Take indirect branch into account as well when folding.

We may not be able to rewrite indirect branch target, but we also want to take it into
account when folding, i.e. if it and all its successor's predecessors go to the same
destination, we can fold, i.e. no need to thread.

llvm-svn: 301816

llvm/lib/Transforms/Scalar/JumpThreading.cpp

@@ -1253,16 +1253,12 @@ bool JumpThreadingPass::ProcessThreadableEdges(Value *Cond, BasicBlock *BB,
   Constant *OnlyVal = nullptr;
   Constant *MultipleVal = (Constant *)(intptr_t)~0ULL;
 
+  unsigned PredWithKnownDest = 0;
   for (const auto &PredValue : PredValues) {
     BasicBlock *Pred = PredValue.second;
     if (!SeenPreds.insert(Pred).second)
       continue;  // Duplicate predecessor entry.
 
-    // If the predecessor ends with an indirect goto, we can't change its
-    // destination.
-    if (isa<IndirectBrInst>(Pred->getTerminator()))
-      continue;
-
     Constant *Val = PredValue.first;
 
     BasicBlock *DestBB;
@@ -1294,6 +1290,14 @@ bool JumpThreadingPass::ProcessThreadableEdges(Value *Cond, BasicBlock *BB,
         OnlyVal = MultipleVal;
     }
 
+    // We know where this predecessor is going.
+    ++PredWithKnownDest;
+
+    // If the predecessor ends with an indirect goto, we can't change its
+    // destination.
+    if (isa<IndirectBrInst>(Pred->getTerminator()))
+      continue;
+
     PredToDestList.push_back(std::make_pair(Pred, DestBB));
   }
 
@@ -1305,7 +1309,7 @@ bool JumpThreadingPass::ProcessThreadableEdges(Value *Cond, BasicBlock *BB,
   // not thread. By doing so, we do not need to duplicate the current block and
   // also miss potential opportunities in case we dont/cant duplicate.
   if (OnlyDest && OnlyDest != MultipleDestSentinel) {
-    if (PredToDestList.size() ==
+    if (PredWithKnownDest ==
         (size_t)std::distance(pred_begin(BB), pred_end(BB))) {
       bool SeenFirstBranchToOnlyDest = false;
       for (BasicBlock *SuccBB : successors(BB)) {

llvm/test/Transforms/JumpThreading/fold-not-thread.ll

@@ -189,4 +189,58 @@ L3:
   ret void
 }
 
+; Make sure we can fold this branch ... We will not be able to thread it as
+; L0 is too big to duplicate.
+; We do not attempt to rewrite the indirectbr target here, but we still take
+; its target after L0 into account and that enables us to fold.
+;
+; L2 is the unreachable block here.
+; 
+; CHECK-LABEL: @test_br_folding_not_threading_indirect_branch(
+; CHECK: L1:
+; CHECK: call i32 @f2()
+; CHECK: call void @f3()
+; CHECK-NEXT: ret void
+; CHECK-NOT: br
+; CHECK: L3:
+define void @test_br_folding_not_threading_indirect_branch(i1 %condx, i1 %cond) nounwind {
+entry:
+  br i1 %condx, label %X0, label %X1
+
+X0:
+  br i1 %cond, label %L0, label %L3
+
+X1:
+  br i1 %cond, label %XX1, label %L3
+
+XX1:
+  indirectbr i8* blockaddress(@test_br_folding_not_threading_indirect_branch, %L0), [label %L0]
+
+L0:
+  call i32 @f2()
+  call i32 @f2()
+  call i32 @f2()
+  call i32 @f2()
+  call i32 @f2()
+  call i32 @f2()
+  call i32 @f2()
+  call i32 @f2()
+  call i32 @f2()
+  call i32 @f2()
+  call i32 @f2()
+  call i32 @f2()
+  call i32 @f2()
+  br i1 %cond, label %L1, label %L2
 
+L1:
+  call void @f3()
+  ret void
+
+L2:
+  call void @f3()
+  ret void
+
+L3:
+  call void @f3()
+  ret void
+}