Merge pull request #15371 from geoffw0/mad

C++: Implement models-as-data

cpp/ql/lib/change-notes/2024-10-04-getc.md

@@ -0,0 +1,4 @@
+---
+category: minorAnalysis
+---
+* Source models have been added for the standard library function `getc` (and variations).

cpp/ql/lib/change-notes/2024-10-04-models-as-data.md

@@ -0,0 +1,4 @@
+---
+category: feature
+---
+* Models-as-Data support has been added for C/C++. This feature allows flow sources, sinks and summaries to be expressed in compact strings as an alternative to modelling each source / sink / summary with explicit QL. See `dataflow/ExternalFlow.qll` for documentation and specification of the model format, and `models/implementations/ZMQ.qll` for a simple example of models. Importing models from `.yml` is not yet supported.

cpp/ql/lib/change-notes/2024-10-04-zmq.md

@@ -0,0 +1,4 @@
+---
+category: minorAnalysis
+---
+* Source, sink and flow models for the ZeroMQ (ZMQ) networking library have been added.

cpp/ql/lib/qlpack.yml

@@ -7,6 +7,7 @@ library: true
 upgrades: upgrades
 dependencies:
   codeql/dataflow: ${workspace}
+  codeql/mad: ${workspace}
   codeql/rangeanalysis: ${workspace}
   codeql/ssa: ${workspace}
   codeql/typeflow: ${workspace}

cpp/ql/lib/semmle/code/cpp/dataflow/ExternalFlow.qll

@@ -0,0 +1,556 @@
+/**
+ * INTERNAL use only. This is an experimental API subject to change without notice.
+ *
+ * Provides classes and predicates for dealing with flow models specified in CSV format.
+ *
+ * The CSV specification has the following columns:
+ * - Sources:
+ *   `namespace; type; subtypes; name; signature; ext; output; kind`
+ * - Sinks:
+ *   `namespace; type; subtypes; name; signature; ext; input; kind`
+ * - Summaries:
+ *   `namespace; type; subtypes; name; signature; ext; input; output; kind`
+ *
+ * The interpretation of a row is similar to API-graphs with a left-to-right
+ * reading.
+ * 1. The `namespace` column selects a namespace.
+ * 2. The `type` column selects a type within that namespace.
+ * 3. The `subtypes` is a boolean that indicates whether to jump to an
+ *    arbitrary subtype of that type. Set this to `false` if leaving the `type`
+ *    blank (for example, a free function).
+ * 4. The `name` column optionally selects a specific named member of the type.
+ * 5. The `signature` column optionally restricts the named member. If
+ *    `signature` is blank then no such filtering is done. The format of the
+ *    signature is a comma-separated list of types enclosed in parentheses. The
+ *    types can be short names or fully qualified names (mixing these two options
+ *    is not allowed within a single signature).
+ * 6. The `ext` column specifies additional API-graph-like edges. Currently
+ *    there is only one valid value: "".
+ * 7. The `input` column specifies how data enters the element selected by the
+ *    first 6 columns, and the `output` column specifies how data leaves the
+ *    element selected by the first 6 columns. An `input` can be either:
+ *    - "": Selects a write to the selected element in case this is a field.
+ *    - "Argument[n]": Selects an argument in a call to the selected element.
+ *      The arguments are zero-indexed, and `-1` specifies the qualifier object,
+ *      that is, `*this`.
+ *      - one or more "*" can be added in front of the argument index to indicate
+ *        indirection, for example, `Argument[*0]` indicates the first indirection
+ *        of the 0th argument.
+ *      - `n1..n2` syntax can be used to indicate a range of arguments, inclusive
+ *        at both ends. One or more "*"s can be added in front of the whole range
+ *        to indicate that every argument in the range is indirect, for example
+ *        `*0..1` is the first indirection of both arguments 0 and 1.
+ *    - "ReturnValue": Selects a value being returned by the selected element.
+ *      One or more "*" can be added as an argument to indicate indirection, for
+ *      example, "ReturnValue[*]" indicates the first indirection of the return
+ *      value.
+ *
+ *    An `output` can be either:
+ *    - "": Selects a read of a selected field.
+ *    - "Argument[n]": Selects the post-update value of an argument in a call to
+ *      the selected element. That is, the value of the argument after the call
+ *      returns. The arguments are zero-indexed, and `-1` specifies the qualifier
+ *      object, that is, `*this`.
+ *      - one or more "*" can be added in front of the argument index to indicate
+ *        indirection, for example, `Argument[*0]` indicates the first indirection
+ *        of the 0th argument.
+ *      - `n1..n2` syntax can be used to indicate a range of arguments, inclusive
+ *        at both ends. One or more "*"s can be added in front of the whole range
+ *        to indicate that every argument in the range is indirect, for example
+ *        `*0..1` is the first indirection of both arguments 0 and 1.
+ *    - "Parameter[n]": Selects the value of a parameter of the selected element.
+ *      The syntax is the same as for "Argument", for example "Parameter[0]",
+ *      "Parameter[*0]", "Parameter[0..2]" etc.
+ *    - "ReturnValue": Selects a value being returned by the selected element.
+ *      One or more "*" can be added as an argument to indicate indirection, for
+ *      example, "ReturnValue[*]" indicates the first indirection of the return
+ *      value.
+ * 8. The `kind` column is a tag that can be referenced from QL to determine to
+ *    which classes the interpreted elements should be added. For example, for
+ *    sources "remote" indicates a default remote flow source, and for summaries
+ *    "taint" indicates a default additional taint step and "value" indicates a
+ *    globally applicable value-preserving step.
+ */
+
+import cpp
+private import new.DataFlow
+private import internal.FlowSummaryImpl
+private import internal.FlowSummaryImpl::Public
+private import internal.FlowSummaryImpl::Private
+private import internal.FlowSummaryImpl::Private::External
+private import codeql.mad.ModelValidation as SharedModelVal
+private import codeql.util.Unit
+
+/**
+ * A unit class for adding additional source model rows.
+ *
+ * Extend this class to add additional source definitions.
+ */
+class SourceModelCsv extends Unit {
+  /** Holds if `row` specifies a source definition. */
+  abstract predicate row(string row);
+}
+
+/**
+ * A unit class for adding additional sink model rows.
+ *
+ * Extend this class to add additional sink definitions.
+ */
+class SinkModelCsv extends Unit {
+  /** Holds if `row` specifies a sink definition. */
+  abstract predicate row(string row);
+}
+
+/**
+ * A unit class for adding additional summary model rows.
+ *
+ * Extend this class to add additional flow summary definitions.
+ */
+class SummaryModelCsv extends Unit {
+  /** Holds if `row` specifies a summary definition. */
+  abstract predicate row(string row);
+}
+
+/** Holds if `row` is a source model. */
+predicate sourceModel(string row) { any(SourceModelCsv s).row(row) }
+
+/** Holds if `row` is a sink model. */
+predicate sinkModel(string row) { any(SinkModelCsv s).row(row) }
+
+/** Holds if `row` is a summary model. */
+predicate summaryModel(string row) { any(SummaryModelCsv s).row(row) }
+
+/** Holds if a source model exists for the given parameters. */
+predicate sourceModel(
+  string namespace, string type, boolean subtypes, string name, string signature, string ext,
+  string output, string kind, string provenance
+) {
+  exists(string row |
+    sourceModel(row) and
+    row.splitAt(";", 0) = namespace and
+    row.splitAt(";", 1) = type and
+    row.splitAt(";", 2) = subtypes.toString() and
+    subtypes = [true, false] and
+    row.splitAt(";", 3) = name and
+    row.splitAt(";", 4) = signature and
+    row.splitAt(";", 5) = ext and
+    row.splitAt(";", 6) = output and
+    row.splitAt(";", 7) = kind
+  ) and
+  provenance = "manual"
+}
+
+/** Holds if a sink model exists for the given parameters. */
+predicate sinkModel(
+  string namespace, string type, boolean subtypes, string name, string signature, string ext,
+  string input, string kind, string provenance
+) {
+  exists(string row |
+    sinkModel(row) and
+    row.splitAt(";", 0) = namespace and
+    row.splitAt(";", 1) = type and
+    row.splitAt(";", 2) = subtypes.toString() and
+    subtypes = [true, false] and
+    row.splitAt(";", 3) = name and
+    row.splitAt(";", 4) = signature and
+    row.splitAt(";", 5) = ext and
+    row.splitAt(";", 6) = input and
+    row.splitAt(";", 7) = kind
+  ) and
+  provenance = "manual"
+}
+
+/** Holds if a summary model exists for the given parameters. */
+predicate summaryModel(
+  string namespace, string type, boolean subtypes, string name, string signature, string ext,
+  string input, string output, string kind, string provenance
+) {
+  exists(string row |
+    summaryModel(row) and
+    row.splitAt(";", 0) = namespace and
+    row.splitAt(";", 1) = type and
+    row.splitAt(";", 2) = subtypes.toString() and
+    subtypes = [true, false] and
+    row.splitAt(";", 3) = name and
+    row.splitAt(";", 4) = signature and
+    row.splitAt(";", 5) = ext and
+    row.splitAt(";", 6) = input and
+    row.splitAt(";", 7) = output and
+    row.splitAt(";", 8) = kind
+  ) and
+  provenance = "manual"
+}
+
+private predicate relevantNamespace(string namespace) {
+  sourceModel(namespace, _, _, _, _, _, _, _, _) or
+  sinkModel(namespace, _, _, _, _, _, _, _, _) or
+  summaryModel(namespace, _, _, _, _, _, _, _, _, _)
+}
+
+private predicate namespaceLink(string shortns, string longns) {
+  relevantNamespace(shortns) and
+  relevantNamespace(longns) and
+  longns.prefix(longns.indexOf("::")) = shortns
+}
+
+private predicate canonicalNamespace(string namespace) {
+  relevantNamespace(namespace) and not namespaceLink(_, namespace)
+}
+
+private predicate canonicalNamespaceLink(string namespace, string subns) {
+  canonicalNamespace(namespace) and
+  (subns = namespace or namespaceLink(namespace, subns))
+}
+
+/**
+ * Holds if CSV framework coverage of `namespace` is `n` api endpoints of the
+ * kind `(kind, part)`.
+ */
+predicate modelCoverage(string namespace, int namespaces, string kind, string part, int n) {
+  namespaces = strictcount(string subns | canonicalNamespaceLink(namespace, subns)) and
+  (
+    part = "source" and
+    n =
+      strictcount(string subns, string type, boolean subtypes, string name, string signature,
+        string ext, string output, string provenance |
+        canonicalNamespaceLink(namespace, subns) and
+        sourceModel(subns, type, subtypes, name, signature, ext, output, kind, provenance)
+      )
+    or
+    part = "sink" and
+    n =
+      strictcount(string subns, string type, boolean subtypes, string name, string signature,
+        string ext, string input, string provenance |
+        canonicalNamespaceLink(namespace, subns) and
+        sinkModel(subns, type, subtypes, name, signature, ext, input, kind, provenance)
+      )
+    or
+    part = "summary" and
+    n =
+      strictcount(string subns, string type, boolean subtypes, string name, string signature,
+        string ext, string input, string output, string provenance |
+        canonicalNamespaceLink(namespace, subns) and
+        summaryModel(subns, type, subtypes, name, signature, ext, input, output, kind, provenance)
+      )
+  )
+}
+
+/** Provides a query predicate to check the CSV data for validation errors. */
+module CsvValidation {
+  private string getInvalidModelInput() {
+    exists(string pred, AccessPath input, string part |
+      sinkModel(_, _, _, _, _, _, input, _, _) and pred = "sink"
+      or
+      summaryModel(_, _, _, _, _, _, input, _, _, _) and pred = "summary"
+    |
+      (
+        invalidSpecComponent(input, part) and
+        not part = "" and
+        not (part = "Argument" and pred = "sink") and
+        not parseArg(part, _)
+        or
+        part = input.getToken(_) and
+        parseParam(part, _)
+      ) and
+      result = "Unrecognized input specification \"" + part + "\" in " + pred + " model."
+    )
+  }
+
+  private string getInvalidModelOutput() {
+    exists(string pred, string output, string part |
+      sourceModel(_, _, _, _, _, _, output, _, _) and pred = "source"
+      or
+      summaryModel(_, _, _, _, _, _, _, output, _, _) and pred = "summary"
+    |
+      invalidSpecComponent(output, part) and
+      not part = "" and
+      not (part = ["Argument", "Parameter"] and pred = "source") and
+      result = "Unrecognized output specification \"" + part + "\" in " + pred + " model."
+    )
+  }
+
+  private module KindValConfig implements SharedModelVal::KindValidationConfigSig {
+    predicate summaryKind(string kind) { summaryModel(_, _, _, _, _, _, _, _, kind, _) }
+
+    predicate sinkKind(string kind) { sinkModel(_, _, _, _, _, _, _, kind, _) }
+
+    predicate sourceKind(string kind) { sourceModel(_, _, _, _, _, _, _, kind, _) }
+  }
+
+  private module KindVal = SharedModelVal::KindValidation<KindValConfig>;
+
+  private string getInvalidModelSubtype() {
+    exists(string pred, string row |
+      sourceModel(row) and pred = "source"
+      or
+      sinkModel(row) and pred = "sink"
+      or
+      summaryModel(row) and pred = "summary"
+    |
+      exists(string b |
+        b = row.splitAt(";", 2) and
+        not b = ["true", "false"] and
+        result = "Invalid boolean \"" + b + "\" in " + pred + " model."
+      )
+    )
+  }
+
+  private string getInvalidModelColumnCount() {
+    exists(string pred, string row, int expect |
+      sourceModel(row) and expect = 8 and pred = "source"
+      or
+      sinkModel(row) and expect = 8 and pred = "sink"
+      or
+      summaryModel(row) and expect = 9 and pred = "summary"
+    |
+      exists(int cols |
+        cols = 1 + max(int n | exists(row.splitAt(";", n))) and
+        cols != expect and
+        result =
+          "Wrong number of columns in " + pred + " model row, expected " + expect + ", got " + cols +
+            "."
+      )
+    )
+  }
+
+  private string getInvalidModelSignature() {
+    exists(string pred, string namespace, string type, string name, string signature, string ext |
+      sourceModel(namespace, type, _, name, signature, ext, _, _, _) and pred = "source"
+      or
+      sinkModel(namespace, type, _, name, signature, ext, _, _, _) and pred = "sink"
+      or
+      summaryModel(namespace, type, _, name, signature, ext, _, _, _, _) and pred = "summary"
+    |
+      not namespace.regexpMatch("[a-zA-Z0-9_\\.]+") and
+      result = "Dubious namespace \"" + namespace + "\" in " + pred + " model."
+      or
+      not type.regexpMatch("[a-zA-Z0-9_<>,\\+]+") and
+      result = "Dubious type \"" + type + "\" in " + pred + " model."
+      or
+      not name.regexpMatch("[a-zA-Z0-9_<>,]*") and
+      result = "Dubious member name \"" + name + "\" in " + pred + " model."
+      or
+      not signature.regexpMatch("|\\([a-zA-Z0-9_<>\\.\\+\\*,\\[\\]]*\\)") and
+      result = "Dubious signature \"" + signature + "\" in " + pred + " model."
+      or
+      not ext.regexpMatch("|Attribute") and
+      result = "Unrecognized extra API graph element \"" + ext + "\" in " + pred + " model."
+    )
+  }
+
+  /** Holds if some row in a CSV-based flow model appears to contain typos. */
+  query predicate invalidModelRow(string msg) {
+    msg =
+      [
+        getInvalidModelSignature(), getInvalidModelInput(), getInvalidModelOutput(),
+        getInvalidModelSubtype(), getInvalidModelColumnCount(), KindVal::getInvalidModelKind()
+      ]
+  }
+}
+
+private predicate elementSpec(
+  string namespace, string type, boolean subtypes, string name, string signature, string ext
+) {
+  sourceModel(namespace, type, subtypes, name, signature, ext, _, _, _) or
+  sinkModel(namespace, type, subtypes, name, signature, ext, _, _, _) or
+  summaryModel(namespace, type, subtypes, name, signature, ext, _, _, _, _)
+}
+
+private string paramsStringPart(Function c, int i) {
+  i = -1 and result = "(" and exists(c)
+  or
+  exists(int n, string p | c.getParameter(n).getType().toString() = p |
+    i = 2 * n and result = p
+    or
+    i = 2 * n - 1 and result = "," and n != 0
+  )
+  or
+  i = 2 * c.getNumberOfParameters() and result = ")"
+}
+
+/**
+ * Gets a parenthesized string containing all parameter types of this callable, separated by a comma.
+ *
+ * Returns the empty string if the callable has no parameters.
+ * Parameter types are represented by their type erasure.
+ */
+cached
+private string paramsString(Function c) {
+  result = concat(int i | | paramsStringPart(c, i) order by i)
+}
+
+bindingset[func]
+private predicate matchesSignature(Function func, string signature) {
+  signature = "" or
+  paramsString(func) = signature
+}
+
+/**
+ * Gets the element in module `namespace` that satisfies the following properties:
+ * 1. If the element is a member of a class-like type, then the class-like type has name `type`
+ * 2. If `subtypes = true` and the element is a member of a class-like type, then overrides of the element
+ *    are also returned.
+ * 3. The element has name `name`
+ * 4. If `signature` is non-empty, then the element has a list of parameter types described by `signature`.
+ *
+ * NOTE: `namespace` is currently not used (since we don't properly extract modules yet).
+ */
+pragma[nomagic]
+private Element interpretElement0(
+  string namespace, string type, boolean subtypes, string name, string signature
+) {
+  elementSpec(namespace, type, subtypes, name, signature, _) and
+  (
+    // Non-member functions
+    exists(Function func |
+      func.hasQualifiedName(namespace, name) and
+      type = "" and
+      matchesSignature(func, signature) and
+      subtypes = false and
+      not exists(func.getDeclaringType()) and
+      result = func
+    )
+    or
+    // Member functions
+    exists(Class namedClass, Class classWithMethod, Function method |
+      classWithMethod = method.getClassAndName(name) and
+      namedClass.hasQualifiedName(namespace, type) and
+      matchesSignature(method, signature) and
+      result = method
+    |
+      // member declared in the named type or a subtype of it
+      subtypes = true and
+      classWithMethod = namedClass.getADerivedClass*()
+      or
+      // member declared directly in the named type
+      subtypes = false and
+      classWithMethod = namedClass
+    )
+    or
+    // Member variables
+    signature = "" and
+    exists(Class namedClass, Class classWithMember, MemberVariable member |
+      member.getName() = name and
+      member = classWithMember.getAMember() and
+      namedClass.hasQualifiedName(namespace, type) and
+      result = member
+    |
+      // field declared in the named type or a subtype of it (or an extension of any)
+      subtypes = true and
+      classWithMember = namedClass.getADerivedClass*()
+      or
+      // field declared directly in the named type (or an extension of it)
+      subtypes = false and
+      classWithMember = namedClass
+    )
+    or
+    // Global or namespace variables
+    signature = "" and
+    type = "" and
+    subtypes = false and
+    result = any(GlobalOrNamespaceVariable v | v.hasQualifiedName(namespace, name))
+  )
+}
+
+/** Gets the source/sink/summary element corresponding to the supplied parameters. */
+Element interpretElement(
+  string namespace, string type, boolean subtypes, string name, string signature, string ext
+) {
+  elementSpec(namespace, type, subtypes, name, signature, ext) and
+  exists(Element e | e = interpretElement0(namespace, type, subtypes, name, signature) |
+    ext = "" and result = e
+  )
+}
+
+cached
+private module Cached {
+  /**
+   * Holds if `node` is specified as a source with the given kind in a CSV flow
+   * model.
+   */
+  cached
+  predicate sourceNode(DataFlow::Node node, string kind) {
+    exists(SourceSinkInterpretationInput::InterpretNode n |
+      isSourceNode(n, kind, _) and n.asNode() = node // TODO
+    )
+  }
+
+  /**
+   * Holds if `node` is specified as a sink with the given kind in a CSV flow
+   * model.
+   */
+  cached
+  predicate sinkNode(DataFlow::Node node, string kind) {
+    exists(SourceSinkInterpretationInput::InterpretNode n |
+      isSinkNode(n, kind, _) and n.asNode() = node // TODO
+    )
+  }
+}
+
+import Cached
+
+private predicate interpretSummary(
+  Function f, string input, string output, string kind, string provenance
+) {
+  exists(
+    string namespace, string type, boolean subtypes, string name, string signature, string ext
+  |
+    summaryModel(namespace, type, subtypes, name, signature, ext, input, output, kind, provenance) and
+    f = interpretElement(namespace, type, subtypes, name, signature, ext)
+  )
+}
+
+// adapter class for converting Mad summaries to `SummarizedCallable`s
+private class SummarizedCallableAdapter extends SummarizedCallable {
+  SummarizedCallableAdapter() { interpretSummary(this, _, _, _, _) }
+
+  private predicate relevantSummaryElementManual(string input, string output, string kind) {
+    exists(Provenance provenance |
+      interpretSummary(this, input, output, kind, provenance) and
+      provenance.isManual()
+    )
+  }
+
+  private predicate relevantSummaryElementGenerated(string input, string output, string kind) {
+    exists(Provenance provenance |
+      interpretSummary(this, input, output, kind, provenance) and
+      provenance.isGenerated()
+    )
+  }
+
+  override predicate propagatesFlow(
+    string input, string output, boolean preservesValue, string model
+  ) {
+    exists(string kind |
+      this.relevantSummaryElementManual(input, output, kind)
+      or
+      not this.relevantSummaryElementManual(_, _, _) and
+      this.relevantSummaryElementGenerated(input, output, kind)
+    |
+      if kind = "value" then preservesValue = true else preservesValue = false
+    ) and
+    model = "" // TODO
+  }
+
+  override predicate hasProvenance(Provenance provenance) {
+    interpretSummary(this, _, _, _, provenance)
+  }
+}
+
+// adapter class for converting Mad neutrals to `NeutralCallable`s
+private class NeutralCallableAdapter extends NeutralCallable {
+  string kind;
+  string provenance_;
+
+  NeutralCallableAdapter() {
+    // Neutral models have not been implemented for CPP.
+    none() and
+    exists(this) and
+    exists(kind) and
+    exists(provenance_)
+  }
+
+  override string getKind() { result = kind }
+
+  override predicate hasProvenance(Provenance provenance) { provenance = provenance_ }
+}

