Merge pull request #14382 from MathiasVP/deduplicate-why

C++: Project away `why` to prevent tuple duplication in `Buffer.qll`

.bazelversion

@@ -1 +1 @@
-6.1.2
+6.3.1

.github/labeler.yml

@@ -45,11 +45,7 @@ documentation:
 
 # Since these are all shared files that need to be synced, just pick _one_ copy of each.
 "DataFlow Library":
-  - "java/ql/lib/semmle/code/java/dataflow/internal/DataFlowImpl.qll"
-  - "java/ql/lib/semmle/code/java/dataflow/internal/DataFlowImplCommon.qll"
-  - "java/ql/lib/semmle/code/java/dataflow/internal/tainttracking1/TaintTrackingImpl.qll"
-  - "java/ql/lib/semmle/code/java/dataflow/internal/DataFlowImplConsistency.qll"
-  - "java/ql/lib/semmle/code/java/dataflow/internal/FlowSummaryImpl.qll"
+  - "shared/dataflow/**/*"
 
 "ATM":
   - javascript/ql/experimental/adaptivethreatmodeling/**/*

.github/workflows/check-implicit-this.yml

@@ -13,7 +13,7 @@ jobs:
   check:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - name: Check that implicit this warnings is enabled for all packs
         shell: bash
         run: |

.github/workflows/check-qldoc.yml

@@ -15,7 +15,7 @@ jobs:
     runs-on: ubuntu-latest
 
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
         with:
           fetch-depth: 2
 

.github/workflows/check-query-ids.yml

@@ -16,6 +16,6 @@ jobs:
     name: Check query IDs
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - name: Check for duplicate query IDs
         run: python3 misc/scripts/check-query-ids.py

.github/workflows/codeql-analysis.yml

@@ -33,7 +33,7 @@ jobs:
         dotnet-version: 7.0.102
 
     - name: Checkout repository
-      uses: actions/checkout@v3
+      uses: actions/checkout@v4
 
     # Initializes the CodeQL tools for scanning.
     - name: Initialize CodeQL

.github/workflows/compile-queries.yml

@@ -13,7 +13,7 @@ jobs:
     runs-on: ubuntu-latest-xl
 
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - name: Setup CodeQL
         uses: ./.github/actions/fetch-codeql
         with:

.github/workflows/csharp-qltest.yml

@@ -29,7 +29,7 @@ jobs:
   qlupgrade:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - uses: ./.github/actions/fetch-codeql
       - name: Check DB upgrade scripts
         run: |
@@ -52,8 +52,7 @@ jobs:
       matrix:
         slice: ["1/2", "2/2"]
     steps:
-      - uses: actions/checkout@v3
-      - uses: ./.github/actions/fetch-codeql
+      - uses: actions/checkout@v4
       - uses: ./csharp/actions/create-extractor-pack
       - name: Cache compilation cache
         id: query-cache
@@ -62,25 +61,41 @@ jobs:
           key: csharp-qltest-${{ matrix.slice }}
       - name: Run QL tests
         run: |
-          CODEQL_PATH=$(gh codeql version --format=json | jq -r .unpackedLocation)
-          # The legacy ASP extractor is not in this repo, so take the one from the nightly build
-          mv "$CODEQL_PATH/csharp/tools/extractor-asp.jar" "${{ github.workspace }}/csharp/extractor-pack/tools"
-          # Safe guard against using the bundled extractor
-          rm -rf "$CODEQL_PATH/csharp"
-          codeql test run --threads=0 --ram 50000 --slice ${{ matrix.slice }} --search-path "${{ github.workspace }}/csharp/extractor-pack" --check-databases --check-undefined-labels --check-repeated-labels --check-redefined-labels --consistency-queries ql/consistency-queries ql/test --compilation-cache "${{ steps.query-cache.outputs.cache-dir }}"
+          codeql test run --threads=0 --ram 50000 --slice ${{ matrix.slice }} --search-path extractor-pack --check-databases --check-undefined-labels --check-repeated-labels --check-redefined-labels --consistency-queries ql/consistency-queries ql/test --compilation-cache "${{ steps.query-cache.outputs.cache-dir }}"
         env:
           GITHUB_TOKEN: ${{ github.token }}
   unit-tests:
-    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        os: [ubuntu-latest, windows-2019]
+    runs-on: ${{ matrix.os }}
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - name: Setup dotnet
         uses: actions/setup-dotnet@v3
         with:
           dotnet-version: 7.0.102
       - name: Extractor unit tests
         run: |
-          dotnet test -p:RuntimeFrameworkVersion=7.0.2 "${{ github.workspace }}/csharp/extractor/Semmle.Util.Tests"
-          dotnet test -p:RuntimeFrameworkVersion=7.0.2 "${{ github.workspace }}/csharp/extractor/Semmle.Extraction.Tests"
-          dotnet test -p:RuntimeFrameworkVersion=7.0.2 "${{ github.workspace }}/csharp/autobuilder/Semmle.Autobuild.CSharp.Tests"
+          dotnet test -p:RuntimeFrameworkVersion=7.0.2 extractor/Semmle.Util.Tests
+          dotnet test -p:RuntimeFrameworkVersion=7.0.2 extractor/Semmle.Extraction.Tests
+          dotnet test -p:RuntimeFrameworkVersion=7.0.2 autobuilder/Semmle.Autobuild.CSharp.Tests
           dotnet test -p:RuntimeFrameworkVersion=7.0.2 "${{ github.workspace }}/cpp/autobuilder/Semmle.Autobuild.Cpp.Tests"
+        shell: bash
+  stubgentest:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: ./csharp/actions/create-extractor-pack
+      - name: Run stub generator tests
+        run: |
+          # Generate (Asp)NetCore stubs
+          STUBS_PATH=stubs_output
+          python3 ql/src/Stubs/make_stubs_nuget.py webapp Swashbuckle.AspNetCore.Swagger latest "$STUBS_PATH"
+          rm -rf ql/test/resources/stubs/_frameworks
+          # Update existing stubs in the repo with the freshly generated ones
+          mv "$STUBS_PATH/output/stubs/_frameworks" ql/test/resources/stubs/
+          git status
+          codeql test run --threads=0 --search-path extractor-pack --check-databases --check-undefined-labels --check-repeated-labels --check-redefined-labels --consistency-queries ql/consistency-queries -- ql/test/library-tests/dataflow/flowsources/aspremote
+        env:
+          GITHUB_TOKEN: ${{ github.token }}

.github/workflows/csv-coverage-metrics.yml

@@ -19,7 +19,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
       - name: Setup CodeQL
         uses: ./.github/actions/fetch-codeql
       - name: Create empty database
@@ -47,7 +47,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
       - name: Setup CodeQL
         uses: ./.github/actions/fetch-codeql
       - name: Create empty database

.github/workflows/csv-coverage-pr-artifacts.yml

@@ -31,11 +31,11 @@ jobs:
           GITHUB_CONTEXT: ${{ toJSON(github.event) }}
         run: echo "$GITHUB_CONTEXT"
       - name: Clone self (github/codeql) - MERGE
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
         with:
           path: merge
       - name: Clone self (github/codeql) - BASE
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
         with:
           fetch-depth: 2
           path: base

.github/workflows/csv-coverage-pr-comment.yml

@@ -20,7 +20,7 @@ jobs:
         GITHUB_CONTEXT: ${{ toJSON(github.event) }}
       run: echo "$GITHUB_CONTEXT"
     - name: Clone self (github/codeql)
-      uses: actions/checkout@v3
+      uses: actions/checkout@v4
     - name: Set up Python 3.8
       uses: actions/setup-python@v4
       with:

.github/workflows/csv-coverage-timeseries.yml

@@ -9,11 +9,11 @@ jobs:
 
     steps:
       - name: Clone self (github/codeql)
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
         with:
           path: script
       - name: Clone self (github/codeql) for analysis
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
         with:
           path: codeqlModels
           fetch-depth: 0

.github/workflows/csv-coverage-update.yml

@@ -17,7 +17,7 @@ jobs:
           GITHUB_CONTEXT: ${{ toJSON(github.event) }}
         run: echo "$GITHUB_CONTEXT"
       - name: Clone self (github/codeql)
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
         with:
           path: ql
           fetch-depth: 0

.github/workflows/csv-coverage.yml

@@ -13,11 +13,11 @@ jobs:
 
     steps:
       - name: Clone self (github/codeql)
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
         with:
           path: script
       - name: Clone self (github/codeql) for analysis
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
         with:
           path: codeqlModels
           ref: ${{ github.event.inputs.qlModelShaOverride || github.ref }}

.github/workflows/fast-forward.yml

@@ -25,7 +25,7 @@ jobs:
           exit 1
 
       - name: Checkout
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
 
       - name: Git config
         shell: bash

.github/workflows/go-tests-other-os.yml

@@ -7,19 +7,21 @@ on:
       - .github/workflows/go-tests-other-os.yml
       - .github/actions/**
       - codeql-workspace.yml
+env:
+  GO_VERSION: '~1.21.0'
 jobs:
   test-mac:
     name: Test MacOS
     runs-on: macos-latest
     steps:
-      - name: Set up Go 1.20
+      - name: Set up Go ${{ env.GO_VERSION }}
         uses: actions/setup-go@v4
         with:
-          go-version: '1.20'
+          go-version: ${{ env.GO_VERSION }}
         id: go
 
       - name: Check out code
-        uses: actions/checkout@v2
+        uses: actions/checkout@v4
 
       - name: Set up CodeQL CLI
         uses: ./.github/actions/fetch-codeql
@@ -47,14 +49,14 @@ jobs:
     name: Test Windows
     runs-on: windows-latest-xl
     steps:
-      - name: Set up Go 1.20
+      - name: Set up Go ${{ env.GO_VERSION }}
         uses: actions/setup-go@v4
         with:
-          go-version: '1.20'
+          go-version: ${{ env.GO_VERSION }}
         id: go
 
       - name: Check out code
-        uses: actions/checkout@v2
+        uses: actions/checkout@v4
 
       - name: Set up CodeQL CLI
         uses: ./.github/actions/fetch-codeql

.github/workflows/go-tests.yml

@@ -15,19 +15,21 @@ on:
       - .github/workflows/go-tests.yml
       - .github/actions/**
       - codeql-workspace.yml
+env:
+  GO_VERSION: '~1.21.0'
 jobs:
   test-linux:
     name: Test Linux (Ubuntu)
     runs-on: ubuntu-latest-xl
     steps:
-      - name: Set up Go 1.20
+      - name: Set up Go ${{ env.GO_VERSION }}
         uses: actions/setup-go@v4
         with:
-          go-version: '1.20'
+          go-version: ${{ env.GO_VERSION }}
         id: go
 
       - name: Check out code
-        uses: actions/checkout@v2
+        uses: actions/checkout@v4
 
       - name: Set up CodeQL CLI
         uses: ./.github/actions/fetch-codeql

.github/workflows/js-ml-tests.yml

@@ -1,65 +0,0 @@
-name: JS ML-powered queries tests
-
-on:
-  push:
-    paths:
-      - "javascript/ql/experimental/adaptivethreatmodeling/**"
-      - .github/workflows/js-ml-tests.yml
-      - .github/actions/fetch-codeql/action.yml
-      - codeql-workspace.yml
-    branches:
-      - main
-      - "rc/*"
-  pull_request:
-    paths:
-      - "javascript/ql/experimental/adaptivethreatmodeling/**"
-      - .github/workflows/js-ml-tests.yml
-      - .github/actions/fetch-codeql/action.yml
-      - codeql-workspace.yml
-  workflow_dispatch:
-
-defaults:
-  run:
-    working-directory: javascript/ql/experimental/adaptivethreatmodeling
-
-jobs:
-  qltest:
-    name: Test QL
-    runs-on: ubuntu-latest-xl
-    steps:
-      - uses: actions/checkout@v3
-
-      - uses: ./.github/actions/fetch-codeql
-
-      - name: Install pack dependencies
-        run: |
-          for pack in modelbuilding src test; do
-            codeql pack install --mode verify -- "${pack}"
-          done
-      
-      - name: Cache compilation cache
-        id: query-cache
-        uses: ./.github/actions/cache-query-compilation
-        with: 
-          key: js-ml-test
-
-      - name: Check QL compilation
-        run: |
-          codeql query compile \
-            --check-only \
-            --ram 50000 \
-            --additional-packs "${{ github.workspace }}" \
-            --threads=0 \
-            --compilation-cache "${{ steps.query-cache.outputs.cache-dir }}" \
-            -- \
-            lib modelbuilding src
-
-      - name: Run QL tests
-        run: |
-          codeql test run \
-            --threads=0 \
-            --ram 50000 \
-            --additional-packs "${{ github.workspace }}" \
-            --compilation-cache "${{ steps.query-cache.outputs.cache-dir }}" \
-            -- \
-            test

.github/workflows/mad_modelDiff.yml

@@ -27,12 +27,12 @@ jobs:
         slug: ${{fromJson(github.event.inputs.projects || '["apache/commons-codec", "apache/commons-io", "apache/commons-beanutils", "apache/commons-logging", "apache/commons-fileupload", "apache/commons-lang", "apache/commons-validator", "apache/commons-csv", "apache/dubbo"]' )}}
     steps:
       - name: Clone github/codeql from PR
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
         if: github.event.pull_request
         with:
           path: codeql-pr
       - name: Clone github/codeql from main
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
         with:
           path: codeql-main
           ref: main

.github/workflows/mad_regenerate-models.yml

@@ -27,11 +27,11 @@ jobs:
             ref: "placeholder"
     steps:
       - name: Clone self (github/codeql)
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
       - name: Setup CodeQL binaries
         uses: ./.github/actions/fetch-codeql
       - name: Clone repositories
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
         with:
           path: repos/${{ matrix.ref }}
           ref: ${{ matrix.ref }}

.github/workflows/qhelp-pr-preview.yml

@@ -43,7 +43,7 @@ jobs:
           if-no-files-found: error
           retention-days: 1
 
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
         with:
           fetch-depth: 2
           persist-credentials: false

.github/workflows/ql-for-ql-build.yml

@@ -14,7 +14,7 @@ jobs:
     runs-on: ubuntu-latest-xl
     steps:
       ### Build the queries ###
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
         with:
           fetch-depth: 0
       - name: Find codeql

.github/workflows/ql-for-ql-dataset_measure.yml

@@ -21,7 +21,7 @@ jobs:
           - github/codeql
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
 
       - name: Find codeql
         id: find-codeql
@@ -42,7 +42,7 @@ jobs:
         env:
           CODEQL: ${{ steps.find-codeql.outputs.codeql-path }}
       - name: Checkout ${{ matrix.repo }}
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
         with:
           repository: ${{ matrix.repo }}
           path: ${{ github.workspace }}/repo
@@ -71,7 +71,7 @@ jobs:
     runs-on: ubuntu-latest
     needs: measure
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - uses: actions/download-artifact@v3
         with:
           name: measurements

