Re: [hatari-devel] VBR as a breakpoint condition

[Thorsten Otto <admin@xxxxxxxxxxx>]

On Freitag, 7. September 2018 07:25:45 CEST Miro Kropáček wrote:
> and for C compiled programs you'd need a compiler which doesn't use an
> assembler as the last step to make this happen (otherwise that assembler
> would complain too). So I think we are safe. :)

Actually, my previous statement wasn't quite right. In a compiler that 
produces elf, you *could* have such a label, that's why you always should use 
"%" as register prefix. In output generated for our target, that won't happen 
since the compiler adds an underscore to the  user label name.

To make it safe, you could use a similar technique when looking up symbols: 
search for a debug symbol first, unless the name to search for is prefixed by 
a "%". That way, you can always get the correct behaviour, even if there are 
conflicting symbols. Another possibility would be to check the names when 
loading the symbols, and warn about it when any of them matches a register 
name.