[ShrinkWrap] Add optimization remarks to the shrink-wrapping pass

Start by emitting remarks for very basic unsupported cases such as
irreducible CFGs and EHFunclets. The end goal is to be able to cover all
the cases where we give up with an explanation.

llvm-svn: 333972

lib/CodeGen/ShrinkWrap.cpp

@@ -63,6 +63,7 @@
 #include "llvm/CodeGen/MachineInstr.h"
 #include "llvm/CodeGen/MachineLoopInfo.h"
 #include "llvm/CodeGen/MachineOperand.h"
+#include "llvm/CodeGen/MachineOptimizationRemarkEmitter.h"
 #include "llvm/CodeGen/MachinePostDominators.h"
 #include "llvm/CodeGen/RegisterClassInfo.h"
 #include "llvm/CodeGen/RegisterScavenging.h"
@@ -130,6 +131,9 @@ class ShrinkWrap : public MachineFunctionPass {
   /// are in the same loop.
   MachineLoopInfo *MLI;
 
+  // Emit remarks.
+  MachineOptimizationRemarkEmitter *ORE = nullptr;
+
   /// Frequency of the Entry block.
   uint64_t EntryFreq;
 
@@ -189,6 +193,7 @@ class ShrinkWrap : public MachineFunctionPass {
     Restore = nullptr;
     MBFI = &getAnalysis<MachineBlockFrequencyInfo>();
     MLI = &getAnalysis<MachineLoopInfo>();
+    ORE = &getAnalysis<MachineOptimizationRemarkEmitterPass>().getORE();
     EntryFreq = MBFI->getEntryFreq();
     const TargetSubtargetInfo &Subtarget = MF.getSubtarget();
     const TargetInstrInfo &TII = *Subtarget.getInstrInfo();
@@ -222,6 +227,7 @@ public:
     AU.addRequired<MachineDominatorTree>();
     AU.addRequired<MachinePostDominatorTree>();
     AU.addRequired<MachineLoopInfo>();
+    AU.addRequired<MachineOptimizationRemarkEmitterPass>();
     MachineFunctionPass::getAnalysisUsage(AU);
   }
 
@@ -248,6 +254,7 @@ INITIALIZE_PASS_DEPENDENCY(MachineBlockFrequencyInfo)
 INITIALIZE_PASS_DEPENDENCY(MachineDominatorTree)
 INITIALIZE_PASS_DEPENDENCY(MachinePostDominatorTree)
 INITIALIZE_PASS_DEPENDENCY(MachineLoopInfo)
+INITIALIZE_PASS_DEPENDENCY(MachineOptimizationRemarkEmitterPass)
 INITIALIZE_PASS_END(ShrinkWrap, DEBUG_TYPE, "Shrink Wrap Pass", false, false)
 
 bool ShrinkWrap::useOrDefCSROrFI(const MachineInstr &MI,
@@ -432,6 +439,19 @@ void ShrinkWrap::updateSaveRestorePoints(MachineBasicBlock &MBB,
   }
 }
 
+static bool giveUpWithRemarks(MachineOptimizationRemarkEmitter *ORE,
+                              StringRef RemarkName, StringRef RemarkMessage,
+                              const DiagnosticLocation &Loc,
+                              const MachineBasicBlock *MBB) {
+  ORE->emit([&]() {
+    return MachineOptimizationRemarkMissed(DEBUG_TYPE, RemarkName, Loc, MBB)
+           << RemarkMessage;
+  });
+
+  LLVM_DEBUG(dbgs() << RemarkMessage << '\n');
+  return false;
+}
+
 bool ShrinkWrap::runOnMachineFunction(MachineFunction &MF) {
   if (skipFunction(MF.getFunction()) || MF.empty() || !isShrinkWrapEnabled(MF))
     return false;
@@ -448,8 +468,9 @@ bool ShrinkWrap::runOnMachineFunction(MachineFunction &MF) {
     // results. Moreover, we may miss that the prologue and
     // epilogue are not in the same loop, leading to unbalanced
     // construction/deconstruction of the stack frame.
-    LLVM_DEBUG(dbgs() << "Irreducible CFGs are not supported yet\n");
-    return false;
+    return giveUpWithRemarks(ORE, "UnsupportedIrreducibleCFG",
+                             "Irreducible CFGs are not supported yet.",
+                             MF.getFunction().getSubprogram(), &MF.front());
   }
 
   const TargetRegisterInfo *TRI = MF.getSubtarget().getRegisterInfo();
@@ -460,10 +481,10 @@ bool ShrinkWrap::runOnMachineFunction(MachineFunction &MF) {
     LLVM_DEBUG(dbgs() << "Look into: " << MBB.getNumber() << ' '
                       << MBB.getName() << '\n');
 
