Merge pull request #1744 from jbj/ast-field-flow-aggregate-init

C++: Field flow through ClassAggregateLiteral

cpp/ql/src/semmle/code/cpp/dataflow/internal/DataFlowPrivate.qll

@@ -6,7 +6,7 @@ private import DataFlowDispatch
 private Node getInstanceArgument(Call call) {
   result.asExpr() = call.getQualifier()
   or
-  result.(PreConstructorCallNode).getConstructorCall() = call
+  result.(PreObjectInitializerNode).getExpr().(ConstructorCall) = call
   // This does not include the implicit `this` argument on auto-generated
   // base class destructor calls as those do not have an AST element.
 }
@@ -169,6 +169,14 @@ private class ArrayContent extends Content, TArrayContent {
  * value of `node1`.
  */
 predicate storeStep(Node node1, Content f, PostUpdateNode node2) {
+  exists(ClassAggregateLiteral aggr, Field field |
+    // The following line requires `node2` to be both an `ExprNode` and a
+    // `PostUpdateNode`, which means it must be an `ObjectInitializerNode`.
+    node2.asExpr() = aggr and
+    f.(FieldContent).getField() = field and
+    aggr.getFieldExpr(field) = node1.asExpr()
+  )
+  or
   exists(FieldAccess fa |
     exists(Assignment a |
       node1.asExpr() = a and

cpp/ql/src/semmle/code/cpp/dataflow/internal/DataFlowUtil.qll

@@ -10,7 +10,11 @@ cached
 private newtype TNode =
   TExprNode(Expr e) or
   TPartialDefinitionNode(PartialDefinition pd) or
-  TPreConstructorCallNode(ConstructorCall call) or
+  TPreObjectInitializerNode(Expr e) {
+    e instanceof ConstructorCall
+    or
+    e instanceof ClassAggregateLiteral
+  } or
   TExplicitParameterNode(Parameter p) { exists(p.getFunction().getBlock()) } or
   TInstanceParameterNode(MemberFunction f) { exists(f.getBlock()) and not f.isStatic() } or
   TUninitializedNode(LocalVariable v) { not v.hasInitializer() }
@@ -256,16 +260,24 @@ class PartialDefinitionNode extends PostUpdateNode, TPartialDefinitionNode {
 }
 
 /**
- * A node representing the object that was just constructed and is identified
- * with the "return value" of the constructor call.
+ * A node representing the temporary value of an object that was just
+ * constructed by a constructor call or an aggregate initializer. This is only
+ * for objects, not for pointers to objects.
+ *
+ * These expressions are their own post-update nodes but instead have synthetic
+ * pre-update nodes.
  */
-private class PostConstructorCallNode extends PostUpdateNode, TExprNode {
-  PostConstructorCallNode() { this = TExprNode(any(ConstructorCall c)) }
+private class ObjectInitializerNode extends PostUpdateNode, TExprNode {
+  PreObjectInitializerNode pre;
 
-  override PreConstructorCallNode getPreUpdateNode() {
-    TExprNode(result.getConstructorCall()) = this
+  ObjectInitializerNode() {
+    // If a `Node` is associated with a `PreObjectInitializerNode`, then it's
+    // an `ObjectInitializerNode`.
+    pre.getExpr() = this.asExpr()
   }
 
+  override PreObjectInitializerNode getPreUpdateNode() { result = pre }
+
   // No override of `toString` since these nodes already have a `toString` from
   // their overlap with `ExprNode`.
 }
@@ -273,19 +285,19 @@ private class PostConstructorCallNode extends PostUpdateNode, TExprNode {
 /**
  * INTERNAL: do not use.
  *
- * A synthetic data-flow node that plays the role of the qualifier (or
- * `this`-argument) to a constructor call.
+ * A synthetic data-flow node that plays the role of a temporary object that
+ * has not yet been initialized.
  */
-class PreConstructorCallNode extends Node, TPreConstructorCallNode {
-  ConstructorCall getConstructorCall() { this = TPreConstructorCallNode(result) }
+class PreObjectInitializerNode extends Node, TPreObjectInitializerNode {
+  Expr getExpr() { this = TPreObjectInitializerNode(result) }
 
