Software status meeting minutes, 2007-06-19.

[Chris Ball]

Attendance:  cjb, Marcelo, Bernie, Andres, Jim, Mitch, J5, Bryan, Luna, Kim,
             Richard, mlj, Dave Woodhouse, Blizzard, Wad, Jordan.

New action items:

AI: jg, Kim: find someone to task with creating a wireless initrd.
AI: jg, Kim: create integration plan for software merges, including testers.
AI: Kim to enter a bug for school server backup/restore scripts
AI: Wad to mail Kim with links to latest school server progress.
AI: Kim and jg to schedule a call with activity developers to start using
    the journal/collaboration features.
AI: jg/cjb to ask Erik/Baktiar for help understanding camera bug.

Status:

Marcelo, for USB suspend/resume:
  Strange things happening to the USB bus on resume, sometimes.  The USB
  analyzer shows that the host issues a command, and then IN requests to
  find out if data is available.  We see an error on the EHCI controller;
  a structure is marked with a fatal error -- timeout/CRC error.  We
  can't see what's going on with our analyzer, we need to look below the
  protocol level.  Marcelo is going to Marvell on Thursday, and will work
  with their hardware engineers.

  Mitch:  Maybe electrical cross-talk?

cjb, for audio problems:
  Two bugs:
    #1420: we're mishandling the audio device -- we aren't powering down
           the hardware once the device is closed.
    #1735: we're getting a "clicking" sound when we power up, either at
           boot time or resume time.
           Mitch:  Easy to locate the code causing the click by adding
                   printk()s surrounded by delays to the driver.

mlj, for B4 build:
  Going great, only problems are mechanical and being solved:
    Bunny ears are a little short;
    Grey bezel is covering part of the camera.
    
  Lots of discussion at Quanta around school server and gang charger.
  
  Our kernel didn't recognise the B4 firmware version (OFW "model"
  property) so Dave Woodhouse pushed a patch to recognize it -- J5
  made a new stable build with it, 406.15, during the call.  The new
  kernel won't be used in the build the B4s arrive to Cambridge with,
  hence the audio won't work, so we'll need to upgrade the build once
  the laptops get here.

  Quanta is using their own OS build for testing; we need to understand
  what's in that build at Cambridge, so that we can understand what's in
  their bugs.

Mitch's status:
  Has started on firmware crypto: it compiled cleanly, moderate footprint.

  Has a school server prototype, has fired it up and is making progress.

  Will be in Cambridge on July 4th.

Request from Quanta:
  Would like an initrd that lets them boot from the network; it needs
  an initrd because userspace wifi configuration.  Is anyone interested
  in doing it?  No hardware knowledge is just required, it's just Fedora
  initrd work.

  AI: jg, Kim: find someone to task with creating a wireless initrd.

Upcoming software changes:
  vserver merge -- has the potential to affect sugar, we don't want to
                   affect activity development.  Should test independently
		   from F7, since both might be responsible for breakage.		   

  F7 merge -- should have a build by Thursday or Friday.

  X 1.3 merge -- Bernie is working on it.

  New Marvell wireless firmware (with an incompatible packet format) is
  showing up soon -- see #1770.

  AI: jg, Kim: create integration plan for these merges, including testers.

Backup/restore:
  cjb is working on image restore using a USB disk.
  We also need backup/restore to the school server.  We already have a
  script to do backup that needs to be modified and packaged.
   
  AI: Kim to enter a bug for school server backup/restore scripts

Power measurement/tinderbox:
  Richard gets back on the 3rd.  His intern Joel will help with power
  measurement.  cjb will move on to this after the upgrade scripts.

Trial-2 blocker updates:
 #1276: Keypad and game buttons: Bernie is working on keyboard definitions
        for game buttons, has sent them to the Xkb maintainer for review.
	They  give us independent keycodes for the left and right sets of
	gamekeys.

 #884:  Arabic and Urdu in web browser: We should check how gecko is doing
        with this.  Blizzard thinks Gecko is behaving and our content
        isn't labelled correctly.

        AI: Jim to find someone to test Arabic with our browser.

 #1339: SD fixes: Believed to work in master, untested by us.

 #1396: Power button should suspend, not shutdown: cjb has a patch for
        this, but we can't take shutdown away without having a replacement
        UI for it.

 #1605: School server registration: Nelson and Christine are working on it,
        are providing code to Scott for integration with the school server.

        AI: Wad to mail Kim with links to latest school server progress.

        AI: Kim and jg to schedule a call with activity developers to
        start using the journal/collaboration features.

 #1710: "Get rid of the SCI queue":  Mitch thinks we should move to the kb
        command queue because the EC protocol is very slow (and racy).
	Dave disagrees; we just need to fix the kernel code to not block
	waiting on EC protocol replies.  They'll take the discussion to
        a Trac bug.

 #1730: AI: jg/cjb to ask Erik/Baktiar for help understanding camera bug.

 
- Chris.
-- 
Chris Ball   <cjb at laptop.org>