-    if (MBB.isEHFuncletEntry()) {
-      LLVM_DEBUG(dbgs() << "EH Funclets are not supported yet.\n");
-      return false;
-    }
+    if (MBB.isEHFuncletEntry())
+      return giveUpWithRemarks(ORE, "UnsupportedEHFunclets",
+                               "EH Funclets are not supported yet.",
+                               MBB.front().getDebugLoc(), &MBB);
 
     if (MBB.isEHPad()) {
       // Push the prologue and epilogue outside of

test/CodeGen/AArch64/O3-pipeline.ll

@@ -129,9 +129,9 @@
 ; CHECK-NEXT:       Machine Natural Loop Construction
 ; CHECK-NEXT:       Machine Block Frequency Analysis
 ; CHECK-NEXT:       MachinePostDominator Tree Construction
-; CHECK-NEXT:       Shrink Wrapping analysis
 ; CHECK-NEXT:       Lazy Machine Block Frequency Analysis
 ; CHECK-NEXT:       Machine Optimization Remark Emitter
+; CHECK-NEXT:       Shrink Wrapping analysis
 ; CHECK-NEXT:       Prologue/Epilogue Insertion & Frame Finalization
 ; CHECK-NEXT:       Control Flow Optimizer
 ; CHECK-NEXT:       Tail Duplication

test/CodeGen/X86/O3-pipeline.ll

@@ -125,9 +125,9 @@
 ; CHECK-NEXT:       PostRA Machine Sink
 ; CHECK-NEXT:       Machine Block Frequency Analysis
 ; CHECK-NEXT:       MachinePostDominator Tree Construction
-; CHECK-NEXT:       Shrink Wrapping analysis
 ; CHECK-NEXT:       Lazy Machine Block Frequency Analysis
 ; CHECK-NEXT:       Machine Optimization Remark Emitter
+; CHECK-NEXT:       Shrink Wrapping analysis
 ; CHECK-NEXT:       Prologue/Epilogue Insertion & Frame Finalization
 ; CHECK-NEXT:       Control Flow Optimizer
 ; CHECK-NEXT:       Tail Duplication

test/CodeGen/X86/late-address-taken.ll

@@ -1,6 +1,7 @@
 ; RUN: llc -mtriple=x86_64-pc-windows-msvc < %s -enable-shrink-wrap=false | FileCheck %s
 ; Make sure shrink-wrapping does not break the lowering of exception handling.
-; RUN: llc -mtriple=x86_64-pc-windows-msvc < %s -enable-shrink-wrap=true | FileCheck %s
+; RUN: llc -mtriple=x86_64-pc-windows-msvc < %s -enable-shrink-wrap=true -pass-remarks-output=%t | FileCheck %s
+; RUN: cat %t | FileCheck %s --check-prefix=REMARKS
 
 ; Repro cases from PR25168
 
@@ -37,6 +38,11 @@ exit:
 ; CHECK: leaq [[Exit]](%rip), %rax
 ; CHECK: retq # CATCHRET
 
+; REMARKS: Pass:            shrink-wrap
+; REMARKS-NEXT: Name:            UnsupportedEHFunclets
+; REMARKS-NEXT: Function:        catchret
+; REMARKS-NEXT: Args:
+; REMARKS-NEXT:   - String:          EH Funclets are not supported yet.
 
 ; test @setjmp - similar to @catchret, but the MBB in question
 ; is the one generated when the setjmp's block is split

test/CodeGen/X86/x86-shrink-wrapping.ll

@@ -1,4 +1,5 @@
-; RUN: llc %s -o - -enable-shrink-wrap=true | FileCheck %s --check-prefix=CHECK --check-prefix=ENABLE
+; RUN: llc %s -o - -enable-shrink-wrap=true -pass-remarks-output=%t | FileCheck %s --check-prefix=CHECK --check-prefix=ENABLE
+; RUN: cat %t | FileCheck %s --check-prefix=REMARKS
 ; RUN: llc %s -o - -enable-shrink-wrap=false | FileCheck %s --check-prefix=CHECK --check-prefix=DISABLE
 ;
 ; Note: Lots of tests use inline asm instead of regular calls.
@@ -940,6 +941,13 @@ attributes #3 = { nounwind }
 ; CHECK: popq
 ; CHECK-NEXT: popq
 ; CHECK-NEXT: retq
+; Make sure we emit missed optimization remarks for this.
+; REMARKS: Pass:            shrink-wrap
+; REMARKS-NEXT: Name:            UnsupportedIrreducibleCFG
+; REMARKS-NEXT: Function:        irreducibleCFG
+; REMARKS-NEXT: Args:
+; REMARKS-NEXT:   - String:          Irreducible CFGs are not supported yet
+
 define i32 @irreducibleCFG() #4 {
 entry:
   %i0 = load i32, i32* @irreducibleCFGa, align 4