cpp/ql/lib/semmle/code/cpp/dataflow/internal/FlowSummaryImpl.qll

@@ -0,0 +1,271 @@
+/**
+ * Provides classes and predicates for defining flow summaries.
+ */
+
+private import cpp as Cpp
+private import codeql.dataflow.internal.FlowSummaryImpl
+private import codeql.dataflow.internal.AccessPathSyntax as AccessPath
+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.DataFlowImplSpecific as DataFlowImplSpecific
+private import semmle.code.cpp.dataflow.ExternalFlow
+private import semmle.code.cpp.ir.IR
+
+module Input implements InputSig<Location, DataFlowImplSpecific::CppDataFlow> {
+  class SummarizedCallableBase = Function;
+
+  ArgumentPosition callbackSelfParameterPosition() { result = TDirectPosition(-1) }
+
+  ReturnKind getStandardReturnValueKind() { result.(NormalReturnKind).getIndirectionIndex() = 0 }
+
+  string encodeParameterPosition(ParameterPosition pos) { result = pos.toString() }
+
+  string encodeArgumentPosition(ArgumentPosition pos) { result = pos.toString() }
+
+  string encodeReturn(ReturnKind rk, string arg) {
+    rk != getStandardReturnValueKind() and
+    result = "ReturnValue" and
+    arg = repeatStars(rk.(NormalReturnKind).getIndirectionIndex())
+  }
+
+  string encodeContent(ContentSet cs, string arg) {
+    exists(FieldContent c |
+      cs.isSingleton(c) and
+      // FieldContent indices have 0 for the address, 1 for content, so we need to subtract one.
+      result = "Field" and
+      arg = repeatStars(c.getIndirectionIndex() - 1) + c.getField().getName()
+    )
+  }
+
+  string encodeWithoutContent(ContentSet c, string arg) {
+    // used for type tracking, not currently used in C/C++.
+    result = "WithoutContent" + c and arg = ""
+  }
+
+  string encodeWithContent(ContentSet c, string arg) {
+    // used for type tracking, not currently used in C/C++.
+    result = "WithContent" + c and arg = ""
+  }
+
+  /**
+   * Decodes an argument / parameter position string, for example the `0` in `Argument[0]`.
+   * Supports ranges (`Argument[x..y]`), qualifiers (`Argument[-1]`), indirections
+   * (`Argument[*x]`) and combinations (such as `Argument[**0..1]`).
+   */
+  bindingset[argString]
+  private TPosition decodePosition(string argString) {
+    exists(int indirection, string posString, int pos |
+      argString = repeatStars(indirection) + posString and
+      pos = AccessPath::parseInt(posString) and
+      (
+        pos >= 0 and indirection = 0 and result = TDirectPosition(pos)
+        or
+        pos >= 0 and indirection > 0 and result = TIndirectionPosition(pos, indirection)
+        or
+        // `Argument[-1]` / `Parameter[-1]` is the qualifier object `*this`, not the `this` pointer itself.
+        pos = -1 and result = TIndirectionPosition(pos, indirection + 1)
+      )
+    )
+  }
+
+  bindingset[token]
+  ParameterPosition decodeUnknownParameterPosition(AccessPath::AccessPathTokenBase token) {
+    token.getName() = "Argument" and
+    result = decodePosition(token.getAnArgument())
+  }
+
+  bindingset[token]
+  ArgumentPosition decodeUnknownArgumentPosition(AccessPath::AccessPathTokenBase token) {
+    token.getName() = "Parameter" and
+    result = decodePosition(token.getAnArgument())
+  }
+
+  bindingset[token]
+  ContentSet decodeUnknownContent(AccessPath::AccessPathTokenBase token) {
+    // field content (no indirection support)
+    exists(FieldContent c |
+      result.isSingleton(c) and
+      token.getName() = c.getField().getName() and
+      not exists(token.getArgumentList()) and
+      c.getIndirectionIndex() = 1
+    )
+    or
+    // field content (with indirection support)
+    exists(FieldContent c |
+      result.isSingleton(c) and
+      token.getName() = c.getField().getName() and
+      // FieldContent indices have 0 for the address, 1 for content, so we need to subtract one.
+      token.getAnArgument() = repeatStars(c.getIndirectionIndex() - 1)
+    )
+  }
+}
+
+private import Make<Location, DataFlowImplSpecific::CppDataFlow, Input> as Impl
+
+private module StepsInput implements Impl::Private::StepsInputSig {
+  DataFlowCall getACall(Public::SummarizedCallable sc) {
+    result.getStaticCallTarget().getUnderlyingCallable() = sc
+  }
+}
+
+module SourceSinkInterpretationInput implements
+  Impl::Private::External::SourceSinkInterpretationInputSig
+{
+  class Element = Cpp::Element;
+
+  class SourceOrSinkElement = Element;
+
+  /**
+   * Holds if an external source specification exists for `e` with output specification
+   * `output`, kind `kind`, and provenance `provenance`.
+   */
+  predicate sourceElement(
+    SourceOrSinkElement e, string output, string kind, Public::Provenance provenance, string model
+  ) {
+    exists(
+      string namespace, string type, boolean subtypes, string name, string signature, string ext
+    |
+      sourceModel(namespace, type, subtypes, name, signature, ext, output, kind, provenance) and
+      e = interpretElement(namespace, type, subtypes, name, signature, ext) and
+      model = "" // TODO
+    )
+  }
+
+  /**
+   * Holds if an external sink specification exists for `e` with input specification
+   * `input`, kind `kind` and provenance `provenance`.
+   */
+  predicate sinkElement(
+    SourceOrSinkElement e, string input, string kind, Public::Provenance provenance, string model
+  ) {
+    exists(
+      string package, string type, boolean subtypes, string name, string signature, string ext
+    |
+      sinkModel(package, type, subtypes, name, signature, ext, input, kind, provenance) and
+      e = interpretElement(package, type, subtypes, name, signature, ext) and
+      model = "" // TODO
+    )
+  }
+
+  private newtype TInterpretNode =
+    TElement_(Element n) or
+    TNode_(Node n)
+
+  /** An entity used to interpret a source/sink specification. */
+  class InterpretNode extends TInterpretNode {
+    /** Gets the element that this node corresponds to, if any. */
+    SourceOrSinkElement asElement() { this = TElement_(result) }
+
+    /** Gets the data-flow node that this node corresponds to, if any. */
+    Node asNode() { this = TNode_(result) }
+
+    /** Gets the call that this node corresponds to, if any. */
+    DataFlowCall asCall() {
+      this.asElement() = result.asCallInstruction().getUnconvertedResultExpression()
+    }
+
+    /** Gets the callable that this node corresponds to, if any. */
+    DataFlowCallable asCallable() { result.getUnderlyingCallable() = this.asElement() }
+
+    /** Gets the target of this call, if any. */
+    Element getCallTarget() { result = this.asCall().getStaticCallTarget().getUnderlyingCallable() }
+
+    /** Gets a textual representation of this node. */
+    string toString() {
+      result = this.asElement().toString()
+      or
+      result = this.asNode().toString()
+      or
+      result = this.asCall().toString()
+    }
+
+    /** Gets the location of this node. */
+    Location getLocation() {
+      result = this.asElement().getLocation()
+      or
+      result = this.asNode().getLocation()
+      or
+      result = this.asCall().getLocation()
+    }
+  }
+
+  /** Provides additional sink specification logic. */
+  bindingset[c]
+  predicate interpretOutput(string c, InterpretNode mid, InterpretNode node) {
+    // Allow variables to be picked as output nodes.
+    exists(Node n, Element ast |
+      n = node.asNode() and
+      ast = mid.asElement()
+    |
+      c = "" and
+      n.asExpr().(VariableAccess).getTarget() = ast
+    )
+  }
+
+  /** Provides additional source specification logic. */
+  bindingset[c]
+  predicate interpretInput(string c, InterpretNode mid, InterpretNode node) {
+    exists(Node n, Element ast, VariableAccess e |
+      n = node.asNode() and
+      ast = mid.asElement() and
+      e.getTarget() = ast
+    |
+      // Allow variables to be picked as input nodes.
+      // We could simply do this as `e = n.asExpr()`, but that would not allow
+      // us to pick `x` as a sink in an example such as `x = source()` (but
+      // only subsequent uses of `x`) since the variable access on `x` doesn't
+      // actually load the value of `x`. So instead, we pick the instruction
+      // node corresponding to the generated `StoreInstruction` and use the
+      // expression associated with the destination instruction. This means
+      // that the `x` in `x = source()` can be marked as an input.
+      c = "" and
+      exists(StoreInstruction store |
+        store.getDestinationAddress().getUnconvertedResultExpression() = e and
+        n.asInstruction() = store
+      )
+    )
+  }
+}
+
+module Private {
+  import Impl::Private
+
+  module Steps = Impl::Private::Steps<StepsInput>;
+
+  module External {
+    import Impl::Private::External
+    import Impl::Private::External::SourceSinkInterpretation<SourceSinkInterpretationInput>
+  }
+
+  /**
+   * Provides predicates for constructing summary components.
+   */
+  module SummaryComponent {
+    private import Impl::Private::SummaryComponent as SC
+
+    predicate parameter = SC::parameter/1;
+
+    predicate argument = SC::argument/1;
+
+    predicate content = SC::content/1;
+
+    predicate withoutContent = SC::withoutContent/1;
+
+    predicate withContent = SC::withContent/1;
+  }
+
+  /**
+   * Provides predicates for constructing stacks of summary components.
+   */
+  module SummaryComponentStack {
+    private import Impl::Private::SummaryComponentStack as SCS
+
+    predicate singleton = SCS::singleton/1;
+
+    predicate push = SCS::push/2;
+
+    predicate argument = SCS::argument/1;
+  }
+}
+
+module Public = Impl::Public;

cpp/ql/lib/semmle/code/cpp/ir/dataflow/ResolveCall.qll

@@ -7,6 +7,7 @@ import cpp
 private import semmle.code.cpp.ir.ValueNumbering
 private import internal.DataFlowDispatch
 private import semmle.code.cpp.ir.IR
+private import semmle.code.cpp.ir.dataflow.internal.DataFlowPrivate
 
 /**
  * Resolve potential target function(s) for `call`.
@@ -16,8 +17,9 @@ private import semmle.code.cpp.ir.IR
  * to identify the possible target(s).
  */
 Function resolveCall(Call call) {
-  exists(CallInstruction callInstruction |
+  exists(DataFlowCall dataFlowCall, CallInstruction callInstruction |
     callInstruction.getAst() = call and
-    result = viableCallable(callInstruction)
+    callInstruction = dataFlowCall.asCallInstruction() and
+    result = viableCallable(dataFlowCall).getUnderlyingCallable()
   )
 }

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

@@ -23,13 +23,13 @@ DataFlowCallable defaultViableCallable(DataFlowCall call) {
   // function with the right signature is present in the database, we return
   // that as a potential callee.
   exists(string qualifiedName, int nparams |
-    callSignatureWithoutBody(qualifiedName, nparams, call) and
-    functionSignatureWithBody(qualifiedName, nparams, result) and
+    callSignatureWithoutBody(qualifiedName, nparams, call.asCallInstruction()) and
+    functionSignatureWithBody(qualifiedName, nparams, result.getUnderlyingCallable()) and
     strictcount(Function other | functionSignatureWithBody(qualifiedName, nparams, other)) = 1
   )
   or
   // Virtual dispatch
-  result = call.(VirtualDispatch::DataSensitiveCall).resolve()
+  result.asSourceCallable() = call.(VirtualDispatch::DataSensitiveCall).resolve()
 }
 
 /**
@@ -40,7 +40,9 @@ DataFlowCallable viableCallable(DataFlowCall call) {
   result = defaultViableCallable(call)
   or
   // Additional call targets
-  result = any(AdditionalCallTarget additional).viableTarget(call.getUnconvertedResultExpression())
+  result.getUnderlyingCallable() =
+    any(AdditionalCallTarget additional)
+        .viableTarget(call.asCallInstruction().getUnconvertedResultExpression())
 }
 
 /**
@@ -150,7 +152,7 @@ private module VirtualDispatch {
     ReturnNode node, ReturnKind kind, DataFlowCallable callable
   ) {
     node.getKind() = kind and
-    node.getEnclosingCallable() = callable
+    node.getEnclosingCallable() = callable.getUnderlyingCallable()
   }
 
   /** Call through a function pointer. */
@@ -176,10 +178,15 @@ private module VirtualDispatch {
   /** Call to a virtual function. */
   private class DataSensitiveOverriddenFunctionCall extends DataSensitiveCall {
     DataSensitiveOverriddenFunctionCall() {
-      exists(this.getStaticCallTarget().(VirtualFunction).getAnOverridingFunction())
+      exists(
+        this.getStaticCallTarget()
+            .getUnderlyingCallable()
+            .(VirtualFunction)
+            .getAnOverridingFunction()
+      )
     }
 
-    override DataFlow::Node getDispatchValue() { result.asInstruction() = this.getThisArgument() }
+    override DataFlow::Node getDispatchValue() { result.asInstruction() = this.getArgument(-1) }
 
     override MemberFunction resolve() {
       exists(Class overridingClass |
@@ -194,7 +201,8 @@ private module VirtualDispatch {
      */
     pragma[noinline]
     private predicate overrideMayAffectCall(Class overridingClass, MemberFunction overridingFunction) {
-      overridingFunction.getAnOverriddenFunction+() = this.getStaticCallTarget().(VirtualFunction) and
+      overridingFunction.getAnOverriddenFunction+() =
+        this.getStaticCallTarget().getUnderlyingCallable().(VirtualFunction) and
       overridingFunction.getDeclaringType() = overridingClass
     }
 
@@ -256,12 +264,12 @@ predicate mayBenefitFromCallContext(DataFlowCall call) { mayBenefitFromCallConte
  * value is given as the `arg`'th argument to `f`.
  */
 private predicate mayBenefitFromCallContext(
-  VirtualDispatch::DataSensitiveCall call, Function f, int arg
+  VirtualDispatch::DataSensitiveCall call, DataFlowCallable f, int arg
 ) {
   f = pragma[only_bind_out](call).getEnclosingCallable() and
   exists(InitializeParameterInstruction init |
     not exists(call.getStaticCallTarget()) and
-    init.getEnclosingFunction() = f and
+    init.getEnclosingFunction() = f.getUnderlyingCallable() and
     call.flowsFrom(DataFlow::instructionNode(init), _) and
     init.getParameter().getIndex() = arg
   )
@@ -273,10 +281,11 @@ private predicate mayBenefitFromCallContext(
  */
 DataFlowCallable viableImplInCallContext(DataFlowCall call, DataFlowCall ctx) {
   result = viableCallable(call) and
-  exists(int i, Function f |
+  exists(int i, DataFlowCallable f |
     mayBenefitFromCallContext(pragma[only_bind_into](call), f, i) and
     f = ctx.getStaticCallTarget() and
-    result = ctx.getArgument(i).getUnconvertedResultExpression().(FunctionAccess).getTarget()
+    result.asSourceCallable() =
+      ctx.getArgument(i).getUnconvertedResultExpression().(FunctionAccess).getTarget()
   )
 }
 

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

@@ -3,6 +3,7 @@ private import DataFlowUtil
 private import semmle.code.cpp.ir.IR
 private import DataFlowDispatch
 private import semmle.code.cpp.ir.internal.IRCppLanguage
+private import semmle.code.cpp.dataflow.internal.FlowSummaryImpl as FlowSummaryImpl
 private import SsaInternals as Ssa
 private import DataFlowImplCommon as DataFlowImplCommon
 private import codeql.util.Unit
@@ -79,15 +80,6 @@ module NodeStars {
     result = n.(FinalParameterNode).getIndirectionIndex()
   }
 
-  private int maxNumberOfIndirections() { result = max(getNumberOfIndirections(_)) }
-
-  private string repeatStars(int n) {
-    n = 0 and result = ""
-    or
-    n = [1 .. maxNumberOfIndirections()] and
-    result = "*" + repeatStars(n - 1)
-  }
-
   /**
    * Gets the number of stars (i.e., `*`s) needed to produce the `toString`
    * output for `n`.
@@ -97,6 +89,11 @@ module NodeStars {
 
 import NodeStars
 
+/**
+ * A cut-down `DataFlow::Node` class that does not depend on the output of SSA.
+ * This can thus be safely used in the SSA computations themselves, as well as
+ * in construction of other node classes (`TIRDataFlowNode`).
+ */
 class Node0Impl extends TIRDataFlowNode0 {
   /**
    * INTERNAL: Do not use.
@@ -333,7 +330,9 @@ private module IndirectInstructions {
 import IndirectInstructions
 
 /** Gets the callable in which this node occurs. */
-DataFlowCallable nodeGetEnclosingCallable(Node n) { result = n.getEnclosingCallable() }
+DataFlowCallable nodeGetEnclosingCallable(Node n) {
+  result.getUnderlyingCallable() = n.getEnclosingCallable()
+}
 
 /** Holds if `p` is a `ParameterNode` of `c` with position `pos`. */
 predicate isParameterNode(ParameterNode p, DataFlowCallable c, ParameterPosition pos) {
@@ -379,12 +378,30 @@ private class SideEffectArgumentNode extends ArgumentNode, SideEffectOperandNode
   override predicate argumentOf(DataFlowCall dfCall, ArgumentPosition pos) {
     exists(int indirectionIndex |
       pos = TIndirectionPosition(argumentIndex, pragma[only_bind_into](indirectionIndex)) and
-      this.getCallInstruction() = dfCall and
+      this.getCallInstruction() = dfCall.asCallInstruction() and
       super.hasAddressOperandAndIndirectionIndex(_, pragma[only_bind_into](indirectionIndex))
     )
   }
 }
 
+/**
+ * An argument node that is part of a summary. These only occur when the
+ * summary contains a synthesized call.
+ */
+class SummaryArgumentNode extends ArgumentNode, FlowSummaryNode {
+  private SummaryCall call_;
+  private ArgumentPosition pos_;
+
+  SummaryArgumentNode() {
+    FlowSummaryImpl::Private::summaryArgumentNode(call_.getReceiver(), this.getSummaryNode(), pos_)
+  }
+
+  override predicate argumentOf(DataFlowCall call, ArgumentPosition pos) {
+    call = call_ and
+    pos = pos_
+  }
+}
+
 /** A parameter position represented by an integer. */
 class ParameterPosition = Position;
 
@@ -432,24 +449,42 @@ class IndirectionPosition extends Position, TIndirectionPosition {
 }
 
 newtype TPosition =
-  TDirectPosition(int index) { exists(any(CallInstruction c).getArgument(index)) } or
+  TDirectPosition(int argumentIndex) { exists(any(CallInstruction c).getArgument(argumentIndex)) } or
   TIndirectionPosition(int argumentIndex, int indirectionIndex) {
-    hasOperandAndIndex(_, any(CallInstruction call).getArgumentOperand(argumentIndex),
+    Ssa::hasIndirectOperand(any(CallInstruction call).getArgumentOperand(argumentIndex),
       indirectionIndex)
   }
 
 private newtype TReturnKind =
-  TNormalReturnKind(int index) {
-    exists(IndirectReturnNode return |
-      return.isNormalReturn() and
-      index = return.getIndirectionIndex() - 1 // We subtract one because the return loads the value.
-    )
+  TNormalReturnKind(int indirectionIndex) {
+    // derive a possible return indirection from SSA
+    // (this is a more durable approach if SSA infers additional indirections for any reason)
+    Ssa::hasIndirectOperand(any(ReturnValueInstruction ret).getReturnAddressOperand(),
+      indirectionIndex + 1) // We subtract one because the return loads the value.
+    or
+    // derive a possible return kind from the AST
+    // (this approach includes functions declared that have no body; they may still have flow summaries)
+    indirectionIndex =
+      [0 .. max(Cpp::Function f |
+          not exists(f.getBlock())
+        |
+          Ssa::getMaxIndirectionsForType(f.getUnspecifiedType()) - 1 // -1 because a returned value is a prvalue not a glvalue
+        )]
   } or
   TIndirectReturnKind(int argumentIndex, int indirectionIndex) {
-    exists(IndirectReturnNode return |
-      return.isParameterReturn(argumentIndex) and
-      indirectionIndex = return.getIndirectionIndex()
+    // derive a possible return argument from SSA
+    exists(Ssa::FinalParameterUse use |
+      use.getIndirectionIndex() = indirectionIndex and
+      use.getArgumentIndex() = argumentIndex
     )
+    or
+    // derive a possible return argument from the AST
+    indirectionIndex =
+      [0 .. max(Cpp::Function f |
+          not exists(f.getBlock())
+        |
+          Ssa::getMaxIndirectionsForType(f.getParameter(argumentIndex).getUnspecifiedType()) - 1 // -1 because an argument is a prvalue not a glvalue
+        )]
   }
 
 /**
@@ -457,29 +492,44 @@ private newtype TReturnKind =
  * from a callable. For C++, this is simply a function return.
  */
 class ReturnKind extends TReturnKind {
+  /**
+   * Gets the indirection index of this return kind.
+   */
+  abstract int getIndirectionIndex();
+
   /** Gets a textual representation of this return kind. */
   abstract string toString();
 }
 
-private class NormalReturnKind extends ReturnKind, TNormalReturnKind {
-  int index;
+/**
+ * A value returned from a callable using a `return` statement, that is, a "normal" return.
+ */
+class NormalReturnKind extends ReturnKind, TNormalReturnKind {
+  int indirectionIndex;
 
-  NormalReturnKind() { this = TNormalReturnKind(index) }
+  NormalReturnKind() { this = TNormalReturnKind(indirectionIndex) }
+
+  override int getIndirectionIndex() { result = indirectionIndex }
 
   override string toString() { result = "indirect return" }
 }
 
+/**
+ * A value returned from a callable through a parameter.
+ */
 private class IndirectReturnKind extends ReturnKind, TIndirectReturnKind {
   int argumentIndex;
   int indirectionIndex;
 
   IndirectReturnKind() { this = TIndirectReturnKind(argumentIndex, indirectionIndex) }
 
+  override int getIndirectionIndex() { result = indirectionIndex }
+
   override string toString() { result = "indirect outparam[" + argumentIndex.toString() + "]" }
 }
 
 /** A data flow node that occurs as the result of a `ReturnStmt`. */
-class ReturnNode extends Node instanceof IndirectReturnNode {
+abstract class ReturnNode extends Node {
   /** Gets the kind of this returned value. */
   abstract ReturnKind getKind();
 }
@@ -510,6 +560,17 @@ class ReturnIndirectionNode extends IndirectReturnNode, ReturnNode {
   }
 }
 
+/**
+ * A return node that is part of a summary.
+ */
+private class SummaryReturnNode extends ReturnNode, FlowSummaryNode {
+  private ReturnKind rk;
+
+  SummaryReturnNode() { FlowSummaryImpl::Private::summaryReturnNode(this.getSummaryNode(), rk) }
+
+  override ReturnKind getKind() { result = rk }
+}
+
 private Operand fullyConvertedCallStepImpl(Operand op) {
   not exists(getANonConversionUse(op)) and
   exists(Instruction instr |
@@ -616,7 +677,10 @@ predicate simpleOutNode(Node node, CallInstruction call) {
   instructionForFullyConvertedCall(node.asInstruction(), call)
 }
 
-/** A data flow node that represents the output of a call. */
+/**
+ * A data flow node that represents the output of a call (for example, a
+ * return value) at the call site.
+ */
 class OutNode extends Node {
   OutNode() {
     // Return values not hidden behind indirections
@@ -627,11 +691,15 @@ class OutNode extends Node {
     or
     // Modified arguments hidden behind indirections
     this instanceof IndirectArgumentOutNode
+    or
+    // Summary node
+    FlowSummaryImpl::Private::summaryOutNode(_, this.(FlowSummaryNode).getSummaryNode(), _)
   }
 
   /** Gets the underlying call. */
   abstract DataFlowCall getCall();
 
+  /** Gets the kind of this out node. */
   abstract ReturnKind getReturnKind();
 }
 
@@ -640,25 +708,44 @@ private class DirectCallOutNode extends OutNode {
 
   DirectCallOutNode() { simpleOutNode(this, call) }
 
-  override DataFlowCall getCall() { result = call }
+  override DataFlowCall getCall() { result.asCallInstruction() = call }
 
   override ReturnKind getReturnKind() { result = TNormalReturnKind(0) }
 }
 
 private class IndirectCallOutNode extends OutNode, IndirectReturnOutNode {
-  override DataFlowCall getCall() { result = this.getCallInstruction() }
+  override DataFlowCall getCall() { result.asCallInstruction() = this.getCallInstruction() }
 
   override ReturnKind getReturnKind() { result = TNormalReturnKind(this.getIndirectionIndex()) }
 }
 
 private class SideEffectOutNode extends OutNode, IndirectArgumentOutNode {
-  override DataFlowCall getCall() { result = this.getCallInstruction() }
+  override DataFlowCall getCall() { result.asCallInstruction() = this.getCallInstruction() }
 
   override ReturnKind getReturnKind() {
     result = TIndirectReturnKind(this.getArgumentIndex(), this.getIndirectionIndex())
   }
 }
 
