commit de91dff5d403387a8f0dafbe8dbbb18ecd416570
author Craig Topper <craig.topper@intel.com> Mon Jan 08 22:37:56 2018 +0000
committer Craig Topper <craig.topper@intel.com> Mon Jan 08 22:37:56 2018 +0000
tree 5778fd276ac1ee3cfa723cd8f337c8e62a3bdab7
parent 53bf4ba69dafdc8d2b97149d8fceaa8a13b671c4

[X86] Replace cvt*2mask intrinsics with native IR using 'icmp slt X, zeroinitializer.

llvm-svn: 322038

clang/lib/CodeGen/CGBuiltin.cpp

diff --git a/clang/lib/CodeGen/CGBuiltin.cpp b/clang/lib/CodeGen/CGBuiltin.cpp
index 631085a..9ffc7de 100644
--- a/clang/lib/CodeGen/CGBuiltin.cpp
+++ b/clang/lib/CodeGen/CGBuiltin.cpp
@@ -7791,7 +7791,9 @@
 }
 
 static Value *EmitX86MaskedCompare(CodeGenFunction &CGF, unsigned CC,
-                                   bool Signed, SmallVectorImpl<Value *> &Ops) {
+                                   bool Signed, ArrayRef<Value *> Ops) {
+  assert((Ops.size() == 2 || Ops.size() == 4) &&
+         "Unexpected number of arguments");
   unsigned NumElts = Ops[0]->getType()->getVectorNumElements();
   Value *Cmp;
 
@@ -7815,9 +7817,11 @@
     Cmp = CGF.Builder.CreateICmp(Pred, Ops[0], Ops[1]);
   }
 
-  const auto *C = dyn_cast<Constant>(Ops.back());
-  if (!C || !C->isAllOnesValue())
-    Cmp = CGF.Builder.CreateAnd(Cmp, getMaskVecValue(CGF, Ops.back(), NumElts));
+  if (Ops.size() == 4) {
+    const auto *C = dyn_cast<Constant>(Ops[3]);
+    if (!C || !C->isAllOnesValue())
+      Cmp = CGF.Builder.CreateAnd(Cmp, getMaskVecValue(CGF, Ops[3], NumElts));
+  }
 
   if (NumElts < 8) {
     uint32_t Indices[8];
@@ -7833,6 +7837,11 @@
                                                     std::max(NumElts, 8U)));
 }
 
+static Value *EmitX86ConvertToMask(CodeGenFunction &CGF, Value *In) {
+  Value *Zero = Constant::getNullValue(In->getType());
+  return EmitX86MaskedCompare(CGF, 1, true, { In, Zero });
+}
+
 static Value *EmitX86Abs(CodeGenFunction &CGF, ArrayRef<Value *> Ops) {
 
   llvm::Type *Ty = Ops[0]->getType();
@@ -8179,6 +8188,20 @@
   case X86::BI__builtin_ia32_cvtmask2q512:
     return EmitX86SExtMask(*this, Ops[0], ConvertType(E->getType()));
 
+  case X86::BI__builtin_ia32_cvtb2mask128:
+  case X86::BI__builtin_ia32_cvtb2mask256:
+  case X86::BI__builtin_ia32_cvtb2mask512:
+  case X86::BI__builtin_ia32_cvtw2mask128:
+  case X86::BI__builtin_ia32_cvtw2mask256:
+  case X86::BI__builtin_ia32_cvtw2mask512:
+  case X86::BI__builtin_ia32_cvtd2mask128:
+  case X86::BI__builtin_ia32_cvtd2mask256:
+  case X86::BI__builtin_ia32_cvtd2mask512:
+  case X86::BI__builtin_ia32_cvtq2mask128:
+  case X86::BI__builtin_ia32_cvtq2mask256:
+  case X86::BI__builtin_ia32_cvtq2mask512:
+    return EmitX86ConvertToMask(*this, Ops[0]);
+
   case X86::BI__builtin_ia32_movdqa32store128_mask:
   case X86::BI__builtin_ia32_movdqa64store128_mask:
   case X86::BI__builtin_ia32_storeaps128_mask: