#12032 NORM Not Tri: XO-1.5s in critical low-battery conditions do not cleanly shutdown or restart

[Zarro Boogs per Child]

#12032: XO-1.5s in critical low-battery conditions do not cleanly shutdown or
restart
-------------------------------------------+--------------------------------
           Reporter:  greenfeld            |       Owner:  rsmith                           
               Type:  defect               |      Status:  closed                           
           Priority:  normal               |   Milestone:  Not Triaged                      
          Component:  embedded controller  |     Version:  Development build as of this date
         Resolution:  invalid              |    Keywords:                                   
        Next_action:  no action            |    Verified:  0                                
Deployment_affected:                       |   Blockedby:                                   
           Blocking:                       |  
-------------------------------------------+--------------------------------

Comment(by rsmith):

 Replying to [comment:6 greenfeld]:

 > Perhaps I should not use the word "clean" in reference to EC land when
 dealing with "hard" OS shutdowns.  My question probably should be if the
 EC should turn off the voltage regulators, etc. before the main processor
 tries to reboot a few times.   This ultimately fails on the main CPU.

 You don't have a choice.  The EC powers off the system when the battery
 voltage reaches the level at which you aren't allowed to operate at any
 more.  This is why powerd would normally shut the system down prior to
 this condition.

 > After this happens the EC may also fail to power on the XO until all
 power is removed (resetting the EC) even with external power present.  One
 XO-1.5 in particular seems prone to this.

 This is because run-in discharge is still active.  External power doesn't
 help you because the EC has it turned off. (Well on 1.5 its not 100% off
 but not enough to sustain running the CPU).  Your EC reset clears run-in
 discharge.  So it sounds like fixing this is worth the effort.  I'll back
 port the change and release a new 1.5 EC firmware.

 > EC's critical shutdown tolerances are below the main CPU's tolerances
 and this is intentionally so, then there is not an issue.

 It's the tolerance of the 5V regulator.  When the battery voltage dips
 below the headroom of the 5V regulator then bad things can happen. If a
 rail fails then back feeding can happen.  So the EC shuts them all down
 when 5V can no longer be 5V.

-- 
Ticket URL: <http://dev.laptop.org/ticket/12032#comment:7>
One Laptop Per Child <http://laptop.org/>
OLPC bug tracking system