Re: [avr-gcc-list] Building AVRGCC on an AMD64-Box fails

[Matthias Weißer]

Joerg Wunsch schrieb:

> address@hidden (Joerg Wunsch) wrote:

> USItype
> __fixunsdfsi (DFtype a)
> {
>   if (a >= - (DFtype) (- ((SItype)(((USItype)1 << ((4 * 8) - 1)) - 1)) - 1))
>     return (SItype) (a + (- ((SItype)(((USItype)1 << ((4 * 8) - 1)) - 1)) - 1)) - (- 
> ((SItype)(((USItype)1 << ((4 * 8) - 1)) - 1)) - 1);
>   return (SItype) a;
> }

Looks exactly the same here.

> USItype is `unsigned int'
> SItype is `int'
> DFtype is `float'
>
> I've added the following piece of code for verification:
>
>  static volatile size_t usi, si, df;
>  usi = sizeof(USItype);
>  si = sizeof(SItype);
>  df = sizeof(DFtype);
>
> ....and analyzed the compiled assembler code.  usi is 4, si is 4, df is
> 4 (so it's sure that all types are evaluated in the target CPU's
> context).

The same here.

For some additional comments see my reply to Eric's mail.


--
Matthias Weißer
address@hidden
http://www.matwei.de