On Tue, Apr 12, 2011 at 12:54:52AM -0400, CAI Qian wrote: > > I wonder if it's possible the USB 3.0 hub refused the Link PM command > > to > > put the port into a lower power state. Can you test that hypothesis by > > unloading the usb-storage driver and turning on autosuspend for the > > USB > > device? I.e. find the device's entry in /sys/bus/usb/ and > > `echo auto > power/control`. Send me dmesg from that attempt. > Sorry, I don't understand. Which device are you talking about here? As > I mentioned before, my setup was, > PCI-E card -> USB3 VIA Hub (9-1) -> USB3 Disk (9-1.2) > > # ./lsusb.py > usb1 1d6b:0002 09 2.00 480MBit/s 0mA 1IFs (ehci_hcd 0000:00:1a.7) hub > usb2 1d6b:0002 09 2.00 480MBit/s 0mA 1IFs (ehci_hcd 0000:00:1d.7) hub > usb3 1d6b:0001 09 1.10 12MBit/s 0mA 1IFs (uhci_hcd 0000:00:1a.0) hub > usb4 1d6b:0001 09 1.10 12MBit/s 0mA 1IFs (uhci_hcd 0000:00:1a.1) hub > usb5 1d6b:0001 09 1.10 12MBit/s 0mA 1IFs (uhci_hcd 0000:00:1d.0) hub > 5-2 04b3:4010 02 2.00 12MBit/s 100mA 2IFs (IBM RNDIS/CDC ETHER) > usb6 1d6b:0001 09 1.10 12MBit/s 0mA 1IFs (uhci_hcd 0000:00:1d.1) hub > usb7 1d6b:0001 09 1.10 12MBit/s 0mA 1IFs (uhci_hcd 0000:00:1d.2) hub > 7-2 0624:0307 00 1.10 1.5MBit/s 100mA 2IFs (Avocent Avocent DSRIQ-USB) > usb8 1d6b:0002 09 2.00 480MBit/s 0mA 1IFs (xhci_hcd 0000:1a:00.0) hub > 8-1 2109:3431 09 2.00 480MBit/s 100mA 1IFs () hub > usb9 1d6b:0003 09 3.005000MBit/s 0mA 1IFs (xhci_hcd 0000:1a:00.0) hub > 9-1 2109:0810 09 3.005000MBit/s 2mA 1IFs (VIA Labs, Inc. 4-Port USB 3.0 Hub) hub > 9-1.2 0bc2:50a0 00 3.005000MBit/s 0mA 1IFs (Seagate FA GoFlex Desk 01234567) > > Do you mean VIA hub? If so, that would be 9-1. No, I didn't mean the VIA hub, sorry I wasn't clear which USB device I meant. I meant that I wanted you to enable auto-suspend for the USB 3.0 storage device. You can't auto-suspend a USB hub unless all the devices under it are auto-suspended, so turning on autosuspend for the USB 3.0 hub would not have any effect. (Especially since it's already on by default.) > So if I do, > # modprobe -r usb-storage > # echo auto >/sys/bus/usb/devices/9-1/power/control > > So in this case, I don't even need to echo disk >/sys/power/state? Anyway, here is > /var/log/dmesg, > http://people.redhat.com/qcai/sarah/dmesg I'm trying to get you to run a different test, unrelated to hibernate. I'm trying to test whether the VIA hub will successfully suspend the USB 3.0 hard drive (this is a USB bus suspend, unrelated to system suspend). I'd like you to try with Andiry's 5 patches applied first. In order to suspend the hard drive, we need to enable auto-suspend for it. So you need to run this command: # echo auto >/sys/bus/usb/devices/9-1.2/power/control However, the usb-storage driver won't allow the device to be suspended until it's unmounted. So you either need to unmount the disk, or unload the usb-storage driver. Please start capturing dmesg, and then unload the usb-storage driver. If auto-suspend is working properly, you should see messages like: "usb 9-1: usb auto-suspend" If it's not auto-suspending properly, then we need to address that issue before we look at the hibernate path. > > I'm still not sure exactly what's going on here. My system is still > > not > > hibernating properly (even with the xHCI driver blacklisted) with > > 2.6.39-rc2, plus the patch revert you mentioned, so I can't test this > > myself. I'll try to compile on top of Linus' latest tree tomorrow. > > The patch reverting is not needed any more, as it was fixed in the > lastest Linus tree already. > http://marc.info/?l=linux-kernel&m=130254808802179&w=2 > > > In the meantime, can you test the hypothesis I had about the VIA hub > > refusing the link PM commands? After that, would you please test > > The second attempt of hibernate/resume is still refused after those > modprobe -r & echo auto commands. > > > Andiry's one work around patch on top of Linus' tree, instead of his > > five patches? It may help if the issue really is the USB 3.0 hub > > refusing to accept link PM commands. > > Are you refering to this patch? > http://marc.info/?l=linux-usb&m=130167678226265&w=2 Yes, I meant that patch. > If so, I had the same problem with this patch only. However, I have not > tested it with modprobe -r & echo auto there yet. Let me know if you'd > like me to test this as well. Hmm, ok, something else might be off then. I'll try with Linus' latest tree and see if I can hibernate my system. 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
