Merge pull request diffblue#2053 from owen-jones-diffblue/owen-jones-…

…diffblue/bugfix/make-callgraph-include-uncalled-functions

Make directed callgraph include nodes with no edges

regression/goto-instrument/call-graph1/main.c

@@ -0,0 +1,81 @@
+
+int x;
+
+void func0()
+{
+  func0();
+}
+
+void func1()
+{
+  x = 1;
+}
+
+void func2()
+{
+  x = 2;
+  func3();
+}
+
+void func3()
+{
+  x = 3;
+}
+
+void func4(int b)
+{
+  x = 4;
+  if(b)
+  {
+    func5(0);
+  }
+}
+
+void func5(int b)
+{
+  x = 5;
+  func4(b);
+}
+
+void func6()
+{
+  x = 6;
+}
+
+void func7(int b)
+{
+  x = 7;
+  if(b)
+  {
+    func8(0);
+  }
+}
+
+void func8(int b)
+{
+  x = 8;
+  func7(b);
+}
+
+void func9()
+{
+  x = 9;
+  funca();
+}
+
+void funca()
+{
+  x = 10;
+  func9();
+}
+
+
+
+int main()
+{
+  func1();
+  func2();
+  func4(1);
+
+  return 0;
+}

regression/goto-instrument/call-graph1/test.desc

@@ -0,0 +1,20 @@
+CORE
+main.c
+--call-graph
+^EXIT=0$
+^SIGNAL=0$
+^main -> func2$
+^main -> func1$
+^main -> func4$
+^funca -> func9$
+^func9 -> funca$
+^func8 -> func7$
+^func7 -> func8$
+^__CPROVER__start -> __CPROVER_initialize$
+^__CPROVER__start -> main$
+^func2 -> func3$
+^func0 -> func0$
+^func4 -> func5$
+^func5 -> func4$
+--
+^warning: ignoring

regression/goto-instrument/reachable-call-graph1/main.c

@@ -0,0 +1,81 @@
+
+int x;
+
+void func0()
+{
+  func0();
+}
+
+void func1()
+{
+  x = 1;
+}
+
+void func2()
+{
+  x = 2;
+  func3();
+}
+
+void func3()
+{
+  x = 3;
+}
+
+void func4(int b)
+{
+  x = 4;
+  if(b)
+  {
+    func5(0);
+  }
+}
+
+void func5(int b)
+{
+  x = 5;
+  func4(b);
+}
+
+void func6()
+{
+  x = 6;
+}
+
+void func7(int b)
+{
+  x = 7;
+  if(b)
+  {
+    func8(0);
+  }
+}
+
+void func8(int b)
+{
+  x = 8;
+  func7(b);
+}
+
+void func9()
+{
+  x = 9;
+  funca();
+}
+
+void funca()
+{
+  x = 10;
+  func9();
+}
+
+
+
+int main()
+{
+  func1();
+  func2();
+  func4(1);
+
+  return 0;
+}

regression/goto-instrument/reachable-call-graph1/test.desc

@@ -0,0 +1,20 @@
+CORE
+main.c
+--reachable-call-graph
+^EXIT=0$
+^SIGNAL=0$
+^main -> func2$
+^main -> func1$
+^main -> func4$
+^__CPROVER__start -> __CPROVER_initialize$
+^__CPROVER__start -> main$
+^func2 -> func3$
+^func4 -> func5$
+^func5 -> func4$
+--
+^funca -> func9$
+^func9 -> funca$
+^func8 -> func7$
+^func7 -> func8$
+^func0 -> func0$
+^warning: ignoring

src/analyses/call_graph.cpp

@@ -41,8 +41,10 @@ call_grapht::call_grapht(
 {
   forall_goto_functions(f_it, goto_functions)
   {
-    const goto_programt &body=f_it->second.body;
-    add(f_it->first, body);
+    const irep_idt &function_name = f_it->first;
+    const goto_programt &body = f_it->second.body;
+    nodes.insert(function_name);
+    add(function_name, body);
   }
 }
 
@@ -84,12 +86,14 @@ call_grapht::call_grapht(
     const goto_programt &goto_program=
       goto_functions.function_map.at(function).body;
 
