#5128 NORM Never A: Handle RTC day and month alarm locations

[Zarro Boogs per Child]

#5128: Handle RTC day and month alarm locations
---------------------------------+------------------------------------------
 Reporter:  wmb at firmworks.com    |       Owner:  wmb at firmworks.com
     Type:  defect               |      Status:  new              
 Priority:  normal               |   Milestone:  Never Assigned   
Component:  ofw - open firmware  |     Version:                   
 Keywords:                       |    Verified:  0                
---------------------------------+------------------------------------------
 The Geode 5536 companion chip has MSRs that define CMOS RAM locations for
 a century byte, the alarm day, and the alarm month.

 OFW versions up to and including q2d04l do not explicitly set those MSRs.
 Empirically, the chip comes up with century offset 0x00, alarm day offset
 0x10, and alarm month offset 0x11.

 The current Linux version, in arch/i386/kernel/olpc-pm.c , thinks that
 those offsets are all 0, which probably means that it ignores those
 registers.  This is a possible explanation for the problem described
 below.

 Regardless, OFW should set those MSRs, and Linux should be aware of the
 values that OFW has set them to.

 {{{
 Our production test program use RTC alarm to test suspend /resume function
 (RTCWAK write by CJB)

 It turns out some of unit won’t wakeup from RTC ,but still can wakeup from
 other wakeup event.

 We can fix this issue by using following steps,

 1.       Discharge RTC battery, the RTC ram will get default setting.

 2.       Use Linux commend “date” and “hwclock –hctosys” setting current
 system time.

 3.       after reboot , the unit can wakeup from RTC.

 Production engineer already implement those steps on assembly line for
 short turn solution,

 We want to know is there have any software command can wipeout RTC ram or
 reset RTC ram,

 I believe we will use new EC code for autowake or slowpingwake to test
 suspend/resume in the nearly future,

 RTC wakeup will no longer useful in that time,
 but if we can get a software support to clear RTC ram in production line
 (before setting system time),
 then we can make sure each unit will get same RTC ram data.

 Gary Chiang 江品逸
 }}}

-- 
Ticket URL: <http://dev.laptop.org/ticket/5128>
One Laptop Per Child <http://dev.laptop.org>
OLPC bug tracking system