[llvm-exegesis] Add explicit support for setting DF in X86 (#115644)

While llvm-exegesis has explicit support for setting EFLAGS which
contains DF, it can be nice sometimes to explicitly set DF, especially
given that it is modeled as a separate register within LLVM. This patch
adds the ability to do that by lowering setting the value to 0 or 1 to
cld and std respectively.

Author: boomanaiden154

llvm/tools/llvm-exegesis/lib/X86/Target.cpp

@@ -537,6 +537,8 @@ struct ConstantInliner {
   std::vector<MCInst> loadImplicitRegAndFinalize(unsigned Opcode,
                                                  unsigned Value);
 
+  std::vector<MCInst> loadDirectionFlagAndFinalize();
+
 private:
   ConstantInliner &add(const MCInst &Inst) {
     Instructions.push_back(Inst);
@@ -612,6 +614,15 @@ ConstantInliner::loadImplicitRegAndFinalize(unsigned Opcode, unsigned Value) {
   return std::move(Instructions);
 }
 
+std::vector<MCInst> ConstantInliner::loadDirectionFlagAndFinalize() {
+  if (Constant_.isZero())
+    add(MCInstBuilder(X86::CLD));
+  else if (Constant_.isOne())
+    add(MCInstBuilder(X86::STD));
+
+  return std::move(Instructions);
+}
+
 void ConstantInliner::initStack(unsigned Bytes) {
   assert(Constant_.getBitWidth() <= Bytes * 8 &&
          "Value does not have the correct size");
@@ -1089,6 +1100,8 @@ std::vector<MCInst> ExegesisX86Target::setRegTo(const MCSubtargetInfo &STI,
         0x1f80);
   if (Reg == X86::FPCW)
     return CI.loadImplicitRegAndFinalize(X86::FLDCW16m, 0x37f);
+  if (Reg == X86::DF)
+    return CI.loadDirectionFlagAndFinalize();
   return {}; // Not yet implemented.
 }
 

llvm/unittests/tools/llvm-exegesis/X86/TargetTest.cpp

@@ -585,6 +585,14 @@ TEST_F(X86Core2TargetTest, SetRegToFP1_4Bits) {
                           OpcodeIs(X86::LD_Fp80m), IsStackDeallocate(10)));
 }
 
+TEST_F(X86Core2TargetTest, SetRegToDf1) {
+  EXPECT_THAT(setRegTo(X86::DF, APInt(1, 1)), ElementsAre(OpcodeIs(X86::STD)));
+}
+
+TEST_F(X86Core2TargetTest, SetRegToDf0) {
+  EXPECT_THAT(setRegTo(X86::DF, APInt(1, 0)), ElementsAre(OpcodeIs(X86::CLD)));
+}
+
 TEST_F(X86Core2Avx512TargetTest, FillMemoryOperands_ADD64rm) {
   const Instruction &I = getInstr(X86::ADD64rm);
   InstructionTemplate IT(&I);