-  override Function getFunction() { result = getConstructorCall().getEnclosingFunction() }
+  override Function getFunction() { result = getExpr().getEnclosingFunction() }
 
-  override Type getType() { result = getConstructorCall().getType() }
+  override Type getType() { result = getExpr().getType() }
 
-  override Location getLocation() { result = getConstructorCall().getLocation() }
+  override Location getLocation() { result = getExpr().getLocation() }
 
-  override string toString() { result = getConstructorCall().toString() + " [pre constructor call]" }
+  override string toString() { result = getExpr().toString() + " [pre init]" }
 }
 
 /**

cpp/ql/test/library-tests/dataflow/dataflow-tests/test.cpp

@@ -142,7 +142,7 @@ void following_pointers(
   twoIntFields s = { source(), source() };
 
 
-  sink(s.m2); // flow (AST dataflow misses this due to limitations of the analysis)
+  sink(s.m2); // flow
 
   twoIntFields sArray[1] = { { source(), source() } };
   // TODO: fix this like above
@@ -150,7 +150,7 @@ void following_pointers(
 
   twoIntFields sSwapped = { .m2 = source(), .m1 = 0 };
 
-  sink(sSwapped.m2); // flow (AST dataflow misses this due to limitations of the analysis)
+  sink(sSwapped.m2); // flow
 
   sink(sourceFunctionPointer()); // no flow
 

cpp/ql/test/library-tests/dataflow/dataflow-tests/test.expected

@@ -19,6 +19,8 @@
 | test.cpp:137:27:137:28 | m1 | test.cpp:136:27:136:32 | call to source |
 | test.cpp:138:27:138:34 | call to getFirst | test.cpp:136:27:136:32 | call to source |
 | test.cpp:140:22:140:23 | m1 | test.cpp:136:27:136:32 | call to source |
+| test.cpp:145:10:145:11 | m2 | test.cpp:142:32:142:37 | call to source |
+| test.cpp:153:17:153:18 | m2 | test.cpp:151:35:151:40 | call to source |
 | test.cpp:188:8:188:8 | y | test.cpp:186:27:186:32 | call to source |
 | test.cpp:192:8:192:8 | s | test.cpp:199:33:199:38 | call to source |
 | test.cpp:200:8:200:8 | y | test.cpp:199:33:199:38 | call to source |

cpp/ql/test/library-tests/dataflow/dataflow-tests/test_diff.expected

@@ -6,8 +6,6 @@
 | test.cpp:136:27:136:32 | test.cpp:137:27:137:28 | AST only |
 | test.cpp:136:27:136:32 | test.cpp:138:27:138:34 | AST only |
 | test.cpp:136:27:136:32 | test.cpp:140:22:140:23 | AST only |
-| test.cpp:142:32:142:37 | test.cpp:145:10:145:11 | IR only |
-| test.cpp:151:35:151:40 | test.cpp:153:17:153:18 | IR only |
 | test.cpp:395:17:395:22 | test.cpp:397:10:397:18 | AST only |
 | test.cpp:407:13:407:18 | test.cpp:413:10:413:14 | AST only |
 | test.cpp:421:13:421:18 | test.cpp:417:10:417:14 | AST only |

