diff --git a/lib/CodeGen/TailDuplication.cpp b/lib/CodeGen/TailDuplication.cpp
index ce4b1be8541..15aed3436c7 100644
--- a/lib/CodeGen/TailDuplication.cpp
+++ b/lib/CodeGen/TailDuplication.cpp
@@ -465,9 +465,12 @@ TailDuplicatePass::TailDuplicate(MachineBasicBlock *TailBB, MachineFunction &MF,
     MaxDuplicateCount = TailDuplicateSize;
 
   if (PreRegAlloc) {
-      // Pre-regalloc tail duplication hurts compile time and doesn't help
-      // much except for indirect branches.
-    if (TailBB->empty() || !TailBB->back().getDesc().isIndirectBranch())
+    if (TailBB->empty())
+      return false;
+    const TargetInstrDesc &TID = TailBB->back().getDesc();
+    // Pre-regalloc tail duplication hurts compile time and doesn't help
+    // much except for indirect branches and returns.
+    if (!TID.isIndirectBranch() && !TID.isReturn())
       return false;
     // If the target has hardware branch prediction that can handle indirect
     // branches, duplicating them can often make them predictable when there
@@ -502,7 +505,7 @@ TailDuplicatePass::TailDuplicate(MachineBasicBlock *TailBB, MachineFunction &MF,
   }
   // Heuristically, don't tail-duplicate calls if it would expand code size,
   // as it's less likely to be worth the extra cost.
-  if (InstrCount > 1 && HasCall)
+  if (InstrCount > 1 && (PreRegAlloc && HasCall))
     return false;
 
   DEBUG(dbgs() << "\n*** Tail-duplicating BB#" << TailBB->getNumber() << '\n');
diff --git a/lib/Transforms/Utils/SimplifyCFG.cpp b/lib/Transforms/Utils/SimplifyCFG.cpp
index f6d7d76dbf6..37e6d28d7b0 100644
--- a/lib/Transforms/Utils/SimplifyCFG.cpp
+++ b/lib/Transforms/Utils/SimplifyCFG.cpp
@@ -28,6 +28,7 @@
 #include "llvm/ADT/Statistic.h"
 #include "llvm/ADT/STLExtras.h"
 #include "llvm/Support/CFG.h"
+#include "llvm/Support/CommandLine.h"
 #include "llvm/Support/ConstantRange.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/raw_ostream.h"
@@ -36,6 +37,10 @@
 #include <map>
 using namespace llvm;
 
+static cl::opt<bool>
+DupRet("simplifycfg-dup-ret", cl::Hidden, cl::init(false),
+       cl::desc("Duplicate return instructions into unconditional branches"));
+
 STATISTIC(NumSpeculations, "Number of speculative executed instructions");
 
 namespace {
@@ -2027,7 +2032,7 @@ bool SimplifyCFGOpt::SimplifyReturn(ReturnInst *RI) {
   }
   
