#1905 NORM Untriag: flash corruption

[Zarro Boogs per Child]

#1905: flash corruption
-----------------------+----------------------------------------------------
  Reporter:  dwmw2     |       Owner:  wad      
      Type:  defect    |      Status:  assigned 
  Priority:  normal    |   Milestone:  Untriaged
 Component:  hardware  |     Version:           
Resolution:            |    Keywords:           
  Verified:  0         |  
-----------------------+----------------------------------------------------
Comment (by wmb at firmworks.com):

 Actually, it's not necessarily PCI writes.  The corruption could just as
 easily occur in memory.  The JFFS2 code writes the information into a
 memory-based write buffer before it is DMA-ed out to the NAND FLASH.

 The failures that have been analyzed so far all show the same bit position
 within an 8-byte chunk.  The memory array is 64 bits wide.  If additional
 analysis continues to show that same bit error position, the memory
 hypothesis is stronger than the PCI hypothesis.

-- 
Ticket URL: <http://dev.laptop.org/ticket/1905#comment:8>
One Laptop Per Child <http://laptop.org/>