.github/workflows/ql-for-ql-tests.yml

@@ -21,7 +21,7 @@ jobs:
   qltest:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - name: Find codeql
         id: find-codeql
         uses: github/codeql-action/init@v2
@@ -61,7 +61,7 @@ jobs:
     needs: [qltest]
     runs-on: ${{ matrix.os }}
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - name: Install GNU tar 
         if: runner.os == 'macOS'
         run: |

.github/workflows/query-list.yml

@@ -20,7 +20,7 @@ jobs:
 
     steps:
     - name: Clone self (github/codeql)
-      uses: actions/checkout@v3
+      uses: actions/checkout@v4
       with:
         path: codeql 
     - name: Set up Python 3.8

.github/workflows/ruby-build.yml

@@ -42,7 +42,7 @@ jobs:
     runs-on: ${{ matrix.os }}
 
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - name: Install GNU tar
         if: runner.os == 'macOS'
         run: |
@@ -113,7 +113,7 @@ jobs:
   compile-queries:
     runs-on: ubuntu-latest-xl
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - name: Fetch CodeQL
         uses: ./.github/actions/fetch-codeql
       - name: Cache compilation cache
@@ -145,7 +145,7 @@ jobs:
     runs-on: ubuntu-latest
     needs: [build, compile-queries]
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - uses: actions/download-artifact@v3
         with:
           name: ruby.dbscheme
@@ -206,7 +206,7 @@ jobs:
     runs-on: ${{ matrix.os }}
     needs: [package]
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - name: Fetch CodeQL
         uses: ./.github/actions/fetch-codeql
 

.github/workflows/ruby-dataset-measure.yml

@@ -27,14 +27,14 @@ jobs:
         repo: [rails/rails, discourse/discourse, spree/spree, ruby/ruby]
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
 
       - uses: ./.github/actions/fetch-codeql
 
       - uses: ./ruby/actions/create-extractor-pack
 
       - name: Checkout ${{ matrix.repo }}
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
         with:
           repository: ${{ matrix.repo }}
           path: ${{ github.workspace }}/repo
@@ -59,7 +59,7 @@ jobs:
     runs-on: ubuntu-latest
     needs: measure
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - uses: actions/download-artifact@v3
         with:
           name: measurements

.github/workflows/ruby-qltest.yml

@@ -14,6 +14,7 @@ on:
   pull_request:
     paths:
       - "ruby/**"
+      - "shared/**"
       - .github/workflows/ruby-qltest.yml
       - .github/actions/fetch-codeql/action.yml
       - codeql-workspace.yml
@@ -32,7 +33,7 @@ jobs:
   qlupgrade:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - uses: ./.github/actions/fetch-codeql
       - name: Check DB upgrade scripts
         run: |
@@ -53,7 +54,7 @@ jobs:
     strategy:
       fail-fast: false
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - uses: ./.github/actions/fetch-codeql
       - uses: ./ruby/actions/create-extractor-pack
       - name: Cache compilation cache

.github/workflows/swift.yml

@@ -39,31 +39,31 @@ jobs:
   build-and-test-macos:
     runs-on: macos-12-xl
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - uses: ./swift/actions/build-and-test
   build-and-test-linux:
     runs-on: ubuntu-latest-xl
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - uses: ./swift/actions/build-and-test
   qltests-linux:
     needs: build-and-test-linux
     runs-on: ubuntu-latest-xl
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - uses: ./swift/actions/run-ql-tests
   qltests-macos:
     if : ${{ github.event_name == 'pull_request' }}
     needs: build-and-test-macos
     runs-on: macos-12-xl
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - uses: ./swift/actions/run-ql-tests
   integration-tests-linux:
     needs: build-and-test-linux
     runs-on: ubuntu-latest-xl
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - uses: ./swift/actions/run-integration-tests
   integration-tests-macos:
     if : ${{ github.event_name == 'pull_request' }}
@@ -71,13 +71,13 @@ jobs:
     runs-on: macos-12-xl
     timeout-minutes: 60
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - uses: ./swift/actions/run-integration-tests
   codegen:
     if : ${{ github.event_name == 'pull_request' }}
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - uses: bazelbuild/setup-bazelisk@v2
       - uses: actions/setup-python@v4
         with:
@@ -102,6 +102,6 @@ jobs:
     if : ${{ github.event_name == 'pull_request' }}
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - uses: ./.github/actions/fetch-codeql
       - uses: ./swift/actions/database-upgrade-scripts

.github/workflows/sync-files.yml

@@ -14,7 +14,7 @@ jobs:
   sync:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - name: Check synchronized files
         run: python config/sync-files.py
       - name: Check dbscheme fragments

.github/workflows/tree-sitter-extractor-test.yml

@@ -27,7 +27,7 @@ jobs:
   test:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - name: Check formatting
         run: cargo fmt --all -- --check
       - name: Run tests
@@ -35,12 +35,12 @@ jobs:
   fmt:
     runs-on: ubuntu-latest  
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - name: Check formatting
         run: cargo fmt --check
   clippy:
     runs-on: ubuntu-latest  
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - name: Run clippy
         run: cargo clippy -- --no-deps -D warnings -A clippy::new_without_default -A clippy::too_many_arguments

.github/workflows/validate-change-notes.yml

@@ -20,7 +20,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout repository
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
 
       - name: Setup CodeQL
         uses: ./.github/actions/fetch-codeql

codeql-workspace.yml

@@ -4,6 +4,8 @@ provide:
   - "*/ql/test/qlpack.yml"
   - "*/ql/examples/qlpack.yml"
   - "*/ql/consistency-queries/qlpack.yml"
+  - "*/ql/automodel/src/qlpack.yml"
+  - "*/ql/automodel/test/qlpack.yml"
   - "shared/*/qlpack.yml"
   - "cpp/ql/test/query-tests/Security/CWE/CWE-190/semmle/tainted/qlpack.yml"
   - "go/ql/config/legacy-support/qlpack.yml"

config/identical-files.json

@@ -1,24 +1,4 @@
 {
-  "DataFlow Java/C++/C#/Go/Python/Ruby/Swift": [
-    "java/ql/lib/semmle/code/java/dataflow/internal/DataFlow.qll",
-    "cpp/ql/lib/semmle/code/cpp/dataflow/internal/DataFlow.qll",
-    "cpp/ql/lib/semmle/code/cpp/ir/dataflow/internal/DataFlow.qll",
-    "csharp/ql/lib/semmle/code/csharp/dataflow/internal/DataFlow.qll",
-    "go/ql/lib/semmle/go/dataflow/internal/DataFlow.qll",
-    "python/ql/lib/semmle/python/dataflow/new/internal/DataFlow.qll",
-    "ruby/ql/lib/codeql/ruby/dataflow/internal/DataFlow.qll",
-    "swift/ql/lib/codeql/swift/dataflow/internal/DataFlow.qll"
-  ],
-  "DataFlowImpl Java/C++/C#/Go/Python/Ruby/Swift": [
-    "java/ql/lib/semmle/code/java/dataflow/internal/DataFlowImpl.qll",
-    "cpp/ql/lib/semmle/code/cpp/dataflow/internal/DataFlowImpl.qll",
-    "cpp/ql/lib/semmle/code/cpp/ir/dataflow/internal/DataFlowImpl.qll",
-    "csharp/ql/lib/semmle/code/csharp/dataflow/internal/DataFlowImpl.qll",
-    "go/ql/lib/semmle/go/dataflow/internal/DataFlowImpl.qll",
-    "python/ql/lib/semmle/python/dataflow/new/internal/DataFlowImpl.qll",
-    "ruby/ql/lib/codeql/ruby/dataflow/internal/DataFlowImpl.qll",
-    "swift/ql/lib/codeql/swift/dataflow/internal/DataFlowImpl.qll"
-  ],
   "DataFlow Java/C++/C#/Go/Python/Ruby/Swift Legacy Configuration": [
     "java/ql/lib/semmle/code/java/dataflow/internal/DataFlowImpl1.qll",
     "java/ql/lib/semmle/code/java/dataflow/internal/DataFlowImpl2.qll",
@@ -42,37 +22,14 @@
     "csharp/ql/lib/semmle/code/csharp/dataflow/internal/DataFlowImpl5.qll",
     "go/ql/lib/semmle/go/dataflow/internal/DataFlowImpl1.qll",
     "go/ql/lib/semmle/go/dataflow/internal/DataFlowImpl2.qll",
-    "go/ql/lib/semmle/go/dataflow/internal/DataFlowImplForStringsNewReplacer.qll",
     "python/ql/lib/semmle/python/dataflow/new/internal/DataFlowImpl1.qll",
     "python/ql/lib/semmle/python/dataflow/new/internal/DataFlowImpl2.qll",
     "python/ql/lib/semmle/python/dataflow/new/internal/DataFlowImpl3.qll",
     "python/ql/lib/semmle/python/dataflow/new/internal/DataFlowImpl4.qll",
     "ruby/ql/lib/codeql/ruby/dataflow/internal/DataFlowImpl1.qll",
     "ruby/ql/lib/codeql/ruby/dataflow/internal/DataFlowImpl2.qll",
-    "ruby/ql/lib/codeql/ruby/dataflow/internal/DataFlowImplForHttpClientLibraries.qll",
-    "ruby/ql/lib/codeql/ruby/dataflow/internal/DataFlowImplForPathname.qll",
     "swift/ql/lib/codeql/swift/dataflow/internal/DataFlowImpl1.qll"
   ],
-  "DataFlow Java/C++/C#/Go/Python/Ruby/Swift Common": [
-    "java/ql/lib/semmle/code/java/dataflow/internal/DataFlowImplCommon.qll",
-    "cpp/ql/lib/semmle/code/cpp/dataflow/internal/DataFlowImplCommon.qll",
-    "cpp/ql/lib/semmle/code/cpp/ir/dataflow/internal/DataFlowImplCommon.qll",
-    "csharp/ql/lib/semmle/code/csharp/dataflow/internal/DataFlowImplCommon.qll",
-    "go/ql/lib/semmle/go/dataflow/internal/DataFlowImplCommon.qll",
-    "python/ql/lib/semmle/python/dataflow/new/internal/DataFlowImplCommon.qll",
-    "ruby/ql/lib/codeql/ruby/dataflow/internal/DataFlowImplCommon.qll",
-    "swift/ql/lib/codeql/swift/dataflow/internal/DataFlowImplCommon.qll"
-  ],
-  "TaintTracking Java/C++/C#/Go/Python/Ruby/Swift": [
-    "cpp/ql/lib/semmle/code/cpp/dataflow/internal/tainttracking1/TaintTracking.qll",
-    "cpp/ql/lib/semmle/code/cpp/ir/dataflow/internal/tainttracking1/TaintTracking.qll",
-    "csharp/ql/lib/semmle/code/csharp/dataflow/internal/tainttracking1/TaintTracking.qll",
-    "go/ql/lib/semmle/go/dataflow/internal/tainttracking1/TaintTracking.qll",
-    "java/ql/lib/semmle/code/java/dataflow/internal/tainttracking1/TaintTracking.qll",
-    "python/ql/lib/semmle/python/dataflow/new/internal/tainttracking1/TaintTracking.qll",
-    "ruby/ql/lib/codeql/ruby/dataflow/internal/tainttracking1/TaintTracking.qll",
-    "swift/ql/lib/codeql/swift/dataflow/internal/tainttracking1/TaintTracking.qll"
-  ],
   "TaintTracking Legacy Configuration Java/C++/C#/Go/Python/Ruby/Swift": [
     "cpp/ql/lib/semmle/code/cpp/dataflow/internal/tainttracking1/TaintTrackingImpl.qll",
     "cpp/ql/lib/semmle/code/cpp/dataflow/internal/tainttracking2/TaintTrackingImpl.qll",
@@ -96,15 +53,6 @@
     "ruby/ql/lib/codeql/ruby/dataflow/internal/tainttracking1/TaintTrackingImpl.qll",
     "swift/ql/lib/codeql/swift/dataflow/internal/tainttracking1/TaintTrackingImpl.qll"
   ],
-  "DataFlow Java/C++/C#/Python/Ruby/Swift Consistency checks": [
-    "java/ql/lib/semmle/code/java/dataflow/internal/DataFlowImplConsistency.qll",
-    "cpp/ql/lib/semmle/code/cpp/dataflow/internal/DataFlowImplConsistency.qll",
-    "cpp/ql/lib/semmle/code/cpp/ir/dataflow/internal/DataFlowImplConsistency.qll",
-    "csharp/ql/lib/semmle/code/csharp/dataflow/internal/DataFlowImplConsistency.qll",
-    "python/ql/lib/semmle/python/dataflow/new/internal/DataFlowImplConsistency.qll",
-    "ruby/ql/lib/codeql/ruby/dataflow/internal/DataFlowImplConsistency.qll",
-    "swift/ql/lib/codeql/swift/dataflow/internal/DataFlowImplConsistency.qll"
-  ],
   "DataFlow Java/C#/Go/Ruby/Python/Swift Flow Summaries": [
     "java/ql/lib/semmle/code/java/dataflow/internal/FlowSummaryImpl.qll",
     "csharp/ql/lib/semmle/code/csharp/dataflow/internal/FlowSummaryImpl.qll",
@@ -514,11 +462,6 @@
     "ruby/ql/lib/codeql/ruby/security/internal/SensitiveDataHeuristics.qll",
     "swift/ql/lib/codeql/swift/security/internal/SensitiveDataHeuristics.qll"
   ],
-  "CFG": [
-    "csharp/ql/lib/semmle/code/csharp/controlflow/internal/ControlFlowGraphImplShared.qll",
-    "ruby/ql/lib/codeql/ruby/controlflow/internal/ControlFlowGraphImplShared.qll",
-    "swift/ql/lib/codeql/swift/controlflow/internal/ControlFlowGraphImplShared.qll"
-  ],
   "TypeTracker": [
     "python/ql/lib/semmle/python/dataflow/new/internal/TypeTracker.qll",
     "ruby/ql/lib/codeql/ruby/typetracking/TypeTracker.qll"
@@ -602,5 +545,9 @@
   "EncryptionKeySizes Python/Java": [
     "python/ql/lib/semmle/python/security/internal/EncryptionKeySizes.qll",
     "java/ql/lib/semmle/code/java/security/internal/EncryptionKeySizes.qll"
+  ],
+  "Python model summaries test extension": [
+    "python/ql/test/experimental/dataflow/model-summaries/InlineTaintTest.ext.yml",
+    "python/ql/test/experimental/dataflow/model-summaries/NormalDataflowTest.ext.yml"
   ]
 }

