#2614 BLOC Trial-3: OFW copy-nand should abort on CRC mismatch

[Zarro Boogs per Child]

#2614: OFW copy-nand should abort on CRC mismatch
----------------------------------+-----------------------------------------
  Reporter:  cscott               |       Owner:  wmb at firmworks.com
      Type:  defect               |      Status:  new              
  Priority:  blocker              |   Milestone:  Trial-3          
 Component:  ofw - open firmware  |     Version:                   
Resolution:                       |    Keywords:                   
  Verified:  0                    |  
----------------------------------+-----------------------------------------
Changes (by cscott):

  * owner:  cscott => wmb at firmworks.com

Comment:

 I fixed the autoreinstallation script to not catch the abort, but a bit
 more is needed: often we get far enough through the copy-nand to create
 /boot/olpc_build, which means that when we get a corrected NAND image on
 the USB key, autoreinstallation will skip the installation because that
 build is "already on NAND".  We need to catch the error and wipe the NAND
 (or at least enough of it so that /boot/olpc_build doesn't appear to
 exist!).

 You can reproduce this failure scenario by randomly altering a few CRC
 bits halfway through the CRC file.

 Mitch, can you help me out with the right OFW words to ensure that we
 leave the NAND in an obviously-broken state?

-- 
Ticket URL: <http://dev.laptop.org/ticket/2614#comment:4>
One Laptop Per Child <http://laptop.org/>