[quandary] allow custom sources/sinks in C++

Reviewed By: akotulski

Differential Revision: D4975414

fbshipit-source-id: 4fd7078

infer/src/quandary/ClangTrace.ml

@@ -12,8 +12,7 @@ open! IStd
 module F = Format
 module L = Logging
 
-
+module SourceKind = struct
-module Kind = struct
   type t =
     | EnvironmentVariable (** source that was read from an environment variable *)
     | Other (** for testing or uncategorized sources *)
@@ -22,19 +21,34 @@ module Kind = struct
 
   let unknown = Unknown
 
+  let of_string = function
+    | "EnvironmentVariable" -> EnvironmentVariable
+    | _ -> Other
+
+  let external_sources =
+    List.map
+      ~f:(fun { QuandaryConfig.Source.procedure; kind; } ->
+          QualifiedCppName.Match.of_fuzzy_qual_names [procedure], kind)
+      (QuandaryConfig.Source.of_json Config.quandary_sources)
+
+  (* return Some(source kind) if [procedure_name] is in the list of externally specified sources *)
+  let get_external_source pname =
+    let qualified_pname = Typ.Procname.get_qualifiers pname in
+    List.find_map
+      ~f:(fun (qualifiers, kind) ->
+          if QualifiedCppName.Match.match_qualifiers qualifiers qualified_pname
+          then Some (of_string kind)
+          else None)
+      external_sources
+
   let get pname _ = match pname with
-    | (Typ.Procname.ObjC_Cpp cpp_pname) as pname ->
+    | Typ.Procname.ObjC_Cpp _ ->
-        begin
+        get_external_source pname
-          match Typ.Procname.objc_cpp_get_class_name cpp_pname, Typ.Procname.get_method pname with
+    | Typ.Procname.C _ ->
-          | "InferTaint", "source" -> Some Other
-          | _ -> None
-        end
-    | (Typ.Procname.C _) as pname ->
         begin
           match Typ.Procname.to_string pname with
           | "getenv" -> Some EnvironmentVariable
-          | "__infer_taint_source" -> Some Other
+          | _ -> get_external_source pname
-          | _ -> None
         end
     | Typ.Procname.Block _ ->
         None
@@ -52,7 +66,7 @@ module Kind = struct
     | Unknown -> F.fprintf fmt "Unknown"
 end
 
-module CppSource = Source.Make(Kind)
+module CppSource = Source.Make(SourceKind)
 
 module SinkKind = struct
 
@@ -61,27 +75,54 @@ module SinkKind = struct
     | Other (** for testing or uncategorized sinks *)
   [@@deriving compare]
 
+  let of_string = function
+    | "ShellExec" -> ShellExec
+    | _ -> Other
+
+  let external_sinks =
+    List.map
+      ~f:(fun { QuandaryConfig.Sink.procedure; kind; index; } ->
+          QualifiedCppName.Match.of_fuzzy_qual_names [procedure], kind, index)
+      (QuandaryConfig.Sink.of_json Config.quandary_sinks)
+
+  (* taint the nth parameter (0-indexed) *)
+  let taint_nth n kind ~report_reachable =
+    [kind, n, report_reachable]
+
+  let taint_all actuals kind ~report_reachable =
+    List.mapi
+      ~f:(fun actual_num _ -> kind, actual_num, report_reachable)
+      actuals
+
+  (* return Some(sink kind) if [procedure_name] is in the list of externally specified sinks *)
+  let get_external_sink pname actuals =
+    let qualified_pname = Typ.Procname.get_qualifiers pname in
+    List.find_map
+      ~f:(fun (qualifiers, kind, index) ->
+          if QualifiedCppName.Match.match_qualifiers qualifiers qualified_pname
+          then
+            let kind = of_string kind in
+            try
+              let n = int_of_string index in
+              Some (taint_nth n kind ~report_reachable:true)
+            with Failure _ ->
+              (* couldn't parse the index, just taint everything *)
+              Some (taint_all actuals kind ~report_reachable:true)
+          else
+            None)
+      external_sinks
+
   let get pname actuals _ =
-    let taint_all actuals kind ~report_reachable =
-      List.mapi
-        ~f:(fun actual_num _ -> kind, actual_num, report_reachable)
-        actuals in
     match pname with
-    | (Typ.Procname.ObjC_Cpp cpp_pname) as pname ->
+    | Typ.Procname.ObjC_Cpp _ ->
-        begin
+        Option.value (get_external_sink pname actuals) ~default:[]
-          match Typ.Procname.objc_cpp_get_class_name cpp_pname, Typ.Procname.get_method pname with
-          | "InferTaint", "sink:" -> taint_all actuals Other ~report_reachable:true
-          | _ -> []
-        end
     | Typ.Procname.C _ ->
         begin
           match Typ.Procname.to_string pname with
           | "execl" | "execlp" | "execle" | "execv" | "execvp" ->
               taint_all actuals ShellExec ~report_reachable:false
-          | "__infer_taint_sink" ->
-              [Other, 0, false]
           | _ ->
-              []
+              Option.value (get_external_sink pname actuals) ~default:[]
         end
     | Typ.Procname.Block _ ->
         []

