newlib and the mno-fpu option

newlib and the mno-fpu option

[Terry Hipp]

I have successfully built a windows hosted cross compiler for a sparc-elf 
target.  The embedded sparc target does not have a fpu, so I am compiling it 
with the mno-fpu option. A test application can successfully complete float 
operations such as floatValue += 0.5; but exceptions when I try to print the 
value using
printf("floatValue = %f\r\n", floatValue);  Is it sufficient to compile the 
application with the mno-fpu option or does newlib have to be built with 
this option as well?

Thanks
Terry



_________________________________________________________________
Get your FREE download of MSN Explorer at http://explorer.msn.com/intl.asp.

Re: newlib and the mno-fpu option

[Andrea 'Fyre Wyzard' Bocci]

At 09.41 11/01/2002 (GMT -0500), Terry Hipp wrote:
>The embedded sparc target does not have a fpu, so I am compiling it with 
>the mno-fpu option. A test application can successfully complete float 
>operations such as floatValue += 0.5; but exceptions when I try to print 
>the value using
>printf("floatValue = %f\r\n", floatValue);  Is it sufficient to compile 
>the application with the mno-fpu option or does newlib have to be built 
>with this option as well?

I think newlib, too, needs be compiled with the option.

fwyzard

Re: newlib and the mno-fpu option

[Terry Hipp]

fyi...
I am no expert but when you build newlib the only option you have is the 
--nfp during the configure phase. However, this option only applies to 
sun-sunos and m68k-isi-bsd machines and has no effect on other systems. I 
rebuilt newlib with the --nfp option to confirm this.  I believe when newlib 
is built a soft-float version is built in a "soft" directory. The library 
version linked into the application depends on whether you use the -mno-fpu 
or -msoft-float option during compilation.

Must be something else going on.


>From: Andrea 'Fyre Wyzard' Bocci <fwyzard@inwind.it>
>To: "Terry Hipp" <terry_hipp@hotmail.com>,help-gcc@gnu.org
>Subject: Re: newlib and the mno-fpu option
>Date: Sat, 12 Jan 2002 00:31:31 +0100
>
>At 09.41 11/01/2002 (GMT -0500), Terry Hipp wrote:
>>The embedded sparc target does not have a fpu, so I am compiling it with
>>the mno-fpu option. A test application can successfully complete float
>>operations such as floatValue += 0.5; but exceptions when I try to print
>>the value using
>>printf("floatValue = %f\r\n", floatValue);  Is it sufficient to compile
>>the application with the mno-fpu option or does newlib have to be built
>>with this option as well?
>
>I think newlib, too, needs be compiled with the option.
>
>fwyzard
>




_________________________________________________________________
Get your FREE download of MSN Explorer at http://explorer.msn.com/intl.asp.