[SPARC] Use lzcnt to implement CTLZ when we have VIS3

Reviewers: s-barannikov, brad0, rorth

Reviewed By: s-barannikov

Pull Request: #135715

Author: koachan

llvm/lib/Target/Sparc/SparcISelLowering.cpp

@@ -27,6 +27,7 @@
 #include "llvm/CodeGen/MachineRegisterInfo.h"
 #include "llvm/CodeGen/SelectionDAG.h"
 #include "llvm/CodeGen/SelectionDAGNodes.h"
+#include "llvm/CodeGen/TargetLowering.h"
 #include "llvm/CodeGen/TargetLoweringObjectFileImpl.h"
 #include "llvm/IR/DerivedTypes.h"
 #include "llvm/IR/DiagnosticInfo.h"
@@ -1752,8 +1753,7 @@ SparcTargetLowering::SparcTargetLowering(const TargetMachine &TM,
 
     setOperationAction(ISD::CTPOP, MVT::i64,
                        Subtarget->usePopc() ? Legal : Expand);
-    setOperationAction(ISD::CTTZ , MVT::i64, Expand);
-    setOperationAction(ISD::CTLZ , MVT::i64, Expand);
+    setOperationAction(ISD::CTTZ, MVT::i64, Expand);
     setOperationAction(ISD::BSWAP, MVT::i64, Expand);
     setOperationAction(ISD::ROTL , MVT::i64, Expand);
     setOperationAction(ISD::ROTR , MVT::i64, Expand);
@@ -1814,8 +1814,7 @@ SparcTargetLowering::SparcTargetLowering(const TargetMachine &TM,
   setOperationAction(ISD::FSINCOS, MVT::f32, Expand);
   setOperationAction(ISD::FREM , MVT::f32, Expand);
   setOperationAction(ISD::FMA  , MVT::f32, Expand);
-  setOperationAction(ISD::CTTZ , MVT::i32, Expand);
-  setOperationAction(ISD::CTLZ , MVT::i32, Expand);
+  setOperationAction(ISD::CTTZ, MVT::i32, Expand);
   setOperationAction(ISD::ROTL , MVT::i32, Expand);
   setOperationAction(ISD::ROTR , MVT::i32, Expand);
   setOperationAction(ISD::BSWAP, MVT::i32, Expand);
@@ -1986,6 +1985,24 @@ SparcTargetLowering::SparcTargetLowering(const TargetMachine &TM,
   if (Subtarget->hasLeonCycleCounter())
     setOperationAction(ISD::READCYCLECOUNTER, MVT::i64, Custom);
 
+  if (Subtarget->isVIS3()) {
+    setOperationAction(ISD::CTLZ, MVT::i32, Legal);
+    setOperationAction(ISD::CTLZ, MVT::i64, Legal);
+    setOperationAction(ISD::CTLZ_ZERO_UNDEF, MVT::i32, Legal);
+    setOperationAction(ISD::CTLZ_ZERO_UNDEF, MVT::i64, Legal);
+  } else if (Subtarget->usePopc()) {
+    setOperationAction(ISD::CTLZ, MVT::i32, Expand);
+    setOperationAction(ISD::CTLZ, MVT::i64, Expand);
+    setOperationAction(ISD::CTLZ_ZERO_UNDEF, MVT::i32, Expand);
+    setOperationAction(ISD::CTLZ_ZERO_UNDEF, MVT::i64, Expand);
+  } else {
+    setOperationAction(ISD::CTLZ, MVT::i32, Expand);
+    setOperationAction(ISD::CTLZ, MVT::i64, Expand);
+    setOperationAction(ISD::CTLZ_ZERO_UNDEF, MVT::i32,
+                       Subtarget->is64Bit() ? Promote : LibCall);
+    setOperationAction(ISD::CTLZ_ZERO_UNDEF, MVT::i64, LibCall);
+  }
+
   setOperationAction(ISD::INTRINSIC_WO_CHAIN, MVT::Other, Custom);
 
   setMinFunctionAlignment(Align(4));
@@ -3571,6 +3588,8 @@ bool SparcTargetLowering::isFPImmLegal(const APFloat &Imm, EVT VT,
          Imm.isZero();
 }
 
+bool SparcTargetLowering::isCtlzFast() const { return Subtarget->isVIS3(); }
+
 // Override to disable global variable loading on Linux.
 void SparcTargetLowering::insertSSPDeclarations(Module &M) const {
   if (!Subtarget->isTargetLinux())

llvm/lib/Target/Sparc/SparcISelLowering.h

@@ -210,6 +210,12 @@ namespace llvm {
     bool isFPImmLegal(const APFloat &Imm, EVT VT,
                       bool ForCodeSize) const override;
 
+    bool isCtlzFast() const override;
+
+    bool isCheapToSpeculateCtlz(Type *Ty) const override {
+      return isCtlzFast();
+    }
+
     bool shouldInsertFencesForAtomic(const Instruction *I) const override {
       // FIXME: We insert fences for each atomics and generate
       // sub-optimal code for PSO/TSO. (Approximately nobody uses any

llvm/lib/Target/Sparc/SparcInstrVIS.td

@@ -294,4 +294,14 @@ def : Pat<(f32 fpnegimm0), (FNEGS (FZEROS))>;
 // VIS3 instruction patterns.
 let Predicates = [HasVIS3] in {
 def : Pat<(i64 (adde i64:$lhs, i64:$rhs)), (ADDXCCC $lhs, $rhs)>;
+
+def : Pat<(i64 (ctlz i64:$src)), (LZCNT $src)>;
+def : Pat<(i64 (ctlz_zero_undef i64:$src)), (LZCNT $src)>;
+// 32-bit LZCNT.
+// The zero extension will leave us with 32 extra leading zeros,
+// so we need to compensate for it.
+// FIXME remove this when the codegen supports using 64-bit values directly
+// in V8+ mode.
+def : Pat<(i32 (ctlz i32:$src)), (ADDri (LZCNT (SRLri $src, 0)), (i32 -32))>;
+def : Pat<(i32 (ctlz_zero_undef i32:$src)), (ADDri (LZCNT (SRLri $src, 0)), (i32 -32))>;
 } // Predicates = [HasVIS3]