+/**
+ * An output node that is part of a summary. An output node is needed when the
+ * model contains a synthesized call (`SummaryCall`) and the return value of
+ * that call is needed by the summary (for example when the model has flow from
+ * `Argument[0].ReturnValue`).
+ */
+private class SummaryOutNode extends OutNode, FlowSummaryNode {
+  private SummaryCall call;
+  private ReturnKind kind_;
+
+  SummaryOutNode() {
+    FlowSummaryImpl::Private::summaryOutNode(call.getReceiver(), this.getSummaryNode(), kind_)
+  }
+
+  override DataFlowCall getCall() { result = call }
+
+  override ReturnKind getReturnKind() { result = kind_ }
+}
+
 /**
  * Gets a node that can read the value returned from `call` with return kind
  * `kind`.
@@ -721,6 +808,10 @@ predicate jumpStep(Node n1, Node n2) {
       v = n1.asIndirectVariable(globalDef.getIndirection())
     )
   )
+  or
+  // models-as-data summarized flow
+  FlowSummaryImpl::Private::Steps::summaryJumpStep(n1.(FlowSummaryNode).getSummaryNode(),
+    n2.(FlowSummaryNode).getSummaryNode())
 }
 
 /**
@@ -729,25 +820,35 @@ predicate jumpStep(Node n1, Node n2) {
  * value of `node1`.
  *
  * The boolean `certain` is true if the destination address does not involve
- * any pointer arithmetic, and false otherwise.
+ * any pointer arithmetic, and false otherwise. This has to do with whether a
+ * store step can be used to clear a field (see `clearsContent`).
  */
-predicate storeStepImpl(Node node1, Content c, PostFieldUpdateNode node2, boolean certain) {
-  exists(int indirectionIndex1, int numberOfLoads, StoreInstruction store |
+predicate storeStepImpl(Node node1, Content c, Node node2, boolean certain) {
+  exists(
+    PostFieldUpdateNode postFieldUpdate, int indirectionIndex1, int numberOfLoads,
+    StoreInstruction store
+  |
+    postFieldUpdate = node2 and
     nodeHasInstruction(node1, store, pragma[only_bind_into](indirectionIndex1)) and
-    node2.getIndirectionIndex() = 1 and
-    numberOfLoadsFromOperand(node2.getFieldAddress(), store.getDestinationAddressOperand(),
-      numberOfLoads, certain)
+    postFieldUpdate.getIndirectionIndex() = 1 and
+    numberOfLoadsFromOperand(postFieldUpdate.getFieldAddress(),
+      store.getDestinationAddressOperand(), numberOfLoads, certain)
   |
     exists(FieldContent fc | fc = c |
-      fc.getField() = node2.getUpdatedField() and
+      fc.getField() = postFieldUpdate.getUpdatedField() and
       fc.getIndirectionIndex() = 1 + indirectionIndex1 + numberOfLoads
     )
     or
     exists(UnionContent uc | uc = c |
-      uc.getAField() = node2.getUpdatedField() and
+      uc.getAField() = postFieldUpdate.getUpdatedField() and
       uc.getIndirectionIndex() = 1 + indirectionIndex1 + numberOfLoads
     )
   )
+  or
+  // models-as-data summarized flow
+  FlowSummaryImpl::Private::Steps::summaryStoreStep(node1.(FlowSummaryNode).getSummaryNode(), c,
+    node2.(FlowSummaryNode).getSummaryNode()) and
+  certain = true
 }
 
 /**
@@ -834,6 +935,10 @@ predicate readStep(Node node1, ContentSet c, Node node2) {
       uc.getIndirectionIndex() = indirectionIndex2 + numberOfLoads
     )
   )
+  or
+  // models-as-data summarized flow
+  FlowSummaryImpl::Private::Steps::summaryReadStep(node1.(FlowSummaryNode).getSummaryNode(), c,
+    node2.(FlowSummaryNode).getSummaryNode())
 }
 
 /**
@@ -907,21 +1012,213 @@ class CastNode extends Node {
   CastNode() { none() } // stub implementation
 }
 
+cached
+private newtype TDataFlowCallable =
+  TSourceCallable(Cpp::Declaration decl) {
+    not decl instanceof FlowSummaryImpl::Public::SummarizedCallable
+  } or
+  TSummarizedCallable(FlowSummaryImpl::Public::SummarizedCallable c)
+
 /**
- * A function that may contain code or a variable that may contain itself. When
- * flow crosses from one _enclosing callable_ to another, the interprocedural
- * data-flow library discards call contexts and inserts a node in the big-step
- * relation used for human-readable path explanations.
+ * A callable, which may be:
+ *  - a function (that may contain code)
+ *  - a summarized function (that may contain only `FlowSummaryNode`s)
+ *  - a variable (this is used as context for global initialization, and also
+ *    for the mid-point in interprocedural data flow between a write and read
+ *    of a global variable in different functions).
+ * When flow crosses from one _enclosing callable_ to another, the
+ * interprocedural data-flow library discards call contexts and inserts a node
+ * in the big-step relation used for human-readable path explanations.
  */
-class DataFlowCallable = Cpp::Declaration;
+class DataFlowCallable extends TDataFlowCallable {
+  /** Gets the location of this callable. */
+  Location getLocation() { none() }
+
+  /** Gets a textual representation of this callable. */
+  string toString() { none() }
+
+  /**
+   * Gets the `Declaration` corresponding to this callable if it exists in the database.
+   * For summarized callables (which may not exist in the database), use `asSummarizedCallable`.
+   */
+  Cpp::Declaration asSourceCallable() { this = TSourceCallable(result) }
+
+  /**
+   * Gets the underlying summarized callable, if
+   * this callable is generated from a models-as-data
+   * model.
+   */
+  FlowSummaryImpl::Public::SummarizedCallable asSummarizedCallable() {
+    this = TSummarizedCallable(result)
+  }
+
+  /**
+   * Gets the underlying `Declaration` of this `DataFlowCallable`. This
+   * predicate returns a result for both source and summarized callables.
+   */
+  Cpp::Declaration getUnderlyingCallable() {
+    result = this.asSummarizedCallable() or // SummarizedCallable = Function (in CPP)
+    result = this.asSourceCallable()
+  }
+}
+
+/**
+ * A source callable, conceptually, a function in the source code for the
+ * purpose of computing data flow. In practice this excludes functions that
+ * are summarized using models-as-data (as we don't want to create
+ * unmodeled flows or duplicate paths), and includes variables (for reasons
+ * explained in `DataFlowCallable`).
+ */
+class SourceCallable extends DataFlowCallable, TSourceCallable {
+  Cpp::Declaration decl;
+
+  SourceCallable() { this = TSourceCallable(decl) }
+
+  override string toString() { result = decl.toString() }
+
+  override Location getLocation() { result = decl.getLocation() }
+}
+
+/**
+ * A summarized callable, that is, a function synthesized from one or more
+ * models-as-data models as a place to contain the corresponding
+ * `FlowSummaryNode`s.
+ */
+class SummarizedCallable extends DataFlowCallable, TSummarizedCallable {
+  FlowSummaryImpl::Public::SummarizedCallable sc;
+
+  SummarizedCallable() { this = TSummarizedCallable(sc) }
+
+  override string toString() { result = sc.toString() }
+
+  override Location getLocation() { result = sc.getLocation() }
+}
 
 class DataFlowExpr = Expr;
 
 class DataFlowType = Type;
 
-/** A function call relevant for data flow. */
-class DataFlowCall extends CallInstruction {
-  DataFlowCallable getEnclosingCallable() { result = this.getEnclosingFunction() }
+cached
+private newtype TDataFlowCall =
+  TNormalCall(CallInstruction call) or
+  TSummaryCall(
+    FlowSummaryImpl::Public::SummarizedCallable c, FlowSummaryImpl::Private::SummaryNode receiver
+  ) {
+    FlowSummaryImpl::Private::summaryCallbackRange(c, receiver)
+  }
+
+/**
+ * A function call relevant for data flow. This includes calls from source
+ * code and calls inside library callables with a flow summary.
+ */
+class DataFlowCall extends TDataFlowCall {
+  /**
+   * Gets the underlying data flow call instruction, if any.
+   */
+  CallInstruction asCallInstruction() { none() }
+
+  /**
+   * Gets the operand the specifies the target function of the call.
+   */
+  CallTargetOperand getCallTargetOperand() { none() }
+
+  /**
+   * Gets the `Function` that the call targets, if this is statically known.
+   */
+  DataFlowCallable getStaticCallTarget() { none() }
+
+  /**
+   * Gets the `index`'th argument operand. The qualifier is considered to have index `-1`.
+   */
+  ArgumentOperand getArgumentOperand(int index) { none() }
+
+  /**
+   * Gets the argument at the specified index, or `this` if `index` is `-1`.
+   */
+  pragma[noinline]
+  final Instruction getArgument(int index) { result = this.getArgumentOperand(index).getDef() }
+
+  /**
+   * Gets the number of arguments of the call, including the `this` pointer, if any.
+   */
+  final int getNumberOfArguments() { result = count(this.getArgumentOperand(_)) }
+
+  /**
+   * Gets the enclosing callable, if any.
+   */
+  DataFlowCallable getEnclosingCallable() { none() }
+
+  /**
+   * Gets a textual representation of this call.
+   */
+  string toString() { none() }
+
+  /**
+   * Gets the location of this call.
+   */
+  Location getLocation() { none() }
+}
+
+/**
+ * A function call relevant for data flow, that exists in source code.
+ */
+private class NormalCall extends DataFlowCall, TNormalCall {
+  private CallInstruction call;
+
+  NormalCall() { this = TNormalCall(call) }
+
+  override CallInstruction asCallInstruction() { result = call }
+
+  override CallTargetOperand getCallTargetOperand() { result = call.getCallTargetOperand() }
+
+  override DataFlowCallable getStaticCallTarget() {
+    result.getUnderlyingCallable() = call.getStaticCallTarget()
+  }
+
+  override ArgumentOperand getArgumentOperand(int index) { result = call.getArgumentOperand(index) }
+
+  override DataFlowCallable getEnclosingCallable() {
+    result.getUnderlyingCallable() = call.getEnclosingFunction()
+  }
+
+  override string toString() { result = call.toString() }
+
+  override Location getLocation() { result = call.getLocation() }
+}
+
+/**
+ * A synthesized call inside a callable with a flow summary.
+ *
+ * For example, consider the function:
+ * ```
+ * int myFunction(int (*funPtr)());
+ * ```
+ * with an accompanying models-as-data flow summary involving `funPtr` (for
+ * example from `Argument[0].ReturnValue` to `ReturnValue`). A `SummaryCall`
+ * will be synthesized representing a call to `funPtr` inside `myFunction`,
+ * so that flow can be connected as described in the model.
+ */
+class SummaryCall extends DataFlowCall, TSummaryCall {
+  private FlowSummaryImpl::Public::SummarizedCallable c;
+  private FlowSummaryImpl::Private::SummaryNode receiver;
+
+  SummaryCall() { this = TSummaryCall(c, receiver) }
+
+  /**
+   * Gets the data flow node that holds the address of the function this call
+   * targets.
+   */
+  FlowSummaryImpl::Private::SummaryNode getReceiver() { result = receiver }
+
+  // no implementation for `getCallTargetOperand()`, `getStaticCallTarget()`
+  // or `getArgumentOperand(int index)`. This is because the flow summary
+  // library is responsible for finding the call target, and there are no
+  // IR nodes available for the call target operand or argument operands.
+  override DataFlowCallable getEnclosingCallable() { result = TSummarizedCallable(c) }
+
+  override string toString() { result = "[summary] call to " + receiver + " in " + c }
+
+  override UnknownLocation getLocation() { any() }
 }
 
 module IsUnreachableInCall {
@@ -1012,10 +1309,16 @@ predicate nodeIsHidden(Node n) {
 class LambdaCallKind = Unit;
 
 /** Holds if `creation` is an expression that creates a lambda of kind `kind` for `c`. */
-predicate lambdaCreation(Node creation, LambdaCallKind kind, DataFlowCallable c) { none() }
+predicate lambdaCreation(Node creation, LambdaCallKind kind, DataFlowCallable c) {
+  creation.asInstruction().(FunctionAddressInstruction).getFunctionSymbol() = c.asSourceCallable() and
+  exists(kind)
+}
 
 /** Holds if `call` is a lambda call of kind `kind` where `receiver` is the lambda expression. */
-predicate lambdaCall(DataFlowCall call, LambdaCallKind kind, Node receiver) { none() }
+predicate lambdaCall(DataFlowCall call, LambdaCallKind kind, Node receiver) {
+  call.(SummaryCall).getReceiver() = receiver.(FlowSummaryNode).getSummaryNode() and
+  exists(kind)
+}
 
 /** Extra data-flow steps needed for lambda flow analysis. */
 predicate additionalLambdaFlowStep(Node nodeFrom, Node nodeTo, boolean preservesValue) { none() }
@@ -1031,7 +1334,15 @@ predicate knownSinkModel(Node sink, string model) { none() }
  * One example would be to allow flow like `p.foo = p.bar;`, which is disallowed
  * by default as a heuristic.
  */
-predicate allowParameterReturnInSelf(ParameterNode p) { p instanceof IndirectParameterNode }
+predicate allowParameterReturnInSelf(ParameterNode p) {
+  p instanceof IndirectParameterNode
+  or
+  // models-as-data summarized flow
+  exists(DataFlowCallable c, ParameterPosition pos |
+    p.isParameterOf(c, pos) and
+    FlowSummaryImpl::Private::summaryAllowParameterReturnInSelf(c.asSummarizedCallable(), pos)
+  )
+}
 
 private predicate fieldHasApproxName(Field f, string s) {
   s = f.getName().charAt(0) and

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

@@ -10,6 +10,7 @@ private import semmle.code.cpp.ir.ValueNumbering
 private import semmle.code.cpp.ir.IR
 private import semmle.code.cpp.controlflow.IRGuards
 private import semmle.code.cpp.models.interfaces.DataFlow
+private import semmle.code.cpp.dataflow.internal.FlowSummaryImpl as FlowSummaryImpl
 private import DataFlowPrivate
 private import ModelUtil
 private import SsaInternals as Ssa
@@ -59,7 +60,8 @@ private newtype TIRDataFlowNode =
     )
   } or
   TFinalGlobalValue(Ssa::GlobalUse globalUse) or
-  TInitialGlobalValue(Ssa::GlobalDef globalUse)
+  TInitialGlobalValue(Ssa::GlobalDef globalUse) or
+  TFlowSummaryNode(FlowSummaryImpl::Private::SummaryNode sn)
 
 /**
  * An operand that is defined by a `FieldAddressInstruction`.
@@ -735,6 +737,36 @@ class InitialGlobalValue extends Node, TInitialGlobalValue {
   override string toStringImpl() { result = globalDef.toString() }
 }
 
+/**
+ * A data-flow node used to model flow summaries. That is, a dataflow node
+ * that is synthesized to represent a parameter, return value, or other part
+ * of a models-as-data modeled function.
+ */
+class FlowSummaryNode extends Node, TFlowSummaryNode {
+  /**
+   * Gets the models-as-data `SummaryNode` associated with this dataflow
+   * `FlowSummaryNode`.
+   */
+  FlowSummaryImpl::Private::SummaryNode getSummaryNode() { this = TFlowSummaryNode(result) }
+
+  /**
+   * Gets the summarized callable that this node belongs to.
+   */
+  FlowSummaryImpl::Public::SummarizedCallable getSummarizedCallable() {
+    result = this.getSummaryNode().getSummarizedCallable()
+  }
+
+  /**
+   * Gets the enclosing callable. For a `FlowSummaryNode` this is always the
+   * summarized function this node is part of.
+   */
+  override Declaration getEnclosingCallable() { result = this.getSummarizedCallable() }
+
+  override Location getLocationImpl() { result = this.getSummarizedCallable().getLocation() }
+
+  override string toStringImpl() { result = this.getSummaryNode().toString() }
+}
+
 /**
  * INTERNAL: do not use.
  *
@@ -826,6 +858,9 @@ class IndirectArgumentOutNode extends PostUpdateNodeImpl {
 
   CallInstruction getCallInstruction() { result.getAnArgumentOperand() = operand }
 
+  /**
+   * Gets the `Function` that the call targets, if this is statically known.
+   */
   Function getStaticCallTarget() { result = this.getCallInstruction().getStaticCallTarget() }
 
   override string toStringImpl() {
@@ -1635,6 +1670,8 @@ class ParameterNode extends Node {
     this.asInstruction() instanceof InitializeParameterInstruction
     or
     this instanceof IndirectParameterNode
+    or
+    FlowSummaryImpl::Private::summaryParameterNode(this.(FlowSummaryNode).getSummaryNode(), _)
   }
 
   /**
@@ -1642,7 +1679,7 @@ class ParameterNode extends Node {
    * implicit `this` parameter is considered to have position `-1`, and
    * pointer-indirection parameters are at further negative positions.
    */
-  predicate isParameterOf(Function f, ParameterPosition pos) { none() } // overridden by subclasses
+  predicate isParameterOf(DataFlowCallable f, ParameterPosition pos) { none() } // overridden by subclasses
 
   /** Gets the `Parameter` associated with this node, if it exists. */
   Parameter getParameter() { none() } // overridden by subclasses
@@ -1664,8 +1701,9 @@ class DirectParameterNode extends InstructionNode {
 private class ExplicitParameterNode extends ParameterNode, DirectParameterNode {
   ExplicitParameterNode() { exists(instr.getParameter()) }
 
-  override predicate isParameterOf(Function f, ParameterPosition pos) {
-    f.getParameter(pos.(DirectPosition).getIndex()) = instr.getParameter()
+  override predicate isParameterOf(DataFlowCallable f, ParameterPosition pos) {
+    f.getUnderlyingCallable().(Function).getParameter(pos.(DirectPosition).getIndex()) =
+      instr.getParameter()
   }
 
   override string toStringImpl() { result = instr.getParameter().toString() }
@@ -1677,13 +1715,32 @@ private class ExplicitParameterNode extends ParameterNode, DirectParameterNode {
 class ThisParameterNode extends ParameterNode, DirectParameterNode {
   ThisParameterNode() { instr.getIRVariable() instanceof IRThisVariable }
 
-  override predicate isParameterOf(Function f, ParameterPosition pos) {
-    pos.(DirectPosition).getIndex() = -1 and instr.getEnclosingFunction() = f
+  override predicate isParameterOf(DataFlowCallable f, ParameterPosition pos) {
+    pos.(DirectPosition).getIndex() = -1 and
+    instr.getEnclosingFunction() = f.getUnderlyingCallable()
   }
 
   override string toStringImpl() { result = "this" }
 }
 
+/**
+ * A parameter node that is part of a summary.
+ */
+class SummaryParameterNode extends ParameterNode, FlowSummaryNode {
+  SummaryParameterNode() {
+    FlowSummaryImpl::Private::summaryParameterNode(this.getSummaryNode(), _)
+  }
+
+  private ParameterPosition getPosition() {
+    FlowSummaryImpl::Private::summaryParameterNode(this.getSummaryNode(), result)
+  }
+
+  override predicate isParameterOf(DataFlowCallable c, ParameterPosition p) {
+    c.getUnderlyingCallable() = this.getSummarizedCallable() and
+    p = this.getPosition()
+  }
+}
+
 pragma[noinline]
 private predicate indirectPositionHasArgumentIndexAndIndex(
   IndirectionPosition pos, int argumentIndex, int indirectionIndex
@@ -1702,8 +1759,8 @@ private predicate indirectParameterNodeHasArgumentIndexAndIndex(
 
 /** A synthetic parameter to model the pointed-to object of a pointer parameter. */
 class ParameterIndirectionNode extends ParameterNode instanceof IndirectParameterNode {
-  override predicate isParameterOf(Function f, ParameterPosition pos) {
-    IndirectParameterNode.super.getEnclosingCallable() = f and
+  override predicate isParameterOf(DataFlowCallable f, ParameterPosition pos) {
+    IndirectParameterNode.super.getEnclosingCallable() = f.getUnderlyingCallable() and
     exists(int argumentIndex, int indirectionIndex |
       indirectPositionHasArgumentIndexAndIndex(pos, argumentIndex, indirectionIndex) and
       indirectParameterNodeHasArgumentIndexAndIndex(this, argumentIndex, indirectionIndex)
@@ -1753,6 +1810,22 @@ abstract private class PartialDefinitionNode extends PostUpdateNode {
   abstract Expr getDefinedExpr();
 }
 
+/**
+ * A `PostUpdateNode` that is part of a flow summary. These are synthesized,
+ * for example, when a models-as-data summary models a write to a field since
+ * the write needs to target a `PostUpdateNode`.
+ */
+class SummaryPostUpdateNode extends FlowSummaryNode, PostUpdateNode {
+  SummaryPostUpdateNode() {
+    FlowSummaryImpl::Private::summaryPostUpdateNode(this.getSummaryNode(), _)
+  }
+
+  override Node getPreUpdateNode() {
+    FlowSummaryImpl::Private::summaryPostUpdateNode(this.getSummaryNode(),
+      result.(FlowSummaryNode).getSummaryNode())
+  }
+}
+
 /**
  * A node that represents the value of a variable after a function call that
  * may have changed the variable because it's passed by reference.
@@ -1889,10 +1962,20 @@ cached
 private module Cached {
   /**
    * Holds if data flows from `nodeFrom` to `nodeTo` in exactly one local
-   * (intra-procedural) step.
+   * (intra-procedural) step. This relation is only used for local dataflow
+   * (for example `DataFlow::localFlow(source, sink)`) so it contains
+   * special cases that should only apply to local dataflow.
    */
   cached
-  predicate localFlowStep(Node nodeFrom, Node nodeTo) { simpleLocalFlowStep(nodeFrom, nodeTo, _) }
+  predicate localFlowStep(Node nodeFrom, Node nodeTo) {
+    // common dataflow steps
+    simpleLocalFlowStep(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::summaryThroughStepValue(nodeFrom, nodeTo, _)
+  }
 
   private predicate indirectionOperandFlow(RawIndirectOperand nodeFrom, Node nodeTo) {
     nodeFrom != nodeTo and
@@ -1958,8 +2041,9 @@ private module Cached {
   /**
    * INTERNAL: do not use.
    *
-   * This is the local flow predicate that's used as a building block in global
-   * data flow. It may have less flow than the `localFlowStep` predicate.
+   * This is the local flow predicate that's used as a building block in both
+   * local and global data flow. It may have less flow than the `localFlowStep`
+   * predicate.
    */
   cached
   predicate simpleLocalFlowStep(Node nodeFrom, Node nodeTo, string model) {
@@ -2001,6 +2085,10 @@ private module Cached {
     // function such as `operator[]`.
     reverseFlow(nodeFrom, nodeTo) and
     model = ""
+    or
+    // models-as-data summarized flow
+    FlowSummaryImpl::Private::Steps::summaryLocalStep(nodeFrom.(FlowSummaryNode).getSummaryNode(),
+      nodeTo.(FlowSummaryNode).getSummaryNode(), true, model)
   }
 
   private predicate simpleInstructionLocalFlowStep(Operand opFrom, Instruction iTo) {
@@ -2242,6 +2330,8 @@ private Field getAFieldWithSize(Union u, int bytes) {
 cached
 private newtype TContent =
   TFieldContent(Field f, int indirectionIndex) {
+    // the indirection index for field content starts at 1 (because `TFieldContent` is thought of as
+    // the address of the field, `FieldAddress` in the IR).
     indirectionIndex = [1 .. Ssa::getMaxIndirectionsForType(f.getUnspecifiedType())] and
     // Reads and writes of union fields are tracked using `UnionContent`.
     not f.getDeclaringType() instanceof Union
@@ -2251,7 +2341,8 @@ private newtype TContent =
       f = u.getAField() and
       bytes = getFieldSize(f) and
       // We key `UnionContent` by the union instead of its fields since a write to one
-      // field can be read by any read of the union's fields.
+      // field can be read by any read of the union's fields. Again, the indirection index
+      // is 1-based (because 0 is considered the address).
       indirectionIndex =
         [1 .. max(Ssa::getMaxIndirectionsForType(getAFieldWithSize(u, bytes).getUnspecifiedType()))]
     )
@@ -2285,16 +2376,14 @@ class Content extends TContent {
   abstract predicate impliesClearOf(Content c);
 }
 
