print floating pt using xmm0

[Peter Johnson]

It appears the AMD64 ABI for varargs functions requires all 8-byte double 
values.  The printf function is varargs, so all floats are passed as 
doubles.  A "normal" function call that has float in the arguments list 
takes it as a 4-byte value.

(above per disassembly of GCC output on Linux/amd64)

Peter

On Thu, 26 Oct 2006, Vikas N Kumar wrote:

> But what is wrong with using cvtsi2ss ?
> I want to print a 4byte floating point as opposed to a 8byte double .
>
> Thanks
> Vikas
>
> On 10/26/06, Peter Johnson <peter at tortall.net> wrote:
>> You need to use cvtsi2sd, not cvtsi2ss.
>>
>> Peter
>>
>> On Thu, 26 Oct 2006, Vikas N Kumar wrote:
>>
>>> Hi All
>>> Below is my code to print a floating point number which has been
>>> loaded into XMM0 register.
>>> For some reason it always prints 0.0000. I am not able to figure it
>>> out.  Could anyone let me know what I am doing wrong here ? It seems
>>> right based on the manual.
>>>
>>> section .data
>>>    fltprmpt db "%f",10,0
>>>
>>> section .text
>>>
>>>    global main
>>>    extern printf
>>>
>>>    main:
>>>        enter 0,0
>>>        mov     rdi, dword fltprmpt
>>>        mov     rdx, 10
>>>        cvtsi2ss    xmm0, edx
>>>        mov     rax, 1
>>>        call    printf
>>>        leave
>>>        ret
>>>
>>> Regards,
>>> Vikas
>>> --
>>> http://www.vikaskumar.org/
>>> _______________________________________________
>>> yasm-devel mailing list
>>> yasm-devel at tortall.net
>>> http://cvs.tortall.net/mailman/listinfo/yasm-devel
>>>
>>
>
>
> -- 
> http://www.vikaskumar.org/
> _______________________________________________
> yasm-devel mailing list
> yasm-devel at tortall.net
> http://cvs.tortall.net/mailman/listinfo/yasm-devel
>