System Software Meeting Minutes, 2006-12-26

[Mitch Bradley]

System Software Meeting Minutes, 2006-12-26

Attendance: Ted, Richard, Mitch, Andres, Jim (sparse attendance
due to holidays)

- Jim installed new firmware (with EC B11) on some machines.

   To test:
     With battery fully charged (battery LED green), press button
     to power off, press button to power on.

     Old EC code starts charging in that case, new EC does not.

- Richard thinks OS build 206 is good except for the pixmap update.

- Jim suggests updating to Q2B11 and 207.

Staged deployment of VSA elimination - first amd_drv.so, then kernel with
VSA spoofing, then firmware with a VSA on/off switch, then fastpath 
firmware,
possibly with resume reinit code somewhere in the list.

/proc/device-tree is working.  It requires a kernel patch and a
(fairly low risk) firmware change.  Andres suggests that we try it out
as-it, without first converting it to sysfs, and publicize it on LKML
to get feedback.  When Corbet reappears, we will ask him for help with
the sysfs conversion if that turns out to be necessary.  Andres will
help Mitch package the patches.

- Expect Chris Ball, Richard, and perhaps J5 back in Boston on Friday.

Quanta needs firmware and OS bits for next week's pre-build by Dec 30.

Quanta requests for network to be first in the boot order.

We will release a new firmware image in the next couple of days
for inclusion in the pre-build board.  It will have the battery charging
and touchpad fixes via the new EC bits, plus OFW changes to support
the B2 GPIO bit changes, and the network-first boot order.

Now that olpc-boot.fth is part of the OS images, we can remove the
two-step (boot-configure, boot) load process in favor of just "boot".
That will require the inclusion of an additional line "boot" at the
end of the olpcboot.fth script.

The firmware version that recognizes the USB network device (AX88772,
VID=0x0b95 PID=0x7720) apparently is not working with that device.
The traffic sniffer diagnostic (watch-net) does not report any traffic.
(However, since there was no report of a "failure to open the device"
error message, Mitch thinks that the driver recognized and bound to the
device.  It is possible that the PHY detection did not work right and
the driver is trying to use the wrong PHY, but that is just a guess.
Ted is going to take a known-working USB network device to Shanghai.)