This is the mail archive of the crossgcc@sourceware.cygnus.com mailing list for the crossgcc project.

See the CrossGCC FAQ for lots more infromation.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Specifics on declaring sections in C code


Hello,

I am converting some code from MRI to GNU (m68k target) and have run
into a couple of issues.


1. In my C source code, how do I direct specific types of information to
certain sections? E.g. what if I want my code and initialized data to go
into one section, uninitialized data to go into another section, and
string literals to go into another? Of course, I want to specify the
sections, since I want to override the default ".bss", ".text", ".data",
etc.

Here is a piece of code that works with MRI:
____________________________________________________________
#include "idents.h"

#pragma option -NCUITBL_ROM
#pragma option -NZUITBL_RAM

DMFL_Ram_t DMFL_Ram[NUM_UI_VAR_LOC];

const DMFL_Rom_t DMFL_Rom [] =
{
/* * index, name * initvalue */
   <cut>
};
____________________________________________________________

If someone that knows about the "-NC" and "-NZ" MRI options I'd
appreciate letting me know what they do (I don't have documentation). I
*think* they say to put code (and possible initialized data?) in the
section "UITBL_ROM" and to put uninitialized data in the section
"UITBL_RAM". I did verify in the map file that "DMFL_Ram" was put into
"UITBL_RAM" and that "DMFL_Rom" was put into ROM. I assume the MRI
compiler made the distinction based on whether or not the variable is
initialized.

So, how can I do something similar with GNU? What I came up with is
inconvenient:
____________________________________________________________
#include "idents.h"

__asm__( ".section UITBL_RAM,\"d\"" );

DMFL_Ram_t DMFL_Ram[NUM_UI_VAR_LOC];

__asm__( ".section UITBL_ROM,\"x\"" );

const DMFL_Rom_t DMFL_Rom [] =
{
/* * index, name * initvalue */
   <cut>
};
____________________________________________________________

This can get very messy in a larger file that has initialized data,
unititialized data, and source code mixed up a bit.. Also, it's much
more prone to error if I add stuff in the future to make sure that it is
added after the proper directive. With MRI's options, I can just put all
relevant options at the top of the file and then forget about it. No
matter where I add various kinds of information they will always get
placed in the proper section (assuming I've interpreted the meanings of
the options accurately).

Does anyone know of a cleaner way to do this with GNU/GCC/AS? I'd be
interested in assembler directives, C directives, command line options,
whatever.


2. Is there a return type from GNU's __asm__()? It appears that MRI's
asm() can be used in the following way:

    pTblRam = asm(void*, " move.l #.STARTOF.(UITBL_RAM),D0 ");

Does GNU's __asm__() allow you to provide two arguments and/or use a
return value? It would appear that MRI's asm() returns the value of the
D0 register, with the type that is given as the first parameter ("void*"
in this case). Again, MRI people please correct me if I'm wrong here.
Anyone know how I should convert this line of MRI-compatible source code
to GNU?

Is GNU's __asm__ directive documented thoroughly somewhere?


I'd appreciate any help! Naturally I've already looked through the GNU
gcc, ld, and as documents, as well as used 'man' and 'info' (which was
mostly the same as what was in the documents).

Chris
begin:vcard 
n:Bahns;Christopher
tel;home:812-342-4714
tel;work:812-342-4714
x-mozilla-html:FALSE
adr:;;;;;;
version:2.1
email;internet:chris@bahns.com
fn:Christopher Bahns
end:vcard

------
Want more information?  See the CrossGCC FAQ, http://www.objsw.com/CrossGCC/
Want to unsubscribe? Send a note to crossgcc-unsubscribe@sourceware.cygnus.com

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]