System Software Teleconference minutes, 2007-03027

[Mitch Bradley]

System Software Teleconference, 20070327

Present: Mitch, Jim, Chris Blizzard, Jordan, Chris Ball, Andres, Richard,
Andrey, David (Quanta), Marcelo, J5

Progress on Test1:

Build 347 is running, new wireless firmware was pushed

Build 343 does not have the right wireless firmware or driver

Dan discovered a libertas deadlock today.  The patch is not yet
available, but is important enough to wait for.

Various new packages will be in the build at various levels of
packaging.  Flash, Java, Helix, ...
Jim suggests pulling memosono from the build.

Continuing sugar startup problems, believed to be Avahi-related.

Wad is making progress on the server.

Michalis has a NAT-ed wireless test network.  Do we want to use
it, or set up another network.

David Woodhouse is chasing a JFFS2 problem.  Debugging is difficult
in the large-NAND world.

cjb has a (somewhat-tedious) process for regenerating a NAND image
using rsync, mkfs, etc.  Will work with Mitch to get the (simpler)
OFW-based procedure working.

Late-breaking build-347 info: We have a JFFS2 build image that can be
tested.  Some builds don't go through because of a bad switch, but that
is being fixed.  That switch may be the cause of a lot of download
problems.

Batteries:

Batteries are working much better now with B81.  LiFePO4 still have
some issues with flashing LEDs, but they seem to be working.

The "batman" recovery program works for recovering some bricked
batteries, but not for others.  

Power Management:

The firmware is doing glitchless display resumption in about 80 mS.

There is a color artifact during the resume.  More testing with
different modes needed.

USB resumption takes several hundred milliseconds.  There are
some 100 mS delays in the USB resume code.

We can't wakeup from USB because the USB power is off.

The SDHCI has some hardware options to go into a lower power
mode when no card is plugged in.

The audio CODEC has some deep sleep modes too.

EC support for wireless wakeup: won't have hardware (whereby
the CPU can wake up the wireless) until B3.


Open AIs:

AI: dcbw and marcopg to investigate mesh/avahi not working.
      done, but we're seeing crashing reports
Still open - need to make a decision soon

AI: Richard to work with Ronak  & Michailis to get him the wireless
wakeup info he needs.
- In progress - waiting for Quanta's input on EC signals.  Wad is making 
a GPIO map.
Need to document it


Closed AIs:

AI: wmb: publish a suspend/resume FW binary.
Done

AI: dilinger, wmb: make sure released kernel works with fastpath FW
Done

AI: wmb - file a trac to figure out how to do LX board id.  assign to wad
Done

AI: Richard - update trac tickets after releasing WLAN_RESET firmware
Done