Brick Insurance

[Mitch Bradley]

Tim Flavin wrote:
> On 1/30/07, Mitch Bradley <wmb at firmworks.com> wrote:
>
>> We shall take reasonable precautions to reduce the probability of
>> reflash-brickage to an acceptable level.  A level of 0 is neither
>> achievable nor cost-effective.
>
> Will the production boards have pads or a header that can be used
> by an SPI programmer as a last resort?
They have pads that can be used for a serial recovery procedure.  The 
current run doesn't have a connector on those pads, but the factory has 
a "stinger" thing that can connect to the pads without the connector.  
Even if the connector was there, you would have to disassemble the unit 
fairly extensively, removing several pieces including a metal shield.
> Shipping one USB based SPI
> programmer for each region that gets 10K to 100k laptops may be
> a cost-effective way of fixing the small number of bricks created by
> than 100% effective measures.  It would also permit recovery from
> a disaster caused by someone incorrectly telling a whole classroom
> full of students the wrong way to re-flash the laptops.
The reflash procedure will be automated based on the presence of a 
cryptographically-signed image that is written to the NAND FLASH at the 
OS level.  So in principle, there is no "telling" involved.
> This would
> require disassembling the laptops but it would be an extremely rare
> requirement.
>
> I guess the question is it cost effective to put the pads or headers
> on if they almost will never be used.
>
It might be better just to treat reflash bricks the same as any other 
non-user-serviceable failure - ship them back to the fully-equipped 
repair center.  If we can make the probability of reflash failure low 
enough, then this is probably the right approach.