System software teleconference meeting minutes - 2007-07-17

[Mitch Bradley]

OLPC System Software Teleconference, 2007-07-17

Participants: wad, kim, wmb, jg, Victor, Scott, rsmith, richard, David 
Lin, shenki, cjb

== Status ==

Kim, Wad, and Jim got shared chat, text editor, and games working via 
the mesh
on the train.

Mitch is close to having Forth running on the EC.

Scott: Activation code is written and will be integrated tomorrow.

Hardware verification:

* The DCON can be turned off (via a power switch transistor)
  when the display is not needed.

* The LCD is taking more power than we expected.

* We might be able to revise the mesh firmware to use less power (in 
mesh mode),
  but the wireless chip will be revved toward the end of the year, which 
might
  consume resources that we could otherwise use for improving the firmware.

* Turning off the Geode dotclk might reduce the power consumption when the
  CPU is on but the display isn't needed.

* We need to solve the NAND corruption problem (#1905)

* There are some keyboard/mouse problems that have been attributed to 
the EC,
  but the current indication is that it might be software.

* Some problems when plugging in the AC adapter when the battery is low.
  - But there is a known problem where power SCIs get treated as wakeups.

* Need to test whether or not we can reduce the CaFe power by turning off SD
AI: jg and wad - review CaFe power consumption



watch-battery on XO that was in the freezer says "abnormal sensor error" 
and reports 244 degrees


Blockers:

23 - School server DNS - we now have a stable build environment for 
school server.

31 - OHM power management - OHM is now in the build and is starting to 
look good.

33 - Problems with ajax's patch - still seeing screen corruption 
(probably LX driver
bug), and also something wrong with DCON being suspended incorrectly.  
Weird corruption
with rapid-fire DCON suspend/resume.

60 - Resume from RAM - New EC code (v19) is not as effective as we would 
hope.

1710 - still some problems with EC timeouts.  More testing needed.  
Richard has some
patches to implement some new EC functions around battery status reporting.

1733 - fixed and works in our tree but not upstream, needs review by Jon 
Corbet

884 - won't make it for Trial-2, but not needed for T2, so reprioritize

1276 - maybe push out to Trial3 ?

1339 - >Trial-3

1396 - still open pending ebook reader explicit suspend-on-idleness patch

1498 - no longer a blocker

1500 - >Trial3 .  We don't want to sign with strong crypto at this stage 
because of
       the likelihood of bugs.  But we will have basic activation.

1505 - Scott will work on this soon.

1555 - Reassign to David.Lin to implement the EC software change for the
       new GPIO location.

1687 - ?

1729 - we think the kernel is "inventing" an SCI.

1748 - duplicate of 827 - close 827 and leave this one.

1755 - still open

1770 - close

1804 - Dan needs to look at it
1814 - needs testing with 406
1815 - in 505 - close (also 2092)
1835 - ?
1837 - Fixed, but there may be better fixes.  >Trial-3
1865 - Fixed?
1875 - fixed (wrong mouse driver)
1905 - no progress
1936 - ?
1974 - ?
2019 - ?
2059 - ?
2062 - duplicate of something?  try to reproduce on anther build
2093 - ?
2096 - should be fixed soon
2122 - jg to look at this one tomorrow
2177 - AI: Kim to lean on Marvell
2180 - fixed
2182 - ?
2184 - hasn't happened recently, but leave open.  Perhaps the idle 
plugin was tickling the xorg plugin.
2196 - Benjamin is working on it
2198 - A workaround is in place >Trial3 pending move to new input stuff
2201 - in progress
2210 - fixed, but not in build yet.  Close.
2313 -
2214 - need to make a sugar change after jabber server running
2221 - in progress
2222 - some possible patches are kicking around.  there is a duplicate 
somewhere
2225 - ?
2228 - journal code is being worked on.  ISC DHCPD is running on the 
mesh interface but is bloated - 20 MB.
       Each python process uses 10-15 MB, and no sharing is happening.  
Possibly F7 related.
2231 - fixed (also duplicate)
2243 - related to the hardware change to the lid switch GPIO assignment 
 >Trial3

QEMU is not working.  It might be related to the new Xorg.conf.