infer/src/quandary/JavaTrace.ml

@@ -29,7 +29,10 @@ module SourceKind = struct
     | "UserControlledURI" -> UserControlledURI
     | _ -> Other
 
-  let external_sources = QuandaryConfig.Source.of_json Config.quandary_sources
+  let external_sources =
+    List.map
+      ~f:(fun { QuandaryConfig.Source.procedure; kind; } -> Str.regexp procedure, kind)
+      (QuandaryConfig.Source.of_json Config.quandary_sources)
 
   let get pname tenv = match pname with
     | Typ.Procname.Java pname ->
@@ -70,9 +73,9 @@ module SourceKind = struct
                     (* check the list of externally specified sources *)
                     let procedure = class_name ^ "." ^ method_name in
                     List.find_map
-                      ~f:(fun (source_spec : QuandaryConfig.Source.t) ->
+                      ~f:(fun (procedure_regex, kind) ->
-                          if Str.string_match source_spec.procedure procedure 0
+                          if Str.string_match procedure_regex procedure 0
-                          then Some (of_string source_spec.kind)
+                          then Some (of_string kind)
                           else None)
                       external_sources
               end
@@ -193,7 +196,11 @@ module SinkKind = struct
     | "StartComponent" -> StartComponent
     | _ -> Other
 
-  let external_sinks = QuandaryConfig.Sink.of_json Config.quandary_sinks
+  let external_sinks =
+    List.map
+      ~f:(fun { QuandaryConfig.Sink.procedure; kind; index; } ->
+          Str.regexp procedure, kind, index)
+      (QuandaryConfig.Sink.of_json Config.quandary_sinks)
 
   let get pname actuals tenv =
     (* taint all the inputs of [pname]. for non-static procedures, taints the "this" parameter only
@@ -278,12 +285,12 @@ module SinkKind = struct
                     (* check the list of externally specified sinks *)
                     let procedure = class_name ^ "." ^ method_name in
                     List.find_map
-                      ~f:(fun (sink_spec : QuandaryConfig.Sink.t) ->
+                      ~f:(fun (procedure_regex, kind, index) ->
-                          if Str.string_match sink_spec.procedure procedure 0
+                          if Str.string_match procedure_regex procedure 0
                           then
-                            let kind = of_string sink_spec.kind in
+                            let kind = of_string kind in
                             try
-                              let n = int_of_string sink_spec.index in
+                              let n = int_of_string index in
                               Some (taint_nth n kind ~report_reachable:true)
                             with Failure _ ->
                               (* couldn't parse the index, just taint everything *)

infer/src/quandary/QuandaryConfig.ml

@@ -14,13 +14,13 @@ module F = Format
 (** utilities for importing JSON specifications of sources/sinks into Quandary *)
 
 module Source = struct
-  type t = { procedure : Str.regexp; kind : string; }
+  type t = { procedure : string; kind : string; }
 
   let of_json = function
     | `List sources ->
         let parse_source json =
           let open Yojson.Basic in
-          let procedure = Util.member "procedure" json |> Util.to_string |> Str.regexp in
+          let procedure = Util.member "procedure" json |> Util.to_string in
           let kind = Util.member "kind" json |> Util.to_string in
           { procedure; kind; } in
         List.map ~f:parse_source sources
@@ -29,13 +29,13 @@ module Source = struct
 end
 
 module Sink = struct
-  type t = { procedure : Str.regexp; kind : string; index : string}
+  type t = { procedure : string; kind : string; index : string}
 
   let of_json = function
     | `List sinks ->
         let parse_sink json =
           let open Yojson.Basic in
-          let procedure = Util.member "procedure" json |> Util.to_string |> Str.regexp in
+          let procedure = Util.member "procedure" json |> Util.to_string in
           let kind = Util.member "kind" json |> Util.to_string in
           let index =
             Util.member "index" json |> Util.to_string_option |> Option.value ~default:"all" in

infer/src/quandary/QuandaryConfig.mli

@@ -12,13 +12,13 @@ open! IStd
 (** utilities for importing JSON specifications of sources/sinks into Quandary*)
 
