C++: Reduce re-evaluation

cpp/ql/lib/semmle/code/cpp/ir/dataflow/internal/ExprNodes.qll

@@ -6,6 +6,7 @@ private import cpp
 private import semmle.code.cpp.ir.IR
 private import DataFlowUtil
 private import DataFlowPrivate
+private import DataFlowNodes
 private import semmle.code.cpp.ir.implementation.raw.internal.TranslatedExpr
 private import semmle.code.cpp.ir.implementation.raw.internal.InstructionTag
 

cpp/ql/lib/semmle/code/cpp/ir/dataflow/internal/ModelUtil.qll

@@ -6,6 +6,7 @@
 private import semmle.code.cpp.ir.IR
 private import semmle.code.cpp.models.interfaces.FunctionInputsAndOutputs
 private import DataFlowUtil
+private import DataFlowNodes
 private import DataFlowPrivate
 private import SsaImpl as Ssa
 

cpp/ql/lib/semmle/code/cpp/ir/dataflow/internal/SsaImpl.qll

@@ -12,6 +12,7 @@ private import semmle.code.cpp.ir.internal.IRCppLanguage
 private import semmle.code.cpp.ir.dataflow.internal.ModelUtil
 private import semmle.code.cpp.ir.implementation.raw.internal.TranslatedInitialization
 private import DataFlowPrivate
+private import DataFlowNodes
 import SsaImplCommon
 
 private module SourceVariables {

cpp/ql/lib/semmle/code/cpp/ir/dataflow/internal/SsaImplCommon.qll

@@ -4,6 +4,7 @@ import semmle.code.cpp.ir.internal.IRCppLanguage
 private import semmle.code.cpp.ir.implementation.raw.internal.SideEffects as SideEffects
 private import DataFlowImplCommon as DataFlowImplCommon
 private import DataFlowUtil
+private import DataFlowNodes
 private import semmle.code.cpp.models.interfaces.PointerWrapper
 private import DataFlowPrivate
 private import TypeFlow
@@ -55,26 +56,6 @@ private CppType getThisType(Cpp::MemberFunction f, boolean isGLValue) {
   result.hasType(f.getTypeOfThis(), isGLValue)
 }
 
-/**
- * Gets the C++ type of the instruction `i`.
- *
- * This is equivalent to `i.getResultLanguageType()` with the exception
- * of instructions that directly references a `this` IRVariable. In this
- * case, `i.getResultLanguageType()` gives an unknown type, whereas the
- * predicate gives the expected type (i.e., a potentially cv-qualified
- * type `A*` where `A` is the declaring type of the member function that
- * contains `i`).
- */
-cached
-CppType getResultLanguageType(Instruction i) {
-  if i.(VariableAddressInstruction).getIRVariable() instanceof IRThisVariable
-  then
-    if i.isGLValue()
-    then result = getThisType(i.getEnclosingFunction(), true)
-    else result = getThisType(i.getEnclosingFunction(), false)
-  else result = i.getResultLanguageType()
-}
-
 /**
  * Gets the C++ type of the operand `operand`.
  * This is equivalent to the type of the operand's defining instruction.
@@ -572,6 +553,26 @@ private class BaseCallInstruction extends BaseSourceVariableInstruction, CallIns
 
 cached
 private module Cached {
+  /**
+   * Gets the C++ type of the instruction `i`.
+   *
+   * This is equivalent to `i.getResultLanguageType()` with the exception
+   * of instructions that directly references a `this` IRVariable. In this
+   * case, `i.getResultLanguageType()` gives an unknown type, whereas the
+   * predicate gives the expected type (i.e., a potentially cv-qualified
+   * type `A*` where `A` is the declaring type of the member function that
+   * contains `i`).
+   */
+  cached
+  CppType getResultLanguageType(Instruction i) {
+    if i.(VariableAddressInstruction).getIRVariable() instanceof IRThisVariable
+    then
+      if i.isGLValue()
+      then result = getThisType(i.getEnclosingFunction(), true)
+      else result = getThisType(i.getEnclosingFunction(), false)
+    else result = i.getResultLanguageType()
+  }
+
   /** Holds if `op` is the only use of its defining instruction, and that op is used in a conversation */
   private predicate isConversion(Operand op) {
     exists(Instruction def, Operand use |

