#5240 NORM Update.: RTC & power button can still cause #1835?

[Zarro Boogs per Child]

#5240: RTC & power button can still cause #1835?
----------------------------------+-----------------------------------------
  Reporter:  dilinger             |       Owner:  rsmith  
      Type:  defect               |      Status:  new     
  Priority:  normal               |   Milestone:  Update.2
 Component:  embedded controller  |     Version:          
Resolution:                       |    Keywords:          
  Verified:  0                    |  
----------------------------------+-----------------------------------------
Changes (by wad):

 * cc: rsmith (added)
  * milestone:  Never Assigned => Update.2


Comment:

 No, the race was caused by a wakeup event happening right after the power
 was shutoff, causing power to brownout and all sorts of resets and reset
 delays not to happen.  This was quickly discovered for the WLAN SCI, and
 fixed early on.   What we didn't realize was that the EC was generating
 keyboard SCIs even on machines with no keyboard.
 (Richard can add his own interpretation as well...)

 Jordan and I convinced ourselves that the RTC (and power button) could
 conceivably cause the same problem, as you describe.  Since on a normal
 system, power is not turned off when suspending, there is no delay built
 into the Southbridge ensuring that nothing wakes the system up for 5 mS
 after it is suspended.

 Since the RTC precision is 1 sec. anyway, I propose that we never use it
 to wake up the system.  One proposed plan is to implement a wakeup timer
 in the EC and use it instead, both providing us with mS precision and
 avoiding any chance of hanging on resume.

 Comments ?

-- 
Ticket URL: <http://dev.laptop.org/ticket/5240#comment:2>
One Laptop Per Child <http://dev.laptop.org>
OLPC bug tracking system