 module Source : sig
-  type t = { procedure : Str.regexp; kind : string; }
+  type t = { procedure : string; kind : string; }
 
   val of_json : [> `List of Yojson.Basic.json list ] -> t list
 end
 
 module Sink : sig
-  type t = { procedure : Str.regexp; kind : string; index : string; }
+  type t = { procedure : string; kind : string; index : string; }
 
   val of_json : [> `List of Yojson.Basic.json list ] -> t list
 end

infer/tests/codetoanalyze/cpp/quandary/.inferconfig

@@ -0,0 +1,37 @@
+{
+  "quandary-sources": [
+    {
+      "procedure": "__infer_taint_source",
+      "kind": "Other"
+    },
+    {
+      "procedure": "basics::Obj::method_source",
+      "kind": "Other"
+    },
+    {
+      "procedure": "basics::Obj::static_source",
+      "kind": "Other"
+    },
+    {
+      "procedure": "basics::template_source",
+      "kind": "Other"
+    }
+  ],
+  "quandary-sinks": [
+    {
+      "procedure": "__infer_taint_sink",
+      "kind": "Other",
+      "index": "0"
+    },
+    {
+      "procedure": "basics::Obj::method_sink",
+      "kind": "Other",
+      "index": "1"
+    },
+    {
+      "procedure": "basics::Obj::static_sink",
+      "kind": "Other",
+      "index": "0"
+    }
+  ]
+}

infer/tests/codetoanalyze/cpp/quandary/basics.cpp

@@ -14,7 +14,10 @@ namespace basics {
 
 class Obj {
  public:
-  int field;
+  void* method_source() { return (void*)0; }
+  void method_sink(void*) {}
+  static void* static_source() { return (void*)0; }
+  static void static_sink(void*) {}
 };
 
 void* returnSource() { return __infer_taint_source(); }
@@ -43,4 +46,26 @@ void propagateBad() {
   void* launderedSource = id(source);
   callSink(launderedSource);
 }
+
+// make sure specifying external sources/sinks as instance methods works
+void object_source_sink_bad(Obj obj) {
+  void* source = obj.method_source();
+  obj.method_sink(source);
+}
+
+// make sure specifying external sources/sinks as static methods works
+void static_source_sink_bad(Obj obj) {
+  void* source = Obj::static_source();
+  Obj::static_sink(source);
+}
+
+template <class T>
+T* template_source() {
+  return nullptr;
+}
+
+void template_source_bad() {
+  void* source = template_source<void*>();
+  __infer_taint_sink(source);
+}
 }

infer/tests/codetoanalyze/cpp/quandary/issues.exp

@@ -1,7 +1,10 @@
+codetoanalyze/cpp/quandary/basics.cpp, basics::object_source_sink_bad, 2, QUANDARY_TAINT_ERROR, [return from basics::Obj_method_source,call to basics::Obj_method_sink]
 codetoanalyze/cpp/quandary/basics.cpp, basics::propagateBad, 3, QUANDARY_TAINT_ERROR, [return from __infer_taint_source,flow through basics::id,call to basics::callSink,call to __infer_taint_sink]
 codetoanalyze/cpp/quandary/basics.cpp, basics::returnSourceToSinkBad, 2, QUANDARY_TAINT_ERROR, [return from __infer_taint_source,return from basics::returnSource,call to __infer_taint_sink]
 codetoanalyze/cpp/quandary/basics.cpp, basics::sourceThenCallSinkBad, 2, QUANDARY_TAINT_ERROR, [return from __infer_taint_source,call to basics::callSink,call to __infer_taint_sink]
 codetoanalyze/cpp/quandary/basics.cpp, basics::sourceToSinkDirectBad, 2, QUANDARY_TAINT_ERROR, [return from __infer_taint_source,call to __infer_taint_sink]
+codetoanalyze/cpp/quandary/basics.cpp, basics::static_source_sink_bad, 2, QUANDARY_TAINT_ERROR, [return from basics::Obj_static_source,call to basics::Obj_static_sink]
+codetoanalyze/cpp/quandary/basics.cpp, basics::template_source_bad, 2, QUANDARY_TAINT_ERROR, [return from basics::template_source<void_*>,call to __infer_taint_sink]
 codetoanalyze/cpp/quandary/execs.cpp, execs::callExecBad, 6, QUANDARY_TAINT_ERROR, [return from getenv,call to execl]
 codetoanalyze/cpp/quandary/execs.cpp, execs::callExecBad, 8, QUANDARY_TAINT_ERROR, [return from getenv,call to execl]
 codetoanalyze/cpp/quandary/execs.cpp, execs::callExecBad, 11, QUANDARY_TAINT_ERROR, [return from getenv,call to execl]