   // If we found some, do the transformation!
-  if (!UncondBranchPreds.empty()) {
+  if (!UncondBranchPreds.empty() && DupRet) {
     while (!UncondBranchPreds.empty()) {
       BasicBlock *Pred = UncondBranchPreds.pop_back_val();
       DEBUG(dbgs() << "FOLDING: " << *BB
diff --git a/test/CodeGen/X86/critical-edge-split.ll b/test/CodeGen/X86/critical-edge-split.ll
deleted file mode 100644
index 96fef0fbfc6..00000000000
--- a/test/CodeGen/X86/critical-edge-split.ll
+++ /dev/null
@@ -1,50 +0,0 @@
-; RUN: llc < %s -mtriple=i386-apple-darwin -o /dev/null -stats -info-output-file - | grep asm-printer | grep 29
-
-	%CC = type { %Register }
-	%II = type { %"struct.XX::II::$_74" }
-	%JITFunction = type %YYValue* (%CC*, %YYValue**)
-	%YYValue = type { i32 (...)** }
-	%Register = type { %"struct.XX::ByteCodeFeatures" }
-	%"struct.XX::ByteCodeFeatures" = type { i32 }
-	%"struct.XX::II::$_74" = type { i8* }
-@llvm.used = appending global [1 x i8*] [ i8* bitcast (%JITFunction* @loop to i8*) ], section "llvm.metadata"		; <[1 x i8*]*> [#uses=0]
-
-define %YYValue* @loop(%CC*, %YYValue**) nounwind {
-; <label>:2
-	%3 = getelementptr %CC* %0, i32 -9		; <%CC*> [#uses=1]
-	%4 = bitcast %CC* %3 to %YYValue**		; <%YYValue**> [#uses=2]
-	%5 = load %YYValue** %4		; <%YYValue*> [#uses=3]
-	%unique_1.i = ptrtoint %YYValue* %5 to i1		; <i1> [#uses=1]
-	br i1 %unique_1.i, label %loop, label %11
-
-loop:		; preds = %6, %2
-	%.1 = phi %YYValue* [ inttoptr (i32 1 to %YYValue*), %2 ], [ %intAddValue, %6 ]		; <%YYValue*> [#uses=3]
-	%immediateCmp = icmp slt %YYValue* %.1, %5		; <i1> [#uses=1]
-	br i1 %immediateCmp, label %6, label %8
-
-; <label>:6		; preds = %loop
-	%lhsInt = ptrtoint %YYValue* %.1 to i32		; <i32> [#uses=1]
-	%7 = call { i32, i1 } @llvm.sadd.with.overflow.i32(i32 %lhsInt, i32 2)		; <{ i32, i1 }> [#uses=2]
-	%intAdd = extractvalue { i32, i1 } %7, 0		; <i32> [#uses=1]
-	%intAddValue = inttoptr i32 %intAdd to %YYValue*		; <%YYValue*> [#uses=1]
-	%intAddOverflow = extractvalue { i32, i1 } %7, 1		; <i1> [#uses=1]
-	br i1 %intAddOverflow, label %.loopexit, label %loop
-
-; <label>:8		; preds = %loop
-	ret %YYValue* inttoptr (i32 10 to %YYValue*)
-
-.loopexit:		; preds = %6
-	%9 = bitcast %CC* %0 to %YYValue**		; <%YYValue**> [#uses=1]
-	store %YYValue* %.1, %YYValue** %9
-	store %YYValue* %5, %YYValue** %4
-	%10 = call fastcc %YYValue* @foobar(%II* inttoptr (i32 3431104 to %II*), %CC* %0, %YYValue** %1)		; <%YYValue*> [#uses=1]
-	ret %YYValue* %10
-
-; <label>:11		; preds = %2
-	%12 = call fastcc %YYValue* @foobar(%II* inttoptr (i32 3431080 to %II*), %CC* %0, %YYValue** %1)		; <%YYValue*> [#uses=1]
-	ret %YYValue* %12
-}
-
-declare fastcc %YYValue* @foobar(%II*, %CC*, %YYValue**) nounwind
-
-declare { i32, i1 } @llvm.sadd.with.overflow.i32(i32, i32) nounwind
diff --git a/test/CodeGen/X86/loop-blocks.ll b/test/CodeGen/X86/loop-blocks.ll
index 354d0820697..faba6300712 100644
--- a/test/CodeGen/X86/loop-blocks.ll
+++ b/test/CodeGen/X86/loop-blocks.ll
@@ -70,6 +70,7 @@ exit:
 
 ; Same as slightly_more_involved, but block_a is now a CFG diamond with
 ; fallthrough edges which should be preserved.
+; "callq block_a_merge_func" is tail duped.
 
 ; CHECK: yet_more_involved:
 ;      CHECK:   jmp .LBB2_1
@@ -78,12 +79,12 @@ exit:
 ; CHECK-NEXT:   callq bar99
 ; CHECK-NEXT:   callq get
 ; CHECK-NEXT:   cmpl $2999, %eax
-; CHECK-NEXT:   jg .LBB2_6
-; CHECK-NEXT:   callq block_a_true_func
-; CHECK-NEXT:   jmp .LBB2_7
-; CHECK-NEXT: .LBB2_6:
+; CHECK-NEXT:   jle .LBB2_5
 ; CHECK-NEXT:   callq block_a_false_func
-; CHECK-NEXT: .LBB2_7:
+; CHECK-NEXT:   callq block_a_merge_func
+; CHECK-NEXT:   jmp .LBB2_1
+; CHECK-NEXT: .LBB2_5:
+; CHECK-NEXT:   callq block_a_true_func
 ; CHECK-NEXT:   callq block_a_merge_func
 ; CHECK-NEXT: .LBB2_1:
 ; CHECK-NEXT:   callq body
diff --git a/test/Transforms/JumpThreading/and-and-cond.ll b/test/Transforms/JumpThreading/and-and-cond.ll
index e6db9ee5a32..765d940cc7c 100644
--- a/test/Transforms/JumpThreading/and-and-cond.ll
+++ b/test/Transforms/JumpThreading/and-and-cond.ll
@@ -1,14 +1,14 @@
-; RUN: opt < %s -jump-threading -mem2reg -instcombine -simplifycfg  -S | grep {ret i32 %v1}
-; There should be no uncond branches left.
-; RUN: opt < %s -jump-threading -mem2reg -instcombine -simplifycfg  -S | not grep {br label}
+; RUN: opt < %s -jump-threading -mem2reg -instcombine -simplifycfg  -S | FileCheck %s
 
 declare i32 @f1()
 declare i32 @f2()
 declare void @f3()
 
 define i32 @test(i1 %cond, i1 %cond2, i1 %cond3) {
+; CHECK: test
 	br i1 %cond, label %T1, label %F1
 
