Re: [avr-libc-dev] Is there interest in an arm-libc?

[Markus Hitter]

Am 11.11.2014 um 14:41 schrieb Bob Paddock:
> I support the goal of arm-libc.

Excellent, thank you.

> Not really seeing how that prevents
> having to port new devices?  arm-libc itself would need updated for
> each new device would it not?

For each new MCU released by MCU manufacturers, yes. But not for every new 
board/device using this MCU.

Having an arm-libc (hopefully) means you do the arm-libc port once for the MCU, 
then every hardware/board using this MCU simply updates arm-libc and uses the 
new -mmcu flag to be done with chip internals. Also porting from one MCU to 
another should become much easier. In my area it's pretty common that many 
devices do essentially the same, but with different flavors of MCUs.

With CMSIS and MBED much of the coding work is already done, of course. I 
expect mostly work like sorting files for unifying and simplifying CMSIS/MBED 
code. Like providing pre-built libraries. Like getting this -mmcu parameter 
into gcc at all (which is what I currently work on). Also more simplifications, 
like perhaps writing initializers assuming a constant CPU clock. Example: 
setting baud rate for UART using MBED functions chimes in at a whopping 880 
bytes for just setting 2 or 3 registers, because it calculates the values going 
into these registers, which depend on CPU clock frequency, at runtime.


Markus

-- 
- - - - - - - - - - - - - - - - - - -
Dipl. Ing. (FH) Markus Hitter
http://www.jump-ing.de/