On Thu, Dec 17, 2009 at 11:05 PM, Alan Stern <stern@xxxxxxxxxxxxxxxxxxx> wrote: > On Thu, 17 Dec 2009, Yuping Luo wrote: > >> > You should try removing the set_freezable() call. It might solve your >> > problem. >> > >> it works while some trivial complaints like "sending command-failure >> status" occur. > > Those may be normal. > >> > But a better approach would be to use register_pm_notifier (see >> > include/linux/suspend.h). When you get a PM_SUSPEND_PREPARE message, >> > call usb_gadget_disconnect(). When you get a PM_POST_SUSPEND message, >> > call usb_gadget_connect(). Then it won't matter that the thread gets >> > woken up late. >> > >> > Alan Stern >> > >> it doesn't work, since when system wakeup, the >> pm_notifier_call_chain() called very late, after the restarting the >> tasks. > > That's the whole point. pm_notifier_call_chain() runs very late, after > the FSG thread is awake. Hence when usb_gadget_connect() is called and > the D+ pullup is enabled, the thread will be running and able to handle > exceptions. > > You will need to modify the controller driver if it doesn't implement > the ops->pullup method correctly. > > Alan Stern > > my pcd_pullup() already supported, it will set the RS bit with 0 for USBCMD. however, my device controller resumed earlier than the pm notifier, and will start the usb controller, the RS bit will be set with 1, so the device enumeration procedure started not as we expected. Maybe I need delay the resume of the device controller. Yuping Luo -- 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
