Hi, I can readily recreate this issue with our device within a few hours of usage. I've seen this issue on kernels 2.6.24 and newer. The one I'm currently running is 2.6.32-rc3 with the aggressive autosuspend patch for cdc_ether. Thank you for spending time on this since it is becoming an issue for us. Please let me know when you have something you want me to test. BR, Rickard -----Original Message----- From: Alan Stern [mailto:stern@xxxxxxxxxxxxxxxxxxx] Sent: den 30 oktober 2009 00:53 To: Oliver Neukum Cc: Rickard Bellini; linux-usb@xxxxxxxxxxxxxxx; Nicolas Chevillot; Torgny Johansson; Greg Heinrich; Carl Nordbeck; Martin Chabot Subject: Re: additional debug output for autosuspend in cdc-ether On Thu, 29 Oct 2009, Oliver Neukum wrote: > Hi Alan, > > we have an issue here where the hub code discards a remote wakeup. > > Regards > Oliver > > > I've been testing the new fix and with additional debug prints. > > It seems I cannot reproduce the issue but I am hitting the other > > known issue where remote wakeup is signaled by the device but kernel > > never acts on it. > > However now this problem is much more evident than ever before. > > Please find two occurrences below. > > > > --------------------- First ----------------------------- ... > > (03:37:18.964000) 340782 ****************** Signalling remote > > wakeup****************** > > > > 2009-10-29T03:37:18.345+09:00 Ubuntu kernel: [ 2158.983913] ehci_hcd > > 0000:00:1d.7: port 4 remote wakeup > > 2009-10-29T03:37:18.354+09:00 Ubuntu kernel: [ 2158.989121] hub 7-0:1.0: > > hub_suspend > > 2009-10-29T03:37:18.354+09:00 Ubuntu kernel: [ 2158.989135] hub 7-0:1.0: > > usb_suspend_interface: status 0 > > 2009-10-29T03:37:18.354+09:00 Ubuntu kernel: [ 2158.989142] usb usb7: > > bus auto-suspend > > 2009-10-29T03:37:18.354+09:00 Ubuntu kernel: [ 2158.989147] ehci_hcd > > 0000:00:1d.7: suspend root hub > > > > [Hub is suspended even if remote wakeup was signalled!] Unforunately there is a race between ehci_bus_suspend() and ehci_irq(); if a remote wakeup notification arrives at the wrong time it will be ignored. I have known about this for a while but haven't done anything because it didn't seem to be particularly urgent or easy to trigger (it's a 20-ms window). The fix involves preventing the EHCI controller from generating IRQs during the suspend and checking to see if a remote wakeup signal has been received recently. I'll try to work on it during the next few days. Which kernel version is this? 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
