[Patch] libffi-MIPS: Fix long double return types in n32/n64.

[Andrew Haley]

David Daney writes:
 > This patch fixes failures in the libffi testsuite for long double return
 > values.  These are the last remaining failures for mips64 in libffi.
 > 
 > The fix is to add special handling for long double return values.
 > 
 > Testing on mips64-unknown-linux-gnu n32 and n64.
 > 
 > OK to commit if no problems found?
 > 
 > 2007-12-07  David Daney  <ddaney@avtrex.com>
 > 
 >     * src/mips/ffi.c (ffi_prep_cif_machdep): Handle long double return
 >     type.
 > 
 > Index: src/mips/ffi.c
 > ===================================================================
 > --- src/mips/ffi.c	(revision 130659)
 > +++ src/mips/ffi.c	(working copy)
 > @@ -462,7 +462,13 @@ ffi_status ffi_prep_cif_machdep(ffi_cif 
 >        case FFI_TYPE_DOUBLE:
 >  	cif->flags += cif->rtype->type << (FFI_FLAG_BITS * 8);
 >  	break;
 > -
 > +      case FFI_TYPE_LONGDOUBLE:
 > +	/* Long double is returned as if it were a struct containing
 > +	   two doubles.  */
 > +	cif->flags += FFI_TYPE_STRUCT << (FFI_FLAG_BITS * 8);
 > +	cif->flags += (FFI_TYPE_DOUBLE + (FFI_TYPE_DOUBLE << FFI_FLAG_BITS))
 > +	  << (4 + (FFI_FLAG_BITS * 8));

I seem to be the indentation police today.  :-)

   cif->flags += ((FFI_TYPE_DOUBLE + (FFI_TYPE_DOUBLE << FFI_FLAG_BITS))
		  << (4 + (FFI_FLAG_BITS * 8)));

Andrew.