Index: llvm/trunk/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
===================================================================
--- llvm/trunk/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
+++ llvm/trunk/lib/Transforms/Instrumentation/PGOInstrumentation.cpp
@@ -329,14 +329,16 @@
 // Visitor class that finds all indirect call sites.
 struct PGOIndirectCallSiteVisitor
     : public InstVisitor<PGOIndirectCallSiteVisitor> {
-  std::vector<CallInst *> IndirectCallInsts;
+  std::vector<Instruction *> IndirectCallInsts;
   PGOIndirectCallSiteVisitor() {}
 
-  void visitCallInst(CallInst &I) {
-    CallSite CS(&I);
-    if (CS.getCalledFunction() || !CS.getCalledValue() || I.isInlineAsm())
+  void visitCallSite(CallSite CS) {
+    Instruction *I = CS.getInstruction();
+    CallInst *CI = dyn_cast<CallInst>(I);
+    if (CS.getCalledFunction() || !CS.getCalledValue() ||
+        (CI && CI->isInlineAsm()))
       return;
-    IndirectCallInsts.push_back(&I);
+    IndirectCallInsts.push_back(I);
   }
 };
 
Index: llvm/trunk/test/Transforms/PGOProfile/indirect_call_profile.ll
===================================================================
--- llvm/trunk/test/Transforms/PGOProfile/indirect_call_profile.ll
+++ llvm/trunk/test/Transforms/PGOProfile/indirect_call_profile.ll
@@ -15,3 +15,45 @@
   call void %tmp()
   ret void
 }
+
+@bar2 = global void ()* null, align 8
+@_ZTIi = external constant i8*
+
+define i32 @foo2(i32 %arg, i8** nocapture readnone %arg1) personality i8* bitcast (i32 (...)* @__gxx_personality_v0 to i8*) {
+bb:
+  %tmp2 = load void ()*, void ()** @bar2, align 8
+  invoke void %tmp2()
+          to label %bb10 unwind label %bb2
+; GEN: [[ICALL_TARGET2:%[0-9]+]] = ptrtoint void ()* %tmp2 to i64
+; GEN-NEXT: call void @llvm.instrprof.value.profile(i8* getelementptr inbounds ([4 x i8], [4 x i8]* @__profn_foo2, i32 0, i32 0), i64 38432627612, i64 [[ICALL_TARGET2]], i32 0, i32 0)
+
+bb2:                                              ; preds = %bb
+  %tmp3 = landingpad { i8*, i32 }
+          catch i8* bitcast (i8** @_ZTIi to i8*)
+  %tmp4 = extractvalue { i8*, i32 } %tmp3, 1
+  %tmp5 = tail call i32 @llvm.eh.typeid.for(i8* bitcast (i8** @_ZTIi to i8*))
+  %tmp6 = icmp eq i32 %tmp4, %tmp5
+  br i1 %tmp6, label %bb7, label %bb11
+
+bb7:                                              ; preds = %bb2
+  %tmp8 = extractvalue { i8*, i32 } %tmp3, 0
+  %tmp9 = tail call i8* @__cxa_begin_catch(i8* %tmp8)
+  tail call void @__cxa_end_catch()
+  br label %bb10
+
+bb10:                                             ; preds = %bb7, %bb
+  ret i32 0
+
+bb11:                                             ; preds = %bb2
+  resume { i8*, i32 } %tmp3
+}
+
+declare i32 @__gxx_personality_v0(...)
+
+; Function Attrs: nounwind readnone
+declare i32 @llvm.eh.typeid.for(i8*) #0
+
+declare i8* @__cxa_begin_catch(i8*)
+
+declare void @__cxa_end_catch()
+