[mono] respect hardfloat/softloat setting in ARM ABI (mono#16)

Mono's LLVM backend was always emitting code like this:

> 000d7234 <double_ToString>:
>    d7234:       e92d4800        push    {fp, lr}
>    d7238:       ed2d8b02        vpush   {d8}
>    d723c:       ed908b00        vldr    d8, [r0]
>    d7240:       eb12e4e2        bl      5905d0 <plt_System_Globalization_NumberFormatInfo_get_CurrentInfo>
>    d7244:       e1a03000        mov     r3, r0
>    d7248:       ec510b18        vmov    r0, r1, d8
>    d724c:       e3a02000        mov     r2, #0
>    d7250:       eb12e91a        bl      5916c0 <plt_System_Number_FormatDouble_double_string_System_Globalization_NumberFormatInfo>
>    d7254:       ecbd8b02        vpop    {d8}
>    d7258:       e8bd8800        pop     {fp, pc}

despite the correct setting passed to `llc`. Note that passing the floating
point value (`d8`) in the integer registers `r0` and `r1` is wrong.

With this change we get the following:

> 000d9934 <double_ToString>:
>    d9934:       e92d4800        push    {fp, lr}
>    d9938:       ed2d8b02        vpush   {d8}
>    d993c:       ed908b00        vldr    d8, [r0]
>    d9940:       eb137c2e        bl      5b8a00 <plt_System_Globalization_NumberFormatInfo_get_CurrentInfo>
>    d9944:       eeb00b48        vmov.f64        d0, d8
>    d9948:       e1a01000        mov     r1, r0
>    d994c:       e3a00000        mov     r0, #0
>    d9950:       eb138066        bl      5b9af0 <plt_System_Number_FormatDouble_double_string_System_Globalization_NumberFormatInfo>
>    d9954:       ecbd8b02        vpop    {d8}
>    d9958:       e8bd4800        pop     {fp, lr}
>    d995c:       e1a0f00e        mov     pc, lr

Which matches with what the Mono JIT emits:

>   0:   e92d4100        push    {r8, lr}
>   4:   e24dd028        sub     sp, sp, mono#40     ; 0x28
>   8:   e58d0018        str     r0, [sp, mono#24]
>   c:   e59d0018        ldr     r0, [sp, mono#24]
>  10:   ed900b00        vldr    d0, [r0]
>  14:   ed8d0b08        vstr    d0, [sp, mono#32]
>  18:   eb00000e        bl      0x58
>  1c:   e1a01000        mov     r1, r0
>  20:   ed9d0b08        vldr    d0, [sp, mono#32]
>  24:   e3a00000        mov     r0, #0
>  28:   eb000007        bl      0x4c
>  2c:   e28dd028        add     sp, sp, mono#40     ; 0x28
>  30:   e8bd8100        pop     {r8, pc}

Both are passing the floating point argument correctly via `d0`.

Fixes mono/mono#11095

lib/Target/ARM/ARMCallingConv.td

@@ -334,3 +334,11 @@ def CC_ARM_Mono_AAPCS : CallingConv<[
   CCDelegateTo<CC_ARM_AAPCS>
 ]>;
 
+def CC_ARM_Mono_AAPCS_VFP : CallingConv<[
+  // Mono marks the parameter it wants to pass in this non-abi register with
+  // the 'inreg' attribute.
+  CCIfInReg<CCAssignToReg<[R8]>>,
+
+  CCDelegateTo<CC_ARM_AAPCS_VFP>
+]>;
+

lib/Target/ARM/ARMISelLowering.cpp

@@ -1623,10 +1623,13 @@ CCAssignFn *ARMTargetLowering::CCAssignFnForNode(CallingConv::ID CC,
     if (Return) {
       return CCAssignFnForNode(CallingConv::C, true, isVarArg);
     } else {
-      if (Subtarget->isAAPCS_ABI())
-        return CC_ARM_Mono_AAPCS;
-      else
-        return CC_ARM_Mono_APCS;
+      if (Subtarget->isAAPCS_ABI()) {
+          if (Subtarget->hasVFP2() && !Subtarget->isThumb1Only() && !isVarArg)
+              return CC_ARM_Mono_AAPCS_VFP;
+          else
+              return CC_ARM_Mono_AAPCS;
+      } else
+          return CC_ARM_Mono_APCS;
     }
   }
 }