cpp/autobuilder/Semmle.Autobuild.Cpp.Tests/BuildScripts.cs

@@ -327,7 +327,7 @@ namespace Semmle.Autobuild.Cpp.Tests
         {
             Actions.RunProcess[@"cmd.exe /C nuget restore C:\Project\test.sln -DisableParallelProcessing"] = 1;
             Actions.RunProcess[@"cmd.exe /C C:\Project\.nuget\nuget.exe restore C:\Project\test.sln -DisableParallelProcessing"] = 0;
-            Actions.RunProcess[@"cmd.exe /C CALL ^""C:\Program Files ^(x86^)\Microsoft Visual Studio 14.0\VC\vcvarsall.bat^"" && set Platform=&& type NUL && msbuild C:\Project\test.sln /t:rebuild /p:Platform=""x86"" /p:Configuration=""Release"""] = 0;
+            Actions.RunProcess[@"cmd.exe /C CALL ^""C:\Program^ Files^ ^(x86^)\Microsoft^ Visual^ Studio^ 14.0\VC\vcvarsall.bat^"" && set Platform=&& type NUL && msbuild C:\Project\test.sln /t:rebuild /p:Platform=""x86"" /p:Configuration=""Release"""] = 0;
             Actions.RunProcessOut[@"C:\Program Files (x86)\Microsoft Visual Studio\Installer\vswhere.exe -prerelease -legacy -property installationPath"] = "";
             Actions.RunProcess[@"C:\Program Files (x86)\Microsoft Visual Studio\Installer\vswhere.exe -prerelease -legacy -property installationPath"] = 1;
             Actions.RunProcess[@"C:\Program Files (x86)\Microsoft Visual Studio\Installer\vswhere.exe -prerelease -legacy -property installationVersion"] = 0;

cpp/downgrades/d77c09d8bdc172c9201dec293de1e14c931d3f05/upgrade.properties

@@ -0,0 +1,2 @@
+description: Remove _Float128 type
+compatibility: full

cpp/downgrades/dbe9c8eb5fc6f54b7ae08c7317d0795b24961564/upgrade.properties

@@ -0,0 +1,2 @@
+description: Make __is_trivial a builtin operation
+compatibility: full

cpp/ql/lib/CHANGELOG.md

@@ -1,3 +1,51 @@
+## 0.10.0
+
+### Minor Analysis Improvements
+
+* Functions that do not return due to calling functions that don't return (e.g. `exit`) are now detected as
+ non-returning in the IR and dataflow.
+* Treat functions that reach the end of the function as returning in the IR.
+  They used to be treated as unreachable but it is allowed in C. 
+* The `DataFlow::asDefiningArgument` predicate now takes its argument from the range starting at `1` instead of `2`. Queries that depend on the single-parameter version of `DataFlow::asDefiningArgument` should have their arguments updated accordingly.
+
+## 0.9.3
+
+No user-facing changes.
+
+## 0.9.2
+
+### Deprecated APIs
+
+* `getAllocatorCall` on `DeleteExpr` and `DeleteArrayExpr` has been deprecated. `getDeallocatorCall` should be used instead.
+
+### New Features
+
+* Added `DeleteOrDeleteArrayExpr` as a super type of `DeleteExpr` and `DeleteArrayExpr`
+
+### Minor Analysis Improvements
+
+* `delete` and `delete[]` are now modeled as calls to the relevant `operator delete` in the IR. In the case of a dynamic delete call a new instruction `VirtualDeleteFunctionAddress` is used to represent a function that dispatches to the correct delete implementation.
+* Only the 2 level indirection of `argv` (corresponding to `**argv`) is consided for `FlowSource`.
+
+## 0.9.1
+
+No user-facing changes.
+
+## 0.9.0
+
+### Breaking Changes
+
+* The `shouldPrintFunction` predicate from `PrintAstConfiguration` has been replaced by `shouldPrintDeclaration`. Users should now override `shouldPrintDeclaration` if they want to limit the declarations that should be printed.
+* The `shouldPrintFunction` predicate from `PrintIRConfiguration` has been replaced by `shouldPrintDeclaration`. Users should now override `shouldPrintDeclaration` if they want to limit the declarations that should be printed.
+
+### Major Analysis Improvements
+
+* The `PrintAST` library now also prints global and namespace variables and their initializers.
+
+### Minor Analysis Improvements
+
+* The `_Float128x` type is no longer exposed as a builtin type. As this type could not occur any code base, this should only affect queries that explicitly looked at the builtin types.
+
 ## 0.8.1
 
 ### Deprecated APIs

cpp/ql/lib/change-notes/released/0.10.0.md

@@ -0,0 +1,9 @@
+## 0.10.0
+
+### Minor Analysis Improvements
+
+* Functions that do not return due to calling functions that don't return (e.g. `exit`) are now detected as
+ non-returning in the IR and dataflow.
+* Treat functions that reach the end of the function as returning in the IR.
+  They used to be treated as unreachable but it is allowed in C. 
+* The `DataFlow::asDefiningArgument` predicate now takes its argument from the range starting at `1` instead of `2`. Queries that depend on the single-parameter version of `DataFlow::asDefiningArgument` should have their arguments updated accordingly.

cpp/ql/lib/change-notes/released/0.9.0.md

@@ -0,0 +1,14 @@
+## 0.9.0
+
+### Breaking Changes
+
+* The `shouldPrintFunction` predicate from `PrintAstConfiguration` has been replaced by `shouldPrintDeclaration`. Users should now override `shouldPrintDeclaration` if they want to limit the declarations that should be printed.
+* The `shouldPrintFunction` predicate from `PrintIRConfiguration` has been replaced by `shouldPrintDeclaration`. Users should now override `shouldPrintDeclaration` if they want to limit the declarations that should be printed.
+
+### Major Analysis Improvements
+
+* The `PrintAST` library now also prints global and namespace variables and their initializers.
+
+### Minor Analysis Improvements
+
+* The `_Float128x` type is no longer exposed as a builtin type. As this type could not occur any code base, this should only affect queries that explicitly looked at the builtin types.

cpp/ql/lib/change-notes/released/0.9.1.md

@@ -0,0 +1,3 @@
+## 0.9.1
+
+No user-facing changes.

cpp/ql/lib/change-notes/released/0.9.2.md

@@ -0,0 +1,14 @@
+## 0.9.2
+
+### Deprecated APIs
+
+* `getAllocatorCall` on `DeleteExpr` and `DeleteArrayExpr` has been deprecated. `getDeallocatorCall` should be used instead.
+
+### New Features
+
+* Added `DeleteOrDeleteArrayExpr` as a super type of `DeleteExpr` and `DeleteArrayExpr`
+
+### Minor Analysis Improvements
+
+* `delete` and `delete[]` are now modeled as calls to the relevant `operator delete` in the IR. In the case of a dynamic delete call a new instruction `VirtualDeleteFunctionAddress` is used to represent a function that dispatches to the correct delete implementation.
+* Only the 2 level indirection of `argv` (corresponding to `**argv`) is consided for `FlowSource`.

cpp/ql/lib/change-notes/released/0.9.3.md

@@ -0,0 +1,3 @@
+## 0.9.3
+
+No user-facing changes.

cpp/ql/lib/codeql-pack.release.yml

@@ -1,2 +1,2 @@
 ---
-lastReleaseVersion: 0.8.1
+lastReleaseVersion: 0.10.0

cpp/ql/lib/printAst.ql

@@ -18,10 +18,10 @@ external string selectedSourceFile();
 
 class Cfg extends PrintAstConfiguration {
   /**
-   * Holds if the AST for `func` should be printed.
-   * Print All functions from the selected file.
+   * Holds if the AST for `decl` should be printed.
+   * Print All declarations from the selected file.
    */
-  override predicate shouldPrintFunction(Function func) {
-    func.getFile() = getFileBySourceArchiveName(selectedSourceFile())
+  override predicate shouldPrintDeclaration(Declaration decl) {
+    decl.getFile() = getFileBySourceArchiveName(selectedSourceFile())
   }
 }

cpp/ql/lib/qlpack.yml

@@ -1,11 +1,12 @@
 name: codeql/cpp-all
-version: 0.8.1
+version: 0.10.0
 groups: cpp
 dbscheme: semmlecode.cpp.dbscheme
 extractor: cpp
 library: true
 upgrades: upgrades
 dependencies:
+  codeql/dataflow: ${workspace}
   codeql/ssa: ${workspace}
   codeql/tutorial: ${workspace}
   codeql/util: ${workspace}

cpp/ql/lib/semmle/code/cpp/File.qll

@@ -5,155 +5,35 @@
 import semmle.code.cpp.Element
 import semmle.code.cpp.Declaration
 import semmle.code.cpp.metrics.MetricFile
+private import codeql.util.FileSystem
+
+private module Input implements InputSig {
+  abstract class ContainerBase extends @container {
+    abstract string getAbsolutePath();
+
+    ContainerBase getParentContainer() {
+      containerparent(unresolveElement(result), underlyingElement(this))
+    }
+
+    string toString() { result = this.getAbsolutePath() }
+  }
+
+  class FolderBase extends ContainerBase, @folder {
+    override string getAbsolutePath() { folders(underlyingElement(this), result) }
+  }
+
+  class FileBase extends ContainerBase, @file {
+    override string getAbsolutePath() { files(underlyingElement(this), result) }
+  }
+
+  predicate hasSourceLocationPrefix = sourceLocationPrefix/1;
+}
+
+private module Impl = Make<Input>;
 
 /** A file or folder. */
-class Container extends Locatable, @container {
-  /**
-   * Gets the absolute, canonical path of this container, using forward slashes
-   * as path separator.
-   *
-   * The path starts with a _root prefix_ followed by zero or more _path
-   * segments_ separated by forward slashes.
-   *
-   * The root prefix is of one of the following forms:
-   *
-   *   1. A single forward slash `/` (Unix-style)
-   *   2. An upper-case drive letter followed by a colon and a forward slash,
-   *      such as `C:/` (Windows-style)
-   *   3. Two forward slashes, a computer name, and then another forward slash,
-   *      such as `//FileServer/` (UNC-style)
-   *
-   * Path segments are never empty (that is, absolute paths never contain two
-   * contiguous slashes, except as part of a UNC-style root prefix). Also, path
-   * segments never contain forward slashes, and no path segment is of the
-   * form `.` (one dot) or `..` (two dots).
-   *
-   * Note that an absolute path never ends with a forward slash, except if it is
-   * a bare root prefix, that is, the path has no path segments. A container
-   * whose absolute path has no segments is always a `Folder`, not a `File`.
-   */
-  string getAbsolutePath() { none() } // overridden by subclasses
-
-  /**
-   * Gets the relative path of this file or folder from the root folder of the
-   * analyzed source location. The relative path of the root folder itself is
-   * the empty string.
-   *
-   * This has no result if the container is outside the source root, that is,
-   * if the root folder is not a reflexive, transitive parent of this container.
-   */
-  string getRelativePath() {
-    exists(string absPath, string pref |
-      absPath = this.getAbsolutePath() and sourceLocationPrefix(pref)
-    |
-      absPath = pref and result = ""
-      or
-      absPath = pref.regexpReplaceAll("/$", "") + "/" + result and
-      not result.matches("/%")
-    )
-  }
-
-  /**
-   * Gets the base name of this container including extension, that is, the last
-   * segment of its absolute path, or the empty string if it has no segments.
-   *
-   * Here are some examples of absolute paths and the corresponding base names
-   * (surrounded with quotes to avoid ambiguity):
-   *
-   * <table border="1">
-   * <tr><th>Absolute path</th><th>Base name</th></tr>
-   * <tr><td>"/tmp/tst.js"</td><td>"tst.js"</td></tr>
-   * <tr><td>"C:/Program Files (x86)"</td><td>"Program Files (x86)"</td></tr>
-   * <tr><td>"/"</td><td>""</td></tr>
-   * <tr><td>"C:/"</td><td>""</td></tr>
-   * <tr><td>"D:/"</td><td>""</td></tr>
-   * <tr><td>"//FileServer/"</td><td>""</td></tr>
-   * </table>
-   */
-  string getBaseName() {
-    result = this.getAbsolutePath().regexpCapture(".*/(([^/]*?)(?:\\.([^.]*))?)", 1)
-  }
-
-  /**
-   * Gets the extension of this container, that is, the suffix of its base name
-   * after the last dot character, if any.
-   *
-   * In particular,
-   *
-   *  - if the name does not include a dot, there is no extension, so this
-   *    predicate has no result;
-   *  - if the name ends in a dot, the extension is the empty string;
-   *  - if the name contains multiple dots, the extension follows the last dot.
-   *
-   * Here are some examples of absolute paths and the corresponding extensions
-   * (surrounded with quotes to avoid ambiguity):
-   *
-   * <table border="1">
-   * <tr><th>Absolute path</th><th>Extension</th></tr>
-   * <tr><td>"/tmp/tst.js"</td><td>"js"</td></tr>
-   * <tr><td>"/tmp/.classpath"</td><td>"classpath"</td></tr>
-   * <tr><td>"/bin/bash"</td><td>not defined</td></tr>
-   * <tr><td>"/tmp/tst2."</td><td>""</td></tr>
-   * <tr><td>"/tmp/x.tar.gz"</td><td>"gz"</td></tr>
-   * </table>
-   */
-  string getExtension() {
-    result = this.getAbsolutePath().regexpCapture(".*/([^/]*?)(\\.([^.]*))?", 3)
-  }
-
-  /**
-   * Gets the stem of this container, that is, the prefix of its base name up to
-   * (but not including) the last dot character if there is one, or the entire
-   * base name if there is not.
-   *
-   * Here are some examples of absolute paths and the corresponding stems
-   * (surrounded with quotes to avoid ambiguity):
-   *
-   * <table border="1">
-   * <tr><th>Absolute path</th><th>Stem</th></tr>
-   * <tr><td>"/tmp/tst.js"</td><td>"tst"</td></tr>
-   * <tr><td>"/tmp/.classpath"</td><td>""</td></tr>
-   * <tr><td>"/bin/bash"</td><td>"bash"</td></tr>
-   * <tr><td>"/tmp/tst2."</td><td>"tst2"</td></tr>
-   * <tr><td>"/tmp/x.tar.gz"</td><td>"x.tar"</td></tr>
-   * </table>
-   */
-  string getStem() {
-    result = this.getAbsolutePath().regexpCapture(".*/([^/]*?)(?:\\.([^.]*))?", 1)
-  }
-
-  /** Gets the parent container of this file or folder, if any. */
-  Container getParentContainer() {
-    containerparent(unresolveElement(result), underlyingElement(this))
-  }
-
-  /** Gets a file or sub-folder in this container. */
-  Container getAChildContainer() { this = result.getParentContainer() }
-
-  /** Gets a file in this container. */
-  File getAFile() { result = this.getAChildContainer() }
-
-  /** Gets the file in this container that has the given `baseName`, if any. */
-  File getFile(string baseName) {
-    result = this.getAFile() and
-    result.getBaseName() = baseName
-  }
-
-  /** Gets a sub-folder in this container. */
-  Folder getAFolder() { result = this.getAChildContainer() }
-
-  /** Gets the sub-folder in this container that has the given `baseName`, if any. */
-  Folder getFolder(string baseName) {
-    result = this.getAFolder() and
-    result.getBaseName() = baseName
-  }
-
-  /**
-   * Gets a textual representation of the path of this container.
-   *
-   * This is the absolute path of the container.
-   */
-  override string toString() { result = this.getAbsolutePath() }
+class Container extends Locatable, Impl::Container {
+  override string toString() { result = Impl::Container.super.toString() }
 }
 