+; CHECK-NOT: T1:
 T1:
 	%v1 = call i32 @f1()
 	br label %Merge
@@ -18,6 +18,10 @@ F1:
 	br label %Merge
 
 Merge:
+; CHECK: Merge:
+; CHECK: %v1 = call i32 @f1()
+; CHECK-NEXT: %D = and i1 %cond2, %cond3
+; CHECK-NEXT: br i1 %D
 	%A = phi i1 [true, %T1], [false, %F1]
 	%B = phi i32 [%v1, %T1], [%v2, %F1]
 	%C = and i1 %A, %cond2
diff --git a/test/Transforms/JumpThreading/and-cond.ll b/test/Transforms/JumpThreading/and-cond.ll
index 58dbec72a76..0159bb3bb76 100644
--- a/test/Transforms/JumpThreading/and-cond.ll
+++ b/test/Transforms/JumpThreading/and-cond.ll
@@ -1,14 +1,14 @@
-; RUN: opt < %s -jump-threading -mem2reg -instcombine -simplifycfg  -S | grep {ret i32 %v1}
-; There should be no uncond branches left.
-; RUN: opt < %s -jump-threading -mem2reg -instcombine -simplifycfg  -S | not grep {br label}
+; RUN: opt < %s -jump-threading -mem2reg -instcombine -simplifycfg  -S | FileCheck %s
 
 declare i32 @f1()
 declare i32 @f2()
 declare void @f3()
 
 define i32 @test(i1 %cond, i1 %cond2) {
+; CHECK: test
 	br i1 %cond, label %T1, label %F1
 
+; CHECK-NOT: T1
 T1:
 	%v1 = call i32 @f1()
 	br label %Merge
@@ -18,6 +18,9 @@ F1:
 	br label %Merge
 
 Merge:
+; CHECK: Merge:
+; CHECK: %v1 = call i32 @f1()
+; CHECK-NEXT: br i1 %cond2
 	%A = phi i1 [true, %T1], [false, %F1]
 	%B = phi i32 [%v1, %T1], [%v2, %F1]
 	%C = and i1 %A, %cond2
diff --git a/test/Transforms/JumpThreading/thread-loads.ll b/test/Transforms/JumpThreading/thread-loads.ll
index 96ba701046d..cce23ea319c 100644
--- a/test/Transforms/JumpThreading/thread-loads.ll
+++ b/test/Transforms/JumpThreading/thread-loads.ll
@@ -1,4 +1,4 @@
-; RUN: opt < %s -jump-threading -simplifycfg -S | grep {ret i32 1}
+; RUN: opt < %s -jump-threading -S | FileCheck %s
 ; rdar://6402033
 
 ; Test that we can thread through the block with the partially redundant load (%2).
@@ -6,12 +6,16 @@ target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f3
 target triple = "i386-apple-darwin7"
 
 define i32 @foo(i32* %P) nounwind {
+; CHECK: foo
 entry:
 	%0 = tail call i32 (...)* @f1() nounwind		; <i32> [#uses=1]
 	%1 = icmp eq i32 %0, 0		; <i1> [#uses=1]
 	br i1 %1, label %bb1, label %bb
 
 bb:		; preds = %entry
+; CHECK: bb1.thread:
+; CHECK: store
+; CHECK: br label %bb3
 	store i32 42, i32* %P, align 4
 	br label %bb1
 
@@ -26,6 +30,9 @@ bb2:		; preds = %bb1
 	ret i32 %res.0
 
 bb3:		; preds = %bb1
+; CHECK: bb3:
+; CHECK: %res.01 = phi i32 [ 1, %bb1.thread ], [ 0, %bb1 ]
+; CHECK: ret i32 %res.01
 	ret i32 %res.0
 }
 
diff --git a/test/Transforms/SimplifyCFG/MagicPointer.ll b/test/Transforms/SimplifyCFG/MagicPointer.ll
index 54e5b14880c..93b9a276eac 100644
--- a/test/Transforms/SimplifyCFG/MagicPointer.ll
+++ b/test/Transforms/SimplifyCFG/MagicPointer.ll
@@ -8,7 +8,6 @@
 ; CHECK: i64 2, label
 ; CHECK: i64 3, label
 ; CHECK: i64 4, label
-; CHECK-NOT: br
 ; CHECK: }
 
 target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64"
diff --git a/test/Transforms/SimplifyCFG/basictest.ll b/test/Transforms/SimplifyCFG/basictest.ll
index b485f6ab05f..052e10667da 100644
--- a/test/Transforms/SimplifyCFG/basictest.ll
+++ b/test/Transforms/SimplifyCFG/basictest.ll
@@ -25,16 +25,6 @@ define void @test3(i1 %T) {
 }
 