+/**
+ * Gets a string consisting of `n` star characters ("*"), where n >= 0. This is
+ * used to represent indirection.
+ */
+bindingset[n]
+string repeatStars(int n) { result = concat(int i | i in [1 .. n] | "*") }
+
 private module ContentStars {
-  private int maxNumberOfIndirections() { result = max(any(Content c).getIndirectionIndex()) }
-
-  private string repeatStars(int n) {
-    n = 0 and result = ""
-    or
-    n = [1 .. maxNumberOfIndirections()] and
-    result = "*" + repeatStars(n - 1)
-  }
-
   /**
    * Gets the number of stars (i.e., `*`s) needed to produce the `toString`
    * output for `c`.
@@ -2374,6 +2463,12 @@ class UnionContent extends Content, TUnionContent {
  * stored into (`getAStoreContent`) or read from (`getAReadContent`).
  */
 class ContentSet instanceof Content {
+  /**
+   * Holds if this content set is the singleton `{c}`. At present, this is
+   * the only kind of content set supported in C/C++.
+   */
+  predicate isSingleton(Content c) { this = c }
+
   /** Gets a content that may be stored into when storing into this set. */
   Content getAStoreContent() { result = this }
 
@@ -2591,5 +2686,5 @@ class AdditionalCallTarget extends Unit {
   /**
    * Gets a viable target for `call`.
    */
-  abstract DataFlowCallable viableTarget(Call call);
+  abstract Declaration viableTarget(Call call);
 }

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

@@ -19,15 +19,6 @@ private module SourceVariables {
       ind = [0 .. countIndirectionsForCppType(base.getLanguageType()) + 1]
     }
 
-  private int maxNumberOfIndirections() { result = max(SourceVariable sv | | sv.getIndirection()) }
-
-  private string repeatStars(int n) {
-    n = 0 and result = ""
-    or
-    n = [1 .. maxNumberOfIndirections()] and
-    result = "*" + repeatStars(n - 1)
-  }
-
   class SourceVariable extends TSourceVariable {
     BaseSourceVariable base;
     int ind;
@@ -74,19 +65,28 @@ private module SourceVariables {
 import SourceVariables
 
 /**
- * Holds if the `(operand, indirectionIndex)` columns should be
- * assigned a `RawIndirectOperand` value.
+ * Holds if `indirectionIndex` is a valid non-zero indirection index for
+ * operand `op`. That is, `indirectionIndex` is between 1 and the maximum
+ * indirection for the operand's type.
  */
-predicate hasRawIndirectOperand(Operand op, int indirectionIndex) {
+predicate hasIndirectOperand(Operand op, int indirectionIndex) {
   exists(CppType type, int m |
     not ignoreOperand(op) and
     type = getLanguageType(op) and
     m = countIndirectionsForCppType(type) and
-    indirectionIndex = [1 .. m] and
-    not hasIRRepresentationOfIndirectOperand(op, indirectionIndex, _, _)
+    indirectionIndex = [1 .. m]
   )
 }
 
+/**
+ * Holds if the `(operand, indirectionIndex)` columns should be
+ * assigned a `RawIndirectOperand` value.
+ */
+predicate hasRawIndirectOperand(Operand op, int indirectionIndex) {
+  hasIndirectOperand(op, indirectionIndex) and
+  not hasIRRepresentationOfIndirectOperand(op, indirectionIndex, _, _)
+}
+
 /**
  * Holds if the `(instr, indirectionIndex)` columns should be
  * assigned a `RawIndirectInstruction` value.
@@ -443,6 +443,8 @@ class FinalParameterUse extends UseImpl, TFinalParameterUse {
 
   Parameter getParameter() { result = p }
 
+  int getArgumentIndex() { result = p.getIndex() }
+
   override Node getNode() { finalParameterNodeHasParameterAndIndex(result, p, ind) }
 
   override int getIndirection() { result = ind + 1 }
@@ -891,7 +893,7 @@ private predicate isArgumentOfCallableInstruction(DataFlowCall call, Instruction
 }
 
 private predicate isArgumentOfCallableOperand(DataFlowCall call, Operand operand) {
-  operand.(ArgumentOperand).getCall() = call
+  operand = call.getArgumentOperand(_)
   or
   exists(FieldAddressInstruction fai |
     fai.getObjectAddressOperand() = operand and

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

@@ -6,16 +6,26 @@ private import semmle.code.cpp.models.interfaces.SideEffect
 private import DataFlowUtil
 private import DataFlowPrivate
 private import SsaInternals 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.
+ * (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, _)
 }
 
 /**
@@ -42,6 +52,10 @@ predicate localAdditionalTaintStep(DataFlow::Node nodeFrom, DataFlow::Node nodeT
   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

cpp/ql/lib/semmle/code/cpp/models/Models.qll

@@ -1,5 +1,6 @@
 private import implementations.Allocation
 private import implementations.Deallocation
+private import implementations.Fopen
 private import implementations.Fread
 private import implementations.Getenv
 private import implementations.Gets
@@ -41,4 +42,4 @@ private import implementations.SqLite3
 private import implementations.PostgreSql
 private import implementations.System
 private import implementations.StructuredExceptionHandling
-private import implementations.Fopen
+private import implementations.ZMQ

cpp/ql/lib/semmle/code/cpp/models/implementations/Gets.qll

@@ -112,3 +112,21 @@ private class GetsFunction extends DataFlowFunction, ArrayFunction, AliasFunctio
 
   override predicate hasArrayOutput(int bufParam) { bufParam = 0 }
 }
+
+/**
+ * A model for `getc` and similar functions that are flow sources.
+ */
+private class GetcSource extends SourceModelCsv {
+  override predicate row(string row) {
+    row =
+      [
+        ";;false;getc;;;ReturnValue;remote", ";;false;getwc;;;ReturnValue;remote",
+        ";;false;_getc_nolock;;;ReturnValue;remote", ";;false;_getwc_nolock;;;ReturnValue;remote",
+        ";;false;getch;;;ReturnValue;local", ";;false;_getch;;;ReturnValue;local",
+        ";;false;_getwch;;;ReturnValue;local", ";;false;_getch_nolock;;;ReturnValue;local",
+        ";;false;_getwch_nolock;;;ReturnValue;local", ";;false;getchar;;;ReturnValue;local",
+        ";;false;getwchar;;;ReturnValue;local", ";;false;_getchar_nolock;;;ReturnValue;local",
+        ";;false;_getwchar_nolock;;;ReturnValue;local",
+      ]
+  }
+}

cpp/ql/lib/semmle/code/cpp/models/implementations/ZMQ.qll

@@ -0,0 +1,45 @@
+/**
+ * Provides implementation classes modeling the ZeroMQ networking library.
+ */
+
+import semmle.code.cpp.models.interfaces.FlowSource
+
+/**
+ * Remote flow sources.
+ */
+private class ZmqSource extends SourceModelCsv {
+  override predicate row(string row) {
+    row =
+      [
+        ";;false;zmq_recv;;;Argument[*1];remote", ";;false;zmq_recvmsg;;;Argument[*1];remote",
+        ";;false;zmq_msg_recv;;;Argument[*0];remote",
+      ]
+  }
+}
+
+/**
+ * Remote flow sinks.
+ */
+private class ZmqSinks extends SinkModelCsv {
+  override predicate row(string row) {
+    row =
+      [
+        ";;false;zmq_send;;;Argument[*1];remote-sink",
+        ";;false;zmq_sendmsg;;;Argument[*1];remote-sink",
+        ";;false;zmq_msg_send;;;Argument[*0];remote-sink",
+      ]
+  }
+}
+
+/**
+ * Flow steps.
+ */
+private class ZmqSummaries extends SummaryModelCsv {
+  override predicate row(string row) {
+    row =
+      [
+        ";;false;zmq_msg_init_data;;;Argument[*1];Argument[*0];taint",
+        ";;false;zmq_msg_data;;;Argument[*0];ReturnValue[*];taint",
+      ]
+  }
+}

cpp/ql/lib/semmle/code/cpp/models/interfaces/FlowSource.qll

@@ -9,6 +9,7 @@
 import cpp
 import FunctionInputsAndOutputs
 import semmle.code.cpp.models.Models
+import semmle.code.cpp.dataflow.ExternalFlow
 
 /**
  * A library function that returns data that may be read from a network connection.

cpp/ql/lib/semmle/code/cpp/security/FlowSources.qll

@@ -20,6 +20,9 @@ abstract class RemoteFlowSource extends FlowSource { }
 /** A data flow source of local user input. */
 abstract class LocalFlowSource extends FlowSource { }
 
+/**
+ * A remote data flow source that is defined through a `RemoteFlowSourceFunction` model.
+ */
 private class RemoteModelSource extends RemoteFlowSource {
   string sourceType;
 
@@ -34,6 +37,9 @@ private class RemoteModelSource extends RemoteFlowSource {
   override string getSourceType() { result = sourceType }
 }
 
+/**
+ * A local data flow source that is defined through a `LocalFlowSourceFunction` model.
+ */
 private class LocalModelSource extends LocalFlowSource {
   string sourceType;
 
@@ -48,6 +54,9 @@ private class LocalModelSource extends LocalFlowSource {
   override string getSourceType() { result = sourceType }
 }
 
+/**
+ * A local data flow source that the `argv` parameter to `main`.
+ */
 private class ArgvSource extends LocalFlowSource {
   ArgvSource() {
     exists(Function main, Parameter argv |
@@ -60,12 +69,33 @@ private class ArgvSource extends LocalFlowSource {
   override string getSourceType() { result = "a command-line argument" }
 }
 
+/**
+ * A remote data flow source that is defined through 'models as data'.
+ */
+private class ExternalRemoteFlowSource extends RemoteFlowSource {
+  ExternalRemoteFlowSource() { sourceNode(this, "remote") }
+
+  override string getSourceType() { result = "external" }
+}
+
+/**
+ * A local data flow source that is defined through 'models as data'.
+ */
+private class ExternalLocalFlowSource extends LocalFlowSource {
+  ExternalLocalFlowSource() { sourceNode(this, "local") }
+
+  override string getSourceType() { result = "external" }
+}
+
 /** A remote data flow sink. */
 abstract class RemoteFlowSink extends DataFlow::Node {
   /** Gets a string that describes the type of this flow sink. */
   abstract string getSinkType();
 }
 
+/**
+ * A remote flow sink derived from the `RemoteFlowSinkFunction` model.
+ */
 private class RemoteParameterSink extends RemoteFlowSink {
   string sourceType;
 
@@ -79,3 +109,12 @@ private class RemoteParameterSink extends RemoteFlowSink {
 
   override string getSinkType() { result = sourceType }
 }
+
+/**
+ * A remote flow sink defined in a CSV model.
+ */
+private class RemoteFlowFromCsvSink extends RemoteFlowSink {
+  RemoteFlowFromCsvSink() { sinkNode(this, "remote-sink") }
+
+  override string getSinkType() { result = "remote flow sink" }
+}

cpp/ql/lib/semmle/code/cpp/security/flowafterfree/UseAfterFree.qll

@@ -139,6 +139,7 @@ private module ParameterSinks {
 }
 
 private import semmle.code.cpp.ir.dataflow.internal.DataFlowImplCommon
+private import semmle.code.cpp.ir.dataflow.internal.DataFlowPrivate
 
 /**
  * Holds if `n` represents the expression `e`, and `e` is a pointer that is
@@ -149,11 +150,11 @@ private import semmle.code.cpp.ir.dataflow.internal.DataFlowImplCommon
 predicate isUse(DataFlow::Node n, Expr e) {
   isUse0(e) and n.asExpr() = e
   or
-  exists(CallInstruction call, InitializeParameterInstruction init |
+  exists(DataFlowCall call, InitializeParameterInstruction init |
     n.asOperand().getDef().getUnconvertedResultExpression() = e and
     pragma[only_bind_into](init) = ParameterSinks::getAnAlwaysDereferencedParameter() and
     viableParamArg(call, DataFlow::instructionNode(init), n) and
     pragma[only_bind_out](init.getEnclosingFunction()) =
-      pragma[only_bind_out](call.getStaticCallTarget())
+      pragma[only_bind_out](call.asCallInstruction().getStaticCallTarget())
   )
 }

cpp/ql/src/Security/CWE/CWE-497/ExposedSystemData.ql

@@ -16,6 +16,7 @@ import cpp
 import semmle.code.cpp.ir.dataflow.TaintTracking
 import semmle.code.cpp.models.interfaces.FlowSource
 import semmle.code.cpp.models.implementations.Memset
+import semmle.code.cpp.security.FlowSources
 import ExposedSystemData::PathGraph
 import SystemData
 
@@ -23,11 +24,11 @@ module ExposedSystemDataConfig implements DataFlow::ConfigSig {
   predicate isSource(DataFlow::Node source) { source = any(SystemData sd).getAnExpr() }
 
   predicate isSink(DataFlow::Node sink) {
-    exists(FunctionCall fc, FunctionInput input, int arg |
-      fc.getTarget().(RemoteFlowSinkFunction).hasRemoteFlowSink(input, _) and
-      input.isParameterDeref(arg) and
-      fc.getArgument(arg).getAChild*() = sink.asIndirectExpr()
-    )
+    sink instanceof RemoteFlowSink
+    or
+    // workaround for cases where the sink contains the tainted thing as a child; this could
+    // probably be handled better with taint inheriting content or similar modeling.
+    exists(RemoteFlowSink sinkNode | sinkNode.asIndirectExpr().getAChild*() = sink.asIndirectExpr())
   }
 
   predicate isBarrier(DataFlow::Node node) {

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

@@ -0,0 +1,76 @@
+| tests.cpp:144:5:144:19 | [summary param] 0 in madArg0ToReturn | ParameterNode | madArg0ToReturn | madArg0ToReturn |
+| tests.cpp:144:5:144:19 | [summary] to write: ReturnValue in madArg0ToReturn | ReturnNode | madArg0ToReturn | madArg0ToReturn |
+| tests.cpp:145:6:145:28 | [summary param] 0 in madArg0ToReturnIndirect | ParameterNode | madArg0ToReturnIndirect | madArg0ToReturnIndirect |
+| tests.cpp:145:6:145:28 | [summary] to write: ReturnValue[*] in madArg0ToReturnIndirect | ReturnNode | madArg0ToReturnIndirect | madArg0ToReturnIndirect |
+| tests.cpp:147:5:147:28 | [summary param] 0 in madArg0ToReturnValueFlow | ParameterNode | madArg0ToReturnValueFlow | madArg0ToReturnValueFlow |
+| tests.cpp:147:5:147:28 | [summary] to write: ReturnValue in madArg0ToReturnValueFlow | ReturnNode | madArg0ToReturnValueFlow | madArg0ToReturnValueFlow |
+| tests.cpp:148:5:148:27 | [summary param] 0 indirection in madArg0IndirectToReturn | ParameterNode | madArg0IndirectToReturn | madArg0IndirectToReturn |
+| tests.cpp:148:5:148:27 | [summary] to write: ReturnValue in madArg0IndirectToReturn | ReturnNode | madArg0IndirectToReturn | madArg0IndirectToReturn |
+| tests.cpp:149:5:149:33 | [summary param] 0 indirection in madArg0DoubleIndirectToReturn | ParameterNode | madArg0DoubleIndirectToReturn | madArg0DoubleIndirectToReturn |
+| tests.cpp:149:5:149:33 | [summary] to write: ReturnValue in madArg0DoubleIndirectToReturn | ReturnNode | madArg0DoubleIndirectToReturn | madArg0DoubleIndirectToReturn |
+| tests.cpp:150:5:150:30 | [summary param] 0 in madArg0NotIndirectToReturn | ParameterNode | madArg0NotIndirectToReturn | madArg0NotIndirectToReturn |
+| tests.cpp:150:5:150:30 | [summary] to write: ReturnValue in madArg0NotIndirectToReturn | ReturnNode | madArg0NotIndirectToReturn | madArg0NotIndirectToReturn |
+| tests.cpp:151:6:151:26 | [summary param] 0 in madArg0ToArg1Indirect | ParameterNode | madArg0ToArg1Indirect | madArg0ToArg1Indirect |
+| tests.cpp:151:6:151:26 | [summary param] 1 indirection in madArg0ToArg1Indirect | ParameterNode | madArg0ToArg1Indirect | madArg0ToArg1Indirect |
+| tests.cpp:151:6:151:26 | [summary] to write: Argument[1 indirection] in madArg0ToArg1Indirect | PostUpdateNode | madArg0ToArg1Indirect | madArg0ToArg1Indirect |
+| tests.cpp:152:6:152:34 | [summary param] 0 indirection in madArg0IndirectToArg1Indirect | ParameterNode | madArg0IndirectToArg1Indirect | madArg0IndirectToArg1Indirect |
+| tests.cpp:152:6:152:34 | [summary param] 1 indirection in madArg0IndirectToArg1Indirect | ParameterNode | madArg0IndirectToArg1Indirect | madArg0IndirectToArg1Indirect |
+| tests.cpp:152:6:152:34 | [summary] to write: Argument[1 indirection] in madArg0IndirectToArg1Indirect | PostUpdateNode | madArg0IndirectToArg1Indirect | madArg0IndirectToArg1Indirect |
+| tests.cpp:153:5:153:18 | [summary param] 0 indirection in madArgsComplex | ParameterNode | madArgsComplex | madArgsComplex |
+| tests.cpp:153:5:153:18 | [summary param] 1 indirection in madArgsComplex | ParameterNode | madArgsComplex | madArgsComplex |
+| tests.cpp:153:5:153:18 | [summary param] 2 in madArgsComplex | ParameterNode | madArgsComplex | madArgsComplex |
+| tests.cpp:153:5:153:18 | [summary] to write: ReturnValue in madArgsComplex | ReturnNode | madArgsComplex | madArgsComplex |
+| tests.cpp:155:5:155:28 | [summary param] 2 in madAndImplementedComplex | ParameterNode | madAndImplementedComplex | madAndImplementedComplex |
+| tests.cpp:155:5:155:28 | [summary] to write: ReturnValue in madAndImplementedComplex | ReturnNode | madAndImplementedComplex | madAndImplementedComplex |
+| tests.cpp:160:5:160:24 | [summary param] 0 in madArg0FieldToReturn | ParameterNode | madArg0FieldToReturn | madArg0FieldToReturn |
+| tests.cpp:160:5:160:24 | [summary] read: Argument[0].Field[value] in madArg0FieldToReturn |  | madArg0FieldToReturn | madArg0FieldToReturn |
+| tests.cpp:160:5:160:24 | [summary] to write: ReturnValue in madArg0FieldToReturn | ReturnNode | madArg0FieldToReturn | madArg0FieldToReturn |
+| tests.cpp:161:5:161:32 | [summary param] 0 indirection in madArg0IndirectFieldToReturn | ParameterNode | madArg0IndirectFieldToReturn | madArg0IndirectFieldToReturn |
+| tests.cpp:161:5:161:32 | [summary] read: Argument[0 indirection].Field[value] in madArg0IndirectFieldToReturn |  | madArg0IndirectFieldToReturn | madArg0IndirectFieldToReturn |
+| tests.cpp:161:5:161:32 | [summary] to write: ReturnValue in madArg0IndirectFieldToReturn | ReturnNode | madArg0IndirectFieldToReturn | madArg0IndirectFieldToReturn |
+| tests.cpp:162:5:162:32 | [summary param] 0 in madArg0FieldIndirectToReturn | ParameterNode | madArg0FieldIndirectToReturn | madArg0FieldIndirectToReturn |
+| tests.cpp:162:5:162:32 | [summary] read: Argument[0].Field[*ptr] in madArg0FieldIndirectToReturn |  | madArg0FieldIndirectToReturn | madArg0FieldIndirectToReturn |
+| tests.cpp:162:5:162:32 | [summary] to write: ReturnValue in madArg0FieldIndirectToReturn | ReturnNode | madArg0FieldIndirectToReturn | madArg0FieldIndirectToReturn |
+| tests.cpp:163:13:163:32 | [summary param] 0 in madArg0ToReturnField | ParameterNode | madArg0ToReturnField | madArg0ToReturnField |
+| tests.cpp:163:13:163:32 | [summary] to write: ReturnValue in madArg0ToReturnField | ReturnNode | madArg0ToReturnField | madArg0ToReturnField |
+| tests.cpp:163:13:163:32 | [summary] to write: ReturnValue.Field[value] in madArg0ToReturnField |  | madArg0ToReturnField | madArg0ToReturnField |
+| tests.cpp:164:14:164:41 | [summary param] 0 in madArg0ToReturnIndirectField | ParameterNode | madArg0ToReturnIndirectField | madArg0ToReturnIndirectField |
+| tests.cpp:164:14:164:41 | [summary] to write: ReturnValue[*] in madArg0ToReturnIndirectField | ReturnNode | madArg0ToReturnIndirectField | madArg0ToReturnIndirectField |
+| tests.cpp:164:14:164:41 | [summary] to write: ReturnValue[*].Field[value] in madArg0ToReturnIndirectField |  | madArg0ToReturnIndirectField | madArg0ToReturnIndirectField |
+| tests.cpp:165:13:165:40 | [summary param] 0 in madArg0ToReturnFieldIndirect | ParameterNode | madArg0ToReturnFieldIndirect | madArg0ToReturnFieldIndirect |
+| tests.cpp:165:13:165:40 | [summary] to write: ReturnValue in madArg0ToReturnFieldIndirect | ReturnNode | madArg0ToReturnFieldIndirect | madArg0ToReturnFieldIndirect |
+| tests.cpp:165:13:165:40 | [summary] to write: ReturnValue.Field[*ptr] in madArg0ToReturnFieldIndirect |  | madArg0ToReturnFieldIndirect | madArg0ToReturnFieldIndirect |
+| tests.cpp:284:7:284:19 | [summary param] 0 in madArg0ToSelf | ParameterNode | madArg0ToSelf | madArg0ToSelf |
+| tests.cpp:284:7:284:19 | [summary param] this indirection in madArg0ToSelf | ParameterNode | madArg0ToSelf | madArg0ToSelf |
+| tests.cpp:284:7:284:19 | [summary] to write: Argument[this indirection] in madArg0ToSelf | PostUpdateNode | madArg0ToSelf | madArg0ToSelf |
+| tests.cpp:285:6:285:20 | [summary param] this indirection in madSelfToReturn | ParameterNode | madSelfToReturn | madSelfToReturn |
+| tests.cpp:285:6:285:20 | [summary] to write: ReturnValue in madSelfToReturn | ReturnNode | madSelfToReturn | madSelfToReturn |
+| tests.cpp:287:7:287:20 | [summary param] 0 in madArg0ToField | ParameterNode | madArg0ToField | madArg0ToField |
+| tests.cpp:287:7:287:20 | [summary param] this indirection in madArg0ToField | ParameterNode | madArg0ToField | madArg0ToField |
+| tests.cpp:287:7:287:20 | [summary] to write: Argument[this indirection] in madArg0ToField | PostUpdateNode | madArg0ToField | madArg0ToField |
+| tests.cpp:287:7:287:20 | [summary] to write: Argument[this indirection].Field[val] in madArg0ToField |  | madArg0ToField | madArg0ToField |
+| tests.cpp:288:6:288:21 | [summary param] this indirection in madFieldToReturn | ParameterNode | madFieldToReturn | madFieldToReturn |
+| tests.cpp:288:6:288:21 | [summary] read: Argument[this indirection].Field[val] in madFieldToReturn |  | madFieldToReturn | madFieldToReturn |
+| tests.cpp:288:6:288:21 | [summary] to write: ReturnValue in madFieldToReturn | ReturnNode | madFieldToReturn | madFieldToReturn |
+| tests.cpp:313:7:313:30 | [summary param] this indirection in namespaceMadSelfToReturn | ParameterNode | namespaceMadSelfToReturn | namespaceMadSelfToReturn |
+| tests.cpp:313:7:313:30 | [summary] to write: ReturnValue in namespaceMadSelfToReturn | ReturnNode | namespaceMadSelfToReturn | namespaceMadSelfToReturn |
+| tests.cpp:434:5:434:29 | [summary param] 0 in madCallArg0ReturnToReturn | ParameterNode | madCallArg0ReturnToReturn | madCallArg0ReturnToReturn |
+| tests.cpp:434:5:434:29 | [summary] read: Argument[0].Parameter[this] in madCallArg0ReturnToReturn | PostUpdateNode | madCallArg0ReturnToReturn | madCallArg0ReturnToReturn |
+| tests.cpp:434:5:434:29 | [summary] read: Argument[0].ReturnValue in madCallArg0ReturnToReturn | OutNode | madCallArg0ReturnToReturn | madCallArg0ReturnToReturn |
+| tests.cpp:434:5:434:29 | [summary] to write: Argument[0].Parameter[this] in madCallArg0ReturnToReturn | ArgumentNode | madCallArg0ReturnToReturn | madCallArg0ReturnToReturn |
+| tests.cpp:434:5:434:29 | [summary] to write: ReturnValue in madCallArg0ReturnToReturn | ReturnNode | madCallArg0ReturnToReturn | madCallArg0ReturnToReturn |
+| tests.cpp:435:9:435:38 | [summary param] 0 in madCallArg0ReturnToReturnFirst | ParameterNode | madCallArg0ReturnToReturnFirst | madCallArg0ReturnToReturnFirst |
+| tests.cpp:435:9:435:38 | [summary] read: Argument[0].Parameter[this] in madCallArg0ReturnToReturnFirst | PostUpdateNode | madCallArg0ReturnToReturnFirst | madCallArg0ReturnToReturnFirst |
+| tests.cpp:435:9:435:38 | [summary] read: Argument[0].ReturnValue in madCallArg0ReturnToReturnFirst | OutNode | madCallArg0ReturnToReturnFirst | madCallArg0ReturnToReturnFirst |
+| tests.cpp:435:9:435:38 | [summary] to write: Argument[0].Parameter[this] in madCallArg0ReturnToReturnFirst | ArgumentNode | madCallArg0ReturnToReturnFirst | madCallArg0ReturnToReturnFirst |
+| tests.cpp:435:9:435:38 | [summary] to write: ReturnValue in madCallArg0ReturnToReturnFirst | ReturnNode | madCallArg0ReturnToReturnFirst | madCallArg0ReturnToReturnFirst |
+| tests.cpp:435:9:435:38 | [summary] to write: ReturnValue.Field[first] in madCallArg0ReturnToReturnFirst |  | madCallArg0ReturnToReturnFirst | madCallArg0ReturnToReturnFirst |
+| tests.cpp:436:6:436:25 | [summary param] 0 in madCallArg0WithValue | ParameterNode | madCallArg0WithValue | madCallArg0WithValue |
+| tests.cpp:436:6:436:25 | [summary param] 1 in madCallArg0WithValue | ParameterNode | madCallArg0WithValue | madCallArg0WithValue |
+| tests.cpp:436:6:436:25 | [summary] read: Argument[0].Parameter[0] in madCallArg0WithValue | PostUpdateNode | madCallArg0WithValue | madCallArg0WithValue |
+| tests.cpp:436:6:436:25 | [summary] read: Argument[0].Parameter[this] in madCallArg0WithValue | PostUpdateNode | madCallArg0WithValue | madCallArg0WithValue |
+| tests.cpp:436:6:436:25 | [summary] to write: Argument[0].Parameter[0] in madCallArg0WithValue | ArgumentNode | madCallArg0WithValue | madCallArg0WithValue |
+| tests.cpp:436:6:436:25 | [summary] to write: Argument[0].Parameter[this] in madCallArg0WithValue | ArgumentNode | madCallArg0WithValue | madCallArg0WithValue |
+| tests.cpp:436:6:436:25 | [summary] to write: Argument[1] in madCallArg0WithValue | PostUpdateNode | madCallArg0WithValue | madCallArg0WithValue |
+| tests.cpp:437:5:437:36 | [summary param] 1 in madCallReturnValueIgnoreFunction | ParameterNode | madCallReturnValueIgnoreFunction | madCallReturnValueIgnoreFunction |
+| tests.cpp:437:5:437:36 | [summary] to write: ReturnValue in madCallReturnValueIgnoreFunction | ReturnNode | madCallReturnValueIgnoreFunction | madCallReturnValueIgnoreFunction |

