Re: [avr-gcc-list] Another eager mega8 user

[Marek Michalkiewicz]

> When compiling for the mega8, I am seeing that interrupts don't make it into 
> the vector table. All the vectors are just relative jumps back to vector 0 
> (the reset vector, which is working quite fine :).

Ack.  There is a problem with all <=8K devices (that's what I get for
being lazy and testing everything only on ATmega323 :) - the assembler
resolves rjmp offsets too early, instead of letting the linker do it.
Use "avr-objdump -r" to see the difference between avr4/crtm8.o and
avr5/crtm163.o, for example - the latter has R_AVR_CALL relocs and
they are fine, the former is missing the R_AVR_13_PCREL relocs for
all interrupt vectors.

Many thanks for testing and reporting it - I'll try to fix it and make
a new avr-libc release soon.

Marek

avr-gcc-list at http://avr1.org