partition, you get what you deserve! :-) > If you want to design a model acceptable to all subsystems _and_ > the core power management system, it needs to be simple above > all. Unfortunately I only have a limited grasp of the issues involved, let alone the special requirements of all the different subsystems... Alan Stern ------------------------------------------------------- This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170 Project Admins to receive an Apple iPod Mini FREE for your judgement on who ports your project to Linux PPC the best. Sponsored by IBM. Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php _______________________________________________ [EMAIL PROTECTED] To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