 /**
@@ -166,9 +46,7 @@ class Container extends Locatable, @container {
  *
  * To get the full path, use `getAbsolutePath`.
  */
-class Folder extends Container, @folder {
-  override string getAbsolutePath() { folders(underlyingElement(this), result) }
-
+class Folder extends Container, Impl::Folder {
   override Location getLocation() {
     result.getContainer() = this and
     result.hasLocationInfo(_, 0, 0, 0, 0)
@@ -189,9 +67,7 @@ class Folder extends Container, @folder {
  * The base name further decomposes into the _stem_ and _extension_ -- see
  * `getStem` and `getExtension`. To get the full path, use `getAbsolutePath`.
  */
-class File extends Container, @file {
-  override string getAbsolutePath() { files(underlyingElement(this), result) }
-
+class File extends Container, Impl::File {
   override string getAPrimaryQlClass() { result = "File" }
 
   override Location getLocation() {

cpp/ql/lib/semmle/code/cpp/Print.qll

@@ -6,11 +6,9 @@ private import PrintAST
  * that requests that function, or no `PrintASTConfiguration` exists.
  */
 private predicate shouldPrintDeclaration(Declaration decl) {
-  not decl instanceof Function
+  not (decl instanceof Function or decl instanceof GlobalOrNamespaceVariable)
   or
-  not exists(PrintAstConfiguration c)
-  or
-  exists(PrintAstConfiguration config | config.shouldPrintFunction(decl))
+  exists(PrintAstConfiguration config | config.shouldPrintDeclaration(decl))
 }
 
 /**

cpp/ql/lib/semmle/code/cpp/PrintAST.ql

@@ -9,13 +9,13 @@ import cpp
 import PrintAST
 
 /**
- * Temporarily tweak this class or make a copy to control which functions are
+ * Temporarily tweak this class or make a copy to control which declarations are
  * printed.
  */
 class Cfg extends PrintAstConfiguration {
   /**
    * TWEAK THIS PREDICATE AS NEEDED.
-   * Holds if the AST for `func` should be printed.
+   * Holds if the AST for `decl` should be printed.
    */
-  override predicate shouldPrintFunction(Function func) { any() }
+  override predicate shouldPrintDeclaration(Declaration decl) { any() }
 }

cpp/ql/lib/semmle/code/cpp/PrintAST.qll

@@ -1,9 +1,9 @@
 /**
  * Provides queries to pretty-print a C++ AST as a graph.
  *
- * By default, this will print the AST for all functions in the database. To change this behavior,
- * extend `PrintASTConfiguration` and override `shouldPrintFunction` to hold for only the functions
- * you wish to view the AST for.
+ * By default, this will print the AST for all functions and global and namespace variables in
+ * the database. To change this behavior, extend `PrintASTConfiguration` and override
+ * `shouldPrintDeclaration` to hold for only the declarations you wish to view the AST for.
  */
 
 import cpp
@@ -12,7 +12,7 @@ private import semmle.code.cpp.Print
 private newtype TPrintAstConfiguration = MkPrintAstConfiguration()
 
 /**
- * The query can extend this class to control which functions are printed.
+ * The query can extend this class to control which declarations are printed.
  */
 class PrintAstConfiguration extends TPrintAstConfiguration {
   /**
@@ -21,14 +21,16 @@ class PrintAstConfiguration extends TPrintAstConfiguration {
   string toString() { result = "PrintASTConfiguration" }
 
   /**
-   * Holds if the AST for `func` should be printed. By default, holds for all
-   * functions.
+   * Holds if the AST for `decl` should be printed. By default, holds for all
+   * functions and global and namespace variables. Currently, does not support any
+   * other declaration types.
    */
-  predicate shouldPrintFunction(Function func) { any() }
+  predicate shouldPrintDeclaration(Declaration decl) { any() }
 }
 
-private predicate shouldPrintFunction(Function func) {
-  exists(PrintAstConfiguration config | config.shouldPrintFunction(func))
+private predicate shouldPrintDeclaration(Declaration decl) {
+  exists(PrintAstConfiguration config | config.shouldPrintDeclaration(decl)) and
+  (decl instanceof Function or decl instanceof GlobalOrNamespaceVariable)
 }
 
 bindingset[s]
@@ -69,7 +71,7 @@ private predicate locationSortKeys(Locatable ast, string file, int line, int col
   )
 }
 
-private Function getEnclosingFunction(Locatable ast) {
+private Declaration getAnEnclosingDeclaration(Locatable ast) {
   result = ast.(Expr).getEnclosingFunction()
   or
   result = ast.(Stmt).getEnclosingFunction()
@@ -78,6 +80,10 @@ private Function getEnclosingFunction(Locatable ast) {
   or
   result = ast.(Parameter).getFunction()
   or
+  result = ast.(Expr).getEnclosingDeclaration()
+  or
+  result = ast.(Initializer).getDeclaration()
+  or
   result = ast
 }
 
@@ -86,21 +92,21 @@ private Function getEnclosingFunction(Locatable ast) {
  * nodes for things like parameter lists and constructor init lists.
  */
 private newtype TPrintAstNode =
-  TAstNode(Locatable ast) { shouldPrintFunction(getEnclosingFunction(ast)) } or
+  TAstNode(Locatable ast) { shouldPrintDeclaration(getAnEnclosingDeclaration(ast)) } or
   TDeclarationEntryNode(DeclStmt stmt, DeclarationEntry entry) {
     // We create a unique node for each pair of (stmt, entry), to avoid having one node with
     // multiple parents due to extractor bug CPP-413.
     stmt.getADeclarationEntry() = entry and
-    shouldPrintFunction(stmt.getEnclosingFunction())
+    shouldPrintDeclaration(stmt.getEnclosingFunction())
   } or
-  TParametersNode(Function func) { shouldPrintFunction(func) } or
+  TParametersNode(Function func) { shouldPrintDeclaration(func) } or
   TConstructorInitializersNode(Constructor ctor) {
     ctor.hasEntryPoint() and
-    shouldPrintFunction(ctor)
+    shouldPrintDeclaration(ctor)
   } or
   TDestructorDestructionsNode(Destructor dtor) {
     dtor.hasEntryPoint() and
-    shouldPrintFunction(dtor)
+    shouldPrintDeclaration(dtor)
   }
 
 /**
@@ -158,10 +164,10 @@ class PrintAstNode extends TPrintAstNode {
 
   /**
    * Holds if this node should be printed in the output. By default, all nodes
-   * within a function are printed, but the query can override
-   * `PrintASTConfiguration.shouldPrintFunction` to filter the output.
+   * within functions and global and namespace variables are printed, but the query
+   * can override `PrintASTConfiguration.shouldPrintDeclaration` to filter the output.
    */
-  final predicate shouldPrint() { shouldPrintFunction(this.getEnclosingFunction()) }
+  final predicate shouldPrint() { shouldPrintDeclaration(this.getEnclosingDeclaration()) }
 
   /**
    * Gets the children of this node.
@@ -229,10 +235,15 @@ class PrintAstNode extends TPrintAstNode {
   abstract string getChildAccessorPredicateInternal(int childIndex);
 
   /**
-   * Gets the `Function` that contains this node.
+   * Gets the `Declaration` that contains this node.
    */
-  private Function getEnclosingFunction() {
-    result = this.getParent*().(FunctionNode).getFunction()
+  private Declaration getEnclosingDeclaration() { result = this.getParent*().getDeclaration() }
+
+  /**
+   * Gets the `Declaration` this node represents.
+   */
+  private Declaration getDeclaration() {
+    result = this.(AstNode).getAst() and shouldPrintDeclaration(result)
   }
 }
 
@@ -571,16 +582,53 @@ class DestructorDestructionsNode extends PrintAstNode, TDestructorDestructionsNo
   final Destructor getDestructor() { result = dtor }
 }
 
+abstract private class FunctionOrGlobalOrNamespaceVariableNode extends AstNode {
+  override string toString() { result = qlClass(ast) + getIdentityString(ast) }
+
+  private int getOrder() {
+    this =
+      rank[result](FunctionOrGlobalOrNamespaceVariableNode node, Declaration decl, string file,
+        int line, int column |
+        node.getAst() = decl and
+        locationSortKeys(decl, file, line, column)
+      |
+        node order by file, line, column, getIdentityString(decl)
+      )
+  }
+
+  override string getProperty(string key) {
+    result = super.getProperty(key)
+    or
+    key = "semmle.order" and result = this.getOrder().toString()
+  }
+}
+
+/**
+ * A node representing a `GlobalOrNamespaceVariable`.
+ */
+class GlobalOrNamespaceVariableNode extends FunctionOrGlobalOrNamespaceVariableNode {
+  GlobalOrNamespaceVariable var;
+
+  GlobalOrNamespaceVariableNode() { var = ast }
+
+  override PrintAstNode getChildInternal(int childIndex) {
+    childIndex = 0 and
+    result.(AstNode).getAst() = var.getInitializer()
+  }
+
+  override string getChildAccessorPredicateInternal(int childIndex) {
+    childIndex = 0 and result = "getInitializer()"
+  }
+}
+
 /**
  * A node representing a `Function`.
  */
-class FunctionNode extends AstNode {
+class FunctionNode extends FunctionOrGlobalOrNamespaceVariableNode {
   Function func;
 
   FunctionNode() { func = ast }
 
-  override string toString() { result = qlClass(func) + getIdentityString(func) }
-
   override PrintAstNode getChildInternal(int childIndex) {
     childIndex = 0 and
     result.(ParametersNode).getFunction() = func
@@ -604,31 +652,10 @@ class FunctionNode extends AstNode {
     or
     childIndex = 3 and result = "<destructions>"
   }
-
-  private int getOrder() {
-    this =
-      rank[result](FunctionNode node, Function function, string file, int line, int column |
-        node.getAst() = function and
-        locationSortKeys(function, file, line, column)
-      |
-        node order by file, line, column, getIdentityString(function)
-      )
-  }
-
-  override string getProperty(string key) {
-    result = super.getProperty(key)
-    or
-    key = "semmle.order" and result = this.getOrder().toString()
-  }
-
-  /**
-   * Gets the `Function` this node represents.
-   */
-  final Function getFunction() { result = func }
 }
 
 private string getChildAccessorWithoutConversions(Locatable parent, Element child) {
-  shouldPrintFunction(getEnclosingFunction(parent)) and
+  shouldPrintDeclaration(getAnEnclosingDeclaration(parent)) and
   (
     exists(Stmt s | s = parent |
       namedStmtChildPredicates(s, child, result)
@@ -647,7 +674,7 @@ private string getChildAccessorWithoutConversions(Locatable parent, Element chil
 }
 
 private predicate namedStmtChildPredicates(Locatable s, Element e, string pred) {
-  shouldPrintFunction(getEnclosingFunction(s)) and
+  shouldPrintDeclaration(getAnEnclosingDeclaration(s)) and
   (
     exists(int n | s.(BlockStmt).getStmt(n) = e and pred = "getStmt(" + n + ")")
     or
@@ -735,7 +762,7 @@ private predicate namedStmtChildPredicates(Locatable s, Element e, string pred)
 }
 
 private predicate namedExprChildPredicates(Expr expr, Element ele, string pred) {
-  shouldPrintFunction(expr.getEnclosingFunction()) and
+  shouldPrintDeclaration(expr.getEnclosingDeclaration()) and
   (
     expr.(Access).getTarget() = ele and pred = "getTarget()"
     or
@@ -799,17 +826,11 @@ private predicate namedExprChildPredicates(Expr expr, Element ele, string pred)
     or
     expr.(Conversion).getExpr() = ele and pred = "getExpr()"
     or
-    expr.(DeleteArrayExpr).getAllocatorCall() = ele and pred = "getAllocatorCall()"
+    expr.(DeleteOrDeleteArrayExpr).getDeallocatorCall() = ele and pred = "getDeallocatorCall()"
     or
-    expr.(DeleteArrayExpr).getDestructorCall() = ele and pred = "getDestructorCall()"
+    expr.(DeleteOrDeleteArrayExpr).getDestructorCall() = ele and pred = "getDestructorCall()"
     or
-    expr.(DeleteArrayExpr).getExpr() = ele and pred = "getExpr()"
-    or
-    expr.(DeleteExpr).getAllocatorCall() = ele and pred = "getAllocatorCall()"
-    or
-    expr.(DeleteExpr).getDestructorCall() = ele and pred = "getDestructorCall()"
-    or
-    expr.(DeleteExpr).getExpr() = ele and pred = "getExpr()"
+    expr.(DeleteOrDeleteArrayExpr).getExpr() = ele and pred = "getExpr()"
     or
     expr.(DestructorFieldDestruction).getExpr() = ele and pred = "getExpr()"
     or

cpp/ql/lib/semmle/code/cpp/Type.qll

@@ -814,9 +814,6 @@ private predicate floatingPointTypeMapping(
   // _Float128
   kind = 49 and base = 2 and domain = TRealDomain() and realKind = 49 and extended = false
   or
-  // _Float128x
-  kind = 50 and base = 2 and domain = TRealDomain() and realKind = 50 and extended = true
-  or
   // _Float16
   kind = 52 and base = 2 and domain = TRealDomain() and realKind = 52 and extended = false
   or

