Handle llvm.fma.* intrinsics. rdar://10914096
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@154439 91177308-0d34-0410-b5e6-96231b3b80d8
diff --git a/lib/Target/ARM/ARMISelLowering.cpp b/lib/Target/ARM/ARMISelLowering.cpp
index fcb4f4f..a103c94 100644
--- a/lib/Target/ARM/ARMISelLowering.cpp
+++ b/lib/Target/ARM/ARMISelLowering.cpp
@@ -769,8 +769,10 @@
setOperationAction(ISD::FPOW, MVT::f64, Expand);
setOperationAction(ISD::FPOW, MVT::f32, Expand);
- setOperationAction(ISD::FMA, MVT::f64, Expand);
- setOperationAction(ISD::FMA, MVT::f32, Expand);
+ if (!Subtarget->hasVFP4()) {
+ setOperationAction(ISD::FMA, MVT::f64, Expand);
+ setOperationAction(ISD::FMA, MVT::f32, Expand);
+ }
// Various VFP goodness
if (!TM.Options.UseSoftFloat && !Subtarget->isThumb1Only()) {
diff --git a/lib/Target/ARM/ARMInstrNEON.td b/lib/Target/ARM/ARMInstrNEON.td
index ae52d38..99dbb95 100644
--- a/lib/Target/ARM/ARMInstrNEON.td
+++ b/lib/Target/ARM/ARMInstrNEON.td
@@ -4133,6 +4133,14 @@
v4f32, fmul_su, fsub_mlx>,
Requires<[HasNEON2,FPContractions]>;
+// Match @llvm.fma.* intrinsics
+def : Pat<(fma (v2f32 DPR:$src1), (v2f32 DPR:$Vn), (v2f32 DPR:$Vm)),
+ (VFMAfd DPR:$src1, DPR:$Vn, DPR:$Vm)>,
+ Requires<[HasNEON, HasVFP4]>;
+def : Pat<(fma (v4f32 QPR:$src1), (v4f32 QPR:$Vn), (v4f32 QPR:$Vm)),
+ (VFMAfq QPR:$src1, QPR:$Vn, QPR:$Vm)>,
+ Requires<[HasNEON, HasVFP4]>;
+
// Vector Subtract Operations.
// VSUB : Vector Subtract (integer and floating-point)
diff --git a/lib/Target/ARM/ARMInstrVFP.td b/lib/Target/ARM/ARMInstrVFP.td
index e9d5720..a9ad4fa 100644
--- a/lib/Target/ARM/ARMInstrVFP.td
+++ b/lib/Target/ARM/ARMInstrVFP.td
@@ -1080,6 +1080,14 @@
(VFMAS SPR:$dstin, SPR:$a, SPR:$b)>,
Requires<[HasVFP4,DontUseNEONForFP,FPContractions]>;
+// Match @llvm.fma.* intrinsics
+def : Pat<(fma (f64 DPR:$Ddin), (f64 DPR:$Dn), (f64 DPR:$Dm)),
+ (VFMAD DPR:$Ddin, DPR:$Dn, DPR:$Dm)>,
+ Requires<[HasVFP4]>;
+def : Pat<(fma (f32 SPR:$Sdin), (f32 SPR:$Sn), (f32 SPR:$Sm)),
+ (VFMAS SPR:$Sdin, SPR:$Sn, SPR:$Sm)>,
+ Requires<[HasVFP4]>;
+
def VFMSD : ADbI<0b11101, 0b10, 1, 0,
(outs DPR:$Dd), (ins DPR:$Ddin, DPR:$Dn, DPR:$Dm),
IIC_fpFMAC64, "vfms", ".f64\t$Dd, $Dn, $Dm",