Re: macro's and local variables

[Ian Lance Taylor <iant at google dot com>]

On Tue, Oct 16, 2012 at 11:37 AM, Mischa Baars <mjbaars1977@gmail.com> wrote:
> On 10/16/2012 08:20 PM, Ian Lance Taylor wrote:
>>
>> On Tue, Oct 16, 2012 at 11:12 AM, Mischa Baars <mjbaars1977@gmail.com>
>> wrote:
>>>>
>>>> Also one uses i and the other uses j, and that matters because
>>>> you are invoking them in a loop that also uses i.
>>>
>>> This is exactly what I'm pointing at. Apart from the register naming, I
>>> was
>>> expecting the same values to be assigned to the respective registers.
>>> This
>>> does not happen, because 'i' is not a local but a global value.
>>
>> Yes.  That is how assembler macros work.
>
> So macro B is the only correct macro, as I understand.

I would say that it's the only one that may be correctly invoked by
macro C.  In isolation both A and B are fine.  Perhaps it is C that is
buggy.


>> Assembler macros are not a high level programming language.  They are
>> macros: bodies of text that are expanded with argument values
>> substituted in textually.  You may want to read about how \@ is
>> expanded in a macro body.
>
> I found the link:
> http://sourceware.org/binutils/docs-2.22/as/Macro.html#Macro
>
> However, I do not understand. Do you have an example?

		.intel_syntax	noprefix
		.global		function
		.code64
.macro		A		arg1
		.set		i\@, 0
.rept		\arg1	
		mov		ax, i\@
		.set		i\@, i\@ + 1
.endr
.endm

.macro		C
		.set		i\@, 0
.rept		4	
		A		i\@
		.set		i\@, i\@ + 1
.endr
.endm
function:
		C	
		ret