This is the mail archive of the
binutils@sources.redhat.com
mailing list for the binutils project.
RE: Can I fix the content of symbol table in GAS???
- From: "pschen" <pschen at pllab dot cs dot nthu dot edu dot tw>
- Cc: <binutils at sources dot redhat dot com>
- Date: Thu, 29 Jul 2004 21:53:11 +0800
- Subject: RE: Can I fix the content of symbol table in GAS???
Dear Nick:
> > Currently I am developing the gas for a new DSP processor.
> > In GAS, I want to add a symbol "tmp" in symbol table and
> > relocate its content.
> > I don't know how to program it and I just can do relocation in text
section.
> > Can I do relocation in symbol table?
>
> Not really. Generally speaking relocations are used to alter the
> contents of code and data sections not the symbol table.
>
> You can create new symbols to insert into the symbol table. For
> examples of this see the file <binutils>/gas/config/tc-ia64.c and search
> for occurrences of "symbol_new".
>
> What exactly do you mean when you say that you want to relocate the
> content of symbol "tmp". Do you mean that you want to be able set its
> address at link-time rather than assemble-time ? Or do you mean that
> you want to alter the contents of the memory location whose address is
> the value of the symbol "tmp" ?
>
Thanks your reply.
I want to support pseudo function "ltoff", but the porting of IA64 and BFD
is hard to read.
My idea is that inserting a symbol to solve the probem of pseudo function
"ltoff".
If @ltoff(b) occurred, a symbol "tmp" whose content is the address of "b"
will
be inserted into symbol table. Then, instruction relocation will be done for
"tmp", not "b".
I have got a success on this idea.
Sincerely,
Peng-Sheng