On Thu, 29 Jan 2009, Daniel Qarras wrote: > > Besides, the real problem isn't autosuspend at all, as > > far as I can tell. The first problem is that the computer wakes up > > immediately after going into hibernation. The second problem is that > > the computer wakes up from suspend immediately when a USB mouse is > > plugged in. > > Yes, this is a correct description. > > > I don't know what's causing the first problem (which has been around > > since 2.6.24). Maybe a firmware bug. Perhaps doing > > "rmmod ehci-hcd" before hibernating will help. > > Correct: > > https://lists.linux-foundation.org/pipermail/linux-pm/2009-January/019251.html > > (Of course for a layman like me it is a bit saddening that this worked earlier but not anymore.) Perhaps Rafael has some idea on what has changed in the meantime to cause this problem. It could have something to do with the way ACPI is now integrated with the PCI drivers, for example. > > To attack the second problem, it would help to see a dmesg > > log showing the immediate wakeup from a kernel with CONFIG_USB_DEBUG > > enabled. > > Please find such a dmesg attached. The log doesn't indicate a real cause. It does show that wakeup was disabled on the EHCI controller, but nevertheless that controller must have been responsible for waking up your system when the mouse was plugged back in. You can test this by unloading ehci-hcd before suspending. The fact that _un_plugging the mouse didn't cause a wakeup has an interesting explanation. The mouse is a low-speed USB device, and as such is managed by the UHCI controller. Evidently that controller really was disabled for wakeup, since unplugging the mouse didn't have any effect. But when a new USB device is plugged in, the system doesn't know what speed it runs at. So the device is attached first to EHCI, and then if it proves not to be high-speed it gets switched electronically over to UHCI. Thus the plug-in event stimulated the EHCI controller to cause a wakeup. This does seem like a firmware problem, but maybe Rafael can suggest a way to work around it. Alan Stern _______________________________________________ linux-pm mailing list linux-pm@xxxxxxxxxxxxxxxxxxxxxxxxxx https://lists.linux-foundation.org/mailman/listinfo/linux-pm
