On Sat, Oct 13, 2012 at 03:27:00PM +0200, Sergio Callegari wrote: > Dear Sarah Sharp, > > please take my apologize in advance if this message should have > been sent elsewhere rather than your email. > From the 'Reporting bugs' doc I got the feeling that bugs should be > reported to subsystem maintainers directly, but maybe I am wrong and > a ML would be more appropriate. It's always good to Cc the mailing list as well, so that other people with the same problems can see the mailing list archive. > My hardware is a system with an ASROCK N68S motherboard, AMD Phenom > II processor and a VIA VL800/801 PCI USB3 adapter. > > My software consists in 64bit linux with the ubuntu precise userspace. > I have tried both the ubuntu 3.2 kernel and the mainline kernels up to 3.5. Not that I think it will help, but have you tried 3.6.2? Did hibernate work with any older kernel version? > The issue that I see is that the xhci driver prevents the system from hibernating. > The behavior is weird: when you ask the system to hibernate, the OS > does everything it should to save the state to disk and also appears > to shut down the system (the fans stop). However, after half a > second the system restarts by itself, resuming from the hibernation. Sounds like your host controller is sending a wakeup event that brings the system out of hibernate. > If I remove the VIA usb3 card, the system hibernates correctly and > resumes from hibernation fine (only when asked to do so!). > Similarly, if I manually unbind the xhci_hcd driver from the usb > card on the pci bus, before hibernation then the system hibernates > and resumes in the correct way. > > I am writing this message to assure that you are aware about this > issue. It has already been reported on the ubuntu bug tracker, but I > am not sure whether ubuntu has forwarded it upstream. Ubuntu doesn't forward me bugs relating to xHCI, although I wish they would. > A hack to solve the issue consists in adding rules to the power > management subsystem to automatically unbind xhci before hibernation > and rebind it at wakeup (using information stored in a temporary > file before hibernation), as proposed at https://bugs.launchpad.net/ubuntu/+source/linux/+bug/522998/comments/61. > However, this solution looks rather inelegant. > > You can get detailed information about my hardware at > https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1058725. It doesn't look like you posted dmesg that includes the failed hibernate attempt. That would be useful. You could also compile with CONFIG_USB_XHCI_HCD_DEBUGGING turned on, which could provide some more insight into what the xHCI driver is doing. Sarah Sharp -- 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
