Disabling GP-addressing in generated source code

Hi Everybody,

I would like to share some thougts about the way to control when the compiler uses GP-addressing to address global data structures.

This is resulted also from a talk with Monkeyboy, that I thanks for having enlightened me :-)

Baically, my knownledge is the following:

- small data that is generated into the .sdata / .sbss section will use GP-addressing

- I can disable GP addressing for the entire compilation unit using the compiler option -G0

- if I want to remove GP addressing from a small data, I have to DECLARE the data putting it in some different section, like for exeample the ".data" section, something like

<div class='quotetop'>QUOTE </div>

--- Quote Start ---

extern volatile int mickey __attribute__ ((section (".data")));[/b]

--- Quote End ---

What I do not like in this method is that I have to put into the declaration (that stays for example inside a .h) the fact that the item will be allocated inside the ".data" section. I do not like it because the resulting object file will not contain any allocation in the data segment for that variable...

I also tried the following piece of code:

<div class='quotetop'>QUOTE </div>

--- Quote Start ---

extern volatile int mydata __attribute__ ((section (".data")));

void goofy(void)

{

mydata++;

}

volatile int mydata __attribute__ ((section (".sdata")));

void mickey(void)

{

mydata++;

} [/b]

--- Quote End ---

...that is,

1) declaring a variable mydata putting it in .data (no GP-relative addressing

2) using mydata into goofy (generating absolute code (the same ontained with -G0))

3) REDEFINING mydata to be put in .sdata (the compiler should give a warning because of the section that has changed?????)

4) using mydata into mickey (generating GP-relative code)

As the result, I obtain the following assembler (I only quoted the most interesting parts)

<div class='quotetop'>QUOTE </div>

--- Quote Start ---

.section .text

.align 2

.global goofy

.type goofy, @function

goofy:

addi sp, sp, -4

stw fp, 0(sp)

mov fp, sp

movhi r3, %hiadj(mydata)

addi r3, r3, %lo(mydata)

movhi r2, %hiadj(mydata)

addi r2, r2, %lo(mydata)

ldw r2, 0(r2)

addi r2, r2, 1

stw r2, 0(r3)

ldw fp, 0(sp)

addi sp, sp, 4

ret

.size goofy, .-goofy

.align 2

.global mickey

.type mickey, @function

mickey:

addi sp, sp, -4

stw fp, 0(sp)

mov fp, sp

ldw r2, %gprel(mydata)(gp)

addi r2, r2, 1

stw r2, %gprel(mydata)(gp)

ldw fp, 0(sp)

addi sp, sp, 4

ret

.size mickey, .-mickey

.global mydata

.section .sdata,"aws",@progbits

.align 2

.type mydata, @object

.size mydata, 4

mydata:

.zero 4

.section .debug_frame,"",@progbits[/b]

--- Quote End ---

That is,

- goofy IS NOT using GP-relative addressing

- mickey IS using GP-relative addressing

- the compiler does not complain about the mismatch of the sections!

What I would like to have is:

- the possibility to DEFINE a data structure putting it in some section (as it is now)

- the possibility to DECLARE a data structure without saying in which section it will be allocated, but just saying "hey, I want to disable GP-addressing"

Questions:

1- is it possible to DECLARE a data structure without saying in which section it will be allocated, but just saying "hey, I want to disable GP-addressing"?

2- Why the compiler do not give a warning about the conflicting sections?

3- it seems that generating GP-relative code is related ONLY to the section where the compiler HOPES the variable will be allocated to. So in some sense having __attribute__ ((section (".data"))) after a data declaration is a sufficent answer to question 1?

ok, that&#39;s all for now,

Paolo