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

@@ -0,0 +1,19 @@
+import testModels
+private import semmle.code.cpp.ir.dataflow.internal.DataFlowPrivate
+private import semmle.code.cpp.ir.dataflow.internal.DataFlowUtil
+
+string describe(DataFlow::Node n) {
+  n instanceof ParameterNode and result = "ParameterNode"
+  or
+  n instanceof PostUpdateNode and result = "PostUpdateNode"
+  or
+  n instanceof ArgumentNode and result = "ArgumentNode"
+  or
+  n instanceof ReturnNode and result = "ReturnNode"
+  or
+  n instanceof OutNode and result = "OutNode"
+}
+
+from FlowSummaryNode n
+select n, concat(describe(n), ", "), concat(n.getSummarizedCallable().toString(), ", "),
+  concat(n.getEnclosingCallable().toString(), ", ")

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

@@ -0,0 +1,220 @@
+summaryCalls
+| file://:0:0:0:0 | [summary] call to [summary param] 0 in madCallArg0ReturnToReturn in madCallArg0ReturnToReturn |
+| file://:0:0:0:0 | [summary] call to [summary param] 0 in madCallArg0ReturnToReturnFirst in madCallArg0ReturnToReturnFirst |
+| file://:0:0:0:0 | [summary] call to [summary param] 0 in madCallArg0WithValue in madCallArg0WithValue |
+summarizedCallables
+| tests.cpp:144:5:144:19 | madArg0ToReturn |
+| tests.cpp:145:6:145:28 | madArg0ToReturnIndirect |
+| tests.cpp:147:5:147:28 | madArg0ToReturnValueFlow |
+| tests.cpp:148:5:148:27 | madArg0IndirectToReturn |
+| tests.cpp:149:5:149:33 | madArg0DoubleIndirectToReturn |
+| tests.cpp:150:5:150:30 | madArg0NotIndirectToReturn |
+| tests.cpp:151:6:151:26 | madArg0ToArg1Indirect |
+| tests.cpp:152:6:152:34 | madArg0IndirectToArg1Indirect |
+| tests.cpp:153:5:153:18 | madArgsComplex |
+| tests.cpp:154:5:154:14 | madArgsAny |
+| tests.cpp:155:5:155:28 | madAndImplementedComplex |
+| tests.cpp:160:5:160:24 | madArg0FieldToReturn |
+| tests.cpp:161:5:161:32 | madArg0IndirectFieldToReturn |
+| tests.cpp:162:5:162:32 | madArg0FieldIndirectToReturn |
+| tests.cpp:163:13:163:32 | madArg0ToReturnField |
+| tests.cpp:164:14:164:41 | madArg0ToReturnIndirectField |
+| tests.cpp:165:13:165:40 | madArg0ToReturnFieldIndirect |
+| tests.cpp:284:7:284:19 | madArg0ToSelf |
+| tests.cpp:285:6:285:20 | madSelfToReturn |
+| tests.cpp:287:7:287:20 | madArg0ToField |
+| tests.cpp:288:6:288:21 | madFieldToReturn |
+| tests.cpp:313:7:313:30 | namespaceMadSelfToReturn |
+| tests.cpp:434:5:434:29 | madCallArg0ReturnToReturn |
+| tests.cpp:435:9:435:38 | madCallArg0ReturnToReturnFirst |
+| tests.cpp:436:6:436:25 | madCallArg0WithValue |
+| tests.cpp:437:5:437:36 | madCallReturnValueIgnoreFunction |
+sourceCallables
+| tests.cpp:3:5:3:10 | source |
+| tests.cpp:4:6:4:14 | sourcePtr |
+| tests.cpp:5:6:5:19 | sourceIndirect |
+| tests.cpp:6:6:6:9 | sink |
+| tests.cpp:6:15:6:17 | val |
+| tests.cpp:7:6:7:9 | sink |
+| tests.cpp:7:16:7:18 | ptr |
+| tests.cpp:11:5:11:18 | localMadSource |
+| tests.cpp:12:5:12:19 | remoteMadSource |
+| tests.cpp:13:5:13:14 | notASource |
+| tests.cpp:14:5:14:22 | localMadSourceVoid |
+| tests.cpp:15:5:15:25 | localMadSourceHasBody |
+| tests.cpp:16:6:16:28 | remoteMadSourceIndirect |
+| tests.cpp:17:7:17:35 | remoteMadSourceDoubleIndirect |
+| tests.cpp:18:6:18:32 | remoteMadSourceIndirectArg0 |
+| tests.cpp:18:39:18:39 | x |
+| tests.cpp:18:47:18:47 | y |
+| tests.cpp:19:6:19:32 | remoteMadSourceIndirectArg1 |
+| tests.cpp:19:39:19:39 | x |
+| tests.cpp:19:47:19:47 | y |
+| tests.cpp:20:5:20:22 | remoteMadSourceVar |
+| tests.cpp:21:6:21:31 | remoteMadSourceVarIndirect |
+| tests.cpp:24:6:24:28 | namespaceLocalMadSource |
+| tests.cpp:25:6:25:31 | namespaceLocalMadSourceVar |
+| tests.cpp:28:7:28:30 | namespace2LocalMadSource |
+| tests.cpp:31:6:31:19 | localMadSource |
+| tests.cpp:33:5:33:27 | namespaceLocalMadSource |
+| tests.cpp:35:6:35:17 | test_sources |
+| tests.cpp:50:6:50:6 | v |
+| tests.cpp:51:7:51:16 | v_indirect |
+| tests.cpp:52:6:52:13 | v_direct |
+| tests.cpp:63:6:63:6 | a |
+| tests.cpp:63:9:63:9 | b |
+| tests.cpp:63:12:63:12 | c |
+| tests.cpp:63:15:63:15 | d |
+| tests.cpp:75:6:75:6 | e |
+| tests.cpp:85:6:85:26 | remoteMadSourceParam0 |
+| tests.cpp:85:32:85:32 | x |
+| tests.cpp:92:6:92:16 | madSinkArg0 |
+| tests.cpp:92:22:92:22 | x |
+| tests.cpp:93:6:93:13 | notASink |
+| tests.cpp:93:19:93:19 | x |
+| tests.cpp:94:6:94:16 | madSinkArg1 |
+| tests.cpp:94:22:94:22 | x |
+| tests.cpp:94:29:94:29 | y |
+| tests.cpp:95:6:95:17 | madSinkArg01 |
+| tests.cpp:95:23:95:23 | x |
+| tests.cpp:95:30:95:30 | y |
+| tests.cpp:95:37:95:37 | z |
+| tests.cpp:96:6:96:17 | madSinkArg02 |
+| tests.cpp:96:23:96:23 | x |
+| tests.cpp:96:30:96:30 | y |
+| tests.cpp:96:37:96:37 | z |
+| tests.cpp:97:6:97:24 | madSinkIndirectArg0 |
+| tests.cpp:97:31:97:31 | x |
+| tests.cpp:98:6:98:30 | madSinkDoubleIndirectArg0 |
+| tests.cpp:98:38:98:38 | x |
+| tests.cpp:99:5:99:14 | madSinkVar |
+| tests.cpp:100:6:100:23 | madSinkVarIndirect |
+| tests.cpp:102:6:102:15 | test_sinks |
+| tests.cpp:116:6:116:6 | a |
+| tests.cpp:117:7:117:11 | a_ptr |
+| tests.cpp:132:6:132:18 | madSinkParam0 |
+| tests.cpp:132:24:132:24 | x |
+| tests.cpp:138:8:138:8 | operator= |
+| tests.cpp:138:8:138:8 | operator= |
+| tests.cpp:138:8:138:18 | MyContainer |
+| tests.cpp:139:6:139:10 | value |
+| tests.cpp:140:6:140:11 | value2 |
+| tests.cpp:141:7:141:9 | ptr |
+| tests.cpp:144:25:144:25 | x |
+| tests.cpp:145:34:145:34 | x |
+| tests.cpp:146:5:146:15 | notASummary |
+| tests.cpp:146:21:146:21 | x |
+| tests.cpp:147:34:147:34 | x |
+| tests.cpp:148:34:148:34 | x |
+| tests.cpp:149:41:149:41 | x |
+| tests.cpp:150:37:150:37 | x |
+| tests.cpp:151:32:151:32 | x |
+| tests.cpp:151:40:151:40 | y |
+| tests.cpp:152:47:152:47 | x |
+| tests.cpp:152:55:152:55 | y |
+| tests.cpp:153:25:153:25 | a |
+| tests.cpp:153:33:153:33 | b |
+| tests.cpp:153:40:153:40 | c |
+| tests.cpp:153:47:153:47 | d |
+| tests.cpp:154:20:154:20 | a |
+| tests.cpp:154:28:154:28 | b |
+| tests.cpp:155:34:155:34 | a |
+| tests.cpp:155:41:155:41 | b |
+| tests.cpp:155:48:155:48 | c |
+| tests.cpp:160:38:160:39 | mc |
+| tests.cpp:161:47:161:48 | mc |
+| tests.cpp:162:46:162:47 | mc |
+| tests.cpp:163:38:163:38 | x |
+| tests.cpp:164:47:164:47 | x |
+| tests.cpp:165:46:165:46 | x |
+| tests.cpp:167:13:167:30 | madFieldToFieldVar |
+| tests.cpp:168:13:168:38 | madFieldToIndirectFieldVar |
+| tests.cpp:169:14:169:39 | madIndirectFieldToFieldVar |
+| tests.cpp:171:6:171:19 | test_summaries |
+| tests.cpp:174:6:174:6 | a |
+| tests.cpp:174:9:174:9 | b |
+| tests.cpp:174:12:174:12 | c |
+| tests.cpp:174:15:174:15 | d |
+| tests.cpp:174:18:174:18 | e |
+| tests.cpp:175:7:175:11 | a_ptr |
+| tests.cpp:218:14:218:16 | mc1 |
+| tests.cpp:218:19:218:21 | mc2 |
+| tests.cpp:237:15:237:18 | rtn1 |
+| tests.cpp:240:14:240:17 | rtn2 |
+| tests.cpp:241:7:241:14 | rtn2_ptr |
+| tests.cpp:267:7:267:7 | operator= |
+| tests.cpp:267:7:267:7 | operator= |
+| tests.cpp:267:7:267:13 | MyClass |
+| tests.cpp:270:6:270:26 | memberRemoteMadSource |
+| tests.cpp:271:7:271:39 | memberRemoteMadSourceIndirectArg0 |
+| tests.cpp:271:46:271:46 | x |
+| tests.cpp:272:6:272:29 | memberRemoteMadSourceVar |
+| tests.cpp:273:7:273:21 | qualifierSource |
+| tests.cpp:274:7:274:26 | qualifierFieldSource |
+| tests.cpp:277:7:277:23 | memberMadSinkArg0 |
+| tests.cpp:277:29:277:29 | x |
+| tests.cpp:278:6:278:21 | memberMadSinkVar |
+| tests.cpp:279:7:279:19 | qualifierSink |
+| tests.cpp:280:7:280:23 | qualifierArg0Sink |
+| tests.cpp:280:29:280:29 | x |
+| tests.cpp:281:7:281:24 | qualifierFieldSink |
+| tests.cpp:284:25:284:25 | x |
+| tests.cpp:286:6:286:16 | notASummary |
+| tests.cpp:287:26:287:26 | x |
+| tests.cpp:290:6:290:8 | val |
+| tests.cpp:293:7:293:7 | MyDerivedClass |
+| tests.cpp:293:7:293:7 | operator= |
+| tests.cpp:293:7:293:7 | operator= |
+| tests.cpp:293:7:293:20 | MyDerivedClass |
+| tests.cpp:295:6:295:28 | subtypeRemoteMadSource1 |
+| tests.cpp:296:6:296:21 | subtypeNonSource |
+| tests.cpp:297:6:297:28 | subtypeRemoteMadSource2 |
+| tests.cpp:300:9:300:15 | source2 |
+| tests.cpp:301:6:301:9 | sink |
+| tests.cpp:301:19:301:20 | mc |
+| tests.cpp:304:8:304:8 | operator= |
+| tests.cpp:304:8:304:8 | operator= |
+| tests.cpp:304:8:304:14 | MyClass |
+| tests.cpp:307:8:307:33 | namespaceMemberMadSinkArg0 |
+| tests.cpp:307:39:307:39 | x |
+| tests.cpp:308:15:308:46 | namespaceStaticMemberMadSinkArg0 |
+| tests.cpp:308:52:308:52 | x |
+| tests.cpp:309:7:309:31 | namespaceMemberMadSinkVar |
+| tests.cpp:310:14:310:44 | namespaceStaticMemberMadSinkVar |
+| tests.cpp:317:22:317:28 | source3 |
+| tests.cpp:319:6:319:23 | test_class_members |
+| tests.cpp:320:10:320:11 | mc |
+| tests.cpp:320:14:320:16 | mc2 |
+| tests.cpp:320:19:320:21 | mc3 |
+| tests.cpp:320:24:320:26 | mc4 |
+| tests.cpp:320:29:320:31 | mc5 |
+| tests.cpp:320:34:320:36 | mc6 |
+| tests.cpp:320:39:320:41 | mc7 |
+| tests.cpp:320:44:320:46 | mc8 |
+| tests.cpp:320:49:320:51 | mc9 |
+| tests.cpp:320:54:320:57 | mc10 |
+| tests.cpp:320:60:320:63 | mc11 |
+| tests.cpp:321:11:321:13 | ptr |
+| tests.cpp:321:17:321:23 | mc4_ptr |
+| tests.cpp:322:17:322:19 | mdc |
+| tests.cpp:323:23:323:25 | mnc |
+| tests.cpp:323:28:323:31 | mnc2 |
+| tests.cpp:324:24:324:31 | mnc2_ptr |
+| tests.cpp:330:6:330:6 | a |
+| tests.cpp:429:8:429:8 | operator= |
+| tests.cpp:429:8:429:8 | operator= |
+| tests.cpp:429:8:429:14 | intPair |
+| tests.cpp:430:6:430:10 | first |
+| tests.cpp:431:6:431:11 | second |
+| tests.cpp:434:37:434:43 | fun_ptr |
+| tests.cpp:435:46:435:52 | fun_ptr |
+| tests.cpp:436:34:436:40 | fun_ptr |
+| tests.cpp:436:53:436:57 | value |
+| tests.cpp:437:45:437:51 | fun_ptr |
+| tests.cpp:437:64:437:68 | value |
+| tests.cpp:439:5:439:14 | getTainted |
+| tests.cpp:440:6:440:13 | useValue |
+| tests.cpp:440:19:440:19 | x |
+| tests.cpp:441:6:441:17 | dontUseValue |
+| tests.cpp:441:23:441:23 | x |
+| tests.cpp:443:6:443:27 | test_function_pointers |

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

@@ -0,0 +1,9 @@
+import testModels
+private import semmle.code.cpp.ir.dataflow.internal.DataFlowPrivate
+private import semmle.code.cpp.ir.dataflow.internal.DataFlowUtil
+
+query predicate summaryCalls(SummaryCall c) { any() }
+
+query predicate summarizedCallables(SummarizedCallable c) { any() }
+
+query predicate sourceCallables(SourceCallable c) { c.getLocation().getFile().toString() != "" }

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

@@ -0,0 +1,2 @@
+testFailures
+failures

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

@@ -0,0 +1,18 @@
+import TestUtilities.InlineExpectationsTest
+import testModels
+
+module InterpretElementTest implements TestSig {
+  string getARelevantTag() { result = "interpretElement" }
+
+  predicate hasActualResult(Location location, string element, string tag, string value) {
+    exists(Element e |
+      e = interpretElement(_, _, _, _, _, _) and
+      location = e.getLocation() and
+      element = e.toString() and
+      tag = "interpretElement" and
+      value = ""
+    )
+  }
+}
+
+import MakeTest<InterpretElementTest>

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

@@ -0,0 +1,2 @@
+testFailures
+failures

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

@@ -0,0 +1,32 @@
+import TestUtilities.dataflow.FlowTestCommon
+import testModels
+
+module IRTest {
+  private import semmle.code.cpp.ir.IR
+  private import semmle.code.cpp.ir.dataflow.TaintTracking
+
+  /** Common data flow configuration to be used by tests. */
+  module TestAllocationConfig implements DataFlow::ConfigSig {
+    predicate isSource(DataFlow::Node source) {
+      source instanceof FlowSource
+      or
+      source.asExpr().(FunctionCall).getTarget().getName() =
+        ["source", "source2", "source3", "sourcePtr"]
+      or
+      source.asIndirectExpr(1).(FunctionCall).getTarget().getName() = "sourceIndirect"
+    }
+
+    predicate isSink(DataFlow::Node sink) {
+      sinkNode(sink, "test-sink")
+      or
+      exists(FunctionCall call |
+        call.getTarget().getName() = "sink" and
+        sink.asExpr() = call.getAnArgument()
+      )
+    }
+  }
+
+  module IRFlow = TaintTracking::Global<TestAllocationConfig>;
+}
+
+import MakeTest<IRFlowTest<IRTest::IRFlow>>