+    nodes.insert(function);
+
     forall_callsites(
       goto_program,
       [&](goto_programt::const_targett i_it, const irep_idt &callee)
       {
         add(function, callee, i_it);
-        if(graph.find(callee)==graph.end())
+        if(edges.find(callee)==edges.end())
           pending_stack.push(callee);
       }
     ); // NOLINT
@@ -129,7 +133,9 @@ void call_grapht::add(
   const irep_idt &caller,
   const irep_idt &callee)
 {
-  graph.insert(std::pair<irep_idt, irep_idt>(caller, callee));
+  edges.insert({caller, callee});
+  nodes.insert(caller);
+  nodes.insert(callee);
 }
 
 /// Add edge with optional callsite information
@@ -152,7 +158,8 @@ void call_grapht::add(
 call_grapht call_grapht::get_inverted() const
 {
   call_grapht result;
-  for(const auto &caller_callee : graph)
+  result.nodes = nodes;
+  for(const auto &caller_callee : edges)
     result.add(caller_callee.second, caller_callee.first);
   return result;
 }
@@ -197,7 +204,12 @@ call_grapht::directed_grapht call_grapht::get_directed_graph() const
   call_grapht::directed_grapht ret;
   function_indicest function_indices(ret);
 
-  for(const auto &edge : graph)
+  // To make sure we include unreachable functions we first create indices
+  // for all nodes in the graph
+  for(const irep_idt &function_name : nodes)
+    function_indices[function_name];
+
+  for(const auto &edge : edges)
   {
     auto a_index=function_indices[edge.first];
     auto b_index=function_indices[edge.second];
@@ -237,7 +249,7 @@ void call_grapht::output_dot(std::ostream &out) const
 {
   out << "digraph call_graph {\n";
 
-  for(const auto &edge : graph)
+  for(const auto &edge : edges)
   {
     out << "  \"" << edge.first << "\" -> "
         << "\"" << edge.second << "\" "
@@ -252,7 +264,7 @@ void call_grapht::output_dot(std::ostream &out) const
 
 void call_grapht::output(std::ostream &out) const
 {
-  for(const auto &edge : graph)
+  for(const auto &edge : edges)
   {
     out << edge.first << " -> " << edge.second << "\n";
     if(collect_callsites)
@@ -267,7 +279,7 @@ void call_grapht::output_xml(std::ostream &out) const
   if(collect_callsites)
     out << "<!-- XML call-graph representation does not document callsites yet."
       " If you need this, edit call_grapht::output_xml -->\n";
-  for(const auto &edge : graph)
+  for(const auto &edge : edges)
   {
     out << "<call_graph_edge caller=\"";
     xmlt::escape_attribute(id2string(edge.first), out);

src/analyses/call_graph.h

@@ -62,11 +62,14 @@ class call_grapht
   void output(std::ostream &out) const;
   void output_xml(std::ostream &out) const;
 
-  /// Type of the call graph. Note parallel edges (e.g. A having two callsites
-  /// both targeting B) result in multiple graph edges.
-  typedef std::multimap<irep_idt, irep_idt> grapht;
+  /// Type of the nodes in the call graph.
+  typedef std::unordered_set<irep_idt, irep_id_hash> nodest;
 
-  /// Type of a call graph edge in `grapht`
+  /// Type of the edges in the call graph. Note parallel edges (e.g. A having
+  /// two callsites both targeting B) result in multiple graph edges.
+  typedef std::multimap<irep_idt, irep_idt> edgest;
+
+  /// Type of a call graph edge in `edgest`
   typedef std::pair<irep_idt, irep_idt> edget;
 
   /// Type of a callsite stored in member `callsites`
@@ -84,7 +87,8 @@ class call_grapht
   /// backward compatibility; use `get_directed_graph()` to get a generic
   /// directed graph representation that provides more graph algorithms
   /// (shortest path, SCCs and so on).
-  grapht graph;
+  edgest edges;
+  nodest nodes;
 
   /// Map from call-graph edges to a set of callsites that make the given call.
   callsitest callsites;