[IR] Reformulate LLVM's EH funclet IR

While we have successfully implemented a funclet-oriented EH scheme on
top of LLVM IR, our scheme has some notable deficiencies:
- catchendpad and cleanupendpad are necessary in the current design
  but they are difficult to explain to others, even to seasoned LLVM
  experts.
- catchendpad and cleanupendpad are optimization barriers.  They cannot
  be split and force all potentially throwing call-sites to be invokes.
  This has a noticable effect on the quality of our code generation.
- catchpad, while similar in some aspects to invoke, is fairly awkward.
  It is unsplittable, starts a funclet, and has control flow to other
  funclets.
- The nesting relationship between funclets is currently a property of
  control flow edges.  Because of this, we are forced to carefully
  analyze the flow graph to see if there might potentially exist illegal
  nesting among funclets.  While we have logic to clone funclets when
  they are illegally nested, it would be nicer if we had a
  representation which forbade them upfront.

Let's clean this up a bit by doing the following:
- Instead, make catchpad more like cleanuppad and landingpad: no control
  flow, just a bunch of simple operands;  catchpad would be splittable.
- Introduce catchswitch, a control flow instruction designed to model
  the constraints of funclet oriented EH.
- Make funclet scoping explicit by having funclet instructions consume
  the token produced by the funclet which contains them.
- Remove catchendpad and cleanupendpad.  Their presence can be inferred
  implicitly using coloring information.

N.B.  The state numbering code for the CLR has been updated but the
veracity of it's output cannot be spoken for.  An expert should take a
look to make sure the results are reasonable.

Reviewers: rnk, JosephTremoulet, andrew.w.kaylor

Differential Revision: http://reviews.llvm.org/D15139

llvm-svn: 255422

1 files changed, 108 insertions, 34 deletions

diff --git a/llvm/lib/Transforms/Utils/InlineFunction.cpp b/llvm/lib/Transforms/Utils/InlineFunction.cpp
index cafd1818fed..74ece385581 100644
--- a/llvm/lib/Transforms/Utils/InlineFunction.cpp
+++ b/llvm/lib/Transforms/Utils/InlineFunction.cpp
@@ -21,6 +21,7 @@
 #include "llvm/Analysis/AssumptionCache.h"
 #include "llvm/Analysis/CallGraph.h"
 #include "llvm/Analysis/CaptureTracking.h"
+#include "llvm/Analysis/EHPersonalities.h"
 #include "llvm/Analysis/InstructionSimplify.h"
 #include "llvm/Analysis/ValueTracking.h"
 #include "llvm/IR/Attributes.h"
@@ -192,8 +193,6 @@ HandleCallsInBlockInlinedThroughInvoke(BasicBlock *BB, BasicBlock *UnwindEdge) {
     // instructions require no special handling.
     CallInst *CI = dyn_cast<CallInst>(I);
 
-    // If this call cannot unwind, don't convert it to an invoke.
-    // Inline asm calls cannot throw.
     if (!CI || CI->doesNotThrow() || isa<InlineAsm>(CI->getCalledValue()))
       continue;
 
@@ -327,40 +326,10 @@ static void HandleInlinedEHPad(InvokeInst *II, BasicBlock *FirstNewBlock,
     }
   };
 
-  // Forward EH terminator instructions to the caller's invoke destination.
-  // This is as simple as connect all the instructions which 'unwind to caller'
-  // to the invoke destination.
+  // This connects all the instructions which 'unwind to caller' to the invoke
+  // destination.
   for (Function::iterator BB = FirstNewBlock->getIterator(), E = Caller->end();
        BB != E; ++BB) {
-    Instruction *I = BB->getFirstNonPHI();
-    if (I->isEHPad()) {
-      if (auto *CEPI = dyn_cast<CatchEndPadInst>(I)) {
-        if (CEPI->unwindsToCaller()) {
-          CatchEndPadInst::Create(CEPI->getContext(), UnwindDest, CEPI);
-          CEPI->eraseFromParent();
-          UpdatePHINodes(&*BB);
-        }
-      } else if (auto *CEPI = dyn_cast<CleanupEndPadInst>(I)) {
-        if (CEPI->unwindsToCaller()) {
-          CleanupEndPadInst::Create(CEPI->getCleanupPad(), UnwindDest, CEPI);
-          CEPI->eraseFromParent();
-          UpdatePHINodes(&*BB);
-        }
-      } else if (auto *TPI = dyn_cast<TerminatePadInst>(I)) {
-        if (TPI->unwindsToCaller()) {
-          SmallVector<Value *, 3> TerminatePadArgs;
-          for (Value *ArgOperand : TPI->arg_operands())
-            TerminatePadArgs.push_back(ArgOperand);
-          TerminatePadInst::Create(TPI->getContext(), UnwindDest,
-                                   TerminatePadArgs, TPI);
-          TPI->eraseFromParent();
-          UpdatePHINodes(&*BB);
-        }
-      } else {
-        assert(isa<CatchPadInst>(I) || isa<CleanupPadInst>(I));
-      }
-    }
-
     if (auto *CRI = dyn_cast<CleanupReturnInst>(BB->getTerminator())) {
       if (CRI->unwindsToCaller()) {
         CleanupReturnInst::Create(CRI->getCleanupPad(), UnwindDest, CRI);
@@ -368,6 +337,40 @@ static void HandleInlinedEHPad(InvokeInst *II, BasicBlock *FirstNewBlock,
         UpdatePHINodes(&*BB);
       }
     }
+
+    Instruction *I = BB->getFirstNonPHI();
+    if (!I->isEHPad())
+      continue;
+
+    Instruction *Replacement = nullptr;
+    if (auto *TPI = dyn_cast<TerminatePadInst>(I)) {
+      if (TPI->unwindsToCaller()) {
+        SmallVector<Value *, 3> TerminatePadArgs;
+        for (Value *ArgOperand : TPI->arg_operands())
+          TerminatePadArgs.push_back(ArgOperand);
+        Replacement = TerminatePadInst::Create(TPI->getParentPad(), UnwindDest,
+                                               TerminatePadArgs, TPI);
+      }
+    } else if (auto *CatchSwitch = dyn_cast<CatchSwitchInst>(I)) {
+      if (CatchSwitch->unwindsToCaller()) {
+        auto *NewCatchSwitch = CatchSwitchInst::Create(
+            CatchSwitch->getParentPad(), UnwindDest,
+            CatchSwitch->getNumHandlers(), CatchSwitch->getName(),
+            CatchSwitch);
+        for (BasicBlock *PadBB : CatchSwitch->handlers())
+          NewCatchSwitch->addHandler(PadBB);
+        Replacement = NewCatchSwitch;
+      }
+    } else if (!isa<FuncletPadInst>(I)) {
+      llvm_unreachable("unexpected EHPad!");
+    }
+
+    if (Replacement) {
+      Replacement->takeName(I);
+      I->replaceAllUsesWith(Replacement);
+      I->eraseFromParent();
+      UpdatePHINodes(&*BB);
+    }
   }
 
