[Trac #1410] Overtemperature warning to users.

[Zarro Boogs per Child]

#1410: Overtemperature warning to users.
---------------------------------+------------------------------------------
 Reporter:  jg                   |        Owner:  rsmith 
     Type:  enhancement          |       Status:  new    
 Priority:  high                 |    Milestone:  BTest-4
Component:  embedded controller  |   Resolution:         
 Keywords:                       |  
---------------------------------+------------------------------------------
Changes (by jg):

  * owner:  jg => rsmith
  * priority:  normal => high
  * component:  new component => embedded controller
  * milestone:  Opportunity => BTest-4

Old description:

> The XO-1 has an ambient temperature sensor.  The CPU has a fixed over
> temperature cut-out (see the LX specifications). Testing will tell us the
> typical rise over ambient temperature inside the main housing, and may
> allow us to warn the user if the over temperature cutout is likely soon.
>
> There may be ways in which we can reduce power dissipation (e.g. reducing
> backlight and/or codec power dissipation).  Whether this will be useful
> in practice is not clear.

New description:

 The XO-1 has an ambient temperature sensor.  The CPU has a fixed over
 temperature cut-out (see the LX specifications). Testing will tell us the
 typical rise over ambient temperature inside the main housing, and may
 allow us to warn the user if the over temperature cutout is likely soon.

 There may be ways in which we can reduce power dissipation (e.g. reducing
 backlight and/or codec power dissipation).  Whether this will be useful in
 practice is not clear.

 Jordan notes:
 And multiple temperature levels would be friggin sweet:
 one for warning, so we could throttle, and one for critical so we could
 shut down.

Comment:

 Jordan notes that this process is not automatic; we'll have to do this
 ourselves, as best we can, from estimating the internal temperature of the
 system.  Having a temperature sensor directly for this was ruled out due
 to cost.

 The first step toward implementing this is to have the EC interrupt on
 ambient temperature change (we don't want a thermal manager, whether in
 the kernel or in user space have to poll).  If suspended, we don't need
 the EC to interrupt, either: by definition, the system should be cooling.

-- 
Ticket URL: <http://dev.laptop.org/ticket/1410#comment:2>
One Laptop Per Child <http://laptop.org/>