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

@@ -0,0 +1,103 @@
+import semmle.code.cpp.security.FlowSources
+
+/**
+ * Models-as-data source models for this test.
+ */
+private class TestSources extends SourceModelCsv {
+  override predicate row(string row) {
+    row =
+      [
+        ";;false;localMadSource;;;ReturnValue;local",
+        ";;false;remoteMadSource;;;ReturnValue;remote",
+        ";;false;localMadSourceVoid;;;ReturnValue;local",
+        ";;false;localMadSourceHasBody;;;ReturnValue;local",
+        ";;false;remoteMadSourceIndirect;;;ReturnValue[*];remote",
+        ";;false;remoteMadSourceDoubleIndirect;;;ReturnValue[**];remote",
+        ";;false;remoteMadSourceIndirectArg0;;;Argument[*0];remote",
+        ";;false;remoteMadSourceIndirectArg1;;;Argument[*1];remote",
+        ";;false;remoteMadSourceVar;;;;remote",
+        ";;false;remoteMadSourceVarIndirect;;;*;remote", // not correctly expressed
+        ";;false;remoteMadSourceParam0;;;Parameter[0];remote",
+        "MyNamespace;;false;namespaceLocalMadSource;;;ReturnValue;local",
+        "MyNamespace;;false;namespaceLocalMadSourceVar;;;;local",
+        "MyNamespace::MyNamespace2;;false;namespace2LocalMadSource;;;ReturnValue;local",
+        ";MyClass;true;memberRemoteMadSource;;;ReturnValue;remote",
+        ";MyClass;true;memberRemoteMadSourceIndirectArg0;;;Argument[*0];remote",
+        ";MyClass;true;memberRemoteMadSourceVar;;;;remote",
+        ";MyClass;true;subtypeRemoteMadSource1;;;ReturnValue;remote",
+        ";MyClass;false;subtypeNonSource;;;ReturnValue;remote", // the tests define this in MyDerivedClass, so it should *not* be recongized as a source
+        ";MyClass;true;qualifierSource;;;Argument[-1];remote",
+        ";MyClass;true;qualifierFieldSource;;;Argument[-1].val;remote",
+        ";MyDerivedClass;false;subtypeRemoteMadSource2;;;ReturnValue;remote",
+      ]
+  }
+}
+
+/**
+ * Models-as-data sink models for this test.
+ */
+private class TestSinks extends SinkModelCsv {
+  override predicate row(string row) {
+    row =
+      [
+        ";;false;madSinkArg0;;;Argument[0];test-sink",
+        ";;false;madSinkArg1;;;Argument[1];test-sink",
+        ";;false;madSinkArg01;;;Argument[0..1];test-sink",
+        ";;false;madSinkArg02;;;Argument[0,2];test-sink",
+        ";;false;madSinkIndirectArg0;;;Argument[*0];test-sink",
+        ";;false;madSinkDoubleIndirectArg0;;;Argument[**0];test-sink",
+        ";;false;madSinkVar;;;;test-sink",
+        ";;false;madSinkVarIndirect;;;*;test-sink", // not correctly expressed
+        ";;false;madSinkParam0;;;Parameter[0];test-sink",
+        ";MyClass;true;memberMadSinkArg0;;;Argument[0];test-sink",
+        ";MyClass;true;memberMadSinkVar;;;;test-sink",
+        ";MyClass;true;qualifierSink;;;Argument[-1];test-sink",
+        ";MyClass;true;qualifierArg0Sink;;;Argument[-1..0];test-sink",
+        ";MyClass;true;qualifierFieldSink;;;Argument[-1].val;test-sink",
+        "MyNamespace;MyClass;true;namespaceMemberMadSinkArg0;;;Argument[0];test-sink",
+        "MyNamespace;MyClass;true;namespaceStaticMemberMadSinkArg0;;;Argument[0];test-sink",
+        "MyNamespace;MyClass;true;namespaceMemberMadSinkVar;;;;test-sink",
+        "MyNamespace;MyClass;true;namespaceStaticMemberMadSinkVar;;;;test-sink",
+      ]
+  }
+}
+
+/**
+ * Models-as-data summary models for this test.
+ */
+private class TestSummaries extends SummaryModelCsv {
+  override predicate row(string row) {
+    row =
+      [
+        ";;false;madArg0ToReturn;;;Argument[0];ReturnValue;taint",
+        ";;false;madArg0ToReturnIndirect;;;Argument[0];ReturnValue[*];taint",
+        ";;false;madArg0ToReturnValueFlow;;;Argument[0];ReturnValue;value",
+        ";;false;madArg0IndirectToReturn;;;Argument[*0];ReturnValue;taint",
+        ";;false;madArg0DoubleIndirectToReturn;;;Argument[**0];ReturnValue;taint",
+        ";;false;madArg0NotIndirectToReturn;;;Argument[0];ReturnValue;taint",
+        ";;false;madArg0ToArg1Indirect;;;Argument[0];Argument[*1];taint",
+        ";;false;madArg0IndirectToArg1Indirect;;;Argument[*0];Argument[*1];taint",
+        ";;false;madArgsComplex;;;Argument[*0..1,2];ReturnValue;taint",
+        ";;false;madAndImplementedComplex;;;Argument[2];ReturnValue;taint",
+        ";;false;madArgsAny;;;Argument;ReturnValue;taint", // (syntax not supported)
+        ";;false;madArg0FieldToReturn;;;Argument[0].value;ReturnValue;taint",
+        ";;false;madArg0IndirectFieldToReturn;;;Argument[*0].value;ReturnValue;taint",
+        ";;false;madArg0FieldIndirectToReturn;;;Argument[0].ptr[*];ReturnValue;taint",
+        ";;false;madArg0ToReturnField;;;Argument[0];ReturnValue.value;taint",
+        ";;false;madArg0ToReturnIndirectField;;;Argument[0];ReturnValue[*].value;taint",
+        ";;false;madArg0ToReturnFieldIndirect;;;Argument[0];ReturnValue.ptr[*];taint",
+        ";;false;madFieldToFieldVar;;;value;value2;taint",
+        ";;false;madFieldToIndirectFieldVar;;;value;ptr[*];taint",
+        ";;false;madIndirectFieldToFieldVar;;;;value;value2;taint", // not correctly expressed
+        ";MyClass;true;madArg0ToSelf;;;Argument[0];Argument[-1];taint",
+        ";MyClass;true;madSelfToReturn;;;Argument[-1];ReturnValue;taint",
+        ";MyClass;true;madArg0ToField;;;Argument[0];Argument[-1].val;taint",
+        ";MyClass;true;madFieldToReturn;;;Argument[-1].val;ReturnValue;taint",
+        "MyNamespace;MyClass;true;namespaceMadSelfToReturn;;;Argument[-1];ReturnValue;taint",
+        ";;false;madCallArg0ReturnToReturn;;;Argument[0].ReturnValue;ReturnValue;value",
+        ";;false;madCallArg0ReturnToReturnFirst;;;Argument[0].ReturnValue;ReturnValue.first;value",
+        ";;false;madCallArg0WithValue;;;Argument[1];Argument[0].Parameter[0];value",
+        ";;false;madCallReturnValueIgnoreFunction;;;Argument[1];ReturnValue;value",
+      ]
+  }
+}

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

@@ -0,0 +1,454 @@
+
+// non-MAD sources / sinks
+int source();
+int *sourcePtr();
+int *sourceIndirect();
+void sink(int val);
+void sink(int *ptr);
+
+// --- global MAD sources ---
+
+int localMadSource(); // $ interpretElement
+int remoteMadSource(); // $ interpretElement
+int notASource();
+int localMadSourceVoid(void); // $ interpretElement
+int localMadSourceHasBody() { return 0; } // $ interpretElement
+int *remoteMadSourceIndirect(); // $ interpretElement
+int **remoteMadSourceDoubleIndirect(); // $ interpretElement
+void remoteMadSourceIndirectArg0(int *x, int *y); // $ interpretElement
+void remoteMadSourceIndirectArg1(int &x, int &y); // $ interpretElement
+int remoteMadSourceVar; // $ interpretElement
+int *remoteMadSourceVarIndirect; // $ interpretElement
+
+namespace MyNamespace {
+	int namespaceLocalMadSource(); // $ interpretElement
+	int namespaceLocalMadSourceVar; // $ interpretElement
+
+	namespace MyNamespace2 {
+		int namespace2LocalMadSource(); // $ interpretElement
+	}
+
+	int localMadSource(); // (not a source)
+}
+int namespaceLocalMadSource(); // (not a source)
+
+void test_sources() {
+	sink(0);
+	sink(source()); // $ ir
+
+	// test sources
+
+	sink(localMadSource()); // $ ir
+	sink(remoteMadSource()); // $ ir
+	sink(notASource());
+	sink(localMadSourceVoid()); // $ ir
+	sink(localMadSourceHasBody()); // $ ir
+
+	sink(sourceIndirect());
+	sink(*sourceIndirect()); // $ ir
+
+	int v = localMadSource();
+	int *v_indirect = &v;
+	int v_direct = *v_indirect;
+	sink(v); // $ ir
+	sink(v_indirect);
+	sink(*v_indirect); // $ ir
+	sink(v_direct); // $ ir
+
+	sink(remoteMadSourceIndirect());
+	sink(*remoteMadSourceIndirect()); // $ MISSING: ir
+	sink(*remoteMadSourceDoubleIndirect());
+	sink(**remoteMadSourceDoubleIndirect()); // $ MISSING: ir
+
+	int a, b, c, d;
+
+	remoteMadSourceIndirectArg0(&a, &b);
+	sink(a); // $ ir
+	sink(b);
+	remoteMadSourceIndirectArg1(c, d);
+	sink(c);
+	sink(d); // $ ir
+
+	sink(remoteMadSourceVar); // $ ir
+	sink(*remoteMadSourceVarIndirect); // $ MISSING: ir
+
+	int e = localMadSource();
+	sink(e); // $ ir
+
+	sink(MyNamespace::namespaceLocalMadSource()); // $: ir
+	sink(MyNamespace::namespaceLocalMadSourceVar); // $ ir
+	sink(MyNamespace::MyNamespace2::namespace2LocalMadSource()); // $ ir
+	sink(MyNamespace::localMadSource()); // $ (the MyNamespace version of this function is not a source)
+	sink(namespaceLocalMadSource()); // (the global namespace version of this function is not a source)
+}
+
+void remoteMadSourceParam0(int x) // $ interpretElement
+{
+	sink(x); // $ ir
+}
+
+// --- global MAD sinks ---
+
+void madSinkArg0(int x); // $ interpretElement
+void notASink(int x);
+void madSinkArg1(int x, int y); // $ interpretElement
+void madSinkArg01(int x, int y, int z); // $ interpretElement
+void madSinkArg02(int x, int y, int z); // $ interpretElement
+void madSinkIndirectArg0(int *x); // $ interpretElement
+void madSinkDoubleIndirectArg0(int **x); // $ interpretElement
+int madSinkVar; // $ interpretElement
+int *madSinkVarIndirect; // $ interpretElement
+
+void test_sinks() {
+	// test sinks
+
+	madSinkArg0(source()); // $ ir
+	notASink(source());
+	madSinkArg1(source(), 0);
+	madSinkArg1(0, source()); // $ ir
+	madSinkArg01(source(), 0, 0); // $ ir
+	madSinkArg01(0, source(), 0); // $ ir
+	madSinkArg01(0, 0, source());
+	madSinkArg02(source(), 0, 0); // $ ir
+	madSinkArg02(0, source(), 0);
+	madSinkArg02(0, 0, source()); // $ ir
+
+	int a = source();
+	int *a_ptr = &a;
+	madSinkIndirectArg0(&a); // $ ir
+	madSinkIndirectArg0(a_ptr); // $ ir
+	madSinkDoubleIndirectArg0(&a_ptr); // $ ir
+
+	madSinkVar = source();  // $ ir
+
+	// test sources + sinks together
+
+	madSinkArg0(localMadSource()); // $ ir
+	madSinkIndirectArg0(remoteMadSourceIndirect()); // $ MISSING: ir
+	madSinkVar = remoteMadSourceVar; // $ ir
+	*madSinkVarIndirect = remoteMadSourceVar; // $ MISSING: ir
+}
+
+void madSinkParam0(int x) { // $ interpretElement
+	x = source(); // $ MISSING: ir
+}
+
+// --- global MAD summaries ---
+
+struct MyContainer {
+	int value;
+	int value2;
+	int *ptr;
+};
+
+int madArg0ToReturn(int x); // $ interpretElement
+int *madArg0ToReturnIndirect(int x); // $ interpretElement
+int notASummary(int x);
+int madArg0ToReturnValueFlow(int x); // $ interpretElement
+int madArg0IndirectToReturn(int *x); // $ interpretElement
+int madArg0DoubleIndirectToReturn(int **x); // $ interpretElement
+int madArg0NotIndirectToReturn(int *x); // $ interpretElement
+void madArg0ToArg1Indirect(int x, int &y); // $ interpretElement
+void madArg0IndirectToArg1Indirect(const int *x, int *y); // $ interpretElement
+int madArgsComplex(int *a, int *b, int c, int d); // $ interpretElement
+int madArgsAny(int a, int *b); // $ interpretElement
+int madAndImplementedComplex(int a, int b, int c) { // $ interpretElement
+	// (`b` can be seen to flow to the return value in code, `c` via the MAD model)
+	return b;
+}
+
+int madArg0FieldToReturn(MyContainer mc); // $ interpretElement
+int madArg0IndirectFieldToReturn(MyContainer *mc); // $ interpretElement
+int madArg0FieldIndirectToReturn(MyContainer mc); // $ interpretElement
+MyContainer madArg0ToReturnField(int x); // $ interpretElement
+MyContainer *madArg0ToReturnIndirectField(int x); // $ interpretElement
+MyContainer madArg0ToReturnFieldIndirect(int x); // $ interpretElement
+
+MyContainer madFieldToFieldVar; // $ interpretElement
+MyContainer madFieldToIndirectFieldVar; // $ interpretElement
+MyContainer *madIndirectFieldToFieldVar; // $ interpretElement
+
+void test_summaries() {
+	// test summaries
+
+	int a, b, c, d, e;
+	int *a_ptr;
+
+	sink(madArg0ToReturn(0));
+	sink(madArg0ToReturn(source())); // $ ir
+	sink(*madArg0ToReturnIndirect(0));
+	sink(*madArg0ToReturnIndirect(source())); // $ ir
+	sink(notASummary(source()));
+	sink(madArg0ToReturnValueFlow(0));
+	sink(madArg0ToReturnValueFlow(source())); // $ ir
+
+	a = source();
+	a_ptr = &a;
+	sink(madArg0IndirectToReturn(&a)); // $ ir
+	sink(madArg0IndirectToReturn(a_ptr)); // $ ir
+	sink(madArg0DoubleIndirectToReturn(&a_ptr)); // $ ir
+	sink(madArg0NotIndirectToReturn(a_ptr));
+	sink(madArg0NotIndirectToReturn(sourcePtr())); // $ ir
+	sink(madArg0NotIndirectToReturn(sourceIndirect()));
+
+	madArg0ToArg1Indirect(source(), b);
+	sink(b); // $ ir
+
+	madArg0IndirectToArg1Indirect(&a, &c);
+	sink(c); // $ ir
+
+	sink(madArgsComplex(0, 0, 0, 0));
+	sink(madArgsComplex(sourceIndirect(), 0, 0, 0)); // $ ir
+	sink(madArgsComplex(0, sourceIndirect(), 0, 0)); // $ ir
+	sink(madArgsComplex(0, 0, source(), 0)); // $ ir
+	sink(madArgsComplex(0, 0, 0, source()));
+
+	sink(madAndImplementedComplex(0, 0, 0));
+	sink(madAndImplementedComplex(source(), 0, 0));
+	sink(madAndImplementedComplex(0, source(), 0)); // $ ir
+	sink(madAndImplementedComplex(0, 0, source())); // $ ir
+
+	sink(madArgsAny(0, 0));
+	sink(madArgsAny(source(), 0)); // (syntax not supported)
+	sink(madArgsAny(0, sourcePtr())); // (syntax not supported)
+	sink(madArgsAny(0, sourceIndirect())); // (syntax not supported)
+
+	// test summaries involving structs / fields
+
+	MyContainer mc1, mc2;
+
+	d = 0;
+	mc1.value = 0;
+	mc1.ptr = &d;
+	sink(madArg0FieldToReturn(mc1));
+	sink(madArg0IndirectFieldToReturn(&mc1));
+	sink(madArg0FieldIndirectToReturn(mc1));
+
+	e = source();
+	mc2.value = source();
+	mc2.ptr = &e;
+	sink(madArg0FieldToReturn(mc2)); // $ ir
+	sink(madArg0IndirectFieldToReturn(&mc2)); // $ ir
+	sink(madArg0FieldIndirectToReturn(mc2)); // $ ir
+
+	sink(madArg0ToReturnField(0).value);
+	sink(madArg0ToReturnField(source()).value); // $ ir
+
+	MyContainer *rtn1 = madArg0ToReturnIndirectField(source());
+	sink(rtn1->value); // $ ir
+
+	MyContainer rtn2 = madArg0ToReturnFieldIndirect(source());
+	int *rtn2_ptr = rtn2.ptr;
+	sink(*rtn2_ptr); // $ ir
+
+	// test global variable summaries
+
+	madFieldToFieldVar.value = source();
+	sink(madFieldToFieldVar.value2); // $ MISSING: ir
+
+	madFieldToIndirectFieldVar.value = source();
+	sink(madFieldToIndirectFieldVar.ptr);
+	sink(*(madFieldToIndirectFieldVar.ptr)); // $ MISSING: ir
+
+	madIndirectFieldToFieldVar->value = source();
+	sink((*madIndirectFieldToFieldVar).value2); // $ MISSING: ir
+	sink(madIndirectFieldToFieldVar->value2); // $ MISSING: ir
+
+	// test source + sinks + summaries together
+
+	madSinkArg0(madArg0ToReturn(remoteMadSource())); // $ ir
+	madSinkArg0(madArg0ToReturnValueFlow(remoteMadSource())); // $ ir
+	madSinkArg0(madArg0IndirectToReturn(sourcePtr()));
+	madSinkArg0(madArg0IndirectToReturn(sourceIndirect())); // $ ir
+}
+
+// --- MAD class members ---
+
+class MyClass {
+public:
+	// sources
+	int memberRemoteMadSource(); // $ interpretElement
+	void memberRemoteMadSourceIndirectArg0(int *x); // $ interpretElement
+	int memberRemoteMadSourceVar; // $ interpretElement
+	void qualifierSource(); // $ interpretElement
+	void qualifierFieldSource(); // $ interpretElement
+
+	// sinks
+	void memberMadSinkArg0(int x); // $ interpretElement
+	int memberMadSinkVar; // $ interpretElement
+	void qualifierSink(); // $ interpretElement
+	void qualifierArg0Sink(int x); // $ interpretElement
+	void qualifierFieldSink(); // $ interpretElement
+
+	// summaries
+	void madArg0ToSelf(int x); // $ interpretElement
+	int madSelfToReturn(); // $ interpretElement
+	int notASummary();
+	void madArg0ToField(int x); // $ interpretElement
+	int madFieldToReturn(); // $ interpretElement
+
+	int val;
+};
+
+class MyDerivedClass : public MyClass {
+public:
+	int subtypeRemoteMadSource1(); // $ interpretElement
+	int subtypeNonSource();
+	int subtypeRemoteMadSource2(); // $ interpretElement
+};
+
+MyClass source2();
+void sink(MyClass mc);
+
+namespace MyNamespace {
+	class MyClass {
+	public:
+		// sinks
+		void namespaceMemberMadSinkArg0(int x); // $ interpretElement
+		static void namespaceStaticMemberMadSinkArg0(int x); // $ interpretElement
+		int namespaceMemberMadSinkVar; // $ interpretElement
+		static int namespaceStaticMemberMadSinkVar; // $ interpretElement
+
+		// summaries
+		int namespaceMadSelfToReturn(); // $ interpretElement
+	};
+}
+
+MyNamespace::MyClass source3();
+
+void test_class_members() {
+	MyClass mc, mc2, mc3, mc4, mc5, mc6, mc7, mc8, mc9, mc10, mc11;
+	MyClass *ptr, *mc4_ptr;
+	MyDerivedClass mdc;
+	MyNamespace::MyClass mnc, mnc2;
+	MyNamespace::MyClass *mnc2_ptr;
+
+	// test class member sources
+
+	sink(mc.memberRemoteMadSource()); // $ ir
+
+	int a;
+	mc.memberRemoteMadSourceIndirectArg0(&a);
+	sink(a); // $ ir
+
+	sink(mc.memberRemoteMadSourceVar); // $ ir
+
+	// test subtype sources
+
+	sink(mdc.memberRemoteMadSource()); // $ ir
+	sink(mdc.subtypeRemoteMadSource1()); // $ ir
+	sink(mdc.subtypeNonSource());
+	sink(mdc.subtypeRemoteMadSource2()); // $ ir
+
+	// test class member sinks
+
+	mc.memberMadSinkArg0(source()); // $ ir
+
+	mc.memberMadSinkVar = source(); // $ ir
+
+	mnc.namespaceMemberMadSinkArg0(source()); // $ ir
+	MyNamespace::MyClass::namespaceStaticMemberMadSinkArg0(source()); // $ ir
+	mnc.namespaceMemberMadSinkVar = source(); // $ ir
+	MyNamespace::MyClass::namespaceStaticMemberMadSinkVar = source(); // $ ir
+
+	// test class member summaries
+
+	sink(mc2);
+	mc2.madArg0ToSelf(0);
+	sink(mc2);
+	mc2.madArg0ToSelf(source());
+	sink(mc2); // $ ir
+
+	ptr = new MyClass();
+	sink(*ptr);
+	ptr->madArg0ToSelf(0);
+	sink(*ptr);
+	ptr->madArg0ToSelf(source());
+	sink(*ptr); // $ ir
+
+	mc3.madArg0ToField(source());
+	sink(mc3.val); // $ ir
+
+	mc4 = source2();
+	mc4_ptr = &mc4;
+	sink(mc4); // $ ir
+	sink(mc4.madSelfToReturn()); // $ ir
+	sink(mc4.notASummary());
+	sink(mc4_ptr->madSelfToReturn()); // $ ir
+	sink(mc4_ptr->notASummary());
+	sink(source2().madSelfToReturn()); // $ ir
+	sink(source2().notASummary());
+
+	mc5.val = source();
+	sink(mc5.madFieldToReturn()); // $ ir
+
+	mnc2 = source3();
+	mnc2_ptr = &mnc2;
+	sink(mnc2.namespaceMadSelfToReturn()); // $ ir
+	sink(mnc2_ptr->namespaceMadSelfToReturn()); // $ ir
+	sink(source3().namespaceMadSelfToReturn()); // $ ir
+
+	// test class member sources + sinks + summaries together
+
+	mc.memberMadSinkArg0(mc.memberRemoteMadSource()); // $ ir
+
+	mc6.madArg0ToSelf(source());
+	sink(mc6.madSelfToReturn()); // $ ir
+
+	mc7.madArg0ToField(source());
+	sink(mc7.madFieldToReturn()); // $ ir
+
+	// test taint involving qualifier
+
+	sink(mc8);
+	mc8.qualifierArg0Sink(0);
+	mc8.qualifierArg0Sink(source()); // $ ir
+
+	mc9 = source2();
+	mc9.qualifierSink(); // $ ir
+	mc9.qualifierArg0Sink(0); // $ ir
+
+	mc8.qualifierSource();
+	sink(mc8); // $ ir
+	mc8.qualifierSink(); // $ ir
+	mc9.qualifierArg0Sink(0); // $ ir
+
+	// test taint involving qualifier field
+
+	sink(mc10.val);
+	mc10.qualifierFieldSource();
+	sink(mc10.val); // $ MISSING: ir
+
+	mc11.val = source();
+	sink(mc11.val); // $ ir
+	mc11.qualifierFieldSink(); // $ MISSING: ir
+}
+
+// --- MAD cases involving function pointers ---
+
+struct intPair {
+	int first;
+	int second;
+};
+
+int madCallArg0ReturnToReturn(int (*fun_ptr)()); // $ interpretElement
+intPair madCallArg0ReturnToReturnFirst(int (*fun_ptr)()); // $ interpretElement
+void madCallArg0WithValue(void (*fun_ptr)(int), int value); // $ interpretElement
+int madCallReturnValueIgnoreFunction(void (*fun_ptr)(int), int value); // $ interpretElement
+
+int getTainted() { return source(); }
+void useValue(int x) { sink(x); } // $ ir
+void dontUseValue(int x) { }
+
+void test_function_pointers() {
+	sink(madCallArg0ReturnToReturn(&notASource));
+	sink(madCallArg0ReturnToReturn(&getTainted)); // $ ir
+	sink(madCallArg0ReturnToReturn(&source)); // $ MISSING: ir
+	sink(madCallArg0ReturnToReturnFirst(&getTainted).first); // $ ir
+	sink(madCallArg0ReturnToReturnFirst(&getTainted).second);
+
+	madCallArg0WithValue(&useValue, source());
+	madCallArg0WithValue(&sink, source()); // $ MISSING: ir
+	madCallReturnValueIgnoreFunction(&sink, source());
+	sink(madCallReturnValueIgnoreFunction(&dontUseValue, source())); // $ ir
+}

