[Openec] checksum of first 64k of updateme.bin not calculated correctly

[Frieder Ferlemann]

Hi Paul,

Am 02.11.2010 21:05, schrieb Paul Fox:
> frieder wrote:
>  > Yes, Where does the limitation to 32K come from?
> 
> ah -- sorry.  it's somewhat likely we'll be using a new, lower-cost
> part:  the IO3730 "Mobile IO Controller".  part of the cost and board
> real-estate savings come from it having internal flash.  unfortunately,
> that flash is only 32KB.

Oh, I see.

I like the current setup with the EC and the CPU sharing
serial flash (and the EC having access to rather plenty of flash)
because of the following (long time and illusionary) prospect:

 Embedded Controllers (tiny as they are) nowadays have lots of
 computational power (million(s) 8-bit instructions per second)
 which is available without the laptop being switched on.

 This potential is completely hidden to laptop owners today.
 (XO owners but also owners of  ALL  other laptops!)
 If a virtual machine / something interpreter would be (sandbox)
 implemented on an EC then functionality (which is available
 with say some 100 high level instructions per second) could
 be available with very low compromise on power consumption.
 (functionality could be a complex wake-up condition for the
 main CPU, beeper/LED stuff, or handling of some external
 1-wire bus devices)

 (I admit I have no direct use case for this and that
 potential use cases might rather have hacker value
 instead of direct educational value)

That said if an IO3730 is attractive from an
engineering standpoint then OLPC should not pay a
premium for a vague vision like the one above.

But 32k is sparse. (you might pay for the 32k in
engineering cost, design risk, missed opportunities,
time to market. (well, guess I'm saying nothing new
if you are at 26k (Keil) right now))



>> I'll implement this soon.
Implemented in SDCC #6047.


Greetings,
Frieder