[svlug] Mars Rover

[Amit D. Chaudhary]

Karl Larsen wrote:
> 	As you have heard, they are having computer problems on Mars. 
> This morning I learned the problem is with "flash memory". I have never 
> run across flash memory so I wonder if anyone here knows what that is.
The immediate question was answered by others.
However, typically flash memory refers to a actual chip or set of chips, 
below is a link to family of 3V chips.
http://www.amd.com/us-en/FlashMemory/ProductInformation/0,,37_1447_1623,00.html

The following link has more details on what they tried to do.
http://planetary.org/news/2004/spirit_sol-20.html

Having worked on systems which boot off or use flash memory, there are 
multiple things which can go wrong. Here are a few:
1a. Since it is continously rebooting, it is probably that the flash 
memory has some program or kernel module that is running into a error 
(segfault equivalent) and resetting the system.
1b. Some of the program is corrupt, due to any number of reasons 
including writing the data (see notes on data in the planetar.org link) 
in the wrong place.
1c. It could also be that it is just reading and sending the wrong data 
and not stopping, hence the suprious transfers.

2. Most of the problems can be fixed by having a backup read-only copy 
and falling back to it, unless the code to fallback ran into a glitch.

3. The 3rd level of fallback using the EEPROM only referred to as 
"cripple mode" seems to be working, but the throughput is pretty low.

Whatever it is, it will surely make for another interesting RISKS posting.

On the OS comment, VxWorks, while a robust OS is only as robust as the 
programs running on it, this is same as a buggy kernel module can bring 
linux down.


Amit