cpp/ql/lib/semmle/code/cpp/ir/dataflow/internal/TaintTrackingUtil.qll

@@ -5,65 +5,77 @@ private import semmle.code.cpp.models.interfaces.DataFlow
 private import semmle.code.cpp.models.interfaces.SideEffect
 private import DataFlowUtil
 private import DataFlowPrivate
+private import DataFlowNodes
 private import SsaImpl as Ssa
 private import semmle.code.cpp.dataflow.internal.FlowSummaryImpl as FlowSummaryImpl
 private import semmle.code.cpp.ir.dataflow.FlowSteps
 
-/**
- * Holds if taint propagates from `nodeFrom` to `nodeTo` in exactly one local
- * (intra-procedural) step. This relation is only used for local taint flow
- * (for example `TaintTracking::localTaint(source, sink)`) so it may contain
- * special cases that should only apply to local taint flow.
- */
-predicate localTaintStep(DataFlow::Node nodeFrom, DataFlow::Node nodeTo) {
-  // dataflow step
-  DataFlow::localFlowStep(nodeFrom, nodeTo)
-  or
-  // taint flow step
-  localAdditionalTaintStep(nodeFrom, nodeTo, _)
-  or
-  // models-as-data summarized flow for local data flow (i.e. special case for flow
-  // through calls to modeled functions, without relying on global dataflow to join
-  // the dots).
-  FlowSummaryImpl::Private::Steps::summaryThroughStepTaint(nodeFrom, nodeTo, _)
-}
-
-/**
- * Holds if taint can flow in one local step from `nodeFrom` to `nodeTo` excluding
- * local data flow steps. That is, `nodeFrom` and `nodeTo` are likely to represent
- * different objects.
- */
 cached
-predicate localAdditionalTaintStep(DataFlow::Node nodeFrom, DataFlow::Node nodeTo, string model) {
-  operandToInstructionTaintStep(nodeFrom.asOperand(), nodeTo.asInstruction()) and
-  model = ""
-  or
-  modeledTaintStep(nodeFrom, nodeTo, model)
-  or
-  // Flow from (the indirection of) an operand of a pointer arithmetic instruction to the
-  // indirection of the pointer arithmetic instruction. This provides flow from `source`
-  // in `x[source]` to the result of the associated load instruction.
-  exists(PointerArithmeticInstruction pai, int indirectionIndex |
-    nodeHasOperand(nodeFrom, pai.getAnOperand(), pragma[only_bind_into](indirectionIndex)) and
-    hasInstructionAndIndex(nodeTo, pai, indirectionIndex + 1)
-  ) and
-  model = ""
-  or
-  any(Ssa::Indirection ind).isAdditionalTaintStep(nodeFrom, nodeTo) and
-  model = ""
-  or
-  // models-as-data summarized flow
-  FlowSummaryImpl::Private::Steps::summaryLocalStep(nodeFrom.(FlowSummaryNode).getSummaryNode(),
-    nodeTo.(FlowSummaryNode).getSummaryNode(), false, model)
-  or
-  // object->field conflation for content that is a `TaintInheritingContent`.
-  exists(DataFlow::ContentSet f |
-    readStep(nodeFrom, f, nodeTo) and
-    f.getAReadContent() instanceof TaintInheritingContent
-  ) and
-  model = ""
+private module Cached {
+  private import DataFlowImplCommon as DataFlowImplCommon
+
+  cached
+  predicate forceCachingInSameStage() { DataFlowImplCommon::forceCachingInSameStage() }
+
+  /**
+   * Holds if taint propagates from `nodeFrom` to `nodeTo` in exactly one local
+   * (intra-procedural) step. This relation is only used for local taint flow
+   * (for example `TaintTracking::localTaint(source, sink)`) so it may contain
+   * special cases that should only apply to local taint flow.
+   */
+  cached
+  predicate localTaintStep(DataFlow::Node nodeFrom, DataFlow::Node nodeTo) {
+    // dataflow step
+    DataFlow::localFlowStep(nodeFrom, nodeTo)
+    or
+    // taint flow step
+    localAdditionalTaintStep(nodeFrom, nodeTo, _)
+    or
+    // models-as-data summarized flow for local data flow (i.e. special case for flow
+    // through calls to modeled functions, without relying on global dataflow to join
+    // the dots).
+    FlowSummaryImpl::Private::Steps::summaryThroughStepTaint(nodeFrom, nodeTo, _)
+  }
+
+  /**
+   * Holds if taint can flow in one local step from `nodeFrom` to `nodeTo` excluding
+   * local data flow steps. That is, `nodeFrom` and `nodeTo` are likely to represent
+   * different objects.
+   */
+  cached
+  predicate localAdditionalTaintStep(DataFlow::Node nodeFrom, DataFlow::Node nodeTo, string model) {
+    operandToInstructionTaintStep(nodeFrom.asOperand(), nodeTo.asInstruction()) and
+    model = ""
+    or
+    modeledTaintStep(nodeFrom, nodeTo, model)
+    or
+    // Flow from (the indirection of) an operand of a pointer arithmetic instruction to the
+    // indirection of the pointer arithmetic instruction. This provides flow from `source`
+    // in `x[source]` to the result of the associated load instruction.
+    exists(PointerArithmeticInstruction pai, int indirectionIndex |
+      nodeHasOperand(nodeFrom, pai.getAnOperand(), pragma[only_bind_into](indirectionIndex)) and
+      hasInstructionAndIndex(nodeTo, pai, indirectionIndex + 1)
+    ) and
+    model = ""
+    or
+    any(Ssa::Indirection ind).isAdditionalTaintStep(nodeFrom, nodeTo) and
+    model = ""
+    or
+    // models-as-data summarized flow
+    FlowSummaryImpl::Private::Steps::summaryLocalStep(nodeFrom.(FlowSummaryNode).getSummaryNode(),
+      nodeTo.(FlowSummaryNode).getSummaryNode(), false, model)
+    or
+    // object->field conflation for content that is a `TaintInheritingContent`.
+    exists(DataFlow::ContentSet f |
+      readStep(nodeFrom, f, nodeTo) and
+      f.getAReadContent() instanceof TaintInheritingContent
+    ) and
+    model = ""
+  }
 }
 
