On Wed, Feb 09, 2011 at 05:10:32PM -0500, Alan Stern wrote: > On Wed, 9 Feb 2011, Sarah Sharp wrote: > > > I've been testing a 3G modem that uses the option driver in 2.6.37, and > > we noticed that the device wasn't going into autosuspend. Upon further > > investigation, we found that remote wakeup is disabled by default, and > > the option driver will only suspend the device between transmissions if > > remote wakeup is enabled by userspace. > > No, you have misunderstood. Autosuspend fails if the device _doesn't > support_ remote wakeup -- this has nothing to do with the user's > setting of the power/wakeup attribute. > > Also, I'm not sure what you mean by "between transmissions". The > option driver requires remote wakeup while the device file is open but > not while the file is closed. Hence there should be nothing > preventing autosuspend when the device is not in use. This particular modem has crappy firmware, and transmits useless messages every 2 seconds. I've just found that out, so that explains why the device wasn't going into suspend. > > Is there a reason remote wakeup is disabled by default? > > The power/wakeup attribute applies to system sleep (suspend or > hibernation), not runtime suspend. It is supposed to be disabled by > default for all devices except those normally expected to wake up a > sleeping computer (power button, lid, keyboard, maybe network > interface). > > It is not enabled by default for USB hubs, for example, even though all > hubs support remote wakeup. Otherwise your sleeping laptop would wake > up the moment you unplugged the USB mouse. Ah, ok, I was confused about the sysfs files. I guess I'll have to look at the Documentation directory again. > > Is there a > > history of devices that use the option driver having issues with > > auto-suspend? It seems like if the USB device advertises remote wakeup > > and the device is self-powered, then we should turn on remote wakeup by > > default. > > This question doesn't mean what you think it means. Besides, are you > really sure the device advertises remote wakeup capability? Can you > post the "lsusb -v" output for the modem? I don't have access to the device right now, but I did eyeball the output and I saw "remote wakeup" in the modem listing. It's definitely self-powered too. My understanding was that when the device receives an incoming message while suspended, it will issue a remote wakeup. When I saw the value of the wakeup sysfs file was disabled, I thought the USB core was disabling remote wakeup from auto-suspend, and thus not letting the device go into autosuspend at all because remote wakeup was required. Hence the confusion. 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
