On Mon, 9 Aug 2010, Oliver Neukum wrote: > Am Sonntag, 8. August 2010, 23:03:55 schrieb Alan Stern: > > The patch > > below is just a proof-of-principle; in the end we'll want something > > more flexible. Maybe a sysfs attribute to select the conditions under > > which autosuspend is allowed. > > Ideally we want autodetection. But I agree there's no reason that would > have live in kernel space and sysfs is the way to communicate this > information. > > However, after a bit of thought, it seems to me that if we are going > for an attribute in sysfs a model based on open/close makes no sense > for sd. > The thing is that I see no device that would benefit from that algorithm. USB flash drives? Maybe you think they would be better off with an idle timeout. > Are there any devices which can suspend with medium and opened > which cannot suspend under all circumstances (while not processing > a command)? Rotating drives that need to be spun down shouldn't be suspended if a command will be issued in the near future. Also, spin-up and spin-down operations should be minimized, as they cause physical wear. It might be a good idea to ask the people on linux-scsi if they have any helpful ideas. They may be more familiar with the issues. Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html