+import Cached
+
 /**
  * Holds if taint propagates from `nodeFrom` to `nodeTo` in exactly one local
  * (intra-procedural) step.

cpp/ql/src/Likely Bugs/Memory Management/AllocaInLoop.ql

@@ -15,6 +15,7 @@
 import cpp
 import semmle.code.cpp.rangeanalysis.RangeAnalysisUtils
 import semmle.code.cpp.ir.dataflow.DataFlow
+private import semmle.code.cpp.ir.dataflow.internal.DataFlowNodes
 
 /** Gets a loop that contains `e`. */
 Loop getAnEnclosingLoopOfExpr(Expr e) { result = getAnEnclosingLoopOfStmt(e.getEnclosingStmt()) }

cpp/ql/src/utils/modelgenerator/internal/CaptureModels.qll

@@ -8,6 +8,7 @@ private import semmle.code.cpp.dataflow.ExternalFlow as ExternalFlow
 private import semmle.code.cpp.ir.dataflow.internal.DataFlowImplCommon as DataFlowImplCommon
 private import semmle.code.cpp.ir.dataflow.internal.DataFlowImplSpecific
 private import semmle.code.cpp.ir.dataflow.internal.DataFlowPrivate as DataFlowPrivate
+private import semmle.code.cpp.ir.dataflow.internal.DataFlowNodes as DataFlowNodes
 private import semmle.code.cpp.dataflow.internal.FlowSummaryImpl as FlowSummaryImpl
 private import semmle.code.cpp.ir.dataflow.internal.TaintTrackingImplSpecific
 private import semmle.code.cpp.dataflow.new.TaintTracking as Tt

cpp/ql/test/library-tests/dataflow/models-as-data/FlowSummaryNode.ql

@@ -1,6 +1,7 @@
 import testModels
 private import semmle.code.cpp.ir.dataflow.internal.DataFlowPrivate
 private import semmle.code.cpp.ir.dataflow.internal.DataFlowUtil
+private import semmle.code.cpp.ir.dataflow.internal.DataFlowNodes
 
 string describe(DataFlow::Node n) {
   n instanceof ParameterNode and result = "ParameterNode"