gcc question for inline assembly.

gcc question for inline assembly.

[Stuart Cracraft]

Hi - someone supplied the following code
in a C++ program to me that g++ compiled.

However, I want to convert it to use it in
a regular C program that gcc can compile.

Needless to say, the construct does not work
for GNU C.

I've tried placing each line of the assembly
in asm("...."); and that also fails to get
gcc to compile it.

What would the equivalent GNU C function for the
below be?

static unsigned int bitScanAndReset(unsigned long long & bb) {

       __asm
       {
               xor     edx, edx
               mov     ebx, [bb]
               xor     eax, eax
               inc     edx
               bsf     ecx, [ebx]
               jnz     found
               bsf     ecx, [ebx+4]
               lea     ebx, [ebx+4]
               xor     eax, 32
       found:
               shl     edx, cl
               xor     eax, ecx
               xor     [ebx], edx
       }
}

Thanks,

Stuart

Re: gcc question for inline assembly.

[Brian Budge]

Hi Stuart -

This is really a C/C++ question, and not really a gcc/c++ question.

However, in C, you can't pass things by reference (the &), you should
pass by address instead, and then dereference to set the values.

  Brian

On 9/11/06, Stuart Cracraft <cracraft@cox.net> wrote:
> Hi - someone supplied the following code
> in a C++ program to me that g++ compiled.
>
> However, I want to convert it to use it in
> a regular C program that gcc can compile.
>
> Needless to say, the construct does not work
> for GNU C.
>
> I've tried placing each line of the assembly
> in asm("...."); and that also fails to get
> gcc to compile it.
>
> What would the equivalent GNU C function for the
> below be?
>
> static unsigned int bitScanAndReset(unsigned long long & bb) {
>
>        __asm
>        {
>                xor     edx, edx
>                mov     ebx, [bb]
>                xor     eax, eax
>                inc     edx
>                bsf     ecx, [ebx]
>                jnz     found
>                bsf     ecx, [ebx+4]
>                lea     ebx, [ebx+4]
>                xor     eax, 32
>        found:
>                shl     edx, cl
>                xor     eax, ecx
>                xor     [ebx], edx
>        }
> }
>
> Thanks,
>
> Stuart
>
>
>

Re: gcc question for inline assembly.

[John Love-Jensen]

Hi Stuart,

Your question is off-topic for this forum.

The routine looks for the lowest bit set in the bb variable, a 64-bit
number.  It then sets the bb variable to that number.  It also returns the
bit position of that lowest bit (1-based: starting at 1 for the "least
significant bit", to 64 for the "most significant bit").

If the number passed in is 0 (i.e., bb == 0ULL), it poor behavior.  Don't
call this routine with 0ULL.

I recommend rewriting the routine in pure C, and avoid assembly language.

If you need help on how to do that, get this book:  Hacker's Delight
http://www.amazon.com/Hackers-Delight-Henry-Warren-Jr/dp/0201914654

HTH,
--Eljay

Re: gcc question for inline assembly.

[Marcelo Slomp]

Stuart Cracraft wrote:
> 
> Hi - someone supplied the following code
> in a C++ program to me that g++ compiled.
> 
> However, I want to convert it to use it in
> a regular C program that gcc can compile.
> 
> Needless to say, the construct does not work
> for GNU C.
> 
> I've tried placing each line of the assembly
> in asm("...."); and that also fails to get
> gcc to compile it.
> 
> What would the equivalent GNU C function for the
> below be?
> 
> static unsigned int bitScanAndReset(unsigned long long & bb) {
> 
>        __asm
>        {
>                xor     edx, edx
>                mov     ebx, [bb]
>                xor     eax, eax
>                inc     edx
>                bsf     ecx, [ebx]
>                jnz     found
>                bsf     ecx, [ebx+4]
>                lea     ebx, [ebx+4]
>                xor     eax, 32
>        found:
>                shl     edx, cl
>                xor     eax, ecx
>                xor     [ebx], edx
>        }
> }
> 
> Thanks,
> 
> Stuart
> 
> 

Notice that the assembler used by gcc (as) uses by default the at&t assembly
syntax, and you have intel code.
You can either, translate the code to at&t syntax or use the -masm=intel
flag to compile (mount) this code.

read more on at&t syntax at:
http://www.ibiblio.org/gferg/ldp/GCC-Inline-Assembly-HOWTO.html

Regards,
Marcelo Slomp
-- 
View this message in context: http://www.nabble.com/gcc-question-for-inline-assembly.-tf2256651.html#a6264969
Sent from the gcc - Help forum at Nabble.com.