cpp/ql/test/library-tests/dataflow/source-sink-tests/local-flow.expected

@@ -1,2 +1,2 @@
-failures
 testFailures
+failures

cpp/ql/test/library-tests/dataflow/source-sink-tests/sources-and-sinks.cpp

@@ -50,3 +50,68 @@ void test_inet(char *hostname, char *servname, struct addrinfo *hints) {
   addrinfo *res;
   int ret = getaddrinfo(hostname, servname, hints, &res); // $ remote_source
 }
+
+typedef unsigned int wint_t;
+
+// getc variants
+int getc(FILE *stream);
+wint_t getwc(FILE *stream);
+int _getc_nolock(FILE *stream);
+wint_t _getwc_nolock(FILE *stream);
+
+int getch(void);
+int _getch(void);
+wint_t _getwch(void);
+int _getch_nolock(void);
+wint_t _getwch_nolock(void);
+int getchar(void);
+wint_t getwchar();
+int _getchar_nolock(void);
+wint_t _getwchar_nolock(void);
+
+void test_getchar(FILE *stream) {
+  int a = getc(stream); // $ remote_source
+  wint_t b = getwc(stream); // $ remote_source
+  int c = _getc_nolock(stream); // $ remote_source
+  wint_t d = _getwc_nolock(stream); // $ remote_source
+
+  int e = getch(); // $ local_source
+  int f = _getch(); // $ local_source
+  wint_t g = _getwch(); // $ local_source
+  int h = _getch_nolock(); // $ local_source
+  wint_t i = _getwch_nolock(); // $ local_source
+  int j = getchar(); // $ local_source
+  wint_t k = getwchar(); // $ local_source
+  int l = _getchar_nolock(); // $ local_source
+  wint_t m = _getwchar_nolock(); // $ local_source
+}
+
+// ZMC networking library
+
+typedef unsigned long size_t;
+
+struct zmq_msg_t {
+};
+int zmq_msg_init(zmq_msg_t *msg);
+int zmq_msg_recv(zmq_msg_t *msg, void *socket, int flags);
+int zmq_recvmsg(void *socket, zmq_msg_t *msg, int flags); // deprecated
+int zmq_recv(void *socket, void *buf, size_t len, int flags);
+
+void test_zmc(void *socket) {
+  zmq_msg_t msg1, msg2;
+  char buffer[1024];
+
+  if (zmq_recv(socket, buffer, sizeof(buffer), 0) >= 0) { // $ remote_source
+    // ...
+  }
+
+  zmq_msg_init(&msg1);
+  if (zmq_msg_recv(&msg1, socket, 0) >= 0) { // $ remote_source
+    // ...
+  }
+
+  zmq_msg_init(&msg2);
+  if (zmq_recvmsg(socket, &msg2, 0) >= 0) { // $ remote_source
+    // ...
+  }
+}

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

@@ -8223,3 +8223,50 @@ WARNING: Module TaintTracking has been deprecated and may be removed in future (
 | vector.cpp:531:9:531:10 | it | vector.cpp:531:8:531:8 | call to operator* | TAINT |
 | vector.cpp:532:8:532:9 | ref arg vs | vector.cpp:533:2:533:2 | vs |  |
 | vector.cpp:532:8:532:9 | vs | vector.cpp:532:10:532:10 | call to operator[] | TAINT |
+| zmq.cpp:17:21:17:26 | socket | zmq.cpp:17:21:17:26 | socket |  |
+| zmq.cpp:17:35:17:46 | message_data | zmq.cpp:17:35:17:46 | message_data |  |
+| zmq.cpp:17:35:17:46 | message_data | zmq.cpp:20:35:20:46 | message_data |  |
+| zmq.cpp:17:35:17:46 | message_data | zmq.cpp:25:3:25:14 | message_data |  |
+| zmq.cpp:17:35:17:46 | message_data | zmq.cpp:26:8:26:19 | message_data |  |
+| zmq.cpp:17:35:17:46 | message_data | zmq.cpp:28:35:28:46 | message_data |  |
+| zmq.cpp:17:56:17:66 | message_len | zmq.cpp:20:49:20:59 | message_len |  |
+| zmq.cpp:17:56:17:66 | message_len | zmq.cpp:28:49:28:59 | message_len |  |
+| zmq.cpp:18:13:18:19 | message | zmq.cpp:20:26:20:32 | message |  |
+| zmq.cpp:18:13:18:19 | message | zmq.cpp:21:10:21:16 | message |  |
+| zmq.cpp:18:13:18:19 | message | zmq.cpp:22:24:22:30 | message |  |
+| zmq.cpp:18:13:18:19 | message | zmq.cpp:28:26:28:32 | message |  |
+| zmq.cpp:18:13:18:19 | message | zmq.cpp:29:10:29:16 | message |  |
+| zmq.cpp:18:13:18:19 | message | zmq.cpp:30:24:30:30 | message |  |
+| zmq.cpp:20:25:20:32 | ref arg & ... | zmq.cpp:20:26:20:32 | message [inner post update] |  |
+| zmq.cpp:20:25:20:32 | ref arg & ... | zmq.cpp:21:10:21:16 | message |  |
+| zmq.cpp:20:25:20:32 | ref arg & ... | zmq.cpp:22:24:22:30 | message |  |
+| zmq.cpp:20:25:20:32 | ref arg & ... | zmq.cpp:28:26:28:32 | message |  |
+| zmq.cpp:20:25:20:32 | ref arg & ... | zmq.cpp:29:10:29:16 | message |  |
+| zmq.cpp:20:25:20:32 | ref arg & ... | zmq.cpp:30:24:30:30 | message |  |
+| zmq.cpp:20:26:20:32 | message | zmq.cpp:20:25:20:32 | & ... |  |
+| zmq.cpp:20:35:20:46 | ref arg message_data | zmq.cpp:17:35:17:46 | message_data |  |
+| zmq.cpp:20:35:20:46 | ref arg message_data | zmq.cpp:25:3:25:14 | message_data |  |
+| zmq.cpp:20:35:20:46 | ref arg message_data | zmq.cpp:26:8:26:19 | message_data |  |
+| zmq.cpp:20:35:20:46 | ref arg message_data | zmq.cpp:28:35:28:46 | message_data |  |
+| zmq.cpp:22:23:22:30 | ref arg & ... | zmq.cpp:22:24:22:30 | message [inner post update] |  |
+| zmq.cpp:22:23:22:30 | ref arg & ... | zmq.cpp:28:26:28:32 | message |  |
+| zmq.cpp:22:23:22:30 | ref arg & ... | zmq.cpp:29:10:29:16 | message |  |
+| zmq.cpp:22:23:22:30 | ref arg & ... | zmq.cpp:30:24:30:30 | message |  |
+| zmq.cpp:22:24:22:30 | message | zmq.cpp:22:23:22:30 | & ... |  |
+| zmq.cpp:25:3:25:14 | message_data | zmq.cpp:25:3:25:17 | access to array | TAINT |
+| zmq.cpp:25:3:25:17 | access to array [post update] | zmq.cpp:17:35:17:46 | message_data |  |
+| zmq.cpp:25:3:25:17 | access to array [post update] | zmq.cpp:25:3:25:14 | message_data [inner post update] |  |
+| zmq.cpp:25:3:25:17 | access to array [post update] | zmq.cpp:26:8:26:19 | message_data |  |
+| zmq.cpp:25:3:25:17 | access to array [post update] | zmq.cpp:28:35:28:46 | message_data |  |
+| zmq.cpp:25:3:25:28 | ... = ... | zmq.cpp:25:3:25:17 | access to array [post update] |  |
+| zmq.cpp:25:16:25:16 | 0 | zmq.cpp:25:3:25:17 | access to array | TAINT |
+| zmq.cpp:25:21:25:26 | call to source | zmq.cpp:25:3:25:28 | ... = ... |  |
+| zmq.cpp:26:8:26:19 | ref arg message_data | zmq.cpp:17:35:17:46 | message_data |  |
+| zmq.cpp:26:8:26:19 | ref arg message_data | zmq.cpp:28:35:28:46 | message_data |  |
+| zmq.cpp:28:25:28:32 | ref arg & ... | zmq.cpp:28:26:28:32 | message [inner post update] |  |
+| zmq.cpp:28:25:28:32 | ref arg & ... | zmq.cpp:29:10:29:16 | message |  |
+| zmq.cpp:28:25:28:32 | ref arg & ... | zmq.cpp:30:24:30:30 | message |  |
+| zmq.cpp:28:26:28:32 | message | zmq.cpp:28:25:28:32 | & ... |  |
+| zmq.cpp:28:35:28:46 | ref arg message_data | zmq.cpp:17:35:17:46 | message_data |  |
+| zmq.cpp:30:23:30:30 | ref arg & ... | zmq.cpp:30:24:30:30 | message [inner post update] |  |
+| zmq.cpp:30:24:30:30 | message | zmq.cpp:30:23:30:30 | & ... |  |

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

@@ -0,0 +1,32 @@
+
+int source();
+void sink(...);
+
+// --- ZMC networking library ---
+
+typedef unsigned long size_t;
+
+struct zmq_msg_t {
+  // ...
+};
+typedef void (*zmq_free_fn)();
+
+int zmq_msg_init_data(zmq_msg_t *msg, void *data, size_t size, zmq_free_fn *ffn, void *hint);
+void *zmq_msg_data(zmq_msg_t *msg);
+
+void test_zmc(void *socket, char *message_data, size_t message_len) {
+  zmq_msg_t message;
+
+  if (zmq_msg_init_data(&message, message_data, message_len, 0, 0)) {
+    sink(message); // $ SPURIOUS: ast
+    sink(zmq_msg_data(&message));
+  }
+
+  message_data[0] = source();
+  sink(message_data); // $ ast,ir
+
+  if (zmq_msg_init_data(&message, message_data, message_len, 0, 0)) {
+    sink(message); // $ ast,ir
+    sink(zmq_msg_data(&message)); // $ ir MISSING: ast
+  }
+}

cpp/ql/test/library-tests/fields/fields/Fields.expected

@@ -1,13 +1,18 @@
 | fields.cpp:3:8:3:12 | Entry | fields.cpp:4:9:4:12 | name | public | CharPointerType | char |
+| fields.cpp:3:8:3:12 | Entry | fields.cpp:4:9:4:12 | name | public | PointerOrArrayOrReferenceType | char |
 | fields.cpp:3:8:3:12 | Entry | fields.cpp:5:8:5:8 | t | public | Enum |  |
 | fields.cpp:3:8:3:12 | Entry | fields.cpp:6:9:6:9 | s | public | CharPointerType | char |
+| fields.cpp:3:8:3:12 | Entry | fields.cpp:6:9:6:9 | s | public | PointerOrArrayOrReferenceType | char |
 | fields.cpp:3:8:3:12 | Entry | fields.cpp:7:7:7:7 | i | public | IntType |  |
 | fields.cpp:3:8:3:12 | Entry | fields.cpp:7:7:7:7 | i | public | MicrosoftInt32Type |  |
 | fields.cpp:3:8:3:12 | Entry | fields.cpp:9:7:9:14 | internal | private | IntType |  |
 | fields.cpp:3:8:3:12 | Entry | fields.cpp:9:7:9:14 | internal | private | MicrosoftInt32Type |  |
+| fields.cpp:12:7:12:10 | Name | fields.cpp:13:15:13:15 | s | private | PointerOrArrayOrReferenceType | const char |
 | fields.cpp:12:7:12:10 | Name | fields.cpp:13:15:13:15 | s | private | PointerType | const char |
+| fields.cpp:16:7:16:11 | Table | fields.cpp:17:9:17:9 | p | private | PointerOrArrayOrReferenceType | Name |
 | fields.cpp:16:7:16:11 | Table | fields.cpp:17:9:17:9 | p | private | PointerType | Name |
 | fields.cpp:16:7:16:11 | Table | fields.cpp:18:7:18:8 | sz | private | IntType |  |
 | fields.cpp:16:7:16:11 | Table | fields.cpp:18:7:18:8 | sz | private | MicrosoftInt32Type |  |
 | fields.cpp:26:7:26:10 | Date | fields.cpp:28:16:28:26 | cache_valid | private | BoolType |  |
 | fields.cpp:26:7:26:10 | Date | fields.cpp:30:17:30:21 | cache | public | CharPointerType | char |
+| fields.cpp:26:7:26:10 | Date | fields.cpp:30:17:30:21 | cache | public | PointerOrArrayOrReferenceType | char |

cpp/ql/test/library-tests/types/__wchar_t/wchar_t.expected

@@ -1,3 +1,3 @@
-| file://:0:0:0:0 | __wchar_t * | IteratorByPointer, PointerType | Wchar_t, WideCharType |
+| file://:0:0:0:0 | __wchar_t * | IteratorByPointer, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection, PointerType | Wchar_t, WideCharType |
 | file://:0:0:0:0 | const __wchar_t | SpecifiedType | Wchar_t, WideCharType |
 | file://:0:0:0:0 | wchar_t | Wchar_t, WideCharType |  |

cpp/ql/test/library-tests/types/wchar_t_typedef/wchar_t.expected

@@ -1,3 +1,3 @@
 | file://:0:0:0:0 | wchar_t | Wchar_t, WideCharType |  |
-| file://:0:0:0:0 | wchar_t * | IteratorByPointer, PointerType | CTypedefType, Wchar_t |
+| file://:0:0:0:0 | wchar_t * | IteratorByPointer, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection, PointerType | CTypedefType, Wchar_t |
 | ms.c:2:24:2:30 | wchar_t | CTypedefType, Wchar_t |  |

cpp/ql/test/library-tests/variables/variables/types.expected

@@ -27,59 +27,59 @@
 | __fp16 | BinaryFloatingPointType, RealNumberType |  |  |  |  |
 | __int128 | Int128Type |  |  |  |  |
 | __va_list_tag | DirectAccessHolder, MetricClass, Struct, StructLikeClass |  |  |  |  |
-| __va_list_tag & | LValueReferenceType |  | __va_list_tag |  |  |
-| __va_list_tag && | RValueReferenceType |  | __va_list_tag |  |  |
+| __va_list_tag & | LValueReferenceType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection |  | __va_list_tag |  |  |
+| __va_list_tag && | PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection, RValueReferenceType |  | __va_list_tag |  |  |
 | address | DirectAccessHolder, MetricClass, Struct, StructLikeClass |  |  |  |  |
-| address & | LValueReferenceType |  | address |  |  |
-| address && | RValueReferenceType |  | address |  |  |
+| address & | LValueReferenceType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection |  | address |  |  |
+| address && | PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection, RValueReferenceType |  | address |  |  |
 | auto | AutoType |  |  |  |  |
 | bool | BoolType |  |  |  |  |
 | char | MicrosoftInt8Type, PlainCharType |  |  |  |  |
 | char8_t | Char8Type |  |  |  |  |
 | char16_t | Char16Type |  |  |  |  |
 | char32_t | Char32Type |  |  |  |  |
-| char * | CharPointerType, IteratorByPointer |  | char |  |  |
-| char *[3] | ArrayType | char * | char * |  |  |
-| char *[32] | ArrayType | char * | char * |  |  |
-| char *[] | ArrayType | char * | char * |  |  |
-| char[2] | ArrayType | char | char |  |  |
-| char[3] | ArrayType | char | char |  |  |
-| char[5] | ArrayType | char | char |  |  |
-| char[6] | ArrayType | char | char |  |  |
-| char[8] | ArrayType | char | char |  |  |
-| char[9] | ArrayType | char | char |  |  |
-| char[10] | ArrayType | char | char |  |  |
-| char[53] | ArrayType | char | char |  |  |
-| char[] | ArrayType | char | char |  |  |
+| char * | CharPointerType, IteratorByPointer, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection |  | char |  |  |
+| char *[3] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | char * | char * |  |  |
+| char *[32] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | char * | char * |  |  |
+| char *[] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | char * | char * |  |  |
+| char[2] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | char | char |  |  |
+| char[3] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | char | char |  |  |
+| char[5] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | char | char |  |  |
+| char[6] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | char | char |  |  |
+| char[8] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | char | char |  |  |
+| char[9] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | char | char |  |  |
+| char[10] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | char | char |  |  |
+| char[53] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | char | char |  |  |
+| char[] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | char | char |  |  |
 | const __va_list_tag | SpecifiedType |  | __va_list_tag |  |  |
-| const __va_list_tag & | LValueReferenceType |  | const __va_list_tag |  |  |
+| const __va_list_tag & | LValueReferenceType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection |  | const __va_list_tag |  |  |
 | const address | SpecifiedType |  | address |  |  |
-| const address & | LValueReferenceType |  | const address |  |  |
+| const address & | LValueReferenceType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection |  | const address |  |  |
 | const char | SpecifiedType |  | char |  |  |
-| const char * | IteratorByPointer, PointerType |  | const char |  |  |
-| const char *[3] | ArrayType | const char * | const char * |  |  |
-| const char *[] | ArrayType | const char * | const char * |  |  |
-| const char[5] | ArrayType | const char | const char |  |  |
-| const char[6] | ArrayType | const char | const char |  |  |
-| const char[8] | ArrayType | const char | const char |  |  |
-| const char[9] | ArrayType | const char | const char |  |  |
-| const char[10] | ArrayType | const char | const char |  |  |
-| const char[53] | ArrayType | const char | const char |  |  |
+| const char * | IteratorByPointer, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection, PointerType |  | const char |  |  |
+| const char *[3] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | const char * | const char * |  |  |
+| const char *[] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | const char * | const char * |  |  |
+| const char[5] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | const char | const char |  |  |
+| const char[6] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | const char | const char |  |  |
+| const char[8] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | const char | const char |  |  |
+| const char[9] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | const char | const char |  |  |
+| const char[10] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | const char | const char |  |  |
+| const char[53] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | const char | const char |  |  |
 | const double | SpecifiedType |  | double |  |  |
 | const int | SpecifiedType |  | int |  |  |
 | decltype(nullptr) | NullPointerType |  |  |  |  |
 | double | DoubleType |  |  |  |  |
 | error | ErroneousType |  |  |  |  |
 | float | FloatType |  |  |  |  |
-| float[3] | ArrayType | float | float |  |  |
+| float[3] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | float | float |  |  |
 | int | IntType, MicrosoftInt32Type |  |  |  |  |
-| int * | IntPointerType, IteratorByPointer |  | int |  |  |
-| int[4] | ArrayType | int | int |  |  |
-| int[8] | ArrayType | int | int |  |  |
-| int[10] | ArrayType | int | int |  |  |
-| int[10][20] | ArrayType | int[20] | int[20] |  |  |
-| int[20] | ArrayType | int | int |  |  |
-| int[] | ArrayType | int | int |  |  |
+| int * | IntPointerType, IteratorByPointer, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection |  | int |  |  |
+| int[4] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | int | int |  |  |
+| int[8] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | int | int |  |  |
+| int[10] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | int | int |  |  |
+| int[10][20] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | int[20] | int[20] |  |  |
+| int[20] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | int | int |  |  |
+| int[] | ArrayType, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection | int | int |  |  |
 | long | LongType |  |  |  |  |
 | long double | LongDoubleType |  |  |  |  |
 | long long | LongLongType, MicrosoftInt64Type |  |  |  |  |
@@ -99,5 +99,5 @@
 | unsigned long long | LongLongType |  |  |  | unsigned integral |
 | unsigned short | ShortType |  |  |  | unsigned integral |
 | void | VoidType |  |  |  |  |
-| void * | IteratorByPointer, VoidPointerType |  | void |  |  |
+| void * | IteratorByPointer, PointerOrArrayOrReferenceType, PointerOrArrayOrReferenceTypeIndirection, VoidPointerType |  | void |  |  |
 | wchar_t | Wchar_t, WideCharType |  |  |  |  |

