Re: [avr-gcc-list] objdump patch

[Erik Christiansen]

On Tue, Sep 07, 2004 at 09:40:41AM +0200, Svein E. Seldal wrote:
> 
> Well... I cant control how the symbol will be printed. I.e. if a symbol 
> is non-exisiting, it seems like it will always backtrack until it finds one:
> 
> 41a:       31 f0           breq    0x428 <main+0x6c>       ; .+12
> 

Ahh, I didn't realise that would happen. To my eye main+0x6c is even
better than ".+12". The <main+0x6c> syntax is reminiscent of gdb, I
think. And your columnar output goal is very attractive, imho:

20c:       02 c0           rjmp    0x212     <.do_copy_data_start>

41a:       31 f0           breq    0x428     <main+0x6c>

> The function for printing the symbol is a built-in (global) function, 
> which I do not want to override.

Well and truly understood.

> Since I do not know how the symbol output will look like, I cannot 
> control if I'm going to need the "; .+xxx" comment or not. So either its 
> on or either its off.

Well, now that we have seen the alternative, maybe the "; .+xxx" won't
be missed. (I for one, tend to match the jmp's absolute address with the
line's address prefix, rather than count my way there.)

Erik