.cpload $25 understanding help

[David Daney]

Deep Visana wrote:
> Hi,
>    
>   Writing this to you after reading mail thread
> discussion under:
>  
> http://sourceware.org/ml/binutils/2004-12/msg00094.html
>    
>   Need your help as I am new to MIPS and assembly code
> both.
>    
>   We have seen that gcc compiler emits .cpload $25.
>    
>   Q1: ".cpload $25" pseudo-ops and needs to be
> interpreted by gnu assembler?
>   
.cpload $25 causes $gp to be set to the proper value given that $25 
contains the address of the same .cpload (which should always be the 
function entry point.
>    
>   Q2: What should be done by assembler to replace it
> with correct instruction. What are the content of $25?
>   

If you read the message that you referenced you would see that the 
answer is:

 	lui	$gp,%hi(_gp_disp)
 	addiu	$gp,$gp,%lo(_gp_disp)
 	addu	$gp,$gp,.cpload argument


$25 contains the address of the function being called.

>   Q3: Is there any good refernce/documentation to such
> mips specific pseudo ops available somewhere?
>   

Yes.  I once found one by using Google.  I forget where I put it.
>    
>   Our basic problem is that we need to feed gcc
> compiled ".s" to Greenhills MIPS assembler and it does
> not understand .cpload pseudo. Any suggestions on how
> to replace .cpload with correct m/c instruction in few
> such .s file? 
>
> Cpu is mips 5k running in 32bit mode.
>   

You don't say which ABI the Greenhills tools use.  If they don't use the 
o32 ABI then you will probably have problems.  GCC allows you to choose 
several different ABIs.  You should choose one that matches what 
Greenhills is using.

David Daney