cpp/ql/lib/semmle/code/cpp/commons/Buffer.qll

@@ -73,6 +73,10 @@ private int isSource(Expr bufferExpr, Element why) {
   )
 }
 
+/** Same as `getBufferSize`, but with the `why` column projected away to prevent large duplications. */
+pragma[nomagic]
+int getBufferSizeProj(Expr bufferExpr) { result = getBufferSize(bufferExpr, _) }
+
 /**
  * Get the size in bytes of the buffer pointed to by an expression (if this can be determined).
  */
@@ -87,7 +91,7 @@ int getBufferSize(Expr bufferExpr, Element why) {
     why = bufferVar and
     parentPtr = bufferExpr.(VariableAccess).getQualifier() and
     parentPtr.getTarget().getUnspecifiedType().(PointerType).getBaseType() = parentClass and
-    result = getBufferSize(parentPtr, _) + bufferSize - parentClass.getSize()
+    result = getBufferSizeProj(parentPtr) + bufferSize - parentClass.getSize()
   |
     if exists(bufferVar.getType().getSize())
     then bufferSize = bufferVar.getType().getSize()
@@ -95,7 +99,6 @@ int getBufferSize(Expr bufferExpr, Element why) {
   )
   or
   // dataflow (all sources must be the same size)
-  result = unique(Expr def | DataFlow::localExprFlowStep(def, bufferExpr) | getBufferSize(def, _)) and
-  // find reason
+  result = unique(Expr def | DataFlow::localExprFlowStep(def, bufferExpr) | getBufferSizeProj(def)) and
   exists(Expr def | DataFlow::localExprFlowStep(def, bufferExpr) | exists(getBufferSize(def, why)))
 }

cpp/ql/lib/semmle/code/cpp/controlflow/Dereferenced.qll

@@ -26,17 +26,18 @@ predicate callDereferences(FunctionCall fc, int i) {
 }
 
 /**
- * Holds if evaluation of `op` dereferences `e`.
+ * Holds if evaluation of `op` dereferences `e` directly.
+ *
+ * This predicate does not recurse through function calls or arithmetic operations. To find
+ * such cases, use `dereferencedByOperation`.
  */
-predicate dereferencedByOperation(Expr op, Expr e) {
+predicate directDereferencedByOperation(Expr op, Expr e) {
   exists(PointerDereferenceExpr deref |
     deref.getAChild() = e and
     deref = op and
     not deref.getParent*() instanceof SizeofOperator
   )
   or
-  exists(CrementOperation crement | dereferencedByOperation(e, op) and crement.getOperand() = e)
-  or
   exists(ArrayExpr ae |
     (
       not ae.getParent() instanceof AddressOfExpr and
@@ -50,6 +51,24 @@ predicate dereferencedByOperation(Expr op, Expr e) {
     )
   )
   or
+  // ptr->Field
+  e = op.(FieldAccess).getQualifier() and isClassPointerType(e.getType())
+  or
+  // ptr->method()
+  e = op.(Call).getQualifier() and isClassPointerType(e.getType())
+}
+
+/**
+ * Holds if evaluation of `op` dereferences `e`.
+ *
+ * This includes the set of operations identified via `directDereferencedByOperation`, as well
+ * as calls to function that are known to dereference an argument.
+ */
+predicate dereferencedByOperation(Expr op, Expr e) {
+  directDereferencedByOperation(op, e)
+  or
+  exists(CrementOperation crement | dereferencedByOperation(e, op) and crement.getOperand() = e)
+  or
   exists(AddressOfExpr addof, ArrayExpr ae |
     dereferencedByOperation(addof, op) and
     addof.getOperand() = ae and
@@ -74,12 +93,6 @@ predicate dereferencedByOperation(Expr op, Expr e) {
     e = fc.getArgument(i) and
     op = fc
   )
-  or
-  // ptr->Field
-  e = op.(FieldAccess).getQualifier() and isClassPointerType(e.getType())
-  or
-  // ptr->method()
-  e = op.(Call).getQualifier() and isClassPointerType(e.getType())
 }
 
 private predicate isClassPointerType(Type t) {

cpp/ql/lib/semmle/code/cpp/controlflow/IRGuards.qll

@@ -240,7 +240,7 @@ private class GuardConditionFromIR extends GuardCondition {
    */
   private predicate controlsBlock(BasicBlock controlled, boolean testIsTrue) {
     exists(IRBlock irb |
-      forex(IRGuardCondition inst | inst = ir | inst.controls(irb, testIsTrue)) and
+      ir.controls(irb, testIsTrue) and
       irb.getAnInstruction().getAst().(ControlFlowNode).getBasicBlock() = controlled and
       not isUnreachedBlock(irb)
     )

cpp/ql/lib/semmle/code/cpp/controlflow/internal/CFG.qll

@@ -332,21 +332,12 @@ private Node getControlOrderChildSparse(Node n, int i) {
   n = any(ConditionDeclExpr cd | i = 0 and result = cd.getInitializingExpr())
   or
   n =
-    any(DeleteExpr del |
+    any(DeleteOrDeleteArrayExpr del |
       i = 0 and result = del.getExpr()
       or
       i = 1 and result = del.getDestructorCall()
       or
-      i = 2 and result = del.getAllocatorCall()
-    )
-  or
-  n =
-    any(DeleteArrayExpr del |
-      i = 0 and result = del.getExpr()
-      or
-      i = 1 and result = del.getDestructorCall()
-      or
-      i = 2 and result = del.getAllocatorCall()
+      i = 2 and result = del.getDeallocatorCall()
     )
   or
   n =

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

@@ -26,6 +26,8 @@ import cpp
  * global (inter-procedural) data flow analyses.
  */
 deprecated module DataFlow {
-  import semmle.code.cpp.dataflow.internal.DataFlow
+  private import semmle.code.cpp.dataflow.internal.DataFlowImplSpecific
+  private import codeql.dataflow.DataFlow
+  import DataFlowMake<CppOldDataFlow>
   import semmle.code.cpp.dataflow.internal.DataFlowImpl1
 }

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

@@ -25,6 +25,10 @@ import semmle.code.cpp.dataflow.DataFlow2
  * global (inter-procedural) taint-tracking analyses.
  */
 deprecated module TaintTracking {
-  import semmle.code.cpp.dataflow.internal.tainttracking1.TaintTracking
+  import semmle.code.cpp.dataflow.internal.tainttracking1.TaintTrackingParameter::Public
+  private import semmle.code.cpp.dataflow.internal.DataFlowImplSpecific
+  private import semmle.code.cpp.dataflow.internal.TaintTrackingImplSpecific
+  private import codeql.dataflow.TaintTracking
+  import TaintFlowMake<CppOldDataFlow, CppOldTaintTracking>
   import semmle.code.cpp.dataflow.internal.tainttracking1.TaintTrackingImpl
 }

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

@@ -1,433 +0,0 @@
-/**
- * Provides an implementation of global (interprocedural) data flow. This file
- * re-exports the local (intraprocedural) data flow analysis from
- * `DataFlowImplSpecific::Public` and adds a global analysis, mainly exposed
- * through the `Global` and `GlobalWithState` modules.
- */
-
-private import DataFlowImplCommon
-private import DataFlowImplSpecific::Private
-import DataFlowImplSpecific::Public
-import DataFlowImplCommonPublic
-private import DataFlowImpl
-
-/** An input configuration for data flow. */
-signature module ConfigSig {
-  /**
-   * Holds if `source` is a relevant data flow source.
-   */
-  predicate isSource(Node source);
-
-  /**
-   * Holds if `sink` is a relevant data flow sink.
-   */
-  predicate isSink(Node sink);
-
-  /**
-   * Holds if data flow through `node` is prohibited. This completely removes
-   * `node` from the data flow graph.
-   */
-  default predicate isBarrier(Node node) { none() }
-
-  /** Holds if data flow into `node` is prohibited. */
-  default predicate isBarrierIn(Node node) { none() }
-
-  /** Holds if data flow out of `node` is prohibited. */
-  default predicate isBarrierOut(Node node) { none() }
-
-  /**
-   * Holds if data may flow from `node1` to `node2` in addition to the normal data-flow steps.
-   */
-  default predicate isAdditionalFlowStep(Node node1, Node node2) { none() }
-
-  /**
-   * Holds if an arbitrary number of implicit read steps of content `c` may be
-   * taken at `node`.
-   */
-  default predicate allowImplicitRead(Node node, ContentSet c) { none() }
-
-  /**
-   * Holds if `node` should never be skipped over in the `PathGraph` and in path
-   * explanations.
-   */
-  default predicate neverSkip(Node node) {
-    isAdditionalFlowStep(node, _) or isAdditionalFlowStep(_, node)
-  }
-
-  /**
-   * Gets the virtual dispatch branching limit when calculating field flow.
-   * This can be overridden to a smaller value to improve performance (a
-   * value of 0 disables field flow), or a larger value to get more results.
-   */
-  default int fieldFlowBranchLimit() { result = 2 }
-
-  /**
-   * Gets a data flow configuration feature to add restrictions to the set of
-   * valid flow paths.
-   *
-   * - `FeatureHasSourceCallContext`:
-   *    Assume that sources have some existing call context to disallow
-   *    conflicting return-flow directly following the source.
-   * - `FeatureHasSinkCallContext`:
-   *    Assume that sinks have some existing call context to disallow
-   *    conflicting argument-to-parameter flow directly preceding the sink.
-   * - `FeatureEqualSourceSinkCallContext`:
-   *    Implies both of the above and additionally ensures that the entire flow
-   *    path preserves the call context.
-   *
-   * These features are generally not relevant for typical end-to-end data flow
-   * queries, but should only be used for constructing paths that need to
-   * somehow be pluggable in another path context.
-   */
-  default FlowFeature getAFeature() { none() }
-
-  /** Holds if sources should be grouped in the result of `flowPath`. */
-  default predicate sourceGrouping(Node source, string sourceGroup) { none() }
-
-  /** Holds if sinks should be grouped in the result of `flowPath`. */
-  default predicate sinkGrouping(Node sink, string sinkGroup) { none() }
-
-  /**
-   * Holds if hidden nodes should be included in the data flow graph.
-   *
-   * This feature should only be used for debugging or when the data flow graph
-   * is not visualized (as it is in a `path-problem` query).
-   */
-  default predicate includeHiddenNodes() { none() }
-}
-
-/** An input configuration for data flow using flow state. */
-signature module StateConfigSig {
-  bindingset[this]
-  class FlowState;
-
-  /**
-   * Holds if `source` is a relevant data flow source with the given initial
-   * `state`.
-   */
-  predicate isSource(Node source, FlowState state);
-
-  /**
-   * Holds if `sink` is a relevant data flow sink accepting `state`.
-   */
-  predicate isSink(Node sink, FlowState state);
-
-  /**
-   * Holds if data flow through `node` is prohibited. This completely removes
-   * `node` from the data flow graph.
-   */
-  default predicate isBarrier(Node node) { none() }
-
-  /**
-   * Holds if data flow through `node` is prohibited when the flow state is
-   * `state`.
-   */
-  default predicate isBarrier(Node node, FlowState state) { none() }
-
-  /** Holds if data flow into `node` is prohibited. */
-  default predicate isBarrierIn(Node node) { none() }
-
-  /** Holds if data flow out of `node` is prohibited. */
-  default predicate isBarrierOut(Node node) { none() }
-
-  /**
-   * Holds if data may flow from `node1` to `node2` in addition to the normal data-flow steps.
-   */
-  default predicate isAdditionalFlowStep(Node node1, Node node2) { none() }
-
-  /**
-   * Holds if data may flow from `node1` to `node2` in addition to the normal data-flow steps.
-   * This step is only applicable in `state1` and updates the flow state to `state2`.
-   */
-  default predicate isAdditionalFlowStep(Node node1, FlowState state1, Node node2, FlowState state2) {
-    none()
-  }
-
-  /**
-   * Holds if an arbitrary number of implicit read steps of content `c` may be
-   * taken at `node`.
-   */
-  default predicate allowImplicitRead(Node node, ContentSet c) { none() }
-
-  /**
-   * Holds if `node` should never be skipped over in the `PathGraph` and in path
-   * explanations.
-   */
-  default predicate neverSkip(Node node) {
-    isAdditionalFlowStep(node, _) or
-    isAdditionalFlowStep(_, node) or
-    isAdditionalFlowStep(node, _, _, _) or
-    isAdditionalFlowStep(_, _, node, _)
-  }
-
-  /**
-   * Gets the virtual dispatch branching limit when calculating field flow.
-   * This can be overridden to a smaller value to improve performance (a
-   * value of 0 disables field flow), or a larger value to get more results.
-   */
-  default int fieldFlowBranchLimit() { result = 2 }
-
-  /**
-   * Gets a data flow configuration feature to add restrictions to the set of
-   * valid flow paths.
-   *
-   * - `FeatureHasSourceCallContext`:
-   *    Assume that sources have some existing call context to disallow
-   *    conflicting return-flow directly following the source.
-   * - `FeatureHasSinkCallContext`:
-   *    Assume that sinks have some existing call context to disallow
-   *    conflicting argument-to-parameter flow directly preceding the sink.
-   * - `FeatureEqualSourceSinkCallContext`:
-   *    Implies both of the above and additionally ensures that the entire flow
-   *    path preserves the call context.
-   *
-   * These features are generally not relevant for typical end-to-end data flow
-   * queries, but should only be used for constructing paths that need to
-   * somehow be pluggable in another path context.
-   */
-  default FlowFeature getAFeature() { none() }
-
-  /** Holds if sources should be grouped in the result of `flowPath`. */
-  default predicate sourceGrouping(Node source, string sourceGroup) { none() }
-
-  /** Holds if sinks should be grouped in the result of `flowPath`. */
-  default predicate sinkGrouping(Node sink, string sinkGroup) { none() }
-
-  /**
-   * Holds if hidden nodes should be included in the data flow graph.
-   *
-   * This feature should only be used for debugging or when the data flow graph
-   * is not visualized (as it is in a `path-problem` query).
-   */
-  default predicate includeHiddenNodes() { none() }
-}
-
-/**
- * Gets the exploration limit for `partialFlow` and `partialFlowRev`
- * measured in approximate number of interprocedural steps.
- */
-signature int explorationLimitSig();
-
-/**
- * The output of a global data flow computation.
- */
-signature module GlobalFlowSig {
-  /**
-   * A `Node` augmented with a call context (except for sinks) and an access path.
-   * Only those `PathNode`s that are reachable from a source, and which can reach a sink, are generated.
-   */
-  class PathNode;
-
-  /**
-   * Holds if data can flow from `source` to `sink`.
-   *
-   * The corresponding paths are generated from the end-points and the graph
-   * included in the module `PathGraph`.
-   */
-  predicate flowPath(PathNode source, PathNode sink);
-
-  /**
-   * Holds if data can flow from `source` to `sink`.
-   */
-  predicate flow(Node source, Node sink);
-
-  /**
-   * Holds if data can flow from some source to `sink`.
-   */
-  predicate flowTo(Node sink);
-
-  /**
-   * Holds if data can flow from some source to `sink`.
-   */
-  predicate flowToExpr(DataFlowExpr sink);
-}
-
-/**
- * Constructs a global data flow computation.
- */
-module Global<ConfigSig Config> implements GlobalFlowSig {
-  private module C implements FullStateConfigSig {
-    import DefaultState<Config>
-    import Config
-  }
-
-  import Impl<C>
-}
-
-/** DEPRECATED: Use `Global` instead. */
-deprecated module Make<ConfigSig Config> implements GlobalFlowSig {
-  import Global<Config>
-}
-
-/**
- * Constructs a global data flow computation using flow state.
- */
-module GlobalWithState<StateConfigSig Config> implements GlobalFlowSig {
-  private module C implements FullStateConfigSig {
-    import Config
-  }
-
-  import Impl<C>
-}
-
-/** DEPRECATED: Use `GlobalWithState` instead. */
-deprecated module MakeWithState<StateConfigSig Config> implements GlobalFlowSig {
-  import GlobalWithState<Config>
-}
-
-signature class PathNodeSig {
-  /** Gets a textual representation of this element. */
-  string toString();
-
-  /**
-   * Holds if this element is at the specified location.
-   * The location spans column `startcolumn` of line `startline` to
-   * column `endcolumn` of line `endline` in file `filepath`.
-   * For more information, see
-   * [Locations](https://codeql.github.com/docs/writing-codeql-queries/providing-locations-in-codeql-queries/).
-   */
-  predicate hasLocationInfo(
-    string filepath, int startline, int startcolumn, int endline, int endcolumn
-  );
-
-  /** Gets the underlying `Node`. */
-  Node getNode();
-}
-
-signature module PathGraphSig<PathNodeSig PathNode> {
-  /** Holds if `(a,b)` is an edge in the graph of data flow path explanations. */
-  predicate edges(PathNode a, PathNode b);
-
-  /** Holds if `n` is a node in the graph of data flow path explanations. */
-  predicate nodes(PathNode n, string key, string val);
-
-  /**
-   * Holds if `(arg, par, ret, out)` forms a subpath-tuple, that is, flow through
-   * a subpath between `par` and `ret` with the connecting edges `arg -> par` and
-   * `ret -> out` is summarized as the edge `arg -> out`.
-   */
-  predicate subpaths(PathNode arg, PathNode par, PathNode ret, PathNode out);
-}
-
-/**
- * Constructs a `PathGraph` from two `PathGraph`s by disjoint union.
- */
-module MergePathGraph<
-  PathNodeSig PathNode1, PathNodeSig PathNode2, PathGraphSig<PathNode1> Graph1,
-  PathGraphSig<PathNode2> Graph2>
-{
-  private newtype TPathNode =
-    TPathNode1(PathNode1 p) or
-    TPathNode2(PathNode2 p)
-
-  /** A node in a graph of path explanations that is formed by disjoint union of the two given graphs. */
-  class PathNode extends TPathNode {
-    /** Gets this as a projection on the first given `PathGraph`. */
-    PathNode1 asPathNode1() { this = TPathNode1(result) }
-
-    /** Gets this as a projection on the second given `PathGraph`. */
-    PathNode2 asPathNode2() { this = TPathNode2(result) }
-
-    /** Gets a textual representation of this element. */
-    string toString() {
-      result = this.asPathNode1().toString() or
-      result = this.asPathNode2().toString()
-    }
-
-    /**
-     * Holds if this element is at the specified location.
-     * The location spans column `startcolumn` of line `startline` to
-     * column `endcolumn` of line `endline` in file `filepath`.
-     * For more information, see
-     * [Locations](https://codeql.github.com/docs/writing-codeql-queries/providing-locations-in-codeql-queries/).
-     */
-    predicate hasLocationInfo(
-      string filepath, int startline, int startcolumn, int endline, int endcolumn
-    ) {
-      this.asPathNode1().hasLocationInfo(filepath, startline, startcolumn, endline, endcolumn) or
-      this.asPathNode2().hasLocationInfo(filepath, startline, startcolumn, endline, endcolumn)
-    }
-
-    /** Gets the underlying `Node`. */
-    Node getNode() {
-      result = this.asPathNode1().getNode() or
-      result = this.asPathNode2().getNode()
-    }
-  }
-
-  /**
-   * Provides the query predicates needed to include a graph in a path-problem query.
-   */
-  module PathGraph implements PathGraphSig<PathNode> {
-    /** Holds if `(a,b)` is an edge in the graph of data flow path explanations. */
-    query predicate edges(PathNode a, PathNode b) {
-      Graph1::edges(a.asPathNode1(), b.asPathNode1()) or
-      Graph2::edges(a.asPathNode2(), b.asPathNode2())
-    }
-
-    /** Holds if `n` is a node in the graph of data flow path explanations. */
-    query predicate nodes(PathNode n, string key, string val) {
-      Graph1::nodes(n.asPathNode1(), key, val) or
-      Graph2::nodes(n.asPathNode2(), key, val)
-    }
-
-    /**
-     * Holds if `(arg, par, ret, out)` forms a subpath-tuple, that is, flow through
-     * a subpath between `par` and `ret` with the connecting edges `arg -> par` and
-     * `ret -> out` is summarized as the edge `arg -> out`.
-     */
-    query predicate subpaths(PathNode arg, PathNode par, PathNode ret, PathNode out) {
-      Graph1::subpaths(arg.asPathNode1(), par.asPathNode1(), ret.asPathNode1(), out.asPathNode1()) or
-      Graph2::subpaths(arg.asPathNode2(), par.asPathNode2(), ret.asPathNode2(), out.asPathNode2())
-    }
-  }
-}
-
-/**
- * Constructs a `PathGraph` from three `PathGraph`s by disjoint union.
- */
-module MergePathGraph3<
-  PathNodeSig PathNode1, PathNodeSig PathNode2, PathNodeSig PathNode3,
-  PathGraphSig<PathNode1> Graph1, PathGraphSig<PathNode2> Graph2, PathGraphSig<PathNode3> Graph3>
-{
-  private module MergedInner = MergePathGraph<PathNode1, PathNode2, Graph1, Graph2>;
-
-  private module Merged =
-    MergePathGraph<MergedInner::PathNode, PathNode3, MergedInner::PathGraph, Graph3>;
-
-  /** A node in a graph of path explanations that is formed by disjoint union of the three given graphs. */
-  class PathNode instanceof Merged::PathNode {
-    /** Gets this as a projection on the first given `PathGraph`. */
-    PathNode1 asPathNode1() { result = super.asPathNode1().asPathNode1() }
-
-    /** Gets this as a projection on the second given `PathGraph`. */
-    PathNode2 asPathNode2() { result = super.asPathNode1().asPathNode2() }
-
-    /** Gets this as a projection on the third given `PathGraph`. */
-    PathNode3 asPathNode3() { result = super.asPathNode2() }
-
-    /** Gets a textual representation of this element. */
-    string toString() { result = super.toString() }
-
-    /**
-     * Holds if this element is at the specified location.
-     * The location spans column `startcolumn` of line `startline` to
-     * column `endcolumn` of line `endline` in file `filepath`.
-     * For more information, see
-     * [Locations](https://codeql.github.com/docs/writing-codeql-queries/providing-locations-in-codeql-queries/).
-     */
-    predicate hasLocationInfo(
-      string filepath, int startline, int startcolumn, int endline, int endcolumn
-    ) {
-      super.hasLocationInfo(filepath, startline, startcolumn, endline, endcolumn)
-    }
-
-    /** Gets the underlying `Node`. */
-    Node getNode() { result = super.getNode() }
-  }
-
-  /**
-   * Provides the query predicates needed to include a graph in a path-problem query.
-   */
-  module PathGraph = Merged::PathGraph;
-}

