Re: Decode function entry mask on VAXen

[Jan-Benedict Glaw <jbglaw at microdata-pos dot de>]

On Tue, Mar 08, 2005 at 10:44:00PM +1030, Alan Modra wrote:

Hi Alan!

> objdump already breaks up calls to disassemble_bytes into the pieces of
> a section between two defined symbols (possibly validated by a target
> symbol_is_valid function).  So if the only symbols in vax code are those
> at the beginning of a function, your idea of using last_symbol_address
> will work.  However, I think it would be better to pass
> "asymbol *last_symbol" instead, which gives your backend disassembler_fn
> a chance to deal with non-function symbols that might appear in code.

I'll try to implement that. For sure, non-function symbols will show up
from time to time... However, in a typical ELF object file, there are
somewhat rare, while wrongly disassembled functions were the standard :)

> Note that gdb also uses the disassembler, so it would be a good idea to
> see how your ideas fit in with gdb too.

I think vax-linux isn't yet a supported gdb target, but I'll try to get
NetBSD running on one of my /90 and have a test with it, too. The BSD
guys should have basically the same problems here...

Thanks again, JBG

-- 
AWEK microdata GmbH -- Am Wellbach 4 -- 33609 Bielefeld