Why no expressions in MEMORY command?

Why no expressions in MEMORY command?

[Charles Manning]

It seems you can only use hard-coded values in a MEMORY command. 

Why can't you use scripts ike:


EXT_RAM_START = ...
EXT_RAM_SIZE = ......

MEMORY
{
    ext_ram:  ORIGIN = EXT_RAM_START, LENGTH = EXT_RAM_SIZE
}

This can make things a lot easier...

Re: Why no expressions in MEMORY command?

[Alan Modra]

On Tue, Apr 12, 2005 at 03:41:19PM +1200, Charles Manning wrote:
> Why can't you use scripts ike:

Because no one has cared enough to write support to handle such symbols
in linker scripts.

-- 
Alan Modra
IBM OzLabs - Linux Technology Centre

RE: Why no expressions in MEMORY command?

[Dave Korn]

----Original Message----
>From: Alan Modra
>Sent: 12 April 2005 09:41

> On Tue, Apr 12, 2005 at 03:41:19PM +1200, Charles Manning wrote:
>> Why can't you use scripts ike:
> 
> Because no one has cared enough to write support to handle such symbols
> in linker scripts.


  FYI:  I looked into it in some depth once; it would require quite a hefty
bit of patching, since the information about symbol values isn't evaulated
yet at the time the parser is reading the MEMORY command.  See PR ld/518 
http://sourceware.org/bugzilla/show_bug.cgi?id=518, also see
http://sources.redhat.com/ml/binutils/2004-03/msg00540.html
for a detailed analysis, and 
http://sources.redhat.com/ml/binutils/2004-03/msg00571.html
for a suggested solution.

    cheers,
      DaveK
-- 
Can't think of a witty .sigline today....

Re: Why no expressions in MEMORY command?

[Linus Nielsen Feltzing]

Dave Korn wrote:
>   FYI:  I looked into it in some depth once; it would require quite a hefty
> bit of patching, since the information about symbol values isn't evaulated
> yet at the time the parser is reading the MEMORY command.

In the Rockbox project, we use the C preprocessor on the linker command
files. That way we can use #define, #ifdef and so on. Very useful.

Linus

Re: Why no expressions in MEMORY command?

[Charles Manning]

> > Because no one has cared enough to write support to handle such symbols
> > in linker scripts.
Maybe I care enough to have a crack at this:-)

>
>   FYI:  I looked into it in some depth once; it would require quite a hefty
> bit of patching, since the information about symbol values isn't evaulated
> yet at the time the parser is reading the MEMORY command.  See PR ld/518
> http://sourceware.org/bugzilla/show_bug.cgi?id=518, also see
> http://sources.redhat.com/ml/binutils/2004-03/msg00540.html
> for a detailed analysis, and
> http://sources.redhat.com/ml/binutils/2004-03/msg00571.html
> for a suggested solution.

Hmmm bummer. It seems to come down to there being no distinction between 
expressions that can be evaluated before symbols are produced and others.

This makes the MEMORY stuff relatively limitied in its utility. There are 
work arounds, but they are far from elegant.


-- Charles

Re: Why no expressions in MEMORY command?

[Peter Barada]

>It seems you can only use hard-coded values in a MEMORY command. 
>
>Why can't you use scripts ike:
>
>
>EXT_RAM_START = ...
>EXT_RAM_SIZE = ......
>
>MEMORY
>{
>    ext_ram:  ORIGIN = EXT_RAM_START, LENGTH = EXT_RAM_SIZE
>}
>
>This can make things a lot easier...

You can always *generate* the linker script using other tools (sh,
cpp, perl, etc).

-- 
Peter Barada
peter@the-baradas.com