cpp/ql/test/library-tests/dataflow/fields/flow.expected

@@ -98,6 +98,14 @@ edges
 | simple.cpp:48:9:48:9 | g [b_, ... (1)] | simple.cpp:26:15:26:15 | f [b_, ... (1)] |
 | simple.cpp:51:9:51:9 | h [a_, ... (1)] | simple.cpp:26:15:26:15 | f [a_, ... (1)] |
 | simple.cpp:51:9:51:9 | h [b_, ... (1)] | simple.cpp:26:15:26:15 | f [b_, ... (1)] |
+| struct_init.c:20:17:20:36 | {...} [a, ... (1)] | struct_init.c:22:8:22:9 | ab [a, ... (1)] |
+| struct_init.c:20:20:20:29 | call to user_input [void] | struct_init.c:20:17:20:36 | {...} [a, ... (1)] |
+| struct_init.c:22:8:22:9 | ab [a, ... (1)] | struct_init.c:22:11:22:11 | a |
+| struct_init.c:26:23:29:3 | {...} [nestedAB, ... (2)] | struct_init.c:31:8:31:12 | outer [nestedAB, ... (2)] |
+| struct_init.c:27:5:27:23 | {...} [a, ... (1)] | struct_init.c:26:23:29:3 | {...} [nestedAB, ... (2)] |
+| struct_init.c:27:7:27:16 | call to user_input [void] | struct_init.c:27:5:27:23 | {...} [a, ... (1)] |
+| struct_init.c:31:8:31:12 | outer [nestedAB, ... (2)] | struct_init.c:31:14:31:21 | nestedAB [a, ... (1)] |
+| struct_init.c:31:14:31:21 | nestedAB [a, ... (1)] | struct_init.c:31:23:31:23 | a |
 #select
 | A.cpp:43:10:43:12 | & ... | A.cpp:41:15:41:21 | new [void] | A.cpp:43:10:43:12 | & ... | & ... flows from $@ | A.cpp:41:15:41:21 | new [void] | new [void] |
 | A.cpp:49:13:49:13 | c | A.cpp:47:12:47:18 | new [void] | A.cpp:49:13:49:13 | c | c flows from $@ | A.cpp:47:12:47:18 | new [void] | new [void] |
@@ -121,3 +129,5 @@ edges
 | simple.cpp:28:12:28:12 | call to a | simple.cpp:41:12:41:21 | call to user_input [void] | simple.cpp:28:12:28:12 | call to a | call to a flows from $@ | simple.cpp:41:12:41:21 | call to user_input [void] | call to user_input [void] |
 | simple.cpp:29:12:29:12 | call to b | simple.cpp:40:12:40:21 | call to user_input [void] | simple.cpp:29:12:29:12 | call to b | call to b flows from $@ | simple.cpp:40:12:40:21 | call to user_input [void] | call to user_input [void] |
 | simple.cpp:29:12:29:12 | call to b | simple.cpp:42:12:42:21 | call to user_input [void] | simple.cpp:29:12:29:12 | call to b | call to b flows from $@ | simple.cpp:42:12:42:21 | call to user_input [void] | call to user_input [void] |
+| struct_init.c:22:11:22:11 | a | struct_init.c:20:20:20:29 | call to user_input [void] | struct_init.c:22:11:22:11 | a | a flows from $@ | struct_init.c:20:20:20:29 | call to user_input [void] | call to user_input [void] |
+| struct_init.c:31:23:31:23 | a | struct_init.c:27:7:27:16 | call to user_input [void] | struct_init.c:31:23:31:23 | a | a flows from $@ | struct_init.c:27:7:27:16 | call to user_input [void] | call to user_input [void] |

cpp/ql/test/library-tests/dataflow/fields/struct_init.c

@@ -0,0 +1,38 @@
+void sink(void *o);
+void *user_input(void);
+
+struct AB {
+  void *a;
+  void *b;
+};
+
+struct Outer {
+  struct AB nestedAB;
+  struct AB *pointerAB;
+};
+
+void absink(struct AB *ab) {
+  sink(ab->a); // flow x3 [NOT DETECTED]
+  sink(ab->b); // no flow
+}
+
+int struct_init(void) {
+  struct AB ab = { user_input(), 0 };
+
+  sink(ab.a); // flow
+  sink(ab.b); // no flow
+  absink(&ab);
+
+  struct Outer outer = {
+    { user_input(), 0 },
+    &ab,
+  };
+
+  sink(outer.nestedAB.a); // flow
+  sink(outer.nestedAB.b); // no flow
+  sink(outer.pointerAB->a); // flow [NOT DETECTED]
+  sink(outer.pointerAB->b); // no flow
+
+  absink(&outer.nestedAB);
+  absink(outer.pointerAB);
+}