ARM32 Vector lowering - scalarize select

src/IceTargetLowering.h

 //===- subzero/src/IceTargetLowering.h - Lowering interface -----*- C++ -*-===//
 //
 //                        The Subzero Code Generator
 //
 // This file is distributed under the University of Illinois Open Source
 // License. See LICENSE.TXT for details.
 //
 //===----------------------------------------------------------------------===//
 ///
 /// \file
@@ skipping 307 matching lines @@
   virtual void addProlog(CfgNode *Node) = 0;
   virtual void addEpilog(CfgNode *Node) = 0;
 
   virtual ~TargetLowering() = default;
 
 private:
   // This control variable is used by AutoBundle (RAII-style bundle
   // locking/unlocking) to prevent nested bundles.
   bool AutoBundling = false;
 
+  /// This indicates whether we are in the genTargetHelperCalls phase, and
+  /// therefore can do things like scalarization.
+  bool GeneratingTargetHelpers = false;
+
   // _bundle_lock(), and _bundle_unlock(), were made private to force subtargets
   // to use the AutoBundle helper.
   void
   _bundle_lock(InstBundleLock::Option BundleOption = InstBundleLock::Opt_None) {
     Context.insert<InstBundleLock>(BundleOption);
   }
   void _bundle_unlock() { Context.insert<InstBundleUnlock>(); }
 
 protected:
   /// AutoBundle provides RIAA-style bundling. Sub-targets are expected to use
@@ skipping 122 matching lines @@
   InstCall *makeHelperCall(const IceString &Name, Variable *Dest,
                            SizeT MaxSrcs);
 
   void _set_dest_redefined() { Context.getLastInserted()->setDestRedefined(); }
 
   bool shouldOptimizeMemIntrins();
 
   void scalarizeArithmetic(InstArithmetic::OpKind K, Variable *Dest,
                            Operand *Src0, Operand *Src1);
 
+  Variable *makeExtract(Operand *Src, Operand *Index);
+
   /// Generalizes scalarizeArithmetic to support other instruction types.
   ///
-  /// MakeInstruction is a function-like object with signature
+  /// insertScalarInstruction is a function-like object with signature
   /// (Variable *Dest, Variable *Src0, Variable *Src1) -> Instr *.
-  template <typename F>
-  void scalarizeInstruction(Variable *Dest, Operand *Src0, Operand *Src1,
-                            F &&MakeInstruction) {
+  void scalarizeInstruction(
+      Variable *Dest,
+      std::function<Inst *(Variable *, Variable *)> insertScalarInstruction,
+      Operand *Src0) {
+    assert(GeneratingTargetHelpers &&
+           "scalarizeInstruction called during incorrect phase");
     const Type DestTy = Dest->getType();
     assert(isVectorType(DestTy));
     const Type DestElementTy = typeElementType(DestTy);
     const SizeT NumElements = typeNumElements(DestTy);
-    const Type Src0ElementTy = typeElementType(Src0->getType());
-    const Type Src1ElementTy = typeElementType(Src1->getType());
-
-    assert(NumElements == typeNumElements(Src0->getType()));
-    assert(NumElements == typeNumElements(Src1->getType()));
 
     Variable *T = Func->makeVariable(DestTy);
     Context.insert<InstFakeDef>(T);
+
     for (SizeT I = 0; I < NumElements; ++I) {
-      Constant *Index = Ctx->getConstantInt32(I);
+      auto *Index = Ctx->getConstantInt32(I);
 
-      // Extract the next two inputs.
-      Variable *Op0 = Func->makeVariable(Src0ElementTy);
-      Context.insert<InstExtractElement>(Op0, Src0, Index);
-      Variable *Op1 = Func->makeVariable(Src1ElementTy);
-      Context.insert<InstExtractElement>(Op1, Src1, Index);
+      auto *Op0 = makeExtract(Src0, Index);
 
       // Perform the operation as a scalar operation.
-      Variable *Res = Func->makeVariable(DestElementTy);
-      auto Arith = MakeInstruction(Res, Op0, Op1);
-      // We might have created an operation that needed a helper call.
+      auto *Res = Func->makeVariable(DestElementTy);
+      auto *Arith = insertScalarInstruction(Res, Op0);
       genTargetHelperCallFor(Arith);
 
