#2432 BLOC First D: Suspend kills USB.

[Zarro Boogs per Child]

#2432: Suspend kills USB.
----------------------+-----------------------------------------------------
  Reporter:  warp     |       Owner:  marcelo         
      Type:  defect   |      Status:  new             
  Priority:  blocker  |   Milestone:  First Deployment
 Component:  kernel   |     Version:                  
Resolution:           |    Keywords:                  
  Verified:  0        |  
----------------------+-----------------------------------------------------
Changes (by warp):

  * priority:  normal => blocker
  * milestone:  Trial-3 => First Deployment

Comment:

 Alright, the above fix is wrong.

 We have three different issues, one that I'm still chasing through the USB
 code, one which can wait, and one which is partly OFW trying to save time.

 The first is that for some reason the struct usb_device state is not
 getting properly set to SUSPEND, resulting in the OHCI resume code never
 getting called.

 The second is that USB is taking way too long to resume in general, up to
 half a second very easily.

 The third is that OFW is trying to power on the ports early in the process
 to save time later, this is sadly confusing the Linux OHCI resume code a
 great deal.  The drivers really should be far more mistrustful of what
 happens during suspend, but since this is a speed optimization, we can
 drop it for the moment to make things work.

 Bumping to FRS as a blocker.

-- 
Ticket URL: <http://dev.laptop.org/ticket/2432#comment:9>
One Laptop Per Child <http://laptop.org/>