 
-define void @test4() {
-  br label %return
-return:
-  ret void
-; CHECK: @test4
-; CHECK-NEXT: ret void
-}
-@test4g = global i8* blockaddress(@test4, %return)
-
-
 ; PR5795
 define void @test5(i32 %A) {
   switch i32 %A, label %return [
diff --git a/test/Transforms/SimplifyCFG/switch_create.ll b/test/Transforms/SimplifyCFG/switch_create.ll
index da7f65a6ca3..4e199bc8596 100644
--- a/test/Transforms/SimplifyCFG/switch_create.ll
+++ b/test/Transforms/SimplifyCFG/switch_create.ll
@@ -147,7 +147,7 @@ UnifiedReturnBlock:             ; preds = %shortcirc_done.4, %shortcirc_next.4
 ; CHECK:       i32 16, label %UnifiedReturnBlock
 ; CHECK:       i32 17, label %UnifiedReturnBlock
 ; CHECK:       i32 18, label %UnifiedReturnBlock
-; CHECK:       i32 19, label %switch.edge
+; CHECK:       i32 19, label %UnifiedReturnBlock
 ; CHECK:     ]
 }
 
@@ -441,3 +441,29 @@ if.end:
 ; CHECK-NOT: switch
 ; CHECK: ret void
 }
+
+; PR8675
+; rdar://5134905
+define zeroext i1 @test16(i32 %x) nounwind {
+entry:
+; CHECK: @test16
+; CHECK: switch i32 %x, label %lor.rhs [
+; CHECK:   i32 1, label %lor.end
+; CHECK:   i32 2, label %lor.end
+; CHECK:   i32 3, label %lor.end
+; CHECK: ]
+  %cmp.i = icmp eq i32 %x, 1
+  br i1 %cmp.i, label %lor.end, label %lor.lhs.false
+
+lor.lhs.false:
+  %cmp.i2 = icmp eq i32 %x, 2
+  br i1 %cmp.i2, label %lor.end, label %lor.rhs
+
+lor.rhs:
+  %cmp.i1 = icmp eq i32 %x, 3
+  br label %lor.end
+
+lor.end:
+  %0 = phi i1 [ true, %lor.lhs.false ], [ true, %entry ], [ %cmp.i1, %lor.rhs ]
+  ret i1 %0
+}
diff --git a/test/Transforms/SimplifyCFG/switch_formation.dbg.ll b/test/Transforms/SimplifyCFG/switch_formation.dbg.ll
index f1c820ec43b..357ffb60e1e 100644
--- a/test/Transforms/SimplifyCFG/switch_formation.dbg.ll
+++ b/test/Transforms/SimplifyCFG/switch_formation.dbg.ll
@@ -1,5 +1,4 @@
-; RUN: opt < %s -simplifycfg -S | not grep br
-
+; RUN: opt < %s -simplifycfg -S | FileCheck %s
 
         %llvm.dbg.anchor.type = type { i32, i32 }
         %llvm.dbg.compile_unit.type = type { i32, { }*, i32, i8*, i8*, i8*, i1, i1, i8* }
@@ -13,7 +12,16 @@
 
 declare void @llvm.dbg.stoppoint(i32, i32, { }*) nounwind
 
-define i1 @_ZN4llvm11SetCondInst7classofEPKNS_11InstructionE({ i32, i32 }* %I) {
+define i1 @t({ i32, i32 }* %I) {
+; CHECK: t
+; CHECK:  switch i32 %tmp.2.i, label %shortcirc_next.4 [
+; CHECK:    i32 14, label %UnifiedReturnBlock
+; CHECK:    i32 15, label %UnifiedReturnBlock
+; CHECK:    i32 16, label %UnifiedReturnBlock
+; CHECK:    i32 17, label %UnifiedReturnBlock
+; CHECK:    i32 18, label %UnifiedReturnBlock
+; CHECK:    i32 19, label %UnifiedReturnBlock
+; CHECK:  ]
 entry:
         %tmp.1.i = getelementptr { i32, i32 }* %I, i64 0, i32 1         ; <i32*> [#uses=1]
         %tmp.2.i = load i32* %tmp.1.i           ; <i32> [#uses=6]