Library support in gcc

[sumanth]

I observed other t-target files.
I declared  "long long" as 32-bit and "double" as 32bit in target.h file.
In my t-target file, i gave the option TARGET_LIBGCC2_CFLAGS= -DSF=SF 
-DDF=SF -DSI=SI -DDI=SI
so that my DF should be SF and my DI should be SI.
Are there any side effects?

More over,  there is no hardware support for float .  How can I tell 
compiler it is software floating calculation.
Is there anything that I can do with -msoft-float? ...

Thanks,
Sumanth G

Ian Lance Taylor wrote:
> sumanth <sumanth.gundapneni@redpinesignals.com> writes:
>
>   
>> I am building a cross-compiler for a 32-bit processor.
>> Gcc version is gcc-4.3.4.
>> There is no support for DImode(or larger) and DFmode(or larger)
>> Are the libraries from libgcc2.c(gcc/libgcc2.c) and fp-bit.c in
>> gcc/config/ directory are sufficient for my compiler.
>>     
>
> Yes.
>
>   
>> How can I avoid building DI/DF mode libraries and where should I specify it.
>>     
>
> If you want to support the "long long" and "double" types at their usual
> sizes, you need to build the libraries.  There are many existing targets
> in this situation--look at the t-* files.  That is the path you should
> take unless you have overwhelming reasons to avoid it.
>
> Ian
>
>
>