#6433 NORM Never A: Distance will die after 30 seconds of inactivity due to suspend

[Zarro Boogs per Child]

#6433: Distance will die after 30 seconds of inactivity due to suspend
----------------------------------------+-----------------------------------
  Reporter:  bemasc                     |       Owner:  bemasc        
      Type:  defect                     |      Status:  new           
  Priority:  normal                     |   Milestone:  Never Assigned
 Component:  acoustic-measure-activity  |     Version:                
Resolution:                             |    Keywords:                
  Verified:  0                          |    Blocking:                
 Blockedby:                             |  
----------------------------------------+-----------------------------------

Comment(by gnu):

 It would be better to break Distance -- or remove it from the release --
 than to kludge either it or Ohm.

 The reason Ohm is suspending is because Ohm doesn't know about cpuidle.
 Every package that does anything in realtime is going to have this problem
 with Ohm.  The fix is not to teach every package to tell Ohm not to
 suspend.  The fix is to teach Ohm how not to suspend when you have runing
 processes.  We shouldn't turn off the CPU when there is demand for the
 CPU.

 Then once we suspend, the reason Ohm keeps us suspended despite incoming
 WiFi traffic from the other side is because the two copies of Distance are
 apparently communicating with multicast packets (nobody has confirmed this
 suspicion) -- and due to another bug, multicast packets won't break us out
 of suspend.  If they were talking unicast TCP, we'd immediately break out
 of suspend when the next packet arrives.

 We will never have time to work on these *fundamental* problems that break
 many things, unless we stop patching up the surface damage, and go back to
 work on the fundamentals of suspend.

-- 
Ticket URL: <http://dev.laptop.org/ticket/6433#comment:2>
One Laptop Per Child <http://dev.laptop.org>
OLPC bug tracking system