#7303 BLOC 8.2.0 (: Libertas disconnects on resume with testing/master kernels.

[Zarro Boogs per Child]

#7303: Libertas disconnects on resume with testing/master kernels.
-------------------------+--------------------------------------------------
   Reporter:  cjb        |       Owner:  jcardona            
       Type:  defect     |      Status:  assigned            
   Priority:  blocker    |   Milestone:  8.2.0 (was Update.2)
  Component:  wireless   |     Version:  Update.1            
 Resolution:             |    Keywords:  8.2.0:?             
Next_action:  never set  |    Verified:  0                   
  Blockedby:             |    Blocking:                      
-------------------------+--------------------------------------------------
Changes (by jcardona):

  * status:  new => assigned


Comment:

 > This leaves the following as possible causes for this problem:
 >
 >  1. power loss at the device
 >  2. driver not issuing the EC command that initiates the GPIO handshake
 >  3. EC not executing the
 [http://www.laptop.org/teamwiki/index.php/Tech:Suspend_Resume_Protocol
 GPIO suspend/resume handshake]

  1. ruled out: LED stays on if associated while suspended.
  2. ruled out: ec command observed on the logs:

  {{{ 1028.682820] olpc-ec:  running cmd 0x24 }}}

  3. ruled out: confirmed at firmware level that handshake is properly
 executed.

 OK, so the first phase of suspend/resume sequence seems to execute
 correctly.  The problem must be later on.

 Ah!  Looks like on resume the driver attempts to reload the firmware but
 notices that the device is alive and well.  That results in the driver
 sending a CMD_802_11_RESET to the device, which triggers a reboot of a
 device that had otherwise bravely survived a suspend/resume cycle:

 {{{
 [  277.436575] usb 2-1: configuration #1 chosen from 1 choice
 [  283.071169] usb8xxx: Firmware already seems alive; resetting
 }}}

 (To observe these messages, you need to {{{echo 0x4000 >
 /sys/module/libertas/parameters/libertas_debug}}}).

 On Monday we'll try to understand what's broken in the USB persistence and
 why we are trying to download new firmware after resume.

-- 
Ticket URL: <http://dev.laptop.org/ticket/7303#comment:11>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system