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

@@ -5,8 +5,8 @@ private import DataFlowUtil
 /**
  * Gets a function that might be called by `call`.
  */
-Function viableCallable(Call call) {
-  result = call.getTarget()
+Function viableCallable(DataFlowCall call) {
+  result = call.(Call).getTarget()
   or
   // If the target of the call does not have a body in the snapshot, it might
   // be because the target is just a header declaration, and the real target
@@ -58,13 +58,13 @@ private predicate functionSignature(Function f, string qualifiedName, int nparam
  * Holds if the set of viable implementations that can be called by `call`
  * might be improved by knowing the call context.
  */
-predicate mayBenefitFromCallContext(Call call, Function f) { none() }
+predicate mayBenefitFromCallContext(DataFlowCall call, Function f) { none() }
 
 /**
  * Gets a viable dispatch target of `call` in the context `ctx`. This is
  * restricted to those `call`s for which a context might make a difference.
  */
-Function viableImplInCallContext(Call call, Call ctx) { none() }
+Function viableImplInCallContext(DataFlowCall call, DataFlowCall ctx) { none() }
 
 /** A parameter position represented by an integer. */
 class ParameterPosition extends int {
@@ -79,13 +79,3 @@ class ArgumentPosition extends int {
 /** Holds if arguments at position `apos` match parameters at position `ppos`. */
 pragma[inline]
 predicate parameterMatch(ParameterPosition ppos, ArgumentPosition apos) { ppos = apos }
-
-/**
- * Holds if flow from `call`'s argument `arg` to parameter `p` is permissible.
- *
- * This is a temporary hook to support technical debt in the Go language; do not use.
- */
-pragma[inline]
-predicate golangSpecificParamArgFilter(DataFlowCall call, ParameterNode p, ArgumentNode arg) {
-  any()
-}

cpp/ql/lib/semmle/code/cpp/dataflow/internal/DataFlowImpl1.qll

@@ -276,6 +276,8 @@ private module Config implements FullStateConfigSig {
     getConfig(state).isSource(source) and getState(state) instanceof FlowStateEmpty
   }
 
+  predicate isSink(Node sink) { none() }
+
   predicate isSink(Node sink, FlowState state) {
     getConfig(state).isSink(sink, getState(state))
     or
@@ -295,6 +297,10 @@ private module Config implements FullStateConfigSig {
 
   predicate isBarrierOut(Node node) { any(Configuration config).isBarrierOut(node) }
 
+  predicate isBarrierIn(Node node, FlowState state) { none() }
+
+  predicate isBarrierOut(Node node, FlowState state) { none() }
+
   predicate isAdditionalFlowStep(Node node1, Node node2) {
     singleConfiguration() and
     any(Configuration config).isAdditionalFlowStep(node1, node2)

cpp/ql/lib/semmle/code/cpp/dataflow/internal/DataFlowImpl2.qll

@@ -276,6 +276,8 @@ private module Config implements FullStateConfigSig {
     getConfig(state).isSource(source) and getState(state) instanceof FlowStateEmpty
   }
 
+  predicate isSink(Node sink) { none() }
+
   predicate isSink(Node sink, FlowState state) {
     getConfig(state).isSink(sink, getState(state))
     or
@@ -295,6 +297,10 @@ private module Config implements FullStateConfigSig {
 
   predicate isBarrierOut(Node node) { any(Configuration config).isBarrierOut(node) }
 
+  predicate isBarrierIn(Node node, FlowState state) { none() }
+
+  predicate isBarrierOut(Node node, FlowState state) { none() }
+
   predicate isAdditionalFlowStep(Node node1, Node node2) {
     singleConfiguration() and
     any(Configuration config).isAdditionalFlowStep(node1, node2)

cpp/ql/lib/semmle/code/cpp/dataflow/internal/DataFlowImpl3.qll

@@ -276,6 +276,8 @@ private module Config implements FullStateConfigSig {
     getConfig(state).isSource(source) and getState(state) instanceof FlowStateEmpty
   }
 
+  predicate isSink(Node sink) { none() }
+
   predicate isSink(Node sink, FlowState state) {
     getConfig(state).isSink(sink, getState(state))
     or
@@ -295,6 +297,10 @@ private module Config implements FullStateConfigSig {
 
   predicate isBarrierOut(Node node) { any(Configuration config).isBarrierOut(node) }
 
+  predicate isBarrierIn(Node node, FlowState state) { none() }
+
+  predicate isBarrierOut(Node node, FlowState state) { none() }
+
   predicate isAdditionalFlowStep(Node node1, Node node2) {
     singleConfiguration() and
     any(Configuration config).isAdditionalFlowStep(node1, node2)

cpp/ql/lib/semmle/code/cpp/dataflow/internal/DataFlowImpl4.qll

@@ -276,6 +276,8 @@ private module Config implements FullStateConfigSig {
     getConfig(state).isSource(source) and getState(state) instanceof FlowStateEmpty
   }
 
+  predicate isSink(Node sink) { none() }
+
   predicate isSink(Node sink, FlowState state) {
     getConfig(state).isSink(sink, getState(state))
     or
@@ -295,6 +297,10 @@ private module Config implements FullStateConfigSig {
 
   predicate isBarrierOut(Node node) { any(Configuration config).isBarrierOut(node) }
 
+  predicate isBarrierIn(Node node, FlowState state) { none() }
+
+  predicate isBarrierOut(Node node, FlowState state) { none() }
+
   predicate isAdditionalFlowStep(Node node1, Node node2) {
     singleConfiguration() and
     any(Configuration config).isAdditionalFlowStep(node1, node2)

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

@@ -3,297 +3,25 @@
  * data-flow classes and predicates.
  */
 
-private import DataFlowImplSpecific::Private
-private import DataFlowImplSpecific::Public
-private import tainttracking1.TaintTrackingParameter::Private
-private import tainttracking1.TaintTrackingParameter::Public
+private import cpp
+private import DataFlowImplSpecific
+private import TaintTrackingImplSpecific
+private import codeql.dataflow.internal.DataFlowImplConsistency
 