cpp/ql/test/library-tests/variables/variables/variable.expected

@@ -6,64 +6,36 @@
 | file://:0:0:0:0 | gp_offset | file://:0:0:0:0 | unsigned int | Field |  |  |
 | file://:0:0:0:0 | overflow_arg_area | file://:0:0:0:0 | void * | Field |  |  |
 | file://:0:0:0:0 | reg_save_area | file://:0:0:0:0 | void * | Field |  |  |
-| variables.cpp:1:12:1:12 | i | file://:0:0:0:0 | int | GlobalVariable |  |  |
-| variables.cpp:1:12:1:12 | i | file://:0:0:0:0 | int | StaticStorageDurationVariable |  |  |
-| variables.cpp:2:12:2:12 | i | file://:0:0:0:0 | int | GlobalVariable |  |  |
-| variables.cpp:2:12:2:12 | i | file://:0:0:0:0 | int | StaticStorageDurationVariable |  |  |
-| variables.cpp:3:12:3:12 | i | file://:0:0:0:0 | int | GlobalVariable |  |  |
-| variables.cpp:3:12:3:12 | i | file://:0:0:0:0 | int | StaticStorageDurationVariable |  |  |
-| variables.cpp:5:11:5:11 | c | file://:0:0:0:0 | const int | GlobalVariable | const | static |
-| variables.cpp:5:11:5:11 | c | file://:0:0:0:0 | const int | StaticStorageDurationVariable | const | static |
-| variables.cpp:6:14:6:15 | pi | file://:0:0:0:0 | const double | GlobalVariable | const | static |
-| variables.cpp:6:14:6:15 | pi | file://:0:0:0:0 | const double | StaticStorageDurationVariable | const | static |
-| variables.cpp:8:10:8:10 | a | file://:0:0:0:0 | unsigned int | GlobalVariable |  |  |
-| variables.cpp:8:10:8:10 | a | file://:0:0:0:0 | unsigned int | StaticStorageDurationVariable |  |  |
-| variables.cpp:10:14:10:14 | b | file://:0:0:0:0 | unsigned int | GlobalVariable |  |  |
-| variables.cpp:10:14:10:14 | b | file://:0:0:0:0 | unsigned int | StaticStorageDurationVariable |  |  |
-| variables.cpp:12:13:12:17 | kings | file://:0:0:0:0 | const char *[] | GlobalVariable |  |  |
-| variables.cpp:12:13:12:17 | kings | file://:0:0:0:0 | const char *[] | StaticStorageDurationVariable |  |  |
-| variables.cpp:14:6:14:6 | p | file://:0:0:0:0 | int * | GlobalVariable |  |  |
-| variables.cpp:14:6:14:6 | p | file://:0:0:0:0 | int * | StaticStorageDurationVariable |  |  |
-| variables.cpp:14:9:14:9 | q | file://:0:0:0:0 | int | GlobalVariable |  |  |
-| variables.cpp:14:9:14:9 | q | file://:0:0:0:0 | int | StaticStorageDurationVariable |  |  |
-| variables.cpp:15:12:15:13 | v1 | file://:0:0:0:0 | int[10] | GlobalVariable |  | static |
-| variables.cpp:15:12:15:13 | v1 | file://:0:0:0:0 | int[10] | StaticStorageDurationVariable |  | static |
-| variables.cpp:15:21:15:22 | pv | file://:0:0:0:0 | int * | GlobalVariable |  | static |
-| variables.cpp:15:21:15:22 | pv | file://:0:0:0:0 | int * | StaticStorageDurationVariable |  | static |
-| variables.cpp:17:7:17:8 | fp | file://:0:0:0:0 | ..(*)(..) | GlobalVariable |  |  |
-| variables.cpp:17:7:17:8 | fp | file://:0:0:0:0 | ..(*)(..) | StaticStorageDurationVariable |  |  |
-| variables.cpp:19:7:19:8 | v2 | file://:0:0:0:0 | float[3] | GlobalVariable |  |  |
-| variables.cpp:19:7:19:8 | v2 | file://:0:0:0:0 | float[3] | StaticStorageDurationVariable |  |  |
-| variables.cpp:20:7:20:8 | v3 | file://:0:0:0:0 | char *[32] | GlobalVariable |  |  |
-| variables.cpp:20:7:20:8 | v3 | file://:0:0:0:0 | char *[32] | StaticStorageDurationVariable |  |  |
-| variables.cpp:22:5:22:6 | d2 | file://:0:0:0:0 | int[10][20] | GlobalVariable |  |  |
-| variables.cpp:22:5:22:6 | d2 | file://:0:0:0:0 | int[10][20] | StaticStorageDurationVariable |  |  |
-| variables.cpp:24:6:24:7 | v4 | file://:0:0:0:0 | char[3] | GlobalVariable |  |  |
-| variables.cpp:24:6:24:7 | v4 | file://:0:0:0:0 | char[3] | StaticStorageDurationVariable |  |  |
-| variables.cpp:26:5:26:6 | v5 | file://:0:0:0:0 | int[8] | GlobalVariable |  |  |
-| variables.cpp:26:5:26:6 | v5 | file://:0:0:0:0 | int[8] | StaticStorageDurationVariable |  |  |
-| variables.cpp:28:7:28:8 | p2 | file://:0:0:0:0 | char * | GlobalVariable |  |  |
-| variables.cpp:28:7:28:8 | p2 | file://:0:0:0:0 | char * | StaticStorageDurationVariable |  |  |
-| variables.cpp:29:6:29:7 | p3 | file://:0:0:0:0 | char[] | GlobalVariable |  |  |
-| variables.cpp:29:6:29:7 | p3 | file://:0:0:0:0 | char[] | StaticStorageDurationVariable |  |  |
-| variables.cpp:31:6:31:10 | alpha | file://:0:0:0:0 | char[] | GlobalVariable |  |  |
-| variables.cpp:31:6:31:10 | alpha | file://:0:0:0:0 | char[] | StaticStorageDurationVariable |  |  |
-| variables.cpp:34:5:34:6 | av | file://:0:0:0:0 | int[] | GlobalVariable |  |  |
-| variables.cpp:34:5:34:6 | av | file://:0:0:0:0 | int[] | StaticStorageDurationVariable |  |  |
-| variables.cpp:35:6:35:8 | ap1 | file://:0:0:0:0 | int * | GlobalVariable |  |  |
-| variables.cpp:35:6:35:8 | ap1 | file://:0:0:0:0 | int * | StaticStorageDurationVariable |  |  |
-| variables.cpp:36:6:36:8 | ap2 | file://:0:0:0:0 | int * | GlobalVariable |  |  |
-| variables.cpp:36:6:36:8 | ap2 | file://:0:0:0:0 | int * | StaticStorageDurationVariable |  |  |
-| variables.cpp:37:6:37:8 | ap3 | file://:0:0:0:0 | int * | GlobalVariable |  |  |
-| variables.cpp:37:6:37:8 | ap3 | file://:0:0:0:0 | int * | StaticStorageDurationVariable |  |  |
-| variables.cpp:41:7:41:11 | local | file://:0:0:0:0 | char[] | LocalVariable |  |  |
-| variables.cpp:41:7:41:11 | local | file://:0:0:0:0 | char[] | SemanticStackVariable |  |  |
-| variables.cpp:43:14:43:18 | local | file://:0:0:0:0 | int | StaticLocalVariable |  | static |
+| variables.cpp:1:12:1:12 | i | file://:0:0:0:0 | int | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  |  |
+| variables.cpp:2:12:2:12 | i | file://:0:0:0:0 | int | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  |  |
+| variables.cpp:3:12:3:12 | i | file://:0:0:0:0 | int | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  |  |
+| variables.cpp:5:11:5:11 | c | file://:0:0:0:0 | const int | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable | const | static |
+| variables.cpp:6:14:6:15 | pi | file://:0:0:0:0 | const double | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable | const | static |
+| variables.cpp:8:10:8:10 | a | file://:0:0:0:0 | unsigned int | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  |  |
+| variables.cpp:10:14:10:14 | b | file://:0:0:0:0 | unsigned int | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  |  |
+| variables.cpp:12:13:12:17 | kings | file://:0:0:0:0 | const char *[] | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  |  |
+| variables.cpp:14:6:14:6 | p | file://:0:0:0:0 | int * | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  |  |
+| variables.cpp:14:9:14:9 | q | file://:0:0:0:0 | int | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  |  |
+| variables.cpp:15:12:15:13 | v1 | file://:0:0:0:0 | int[10] | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  | static |
+| variables.cpp:15:21:15:22 | pv | file://:0:0:0:0 | int * | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  | static |
+| variables.cpp:17:7:17:8 | fp | file://:0:0:0:0 | ..(*)(..) | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  |  |
+| variables.cpp:19:7:19:8 | v2 | file://:0:0:0:0 | float[3] | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  |  |
+| variables.cpp:20:7:20:8 | v3 | file://:0:0:0:0 | char *[32] | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  |  |
+| variables.cpp:22:5:22:6 | d2 | file://:0:0:0:0 | int[10][20] | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  |  |
+| variables.cpp:24:6:24:7 | v4 | file://:0:0:0:0 | char[3] | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  |  |
+| variables.cpp:26:5:26:6 | v5 | file://:0:0:0:0 | int[8] | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  |  |
+| variables.cpp:28:7:28:8 | p2 | file://:0:0:0:0 | char * | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  |  |
+| variables.cpp:29:6:29:7 | p3 | file://:0:0:0:0 | char[] | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  |  |
+| variables.cpp:31:6:31:10 | alpha | file://:0:0:0:0 | char[] | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  |  |
+| variables.cpp:34:5:34:6 | av | file://:0:0:0:0 | int[] | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  |  |
+| variables.cpp:35:6:35:8 | ap1 | file://:0:0:0:0 | int * | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  |  |
+| variables.cpp:36:6:36:8 | ap2 | file://:0:0:0:0 | int * | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  |  |
+| variables.cpp:37:6:37:8 | ap3 | file://:0:0:0:0 | int * | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  |  |
+| variables.cpp:41:7:41:11 | local | file://:0:0:0:0 | char[] | LocalVariable, SemanticStackVariable |  |  |
+| variables.cpp:43:14:43:18 | local | file://:0:0:0:0 | int | GlobalLikeVariable, StaticLocalVariable |  | static |
 | variables.cpp:48:9:48:12 | name | file://:0:0:0:0 | char * | Field |  |  |
 | variables.cpp:49:12:49:17 | number | file://:0:0:0:0 | long | Field |  |  |
 | variables.cpp:50:9:50:14 | street | file://:0:0:0:0 | char * | Field |  |  |
 | variables.cpp:51:9:51:12 | town | file://:0:0:0:0 | char * | Field |  |  |
-| variables.cpp:52:16:52:22 | country | file://:0:0:0:0 | char * | MemberVariable |  | static |
-| variables.cpp:52:16:52:22 | country | file://:0:0:0:0 | char * | StaticStorageDurationVariable |  | static |
-| variables.cpp:56:14:56:29 | externInFunction | file://:0:0:0:0 | int | GlobalVariable |  |  |
-| variables.cpp:56:14:56:29 | externInFunction | file://:0:0:0:0 | int | StaticStorageDurationVariable |  |  |
+| variables.cpp:52:16:52:22 | country | file://:0:0:0:0 | char * | MemberVariable, StaticStorageDurationVariable |  | static |
+| variables.cpp:56:14:56:29 | externInFunction | file://:0:0:0:0 | int | GlobalLikeVariable, GlobalVariable, StaticStorageDurationVariable |  |  |

cpp/ql/test/library-tests/variables/variables/variable.ql

@@ -1,10 +1,17 @@
 import cpp
 
-from Variable v, Type t, string qlClass, string const, string static
+string interestingQlClass(Variable v) {
+  result = v.getAQlClass() and
+  (
+    result.matches("%Variable%")
+    or
+    result.matches("%Field%")
+  )
+}
+
+from Variable v, Type t, string const, string static
 where
   t = v.getType() and
-  qlClass = v.getAQlClass() and
-  (qlClass.matches("%Variable%") or qlClass.matches("%Field%")) and
   (if v.isConst() then const = "const" else const = "") and
   if v.isStatic() then static = "static" else static = ""
-select v, t, qlClass, const, static
+select v, t, concat(interestingQlClass(v), ", "), const, static

cpp/ql/test/query-tests/Security/CWE/CWE-497/semmle/tests/ExposedSystemData.expected

@@ -10,6 +10,15 @@ edges
 | tests2.cpp:111:14:111:15 | *c1 [*ptr] | tests2.cpp:111:14:111:19 | *ptr | provenance |  |
 | tests2.cpp:111:14:111:15 | *c1 [*ptr] | tests2.cpp:111:17:111:19 | *ptr | provenance |  |
 | tests2.cpp:111:17:111:19 | *ptr | tests2.cpp:111:14:111:19 | *ptr | provenance |  |
+| tests2.cpp:120:5:120:21 | [summary param] 1 indirection in zmq_msg_init_data | tests2.cpp:120:5:120:21 | [summary] to write: Argument[0 indirection] in zmq_msg_init_data | provenance |  |
+| tests2.cpp:134:17:134:22 | *call to getenv | tests2.cpp:138:23:138:34 | *message_data | provenance |  |
+| tests2.cpp:134:17:134:22 | *call to getenv | tests2.cpp:143:34:143:45 | *message_data | provenance |  |
+| tests2.cpp:143:24:143:31 | zmq_msg_init_data output argument | tests2.cpp:144:33:144:40 | *& ... | provenance |  |
+| tests2.cpp:143:24:143:31 | zmq_msg_init_data output argument | tests2.cpp:147:20:147:27 | *& ... | provenance |  |
+| tests2.cpp:143:24:143:31 | zmq_msg_init_data output argument | tests2.cpp:155:32:155:39 | *& ... | provenance |  |
+| tests2.cpp:143:24:143:31 | zmq_msg_init_data output argument | tests2.cpp:158:20:158:27 | *& ... | provenance |  |
+| tests2.cpp:143:34:143:45 | *message_data | tests2.cpp:120:5:120:21 | [summary param] 1 indirection in zmq_msg_init_data | provenance |  |
+| tests2.cpp:143:34:143:45 | *message_data | tests2.cpp:143:24:143:31 | zmq_msg_init_data output argument | provenance |  |
 | tests_sockets.cpp:26:15:26:20 | *call to getenv | tests_sockets.cpp:39:19:39:22 | *path | provenance |  |
 | tests_sockets.cpp:26:15:26:20 | *call to getenv | tests_sockets.cpp:43:20:43:23 | *path | provenance |  |
 | tests_sockets.cpp:63:15:63:20 | *call to getenv | tests_sockets.cpp:76:19:76:22 | *path | provenance |  |
@@ -36,6 +45,16 @@ nodes
 | tests2.cpp:111:14:111:15 | *c1 [*ptr] | semmle.label | *c1 [*ptr] |
 | tests2.cpp:111:14:111:19 | *ptr | semmle.label | *ptr |
 | tests2.cpp:111:17:111:19 | *ptr | semmle.label | *ptr |
+| tests2.cpp:120:5:120:21 | [summary param] 1 indirection in zmq_msg_init_data | semmle.label | [summary param] 1 indirection in zmq_msg_init_data |
+| tests2.cpp:120:5:120:21 | [summary] to write: Argument[0 indirection] in zmq_msg_init_data | semmle.label | [summary] to write: Argument[0 indirection] in zmq_msg_init_data |
+| tests2.cpp:134:17:134:22 | *call to getenv | semmle.label | *call to getenv |
+| tests2.cpp:138:23:138:34 | *message_data | semmle.label | *message_data |
+| tests2.cpp:143:24:143:31 | zmq_msg_init_data output argument | semmle.label | zmq_msg_init_data output argument |
+| tests2.cpp:143:34:143:45 | *message_data | semmle.label | *message_data |
+| tests2.cpp:144:33:144:40 | *& ... | semmle.label | *& ... |
+| tests2.cpp:147:20:147:27 | *& ... | semmle.label | *& ... |
+| tests2.cpp:155:32:155:39 | *& ... | semmle.label | *& ... |
+| tests2.cpp:158:20:158:27 | *& ... | semmle.label | *& ... |
 | tests_sockets.cpp:26:15:26:20 | *call to getenv | semmle.label | *call to getenv |
 | tests_sockets.cpp:39:19:39:22 | *path | semmle.label | *path |
 | tests_sockets.cpp:43:20:43:23 | *path | semmle.label | *path |
@@ -45,6 +64,7 @@ nodes
 | tests_sysconf.cpp:36:21:36:27 | confstr output argument | semmle.label | confstr output argument |
 | tests_sysconf.cpp:39:19:39:25 | *pathbuf | semmle.label | *pathbuf |
 subpaths
+| tests2.cpp:143:34:143:45 | *message_data | tests2.cpp:120:5:120:21 | [summary param] 1 indirection in zmq_msg_init_data | tests2.cpp:120:5:120:21 | [summary] to write: Argument[0 indirection] in zmq_msg_init_data | tests2.cpp:143:24:143:31 | zmq_msg_init_data output argument |
 #select
 | tests2.cpp:63:13:63:26 | *call to getenv | tests2.cpp:63:13:63:26 | *call to getenv | tests2.cpp:63:13:63:26 | *call to getenv | This operation exposes system data from $@. | tests2.cpp:63:13:63:26 | *call to getenv | *call to getenv |
 | tests2.cpp:64:13:64:26 | *call to getenv | tests2.cpp:64:13:64:26 | *call to getenv | tests2.cpp:64:13:64:26 | *call to getenv | This operation exposes system data from $@. | tests2.cpp:64:13:64:26 | *call to getenv | *call to getenv |
@@ -56,6 +76,11 @@ subpaths
 | tests2.cpp:93:14:93:17 | *str1 | tests2.cpp:91:42:91:45 | *str1 | tests2.cpp:93:14:93:17 | *str1 | This operation exposes system data from $@. | tests2.cpp:91:42:91:45 | *str1 | *str1 |
 | tests2.cpp:102:14:102:15 | *pw | tests2.cpp:101:8:101:15 | *call to getpwuid | tests2.cpp:102:14:102:15 | *pw | This operation exposes system data from $@. | tests2.cpp:101:8:101:15 | *call to getpwuid | *call to getpwuid |
 | tests2.cpp:111:14:111:19 | *ptr | tests2.cpp:109:12:109:17 | *call to getenv | tests2.cpp:111:14:111:19 | *ptr | This operation exposes system data from $@. | tests2.cpp:109:12:109:17 | *call to getenv | *call to getenv |
+| tests2.cpp:138:23:138:34 | *message_data | tests2.cpp:134:17:134:22 | *call to getenv | tests2.cpp:138:23:138:34 | *message_data | This operation exposes system data from $@. | tests2.cpp:134:17:134:22 | *call to getenv | *call to getenv |
+| tests2.cpp:144:33:144:40 | *& ... | tests2.cpp:134:17:134:22 | *call to getenv | tests2.cpp:144:33:144:40 | *& ... | This operation exposes system data from $@. | tests2.cpp:134:17:134:22 | *call to getenv | *call to getenv |
+| tests2.cpp:147:20:147:27 | *& ... | tests2.cpp:134:17:134:22 | *call to getenv | tests2.cpp:147:20:147:27 | *& ... | This operation exposes system data from $@. | tests2.cpp:134:17:134:22 | *call to getenv | *call to getenv |
+| tests2.cpp:155:32:155:39 | *& ... | tests2.cpp:134:17:134:22 | *call to getenv | tests2.cpp:155:32:155:39 | *& ... | This operation exposes system data from $@. | tests2.cpp:134:17:134:22 | *call to getenv | *call to getenv |
+| tests2.cpp:158:20:158:27 | *& ... | tests2.cpp:134:17:134:22 | *call to getenv | tests2.cpp:158:20:158:27 | *& ... | This operation exposes system data from $@. | tests2.cpp:134:17:134:22 | *call to getenv | *call to getenv |
 | tests_sockets.cpp:39:19:39:22 | *path | tests_sockets.cpp:26:15:26:20 | *call to getenv | tests_sockets.cpp:39:19:39:22 | *path | This operation exposes system data from $@. | tests_sockets.cpp:26:15:26:20 | *call to getenv | *call to getenv |
 | tests_sockets.cpp:43:20:43:23 | *path | tests_sockets.cpp:26:15:26:20 | *call to getenv | tests_sockets.cpp:43:20:43:23 | *path | This operation exposes system data from $@. | tests_sockets.cpp:26:15:26:20 | *call to getenv | *call to getenv |
 | tests_sockets.cpp:76:19:76:22 | *path | tests_sockets.cpp:63:15:63:20 | *call to getenv | tests_sockets.cpp:76:19:76:22 | *path | This operation exposes system data from $@. | tests_sockets.cpp:63:15:63:20 | *call to getenv | *call to getenv |

cpp/ql/test/query-tests/Security/CWE/CWE-497/semmle/tests/tests2.cpp

@@ -1,15 +1,15 @@
 // Semmle test cases for rule CWE-497
 
-// library functions etc
+// --- library functions etc ---
 
 #include "tests.h"
 
+typedef unsigned long size_t;
+
+void *memcpy(void *dest, const void *src, size_t count);
 char *getenv(const char *name);
 char *strcpy(char *s1, const char *s2);
-
-
-
-
+size_t strlen(const char *s);
 
 
 
@@ -45,7 +45,7 @@ passwd *getpwuid(int uid);
 
 int val();
 
-// test cases
+// --- test cases ---
 
 const char *global1 = mysql_get_client_info();
 const char *global2 = "abc";
@@ -112,3 +112,51 @@ void test1()
 		send(sock, c2.ptr, val(), val()); // GOOD: not system data
 	}
 }
+
+struct zmq_msg_t {
+};
+typedef void (*zmq_free_fn)();
+
+int zmq_msg_init_data(zmq_msg_t *msg, void *data, size_t size, zmq_free_fn *ffn, void *hint);
+int zmq_msg_init_size(zmq_msg_t *msg, size_t size);
+void *zmq_msg_data(zmq_msg_t *msg);
+int zmq_send(void *socket, const void *buf, size_t len, int flags);
+int zmq_sendmsg(void *socket, zmq_msg_t *msg, int flags); // deprecated
+int zmq_msg_send(zmq_msg_t *msg, void *socket, int flags);
+
+void test_zmq(void *remoteSocket)
+{
+	zmq_msg_t message;
+	char *message_data;
+	size_t message_len;
+
+	// prepare data
+	message_data = getenv("HOME");
+	message_len = strlen(message_data) + 1;
+
+	// send as data
+	if (zmq_send(socket, message_data, message_len, 0) >= 0) { // BAD: outputs HOME environment variable
+		// ...
+	}
+
+	// send as message
+	if (zmq_msg_init_data(&message, message_data, message_len, 0, 0)) {
+		if (zmq_sendmsg(remoteSocket, &message, message_len)) { // BAD: outputs HOME environment variable
+			// ...
+		}
+		if (zmq_msg_send(&message, remoteSocket, message_len)) { // BAD: outputs HOME environment variable
+			// ...
+		}
+	}
+
+	// send as message (alternative path)
+	if (zmq_msg_init_size(&message, message_len) == 0) {
+		memcpy(zmq_msg_data(&message), message_data, message_len);
+		if (zmq_sendmsg(remoteSocket,&message, message_len)) { // BAD: outputs HOME environment variable
+			// ...
+		}
+		if (zmq_msg_send(&message, remoteSocket, message_len)) { // BAD: outputs HOME environment variable
+			// ...
+		}
+	}
+}

swift/ql/lib/codeql/swift/dataflow/ExternalFlow.qll

@@ -46,7 +46,6 @@
  *      The arguments are zero-indexed, and `-1` specifies the qualifier.
  *    - "Argument[n1..n2]": Similar to "Argument[n]" but select any argument in
  *      the given range. The range is inclusive at both ends.
- *    - "Parameter": Selects the value of a parameter of the selected element.
  *    - "Parameter[n]": Similar to "Parameter" but restricted to a specific
  *      numbered parameter (zero-indexed, and `-1` specifies the value of `this`).
  *    - "Parameter[n1..n2]": Similar to "Parameter[n]" but selects any parameter

swift/ql/lib/codeql/swift/dataflow/internal/DataFlowPrivate.qll

@@ -323,7 +323,9 @@ private module Cached {
   /** This is the local flow predicate that is exposed. */
   cached
   predicate localFlowStepImpl(Node nodeFrom, Node nodeTo) {
-    localFlowStepCommon(nodeFrom, nodeTo, _) or
+    localFlowStepCommon(nodeFrom, nodeTo, _)
+    or
+    // models-as-data summarized flow
     FlowSummaryImpl::Private::Steps::summaryThroughStepValue(nodeFrom, nodeTo, _)
   }
 
@@ -1027,6 +1029,7 @@ predicate captureValueStep(Node node1, Node node2) {
 }
 
 predicate jumpStep(Node pred, Node succ) {
+  // models-as-data summarized flow
   FlowSummaryImpl::Private::Steps::summaryJumpStep(pred.(FlowSummaryNode).getSummaryNode(),
     succ.(FlowSummaryNode).getSummaryNode())
 }
@@ -1136,6 +1139,7 @@ predicate storeStep(Node node1, ContentSet c, Node node2) {
     c.isSingleton(any(Content::CollectionContent cc))
   )
   or
+  // models-as-data summarized flow
   FlowSummaryImpl::Private::Steps::summaryStoreStep(node1.(FlowSummaryNode).getSummaryNode(), c,
     node2.(FlowSummaryNode).getSummaryNode())
   or
@@ -1241,6 +1245,7 @@ predicate readStep(Node node1, ContentSet c, Node node2) {
     c instanceof OptionalSomeContentSet
   )
   or
+  // models-as-data summarized flow
   FlowSummaryImpl::Private::Steps::summaryReadStep(node1.(FlowSummaryNode).getSummaryNode(), c,
     node2.(FlowSummaryNode).getSummaryNode())
   or

swift/ql/lib/codeql/swift/dataflow/internal/FlowSummaryImpl.qll

@@ -209,10 +209,6 @@ module SourceSinkInterpretationInput implements
       ast = mid.asElement() and
       e.getMember() = ast
     |
-      // Allow fields to be picked as input nodes.
-      c = "" and
-      e.getBase() = n.asExpr()
-      or
       // Allow post update nodes to be picked as input nodes when the `input` column
       // of the row is `PostUpdate`.
       c = "PostUpdate" and