-      // Insert the result into position.
       Variable *DestT = Func->makeVariable(DestTy);
       Context.insert<InstInsertElement>(DestT, T, Res, Index);
       T = DestT;
     }
     Context.insert<InstAssign>(Dest, T);
   }
 
-  template <typename F>
-  void scalarizeUnaryInstruction(Variable *Dest, Operand *Src0,
-                                 F &&MakeInstruction) {
+  void
+  scalarizeInstruction(Variable *Dest,
+                       std::function<Inst *(Variable *, Variable *, Variable *)>
+                           insertScalarInstruction,
+                       Operand *Src0, Operand *Src1) {
+    assert(GeneratingTargetHelpers &&
+           "scalarizeInstruction called during incorrect phase");
     const Type DestTy = Dest->getType();
     assert(isVectorType(DestTy));
     const Type DestElementTy = typeElementType(DestTy);
     const SizeT NumElements = typeNumElements(DestTy);
-    const Type Src0ElementTy = typeElementType(Src0->getType());
-
-    assert(NumElements == typeNumElements(Src0->getType()));
 
     Variable *T = Func->makeVariable(DestTy);
     Context.insert<InstFakeDef>(T);
+
     for (SizeT I = 0; I < NumElements; ++I) {
-      Constant *Index = Ctx->getConstantInt32(I);
+      auto *Index = Ctx->getConstantInt32(I);
 
-      // Extract the next two inputs.
-      Variable *Op0 = Func->makeVariable(Src0ElementTy);
-      Context.insert<InstExtractElement>(Op0, Src0, Index);
+      auto *Op0 = makeExtract(Src0, Index);
+      auto *Op1 = makeExtract(Src1, Index);
 
       // Perform the operation as a scalar operation.
-      Variable *Res = Func->makeVariable(DestElementTy);
-      auto Arith = MakeInstruction(Res, Op0);
-      // We might have created an operation that needed a helper call.
+      auto *Res = Func->makeVariable(DestElementTy);
+      auto *Arith = insertScalarInstruction(Res, Op0, Op1);
       genTargetHelperCallFor(Arith);
 
-      // Insert the result into position.
       Variable *DestT = Func->makeVariable(DestTy);
       Context.insert<InstInsertElement>(DestT, T, Res, Index);
       T = DestT;
+    }
+    Context.insert<InstAssign>(Dest, T);
+  }
+
+  void scalarizeInstruction(
+      Variable *Dest, std::function<Inst *(Variable *, Variable *, Variable *,
+                                           Variable *)> insertScalarInstruction,
+      Operand *Src0, Operand *Src1, Operand *Src2) {
+    assert(GeneratingTargetHelpers &&
+           "scalarizeInstruction called during incorrect phase");
+    const Type DestTy = Dest->getType();
+    assert(isVectorType(DestTy));
+    const Type DestElementTy = typeElementType(DestTy);
+    const SizeT NumElements = typeNumElements(DestTy);
+
+    Variable *T = Func->makeVariable(DestTy);
+    Context.insert<InstFakeDef>(T);
+
+    for (SizeT I = 0; I < NumElements; ++I) {
+      auto *Index = Ctx->getConstantInt32(I);
+
+      auto *Op0 = makeExtract(Src0, Index);
+      auto *Op1 = makeExtract(Src1, Index);
+      auto *Op2 = makeExtract(Src2, Index);
+
+      // Perform the operation as a scalar operation.
+      auto *Res = Func->makeVariable(DestElementTy);
+      auto *Arith = insertScalarInstruction(Res, Op0, Op1, Op2);
+      genTargetHelperCallFor(Arith);
+
+      Variable *DestT = Func->makeVariable(DestTy);
+      Context.insert<InstInsertElement>(DestT, T, Res, Index);
+      T = DestT;
     }
     Context.insert<InstAssign>(Dest, T);
   }
 
   /// SandboxType enumerates all possible sandboxing strategies that
   enum SandboxType {
     ST_None,
     ST_NaCl,
     ST_Nonsfi,
   };
@@ skipping 95 matching lines @@
   virtual void lower() {}
 
 protected:
   explicit TargetHeaderLowering(GlobalContext *Ctx) : Ctx(Ctx) {}
   GlobalContext *Ctx;
 };
 
 } // end of namespace Ice
 
 #endif // SUBZERO_SRC_ICETARGETLOWERING_H