   if (InlinedCodeInfo.ContainsCalls)
@@ -1090,6 +1093,53 @@ bool llvm::InlineFunction(CallSite CS, InlineFunctionInfo &IFI,
       return false;
   }
 
+  // We need to figure out which funclet the callsite was in so that we may
+  // properly nest the callee.
+  Instruction *CallSiteEHPad = nullptr;
+  if (CalledPersonality && CallerPersonality) {
+    EHPersonality Personality = classifyEHPersonality(CalledPersonality);
+    if (isFuncletEHPersonality(Personality)) {
+      DenseMap<BasicBlock *, ColorVector> CallerBlockColors =
+          colorEHFunclets(*Caller);
+      ColorVector &CallSiteColors = CallerBlockColors[OrigBB];
+      size_t NumColors = CallSiteColors.size();
+      // There is no single parent, inlining will not succeed.
+      if (NumColors > 1)
+        return false;
+      if (NumColors == 1) {
+        BasicBlock *CallSiteFuncletBB = CallSiteColors.front();
+        if (CallSiteFuncletBB != Caller->begin()) {
+          CallSiteEHPad = CallSiteFuncletBB->getFirstNonPHI();
+          assert(CallSiteEHPad->isEHPad() && "Expected an EHPad!");
+        }
+      }
+
+      // OK, the inlining site is legal.  What about the target function?
+
+      if (CallSiteEHPad) {
+        if (Personality == EHPersonality::MSVC_CXX) {
+          // The MSVC personality cannot tolerate catches getting inlined into
+          // cleanup funclets.
+          if (isa<CleanupPadInst>(CallSiteEHPad)) {
+            // Ok, the call site is within a cleanuppad.  Let's check the callee
+            // for catchpads.
+            for (const BasicBlock &CalledBB : *CalledFunc) {
+              if (isa<CatchPadInst>(CalledBB.getFirstNonPHI()))
+                return false;
+            }
+          }
+        } else if (isAsynchronousEHPersonality(Personality)) {
+          // SEH is even less tolerant, there may not be any sort of exceptional
+          // funclet in the callee.
+          for (const BasicBlock &CalledBB : *CalledFunc) {
+            if (CalledBB.isEHPad())
+              return false;
+          }
+        }
+      }
+    }
+  }
+
   // Get an iterator to the last basic block in the function, which will have
   // the new function inlined after it.
   Function::iterator LastBlock = --Caller->end();
@@ -1381,6 +1431,30 @@ bool llvm::InlineFunction(CallSite CS, InlineFunctionInfo &IFI,
     }
   }
 
+  // Update the lexical scopes of the new funclets.  Anything that had 'none' as
+  // its parent is now nested inside the callsite's EHPad.
+  if (CallSiteEHPad) {
+    for (Function::iterator BB = FirstNewBlock->getIterator(),
+                            E = Caller->end();
+         BB != E; ++BB) {
+      Instruction *I = BB->getFirstNonPHI();
+      if (!I->isEHPad())
+        continue;
+
+      if (auto *TPI = dyn_cast<TerminatePadInst>(I)) {
+        if (isa<ConstantTokenNone>(TPI->getParentPad()))
+          TPI->setParentPad(CallSiteEHPad);
+      } else if (auto *CatchSwitch = dyn_cast<CatchSwitchInst>(I)) {
+        if (isa<ConstantTokenNone>(CatchSwitch->getParentPad()))
+          CatchSwitch->setParentPad(CallSiteEHPad);
+      } else {
+        auto *FPI = cast<FuncletPadInst>(I);
+        if (isa<ConstantTokenNone>(FPI->getParentPad()))
+          FPI->setParentPad(CallSiteEHPad);
+      }
+    }
+  }
+
   // If we are inlining for an invoke instruction, we must make sure to rewrite
   // any call instructions into invoke instructions.
   if (auto *II = dyn_cast<InvokeInst>(TheCall)) {