Re: sleep works in 2.6.12-rc2 (was Re: applying patch)
On Sat, 2005-04-09 at 17:25 +1000, Cedric Pradalier wrote:
> According to Tamas K Papp, on Fri, 8 Apr 2005 23:24:22 -0400,
> >On Fri, Apr 08, 2005 at 11:05:36PM -0400, Tamas K Papp wrote:
> >> Thanks for your help. I compiled a 2.6.12-rc2 kernel and I am happy
> >> to share that sleep now works on my powerbook5,4: the machine sleeps
> >> perfectly, wakes up perfectly, and usb devices are powered up every
> >> time. I will test it more tomorrow, but it seems very sound.
> >OK, that was a bit premature ;-) I already found a glitch: if I
> >generate an USB event (eg click the buttons on the mouse) while the
> >machine is sleeping, it will try to wake up but the screen remains
> >blank. If somebody on this list is on the kernel list too, and they
> >don't know already about this, please forward this (but I expect that
> >they would).
> Just a side comment on that.
> For me this means that all the usb devices are still powered when the machine is asleep.
> They can indeed consume a lot of energy and greatly reduce battery life.
> I think I'll continue to unload/reload all the module, except if there is an option
> telling the modules to switch the USB system off completely when going to sleep.
It is not a good idea. If you unload the OHCI or EHCI module during
sleep, the controller will not be put to sleep _at_all_
The USB standard specifies a specific "suspended" state that is using
during machine sleep. In this state, only a little bit of power is
provided to the bus, enough to let some USB devices to trigger a wakeup.
The kernel drivers do suspend the root hubs.