#9896 HIGH 1.5-sof: Ad-hoc network fails when XO goes into sleep

[Zarro Boogs per Child]

#9896: Ad-hoc network fails when XO goes into sleep
------------------------------------+---------------------------------------
           Reporter:  edmcnierney   |       Owner:                    
               Type:  defect        |      Status:  new               
           Priority:  high          |   Milestone:  1.5-software-later
          Component:  not assigned  |     Version:  1.5-B3            
         Resolution:                |    Keywords:  network ad-hoc    
        Next_action:  diagnose      |    Verified:  0                 
Deployment_affected:                |   Blockedby:                    
           Blocking:                |  
------------------------------------+---------------------------------------

Comment(by Quozl):

 Tests on os116.

 Three laptops were set up.  One was the master that was used to create a
 wireless network.  One was the slave that only ever attempted to join on
 request.  One was a slave that only ever scanned.  The master was set to
 various states; on, suspend, or sleep.  The following table summarises the
 results.

 ||''master state''||''network nodes''||''slave can join''||''slave
 behaviour''||''master behaviour''||
 ||on||1||yes||good||good||
 ||on||2||yes||good||good||
 ||sleep||1||no||good||good||
 ||sleep||2||yes||good||good||
 ||suspend||1||no||bad||good||
 ||suspend||2||yes||good||good||

 Sleep (directed suspend) now properly dismantles a single node ad-hoc
 network, during sleep it cannot be found by other laptops scanning, and on
 resume the laptop that created the ad-hoc network has no access point icon
 for it.  It must be created again.

 Sleep now properly handles a multi-node ad-hoc network that it had begun,
 during sleep it can be found by other laptops scanning, and on resume the
 laptop that created the network discovers it in subsequent scanning.

 Suspend (idle suspend) does not properly provide a usable single-node ad-
 hoc network, in that the ad-hoc network can be found by other laptops
 scanning, can be selected, but does not assign an IP address.  The laptop
 attempting to join the network can't join it again until another access
 point is selected, followed by the ad-hoc icon, and only if the laptop
 that created the network is awake.  On resume the laptop that created the
 network continues to operate as expected.

 Suspend does properly work with a multi-node ad-hoc network, in that the
 ad-hoc network survives, and remains "connected" on resume.

 Does anyone have an opinion on that one bad condition; is it worth
 addressing?  It would require wake on link-local address establishment
 sequence heard from another laptop.

-- 
Ticket URL: <http://dev.laptop.org/ticket/9896#comment:8>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system