On Mon, Oct 11, 2010 at 08:42:55PM +0800, Xu, Andiry wrote: > Hi Sarah, > > I've tested on 2.6.36-rc7 with Buffalo USB3.0 HDD, playing a movie on it > and then use pm-hibernate to enter S4. I've try several times and the > movie resumes playing when the system wake up. It may take several > seconds for the USB3.0 HDD to be re-initialized, after that it will > resume working. The config is set with CONFIG_PM_RUNTIME=y and > CONFIG_USB_SUSPEND=n. Ok. I tried the patches against 2.6.36-rc7 on my other computer (desktop) and the USB 3.0 hard drive worked fine playing a movie across suspend and hibernate. So the issue is specific to my laptop. In fact, I was able to trigger an oops on shutdown after a hibernate in the gnome-power-man process with just the NEC host plugged in, but without the xHCI driver loaded. I've also seen crashes after segfaults in other gnome apps while the driver was loaded, but I don't think it was related at all to the xHCI driver: Oct 11 14:27:37 xanatos kernel: [ 149.537707] console-kit-dae[1072]: segfault at 0 ip 00007f3d06032890 sp 00007f3cfffadbe8 error 6 in libpthread-2.10.1.so[7f3d06025000+17000] Oct 11 14:27:37 xanatos kernel: [ 149.615147] gvfs-fuse-daemo[1369]: segfault at 0 ip 00007fcde0ec0890 sp 00007fcddfecd6e8 error 6 Oct 11 14:27:37 xanatos kernel: [ 149.615156] gvfs-fuse-daemo[1370]: segfault at 0 ip 00007fcde0ec0890 sp 00007fcddf6cc6e8 error 6 in libpthread-2.10.1.so[7fcde0eb3000+17000] Oct 11 14:27:37 xanatos kernel: [ 149.620030] in libpthread-2.10.1.so[7fcde0eb3000+17000] So I think there's something f***ed up in either my xserver, i915 driver or gnome apps, and I don't think it's the xHCI driver's fault. I'm willing to call the PCI/bus suspend patches tested, but I still need to test the USB device suspend patch. > Does this issue occurs with suspend and CONFIG_USB_SUSPEND=y? I haven't tried that yet. I'm not really sure there's any value in testing this, since I can trigger an oops without the xHCI driver loaded. > Do the two patches you attached help to resolve this issue? No, those patches solve other issues I noticed, but they do not solve the graphics/xserver issues after resume from hibernate. Please include them (or merge them) into your next patchset. > Is it the same issue with you another mail mentioning call trace in > xhci_endpoint_init()? I'm not really able to root cause the issue right now. I'm still concerned about that general protection fault oops, but I haven't been able to trigger it again. A co-worker looked at the markup, and suggested there might have been a double free of a segment from the DMA pool. But looking at the xHCI driver code, I don't see where that could be possible. > I will do more tests on my side and try to reproduce this issue. Don't bother. I'll see if upgrading my xserver or gnome environment helps. 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
