[Bug target/64687] powerpc: GCC 5.0 (trunk@219851) build failure for powerpc with hardware decimal floating point disabled

[bergner at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64687

--- Comment #4 from Peter Bergner <bergner at gcc dot gnu.org> ---
Here's a simplified test case and simplified options:

[bergner@makalu-lp1 PR64687]$ cat decode.i 
union ieee754r_Decimal32
{
  _Decimal32 sd;
  unsigned int cc0;
};

unsigned int
__decoded32 (_Decimal32 a)
{
    union ieee754r_Decimal32 d;
    d.sd = a;
    return d.cc0;
}

[bergner@makalu-lp1 PR64687]$
/home/bergner/gcc/build/gcc-fsf-mainline-pr64687-debug/gcc/xgcc
-B/home/bergner/gcc/build/gcc-fsf-mainline-pr64687-debug/gcc -O1 -mno-hard-dfp
-mcpu=power7 -S decode.i 
decode.i: In function ‘__decoded32’:
decode.i:13:1: error: insn does not satisfy its constraints:
 }
 ^
(insn 16 4 3 2 (set (mem/c:SD (plus:DI (reg/f:DI 1 1)
                (const_int 48 [0x30])) [0 a+0 S4 A64])
        (reg:SD 33 1)) decode.i:9 492 {movsd_hardfloat}
     (nil))
decode.i:13:1: internal compiler error: in extract_constrain_insn, at
recog.c:2246