#4536 HIGH Never A: Current suspend/resume count should be encoded in CMOS for mfgtest.

[Zarro Boogs per Child]

#4536: Current suspend/resume count should be encoded in CMOS for mfgtest.
-----------------------------------+----------------------------------------
 Reporter:  cjb                    |       Owner:  cjb           
     Type:  defect                 |      Status:  new           
 Priority:  high                   |   Milestone:  Never Assigned
Component:  manufacturing process  |     Version:                
 Keywords:                         |    Verified:  0             
-----------------------------------+----------------------------------------
 We want to test for #1835 by doing an aggressive resume-on-broadcast-wifi
 test on each laptop in the manufacturing process.  If the test fails,
 Quanta would like some way of finding out how many cycles the test passed.
 (If a laptop dies on its first suspend, it should obviously fail.  If it
 dies after 50k suspends, it's less clear.)

 The task for this bug is to modify the "./run" script for suspend/resume
 wakeups to write the current cycle number to CMOS before going down,
 possibly with "dd".  Mitch says:

 "CMOS RAM indices 0x30-0x4f are readily available.  There are others but I
 expect that those will suffice for you.

 /dev/nvram offsets the index so that seeking to 0 in /dev/nvram accesses
 CMOS RAM index 0x10.  That is because CMOS RAM indices 0-0xf are actually
 the RTC registers; /dev/nvram skips the RTC and just maps the RAM portion
 of the device.  Therefore the available locations from the /dev/nvram
 perspective are 0x20-0x2f."

-- 
Ticket URL: <https://dev.laptop.org/ticket/4536>
One Laptop Per Child <https://dev.laptop.org>
OLPC bug tracking system