-module Consistency {
-  private newtype TConsistencyConfiguration = MkConsistencyConfiguration()
-
-  /** A class for configuring the consistency queries. */
-  class ConsistencyConfiguration extends TConsistencyConfiguration {
-    string toString() { none() }
-
-    /** Holds if `n` should be excluded from the consistency test `uniqueEnclosingCallable`. */
-    predicate uniqueEnclosingCallableExclude(Node n) { none() }
-
-    /** Holds if `call` should be excluded from the consistency test `uniqueCallEnclosingCallable`. */
-    predicate uniqueCallEnclosingCallableExclude(DataFlowCall call) { none() }
-
-    /** Holds if `n` should be excluded from the consistency test `uniqueNodeLocation`. */
-    predicate uniqueNodeLocationExclude(Node n) { none() }
-
-    /** Holds if `n` should be excluded from the consistency test `missingLocation`. */
-    predicate missingLocationExclude(Node n) { none() }
-
-    /** Holds if `n` should be excluded from the consistency test `postWithInFlow`. */
-    predicate postWithInFlowExclude(Node n) { none() }
-
-    /** Holds if `n` should be excluded from the consistency test `argHasPostUpdate`. */
-    predicate argHasPostUpdateExclude(ArgumentNode n) { none() }
-
-    /** Holds if `n` should be excluded from the consistency test `reverseRead`. */
-    predicate reverseReadExclude(Node n) { none() }
-
-    /** Holds if `n` should be excluded from the consistency test `postHasUniquePre`. */
-    predicate postHasUniquePreExclude(PostUpdateNode n) { none() }
-
-    /** Holds if `n` should be excluded from the consistency test `uniquePostUpdate`. */
-    predicate uniquePostUpdateExclude(Node n) { none() }
-
-    /** Holds if `(call, ctx)` should be excluded from the consistency test `viableImplInCallContextTooLargeExclude`. */
-    predicate viableImplInCallContextTooLargeExclude(
-      DataFlowCall call, DataFlowCall ctx, DataFlowCallable callable
-    ) {
-      none()
-    }
-
-    /** Holds if `(c, pos, p)` should be excluded from the consistency test `uniqueParameterNodeAtPosition`. */
-    predicate uniqueParameterNodeAtPositionExclude(DataFlowCallable c, ParameterPosition pos, Node p) {
-      none()
-    }
-
-    /** Holds if `(c, pos, p)` should be excluded from the consistency test `uniqueParameterNodePosition`. */
-    predicate uniqueParameterNodePositionExclude(DataFlowCallable c, ParameterPosition pos, Node p) {
-      none()
-    }
-
-    /** Holds if `n` should be excluded from the consistency test `identityLocalStep`. */
-    predicate identityLocalStepExclude(Node n) { none() }
-  }
-
-  private class RelevantNode extends Node {
-    RelevantNode() {
-      this instanceof ArgumentNode or
-      this instanceof ParameterNode or
-      this instanceof ReturnNode or
-      this = getAnOutNode(_, _) or
-      simpleLocalFlowStep(this, _) or
-      simpleLocalFlowStep(_, this) or
-      jumpStep(this, _) or
-      jumpStep(_, this) or
-      storeStep(this, _, _) or
-      storeStep(_, _, this) or
-      readStep(this, _, _) or
-      readStep(_, _, this) or
-      defaultAdditionalTaintStep(this, _) or
-      defaultAdditionalTaintStep(_, this)
-    }
-  }
-
-  query predicate uniqueEnclosingCallable(Node n, string msg) {
-    exists(int c |
-      n instanceof RelevantNode and
-      c = count(nodeGetEnclosingCallable(n)) and
-      c != 1 and
-      not any(ConsistencyConfiguration conf).uniqueEnclosingCallableExclude(n) and
-      msg = "Node should have one enclosing callable but has " + c + "."
-    )
-  }
-
-  query predicate uniqueCallEnclosingCallable(DataFlowCall call, string msg) {
-    exists(int c |
-      c = count(call.getEnclosingCallable()) and
-      c != 1 and
-      not any(ConsistencyConfiguration conf).uniqueCallEnclosingCallableExclude(call) and
-      msg = "Call should have one enclosing callable but has " + c + "."
-    )
-  }
-
-  query predicate uniqueType(Node n, string msg) {
-    exists(int c |
-      n instanceof RelevantNode and
-      c = count(getNodeType(n)) and
-      c != 1 and
-      msg = "Node should have one type but has " + c + "."
-    )
-  }
-
-  query predicate uniqueNodeLocation(Node n, string msg) {
-    exists(int c |
-      c =
-        count(string filepath, int startline, int startcolumn, int endline, int endcolumn |
-          n.hasLocationInfo(filepath, startline, startcolumn, endline, endcolumn)
-        ) and
-      c != 1 and
-      not any(ConsistencyConfiguration conf).uniqueNodeLocationExclude(n) and
-      msg = "Node should have one location but has " + c + "."
-    )
-  }
-
-  query predicate missingLocation(string msg) {
-    exists(int c |
-      c =
-        strictcount(Node n |
-          not n.hasLocationInfo(_, _, _, _, _) and
-          not any(ConsistencyConfiguration conf).missingLocationExclude(n)
-        ) and
-      msg = "Nodes without location: " + c
-    )
-  }
-
-  query predicate uniqueNodeToString(Node n, string msg) {
-    exists(int c |
-      c = count(n.toString()) and
-      c != 1 and
-      msg = "Node should have one toString but has " + c + "."
-    )
-  }
-
-  query predicate missingToString(string msg) {
-    exists(int c |
-      c = strictcount(Node n | not exists(n.toString())) and
-      msg = "Nodes without toString: " + c
-    )
-  }
-
-  query predicate parameterCallable(ParameterNode p, string msg) {
-    exists(DataFlowCallable c | isParameterNode(p, c, _) and c != nodeGetEnclosingCallable(p)) and
-    msg = "Callable mismatch for parameter."
-  }
-
-  query predicate localFlowIsLocal(Node n1, Node n2, string msg) {
-    simpleLocalFlowStep(n1, n2) and
-    nodeGetEnclosingCallable(n1) != nodeGetEnclosingCallable(n2) and
-    msg = "Local flow step does not preserve enclosing callable."
-  }
-
-  query predicate readStepIsLocal(Node n1, Node n2, string msg) {
-    readStep(n1, _, n2) and
-    nodeGetEnclosingCallable(n1) != nodeGetEnclosingCallable(n2) and
-    msg = "Read step does not preserve enclosing callable."
-  }
-
-  query predicate storeStepIsLocal(Node n1, Node n2, string msg) {
-    storeStep(n1, _, n2) and
-    nodeGetEnclosingCallable(n1) != nodeGetEnclosingCallable(n2) and
-    msg = "Store step does not preserve enclosing callable."
-  }
-
-  private DataFlowType typeRepr() { result = getNodeType(_) }
-
-  query predicate compatibleTypesReflexive(DataFlowType t, string msg) {
-    t = typeRepr() and
-    not compatibleTypes(t, t) and
-    msg = "Type compatibility predicate is not reflexive."
-  }
-
-  query predicate unreachableNodeCCtx(Node n, DataFlowCall call, string msg) {
-    isUnreachableInCall(n, call) and
-    exists(DataFlowCallable c |
-      c = nodeGetEnclosingCallable(n) and
-      not viableCallable(call) = c
-    ) and
-    msg = "Call context for isUnreachableInCall is inconsistent with call graph."
-  }
-
-  query predicate localCallNodes(DataFlowCall call, Node n, string msg) {
-    (
-      n = getAnOutNode(call, _) and
-      msg = "OutNode and call does not share enclosing callable."
-      or
-      n.(ArgumentNode).argumentOf(call, _) and
-      msg = "ArgumentNode and call does not share enclosing callable."
-    ) and
-    nodeGetEnclosingCallable(n) != call.getEnclosingCallable()
-  }
-
-  // This predicate helps the compiler forget that in some languages
-  // it is impossible for a result of `getPreUpdateNode` to be an
-  // instance of `PostUpdateNode`.
-  private Node getPre(PostUpdateNode n) {
-    result = n.getPreUpdateNode()
+private module Input implements InputSig<CppOldDataFlow> {
+  predicate argHasPostUpdateExclude(Private::ArgumentNode n) {
+    // Is the null pointer (or something that's not really a pointer)
+    exists(n.asExpr().getValue())
     or
-    none()
-  }
-
-  query predicate postIsNotPre(PostUpdateNode n, string msg) {
-    getPre(n) = n and
-    msg = "PostUpdateNode should not equal its pre-update node."
-  }
-
-  query predicate postHasUniquePre(PostUpdateNode n, string msg) {
-    not any(ConsistencyConfiguration conf).postHasUniquePreExclude(n) and
-    exists(int c |
-      c = count(n.getPreUpdateNode()) and
-      c != 1 and
-      msg = "PostUpdateNode should have one pre-update node but has " + c + "."
+    // Isn't a pointer or is a pointer to const
+    forall(DerivedType dt | dt = n.asExpr().getActualType() |
+      dt.getBaseType().isConst()
+      or
+      dt.getBaseType() instanceof RoutineType
     )
-  }
-
-  query predicate uniquePostUpdate(Node n, string msg) {
-    not any(ConsistencyConfiguration conf).uniquePostUpdateExclude(n) and
-    1 < strictcount(PostUpdateNode post | post.getPreUpdateNode() = n) and
-    msg = "Node has multiple PostUpdateNodes."
-  }
-
-  query predicate postIsInSameCallable(PostUpdateNode n, string msg) {
-    nodeGetEnclosingCallable(n) != nodeGetEnclosingCallable(n.getPreUpdateNode()) and
-    msg = "PostUpdateNode does not share callable with its pre-update node."
-  }
-
-  private predicate hasPost(Node n) { exists(PostUpdateNode post | post.getPreUpdateNode() = n) }
-
-  query predicate reverseRead(Node n, string msg) {
-    exists(Node n2 | readStep(n, _, n2) and hasPost(n2) and not hasPost(n)) and
-    not any(ConsistencyConfiguration conf).reverseReadExclude(n) and
-    msg = "Origin of readStep is missing a PostUpdateNode."
-  }
-
-  query predicate argHasPostUpdate(ArgumentNode n, string msg) {
-    not hasPost(n) and
-    not any(ConsistencyConfiguration c).argHasPostUpdateExclude(n) and
-    msg = "ArgumentNode is missing PostUpdateNode."
-  }
-
-  // This predicate helps the compiler forget that in some languages
-  // it is impossible for a `PostUpdateNode` to be the target of
-  // `simpleLocalFlowStep`.
-  private predicate isPostUpdateNode(Node n) { n instanceof PostUpdateNode or none() }
-
-  query predicate postWithInFlow(Node n, string msg) {
-    isPostUpdateNode(n) and
-    not clearsContent(n, _) and
-    simpleLocalFlowStep(_, n) and
-    not any(ConsistencyConfiguration c).postWithInFlowExclude(n) and
-    msg = "PostUpdateNode should not be the target of local flow."
-  }
-
-  query predicate viableImplInCallContextTooLarge(
-    DataFlowCall call, DataFlowCall ctx, DataFlowCallable callable
-  ) {
-    callable = viableImplInCallContext(call, ctx) and
-    not callable = viableCallable(call) and
-    not any(ConsistencyConfiguration c).viableImplInCallContextTooLargeExclude(call, ctx, callable)
-  }
-
-  query predicate uniqueParameterNodeAtPosition(
-    DataFlowCallable c, ParameterPosition pos, Node p, string msg
-  ) {
-    not any(ConsistencyConfiguration conf).uniqueParameterNodeAtPositionExclude(c, pos, p) and
-    isParameterNode(p, c, pos) and
-    not exists(unique(Node p0 | isParameterNode(p0, c, pos))) and
-    msg = "Parameters with overlapping positions."
-  }
-
-  query predicate uniqueParameterNodePosition(
-    DataFlowCallable c, ParameterPosition pos, Node p, string msg
-  ) {
-    not any(ConsistencyConfiguration conf).uniqueParameterNodePositionExclude(c, pos, p) and
-    isParameterNode(p, c, pos) and
-    not exists(unique(ParameterPosition pos0 | isParameterNode(p, c, pos0))) and
-    msg = "Parameter node with multiple positions."
-  }
-
-  query predicate uniqueContentApprox(Content c, string msg) {
-    not exists(unique(ContentApprox approx | approx = getContentApprox(c))) and
-    msg = "Non-unique content approximation."
-  }
-
-  query predicate identityLocalStep(Node n, string msg) {
-    simpleLocalFlowStep(n, n) and
-    not any(ConsistencyConfiguration c).identityLocalStepExclude(n) and
-    msg = "Node steps to itself"
+    // The above list of cases isn't exhaustive, but it narrows down the
+    // consistency alerts enough that most of them are interesting.
   }
 }
+
+module Consistency = MakeConsistency<CppOldDataFlow, CppOldTaintTracking, Input>;

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

@@ -276,6 +276,8 @@ private module Config implements FullStateConfigSig {
     getConfig(state).isSource(source) and getState(state) instanceof FlowStateEmpty
   }
 
+  predicate isSink(Node sink) { none() }
+
   predicate isSink(Node sink, FlowState state) {
     getConfig(state).isSink(sink, getState(state))
     or
@@ -295,6 +297,10 @@ private module Config implements FullStateConfigSig {
 
   predicate isBarrierOut(Node node) { any(Configuration config).isBarrierOut(node) }
 
+  predicate isBarrierIn(Node node, FlowState state) { none() }
+
+  predicate isBarrierOut(Node node, FlowState state) { none() }
+
   predicate isAdditionalFlowStep(Node node1, Node node2) {
     singleConfiguration() and
     any(Configuration config).isAdditionalFlowStep(node1, node2)

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

@@ -1,6 +1,9 @@
 /**
  * Provides C++-specific definitions for use in the data flow library.
  */
+
+private import codeql.dataflow.DataFlow
+
 module Private {
   import DataFlowPrivate
   import DataFlowDispatch
@@ -9,3 +12,10 @@ module Private {
 module Public {
   import DataFlowUtil
 }
+
+module CppOldDataFlow implements InputSig {
+  import Private
+  import Public
+
+  Node exprNode(DataFlowExpr e) { result = Public::exprNode(e) }
+}

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

@@ -2,7 +2,6 @@ private import cpp
 private import DataFlowUtil
 private import DataFlowDispatch
 private import FlowVar
-private import DataFlowImplConsistency
 private import codeql.util.Unit
 
 /** Gets the callable in which this node occurs. */
@@ -153,10 +152,11 @@ predicate jumpStep(Node n1, Node n2) { none() }
  * Thus, `node2` references an object with a field `f` that contains the
  * value of `node1`.
  */
-predicate storeStep(Node node1, Content f, PostUpdateNode node2) {
+predicate storeStep(Node node1, ContentSet f, Node node2) {
   exists(ClassAggregateLiteral aggr, Field field |
-    // The following line requires `node2` to be both an `ExprNode` and a
+    // The following lines requires `node2` to be both an `ExprNode` and a
     // `PostUpdateNode`, which means it must be an `ObjectInitializerNode`.
+    node2 instanceof PostUpdateNode and
     node2.asExpr() = aggr and
     f.(FieldContent).getField() = field and
     aggr.getAFieldExpr(field) = node1.asExpr()
@@ -167,12 +167,13 @@ predicate storeStep(Node node1, Content f, PostUpdateNode node2) {
       node1.asExpr() = a and
       a.getLValue() = fa
     ) and
-    node2.getPreUpdateNode().asExpr() = fa.getQualifier() and
+    node2.(PostUpdateNode).getPreUpdateNode().asExpr() = fa.getQualifier() and
     f.(FieldContent).getField() = fa.getTarget()
   )
   or
   exists(ConstructorFieldInit cfi |
-    node2.getPreUpdateNode().(PreConstructorInitThis).getConstructorFieldInit() = cfi and
+    node2.(PostUpdateNode).getPreUpdateNode().(PreConstructorInitThis).getConstructorFieldInit() =
+      cfi and
     f.(FieldContent).getField() = cfi.getTarget() and
     node1.asExpr() = cfi.getExpr()
   )
@@ -183,7 +184,7 @@ predicate storeStep(Node node1, Content f, PostUpdateNode node2) {
  * Thus, `node1` references an object with a field `f` whose value ends up in
  * `node2`.
  */
-predicate readStep(Node node1, Content f, Node node2) {
+predicate readStep(Node node1, ContentSet f, Node node2) {
   exists(FieldAccess fr |
     node1.asExpr() = fr.getQualifier() and
     fr.getTarget() = f.(FieldContent).getField() and
@@ -195,7 +196,7 @@ predicate readStep(Node node1, Content f, Node node2) {
 /**
  * Holds if values stored inside content `c` are cleared at node `n`.
  */
-predicate clearsContent(Node n, Content c) {
+predicate clearsContent(Node n, ContentSet c) {
   none() // stub implementation
 }
 
