On Mon, Jul 26, 2010 at 06:24:17PM +0800, Andiry Xu wrote: > Hi Sarah, > > This is xHCI pm patchset v5. > > Changelog from v4: > > 1. Remove atomic operation in port pm and remote wakeup patch. > 2. Refine xhci_cmd_ring_init() and xhci_event_ring_init(). > 3. Free all virt devices if the xHC restore error and reset. > 4. Remove redundant changes to xhci.h. > 5. Some semantic refine. > > The patchset is based on your master branch, but I cannot enter S3 with > 2.6.35-rc4, even with xhci-hcd unloaded. so I use 2.6.34 for test. > > I have tested port PM, remote wakeup and S3. I've tested USB3.0/2.0/1.1 > device, plug in before/during S3 suspend and after S3 resume, and run S3 > for several times. In most cases the devices will be recognized > successfully and work fine. But there are some issues listed below. > > 1. Every time the system resume from S3, the STS_SRE bit is set, which > indicates the xHC restore error, so the xHC should be reset. The hub > driver will try to resume the devices, but it will found the devices > lost and then free and re-initialize them. It will show "cannot reset > HCD device state" in the dmesg. I'm not quite sure how to fix this. Do you use the Save and Restore Operations (section 4.23.2.1)? If so, the xHCI host controller should not indicate that it needs a host controller reset. If you are using those operations, and the host still says there is a restore error, then it's likely that there's something wrong in the code. The NEC host is very good about giving back errors if the slightest field is set incorrectly. 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
