gnupic: Thread: emulating linker script functionality?


[<<] [<] Page 1 of 1 [>] [>>]
Subject: emulating linker script functionality?
From: Charles Lepple ####@####.####
Date: 16 Dec 2002 04:48:22 -0000
Message-Id: <3DFD5999.2030008@ghz.cc>

Please bear with me as I try and explain what I want to do. I'm a little 
rusty when it comes to PIC programming, and so my terminology might be 
slightly off.

I downloaded some sample code from Microchip which is intended to 
demonstrate the use of the USB peripheral on the 16C745/765 chips. It is 
designed to be assembled and linked with MPASM, and it uses several .ASM 
files and a linker script.

Since I realize that the gputils linker is not finished, I concatenated 
the files in the link order given in the project file, commented out the 
'global' directives, and replaced the first of each 'extern' directive 
with the symbol definition itself.

That worked fine, except that since the linker script is never 
consulted, the vectors are not placed at the beginning of program space, 
etc. (from "CODEPAGE   NAME=vectors  START=0x0      END=0x4F 
PROTECTED" and "SECTION    NAME=STARTUP  ROM=vectors" in the linker script).

Has anyone else run into this before? I think I can move most of the 
code around by hand, but I was looking for a more formal method. Also, 
I'm kind of in the dark since I didn't write the original code, and I'm 
sure there are things that will take a few crash-and-burn cycles to work 
out.

Also, if you've read this far, then maybe this next suggestion might 
interest you: has anyone considered integration with GNU binutils? I 
realize that a lot of work has been put into gputils, but the GNU linker 
is pretty advanced, and would make a nice framework. Plus, it already 
has a modular COFF support system which could be adapted to work with 
the Microchip object format (if it is indeed a variant of COFF). I have 
had some experience porting binutils, but at the moment my hands are 
tied (the FSF needs a copyright assignment from me, and my university is 
being a pain about signing it) so I can only answer porting questions at 
the moment (no code writing). Food for thought, anyway...

-- 
Charles Lepple ####@####.####
http://www.ghz.cc/charles/

Subject: Re: emulating linker script functionality?
From: Craig Franklin ####@####.####
Date: 17 Dec 2002 02:42:00 -0000
Message-Id: <3DFE8E2F.D251CAD2@attbi.com>

Charles Lepple wrote:
> Also, if you've read this far, then maybe this next suggestion might
> interest you: has anyone considered integration with GNU binutils? I
> realize that a lot of work has been put into gputils, but the GNU linker
> is pretty advanced, and would make a nice framework. Plus, it already
> has a modular COFF support system which could be adapted to work with
> the Microchip object format (if it is indeed a variant of COFF). I have
> had some experience porting binutils, but at the moment my hands are
> tied (the FSF needs a copyright assignment from me, and my university is
> being a pain about signing it) so I can only answer porting questions at
> the moment (no code writing). Food for thought, anyway...

I considered using binutils, but decided against it.  I reviewed the
hc11 and avr ports to get a feel for what challenges we would have.  I
spoke with the people who did the work.  I wasn't convienced it was the
best solution for us.  We have a functioning assembler, COFF is straight
forward, and "rolling our own" allowed the greatest flexiblity.  Don't
get me wrong, binutils is a great package, but it wasn't what we needed.

This weekend I got gpasm generating COFF objects.  I successfully linked
the objects with mplink.  I hope to commit the changes this weekend. 
Once I do, I will finish up gplink.

> 
> --
> Charles Lepple ####@####.####
> http://www.ghz.cc/charles/
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: ####@####.####
> For additional commands, e-mail: ####@####.####
[<<] [<] Page 1 of 1 [>] [>>]


Powered by ezmlm-browse 0.20.