ECO/7.1.2

From OLPC
< ECO
Revision as of 15:38, 4 January 2008 by Kimquirk (talk | contribs) (Test Results)
Jump to: navigation, search
  This page is monitored by the OLPC team.

Software ECO, 2008-01-02

Title: Memory over-run; touchpad issues; include OFW


Trac items:

Comments

Is 5744 part of this patch? In the ticket, it refers to Update.1.

What is the relationship between 5317 and 5744 in regard to patches? Are any of these patches tested?

What about the disappearing Journal patch (5494)? Presumably that is part of this ECO?

What about WPA support? Which tickets?

Also, we should explicitly mention Q2D07.

Description

  1. If you create a large file that effectively fills the NAND memory, you will not be able to boot the laptop.
    • Critical bug as data is lost
    • Root cause:
  2. Sometimes when you touch the touchpad, the mouse will jump to the lower right of the display. (Note that this bug fix will not fix ALL problems with jumpy mouse or touchpad issues).
    • Some laptops touchpads are so sensitive as to be unusable.
    • Root cause:
  3. We inadvertently included an earlier version of the firmware with the OS image in 653; so this build will include the latest firmware, which helps to recover if the internal battery is dislodged.
    • When the laptop is moved or shipped there is a possibility that the battery will be dislodged and the laptop will no longer boot.

Proposed Fix(es)

Patches for Memory and Touchpad issues are included in the trac items.

Reviewers should include: Bernie, Andres, Jim, Scott, Dave Woodhouse, Tomeu, WMB, Ivan


Testing:

  1. Specific tests include creating the NAND memory problem following the description in the Trac item.
  2. The touchpad must continue to work after upgrade; and should NOT show any jumping to lower right of display. This can only be a negative test since it is not possible to force this behavior.
  3. Automated upgrading - this needs to be tested on a small number of laptops in house and then on a larger number of developers laptops; then finally rolled out, possibly in a controlled fashion.
    • Set the timezone and language parameters from sugar-control-panel before upgrade; ensure they are still there after
    • Use activities that save files before upgrade; ensure they are still there after
    • Download flash before upgrade: Adobe_Flash; ensure that it is still there after
    • Install/test a printer before upgrading: Enabling_CUPS; is it still there afterwards
  4. The build must be installed on MP, CTest and B4 systems
  5. Upgrades from the previous release.
  6. A Fresh install (as mfg).
  7. 1 Hour Smoke Test both using a fresh installation and an upgraded installation, looking specifically for regressions from the release reports.
  8. More than one SKU and keyboard type are to be used during this testing, to catch regressions in keyboard identification.


Rollout:

In manufacturing this should be dropped in as quickly as possible.
In the field this should be part of an automated upgrade when released.

Test Results

NAND Memory test

  1. Following the same steps as found in track 5719, on build 653 first.
    • I was not able to shutdown after filling NAND. I held the power button to shut it down.
    • On boot up, it was not successful in booting sugar and the final text msg was: 'No space left on device'. I could not get a prompt or do anything else other than cleaninstall. I believe on a WP laptop this would have looked like a hang in boot up.
    • I repeated the test after cleaninstall to 656, and had to use the power button to shut down the laptop in the same way.
    • On boot up, it failed to boot into sugar, but cycled on restarting X. When it got tired of that I was able to get a linux prompt with Ctl-Alt-mesh, which I believe would be the same response on a WP laptop, but I can't test that since 656 is not signed.
    • Next I removed the files in /home/olpc/.sugar/default/data and rebooted. Now boots into sugar and the data is still available in the journal.
  2. Laptops with Spanish keyboard comes up in Spanish
  3. One hour smoke test doesn't reveal any new regressions