[Tinyos Core WG] hardware_bootstrap

Philip Levis pal at cs.stanford.edu
Sat Feb 10 15:19:40 PST 2007


One issue that came up in the review of TEP 107 was the case where a  
platform needs to perform initialization before it can, say, access  
memory. We haven't run into a case of this yet, but I could imagine  
situations of configuring memory bus speed, enabling external memory  
banks, etc. The issue is that initializing the scheduler (currently  
the first step) requires accessing memory.

In the current draft of 107, I've put a proposal of having a

hardware_boostrap()

function which is the first thing called in main(). For all existing  
platforms, this function will be null. But its presence means that  
future platforms that need this kind of functionality can use the  
default boot sequence. Does anyone object to adding this? I can put  
default null functions into the existing platforms.

Once we resolve his issue, 107 should be ready to be finalized.

Phil


More information about the Tinyos-2.0wg mailing list