Re: [avr-gcc-list] RE: avr-gcc-list Digest, Vol 6, Issue 10

[Joerg Wunsch]

As Iztok Zupet wrote:

>  Sorry, but this happened also to me, bout only when working with a program 
> full of interrupts. When programmed with uisp, it worked fine and programmed 
> with AvrStudio via hex file it also worked. But when I converted it to coff 
> file (both via objcopy or elfcoff) and downloaded via JTAG with AvrStudio, 
> both serial interrupt routines stopped working, like if the vectors werent 
> set up right.

Sure you read the last paragraphs of the beta README file?

All released versions of AVR Studio so far cannot handle avr-gcc's way
of initializing the contents of .data!  There's a fix for this already
in Atmel's sources (which will only work for the COFF files as
generated by the binutils patch, and only if coff-ext-avr is used),
but this new COFF parser DLL has not yet been officially released by
Atmel.  (All the other compilers for the AVR handle .data
initialization in a completely different way, so that's never been a
problem for AVR Studio.  In fact, they don't even pretend to have a
.data but only a .bss, and the initialization values for the
initialized variables are embedded into their .text section.)

Btw., the way AVR Studio handles the COFF files, they'll probably
never be able to handle user-introduced additional sections in the
COFF file.  So if you invent your own .mycode section, you could never
expect AVR Studio to ever take notice of it.  VMLAB is different here
since they use the COFF file only for obtaining the debugging
information, but require the ihex file for the ROM image.  In
retrospect, while this approach looked more complicated to me at
first, it avoids many of these problems...
-- 
J"org Wunsch                                           Unix support engineer
address@hidden        http://www.interface-systems.de/~j/