Index: lib/Target/AMDGPU/SILowerControlFlow.cpp
===================================================================
--- lib/Target/AMDGPU/SILowerControlFlow.cpp
+++ lib/Target/AMDGPU/SILowerControlFlow.cpp
@@ -125,29 +125,44 @@
   return new SILowerControlFlow();
 }
 
+static bool opcodeEmitsNoInsts(unsigned Opc) {
+  switch (Opc) {
+  case TargetOpcode::IMPLICIT_DEF:
+  case TargetOpcode::KILL:
+  case TargetOpcode::BUNDLE:
+  case TargetOpcode::CFI_INSTRUCTION:
+  case TargetOpcode::EH_LABEL:
+  case TargetOpcode::GC_LABEL:
+  case TargetOpcode::DBG_VALUE:
+    return true;
+  default:
+    return false;
+  }
+}
+
 bool SILowerControlFlow::shouldSkip(MachineBasicBlock *From,
                                     MachineBasicBlock *To) {
 
   unsigned NumInstr = 0;
+  MachineFunction *MF = From->getParent();
 
-  for (MachineFunction::iterator MBBI = MachineFunction::iterator(From),
-                                 ToI = MachineFunction::iterator(To); MBBI != ToI; ++MBBI) {
-
+  for (MachineFunction::iterator MBBI(From), ToI(To), End = MF->end();
+       MBBI != End && MBBI != ToI; ++MBBI) {
     MachineBasicBlock &MBB = *MBBI;
 
     for (MachineBasicBlock::iterator I = MBB.begin(), E = MBB.end();
          NumInstr < SkipThreshold && I != E; ++I) {
+      if (opcodeEmitsNoInsts(I->getOpcode()))
+        continue;
 
-      if (I->isBundle() || !I->isBundled()) {
-        // When a uniform loop is inside non-uniform control flow, the branch
-        // leaving the loop might be an S_CBRANCH_VCCNZ, which is never taken
-        // when EXEC = 0. We should skip the loop lest it becomes infinite.
-        if (I->getOpcode() == AMDGPU::S_CBRANCH_VCCNZ)
-          return true;
+      // When a uniform loop is inside non-uniform control flow, the branch
+      // leaving the loop might be an S_CBRANCH_VCCNZ, which is never taken
+      // when EXEC = 0. We should skip the loop lest it becomes infinite.
+      if (I->getOpcode() == AMDGPU::S_CBRANCH_VCCNZ)
+        return true;
 
-        if (++NumInstr >= SkipThreshold)
-          return true;
-      }
+      if (++NumInstr >= SkipThreshold)
+        return true;
     }
   }
 
Index: test/CodeGen/AMDGPU/si-lower-control-flow-unreachable-block.ll
===================================================================
--- /dev/null
+++ test/CodeGen/AMDGPU/si-lower-control-flow-unreachable-block.ll
@@ -0,0 +1,56 @@
+; RUN: llc -march=amdgcn -verify-machineinstrs < %s | FileCheck -check-prefix=GCN %s
+
+; GCN-LABEL: {{^}}lower_control_flow_unreachable_terminator:
+; GCN: v_cmp_eq_i32
+; GCN: s_and_saveexec_b64
+; GCN: s_xor_b64
+; GCN: s_branch BB0_1
+
+; GCN: s_or_b64 exec, exec
+; GCN: s_endpgm
+
+; GCN: ds_write_b32
+; GCN: s_waitcnt
+define void @lower_control_flow_unreachable_terminator() #0 {
+bb:
+  %tmp15 = tail call i32 @llvm.amdgcn.workitem.id.y()
+  %tmp63 = icmp eq i32 %tmp15, 32
+  br i1 %tmp63, label %bb64, label %bb68
+
+bb64:
+  store volatile i32 0, i32 addrspace(3)* undef, align 4
+  unreachable
+
+bb68:
+  ret void
+}
+
+; GCN-LABEL: {{^}}lower_control_flow_unreachable_terminator_swap_block_order:
+; GCN: v_cmp_eq_i32
+; GCN: s_and_saveexec_b64
+; GCN: s_xor_b64
+; GCN: s_endpgm
+
+; GCN: s_or_b64 exec, exec
+; GCN: ds_write_b32
+; GCN: s_waitcnt
+define void @lower_control_flow_unreachable_terminator_swap_block_order() #0 {
+bb:
+  %tmp15 = tail call i32 @llvm.amdgcn.workitem.id.y()
+  %tmp63 = icmp eq i32 %tmp15, 32
+  br i1 %tmp63, label %bb68, label %bb64
+
+bb68:
+  ret void
+
+bb64:
+  store volatile i32 0, i32 addrspace(3)* undef, align 4
+  unreachable
+}
+
+; Function Attrs: nounwind readnone
+declare i32 @llvm.amdgcn.workitem.id.y() #1
+
+attributes #0 = { nounwind }
+attributes #1 = { nounwind readnone }
+attributes #2 = { nounwind }