On Fri, 18 Jan 2013, Oliver Neukum wrote: > On Friday 18 January 2013 14:09:26 Alan Stern wrote: > > > The autosuspend delay at the USB level probably should be set to 0; > > then the autosuspend delay at the SCSI disk level will control the > > actual power changes. > > What about things like lsusb? Should we better not go down quite > to zero? I don't think lsusb will matter. usbfs prevents autosuspend for as long as the USB device file is open. > > Note that drives with removable media are polled automatically by the > > kernel at intervals of 2 seconds by default. If the autosuspend delay > > is longer than that, it will never have a chance to take effect. > > Unfortunately most storage devices in the usual dongle shape, even if > they do not have a removable medium claim to have a removable medium. Some do and some don't. > Actually I consider this a bug in the spec. It should specify a method > for the device to tell the driver whether medium removal or insertation > will trigger remote wakeup. Neither the USB mass-storage spec nor the SCSI spec includes any support for remote wakeup at all. So the lack of a method doesn't matter; medium changes _never_ trigger remote wakeup! You might conclude that _this_ is a bug -- they should cause remote wakeups. The problem is that this isn't possible; a USB mass storage device might be a real SCSI host, communicating with real SCSI devices over a real SCSI bus. In such a situation, the SCSI host has no way to know when a device experiences a medium change. 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