@@ -207,6 +208,8 @@ predicate expectsContent(Node n, ContentSet c) { none() }
 
 predicate typeStrongerThan(DataFlowType t1, DataFlowType t2) { none() }
 
+predicate localMustFlowStep(Node node1, Node node2) { none() }
+
 /** Gets the type of `n` used for type pruning. */
 Type getNodeType(Node n) {
   suppressUnusedNode(n) and
@@ -235,12 +238,6 @@ class CastNode extends Node {
   CastNode() { none() } // stub implementation
 }
 
-/**
- * Holds if `n` should never be skipped over in the `PathGraph` and in path
- * explanations.
- */
-predicate neverSkipInPathGraph(Node n) { none() }
-
 class DataFlowCallable = Function;
 
 class DataFlowExpr = Expr;
@@ -265,8 +262,6 @@ class DataFlowCall extends Expr instanceof Call {
 
 predicate isUnreachableInCall(Node n, DataFlowCall call) { none() } // stub implementation
 
-int accessPathLimit() { result = 5 }
-
 /**
  * Holds if access paths with `c` at their head always should be tracked at high
  * precision. This disables adaptive access path precision for such access paths.
@@ -302,28 +297,3 @@ class ContentApprox = Unit;
 /** Gets an approximated value for content `c`. */
 pragma[inline]
 ContentApprox getContentApprox(Content c) { any() }
-
-private class MyConsistencyConfiguration extends Consistency::ConsistencyConfiguration {
-  override predicate argHasPostUpdateExclude(ArgumentNode n) {
-    // Is the null pointer (or something that's not really a pointer)
-    exists(n.asExpr().getValue())
-    or
-    // Isn't a pointer or is a pointer to const
-    forall(DerivedType dt | dt = n.asExpr().getActualType() |
-      dt.getBaseType().isConst()
-      or
-      dt.getBaseType() instanceof RoutineType
-    )
-    // The above list of cases isn't exhaustive, but it narrows down the
-    // consistency alerts enough that most of them are interesting.
-  }
-}
-
-/**
- * Gets an additional term that is added to the `join` and `branch` computations to reflect
- * an additional forward or backwards branching factor that is not taken into account
- * when calculating the (virtual) dispatch cost.
- *
- * Argument `arg` is part of a path from a source to a sink, and `p` is the target parameter.
- */
-int getAdditionalFlowIntoCallNodeTerm(ArgumentNode arg, ParameterNode p) { none() }

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

@@ -0,0 +1,10 @@
+/**
+ * Provides C++-specific definitions for use in the taint tracking library.
+ */
+
+private import codeql.dataflow.TaintTracking
+private import DataFlowImplSpecific
+
+module CppOldTaintTracking implements InputSig<CppOldDataFlow> {
+  import TaintTrackingUtil
+}

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

@@ -39,7 +39,7 @@ predicate defaultAdditionalTaintStep(DataFlow::Node src, DataFlow::Node sink) {
  * of `c` at sinks and inputs to additional taint steps.
  */
 bindingset[node]
-predicate defaultImplicitTaintRead(DataFlow::Node node, DataFlow::Content c) { none() }
+predicate defaultImplicitTaintRead(DataFlow::Node node, DataFlow::ContentSet c) { none() }
 
 /**
  * Holds if `node` should be a sanitizer in all global taint flow configurations

cpp/ql/lib/semmle/code/cpp/dataflow/internal/tainttracking1/TaintTracking.qll

@@ -1,74 +0,0 @@
-/**
- * Provides classes for performing local (intra-procedural) and
- * global (inter-procedural) taint-tracking analyses.
- */
-
-import TaintTrackingParameter::Public
-private import TaintTrackingParameter::Private
-
-private module AddTaintDefaults<DataFlowInternal::FullStateConfigSig Config> implements
-  DataFlowInternal::FullStateConfigSig
-{
-  import Config
-
-  predicate isBarrier(DataFlow::Node node) {
-    Config::isBarrier(node) or defaultTaintSanitizer(node)
-  }
-
-  predicate isAdditionalFlowStep(DataFlow::Node node1, DataFlow::Node node2) {
-    Config::isAdditionalFlowStep(node1, node2) or
-    defaultAdditionalTaintStep(node1, node2)
-  }
-
-  predicate allowImplicitRead(DataFlow::Node node, DataFlow::ContentSet c) {
-    Config::allowImplicitRead(node, c)
-    or
-    (
-      Config::isSink(node, _) or
-      Config::isAdditionalFlowStep(node, _) or
-      Config::isAdditionalFlowStep(node, _, _, _)
-    ) and
-    defaultImplicitTaintRead(node, c)
-  }
-}
-
-/**
- * Constructs a global taint tracking computation.
- */
-module Global<DataFlow::ConfigSig Config> implements DataFlow::GlobalFlowSig {
-  private module Config0 implements DataFlowInternal::FullStateConfigSig {
-    import DataFlowInternal::DefaultState<Config>
-    import Config
-  }
-
-  private module C implements DataFlowInternal::FullStateConfigSig {
-    import AddTaintDefaults<Config0>
-  }
-
-  import DataFlowInternal::Impl<C>
-}
-
-/** DEPRECATED: Use `Global` instead. */
-deprecated module Make<DataFlow::ConfigSig Config> implements DataFlow::GlobalFlowSig {
-  import Global<Config>
-}
-
-/**
- * Constructs a global taint tracking computation using flow state.
- */
-module GlobalWithState<DataFlow::StateConfigSig Config> implements DataFlow::GlobalFlowSig {
-  private module Config0 implements DataFlowInternal::FullStateConfigSig {
-    import Config
-  }
-
-  private module C implements DataFlowInternal::FullStateConfigSig {
-    import AddTaintDefaults<Config0>
-  }
-
-  import DataFlowInternal::Impl<C>
-}
-
-/** DEPRECATED: Use `GlobalWithState` instead. */
-deprecated module MakeWithState<DataFlow::StateConfigSig Config> implements DataFlow::GlobalFlowSig {
-  import GlobalWithState<Config>
-}

cpp/ql/lib/semmle/code/cpp/dataflow/new/DataFlow.qll

@@ -26,6 +26,8 @@ import cpp
  * global (inter-procedural) data flow analyses.
  */
 module DataFlow {
-  import semmle.code.cpp.ir.dataflow.internal.DataFlow
+  private import semmle.code.cpp.ir.dataflow.internal.DataFlowImplSpecific
+  private import codeql.dataflow.DataFlow
+  import DataFlowMake<CppDataFlow>
   import semmle.code.cpp.ir.dataflow.internal.DataFlowImpl1
 }

cpp/ql/lib/semmle/code/cpp/dataflow/new/TaintTracking.qll

@@ -23,6 +23,10 @@ import semmle.code.cpp.dataflow.new.DataFlow2
  * global (inter-procedural) taint-tracking analyses.
  */
 module TaintTracking {
-  import semmle.code.cpp.ir.dataflow.internal.tainttracking1.TaintTracking
+  import semmle.code.cpp.ir.dataflow.internal.tainttracking1.TaintTrackingParameter::Public
+  private import semmle.code.cpp.ir.dataflow.internal.DataFlowImplSpecific
+  private import semmle.code.cpp.ir.dataflow.internal.TaintTrackingImplSpecific
+  private import codeql.dataflow.TaintTracking
+  import TaintFlowMake<CppDataFlow, CppTaintTracking>
   import semmle.code.cpp.ir.dataflow.internal.tainttracking1.TaintTrackingImpl
 }

cpp/ql/lib/semmle/code/cpp/exprs/BuiltInOperations.qll

@@ -1547,3 +1547,21 @@ class BuiltInBitCast extends BuiltInOperation, @builtinbitcast {
 
   override string getAPrimaryQlClass() { result = "BuiltInBitCast" }
 }
+
+/**
+ * A C++ `__is_trivial` built-in operation (used by some implementations of the
+ * `<type_traits>` header).
+ *
+ * Returns `true` if a type is a trivial type.
+ * ```
+ *  template<typename _Tp>
+ *    struct is_trivial
+ *    : public integral_constant<bool, __is_trivial(_Tp)>
+ *    {};
+ * ```
+ */
+class BuiltInIsTrivial extends BuiltInOperation, @istrivialexpr {
+  override string toString() { result = "__is_trivial" }
+
+  override string getAPrimaryQlClass() { result = "BuiltInIsTrivial" }
+}

cpp/ql/lib/semmle/code/cpp/exprs/Expr.qll

@@ -152,7 +152,19 @@ class Expr extends StmtParent, @expr {
     else result = this.getValue()
   }
 
-  /** Holds if this expression has a value that can be determined at compile time. */
+  /**
+   * Holds if this expression has a value that can be determined at compile time.
+   *
+   * An expression has a value that can be determined at compile time when:
+   * - it is a compile-time constant, e.g., a literal value or the result of a constexpr
+   *   compile-time constant;
+   * - it is an address of a (member) function, an address of a constexpr variable
+   *   initialized to a constant address, or an address of an lvalue, or any of the
+   *   previous with a constant value added to or subtracted from the address;
+   * - it is a reference to a (member) function, a reference to a constexpr variable
+   *   initialized to a constant address, or a reference to an lvalue;
+   * - it is a non-template parameter of a uninstantiated template.
+   */
   cached
   predicate isConstant() {
     valuebind(_, underlyingElement(this))
@@ -920,19 +932,91 @@ class NewArrayExpr extends NewOrNewArrayExpr, @new_array_expr {
   Expr getExtent() { result = this.getChild(2) }
 }
 
+private class TDeleteOrDeleteArrayExpr = @delete_expr or @delete_array_expr;
+
+/**
+ * A C++ `delete` or `delete[]` expression.
+ */
+class DeleteOrDeleteArrayExpr extends Expr, TDeleteOrDeleteArrayExpr {
+  override int getPrecedence() { result = 16 }
+
+  /**
+   * Gets the call to a destructor that occurs prior to the object's memory being deallocated, if any.
+   *
+   * In the case of `delete[]` at runtime, the destructor will be called once for each element in the array, but the
+   * destructor call only exists once in the AST.
+   */
+  DestructorCall getDestructorCall() { result = this.getChild(1) }
+
+  /**
+   * Gets the destructor to be called to destroy the object or array, if any.
+   */
+  Destructor getDestructor() { result = this.getDestructorCall().getTarget() }
+
+  /**
+   * Gets the `operator delete` or `operator delete[]` that deallocates storage.
+   * Does not hold if the type being destroyed has a virtual destructor. In that case, the
+   * `operator delete` that will be called is determined at runtime based on the
+   * dynamic type of the object.
+   */
+  Function getDeallocator() {
+    expr_deallocator(underlyingElement(this), unresolveElement(result), _)
+  }
+
+  /**
+   * DEPRECATED: use `getDeallocatorCall` instead.
+   */
+  deprecated FunctionCall getAllocatorCall() { result = this.getChild(0) }
+
+  /**
+   * Gets the call to a non-default `operator delete`/`delete[]` that deallocates storage, if any.
+   *
+   * This will only be present when the type being deleted has a custom `operator delete` and
+   * does not have a virtual destructor.
+   */
+  FunctionCall getDeallocatorCall() { result = this.getChild(0) }
+
+  /**
+   * Holds if the deallocation function expects a size argument.
+   */
+  predicate hasSizedDeallocation() {
+    exists(int form |
+      expr_deallocator(underlyingElement(this), _, form) and
+      form.bitAnd(1) != 0 // Bit zero is the "size" bit
+    )
+  }
+
+  /**
+   * Holds if the deallocation function expects an alignment argument.
+   */
+  predicate hasAlignedDeallocation() {
+    exists(int form |
+      expr_deallocator(underlyingElement(this), _, form) and
+      form.bitAnd(2) != 0 // Bit one is the "alignment" bit
+    )
+  }
+
+  /**
+   * Gets the object or array being deleted.
+   */
+  Expr getExpr() {
+    // If there is a destructor call, the object being deleted is the qualifier
+    // otherwise it is the third child.
+    result = this.getChild(3) or result = this.getDestructorCall().getQualifier()
+  }
+}
+
 /**
  * A C++ `delete` (non-array) expression.
  * ```
  * delete ptr;
  * ```
  */
-class DeleteExpr extends Expr, @delete_expr {
+class DeleteExpr extends DeleteOrDeleteArrayExpr, @delete_expr {
   override string toString() { result = "delete" }
 
   override string getAPrimaryQlClass() { result = "DeleteExpr" }
 
-  override int getPrecedence() { result = 16 }
-
   /**
    * Gets the compile-time type of the object being deleted.
    */
@@ -945,58 +1029,6 @@ class DeleteExpr extends Expr, @delete_expr {
           .(PointerType)
           .getBaseType()
   }
-
-  /**
-   * Gets the call to a destructor that occurs prior to the object's memory being deallocated, if any.
-   */
-  DestructorCall getDestructorCall() { result = this.getChild(1) }
-
-  /**
-   * Gets the destructor to be called to destroy the object, if any.
-   */
-  Destructor getDestructor() { result = this.getDestructorCall().getTarget() }
-
-  /**
-   * Gets the `operator delete` that deallocates storage. Does not hold
-   * if the type being destroyed has a virtual destructor. In that case, the
-   * `operator delete` that will be called is determined at runtime based on the
-   * dynamic type of the object.
-   */
-  Function getDeallocator() {
-    expr_deallocator(underlyingElement(this), unresolveElement(result), _)
-  }
-
-  /**
-   * Holds if the deallocation function expects a size argument.
-   */
-  predicate hasSizedDeallocation() {
-    exists(int form |
-      expr_deallocator(underlyingElement(this), _, form) and
-      form.bitAnd(1) != 0 // Bit zero is the "size" bit
-    )
-  }
-
-  /**
-   * Holds if the deallocation function expects an alignment argument.
-   */
-  predicate hasAlignedDeallocation() {
-    exists(int form |
-      expr_deallocator(underlyingElement(this), _, form) and
-      form.bitAnd(2) != 0 // Bit one is the "alignment" bit
-    )
-  }
-
-  /**
-   * Gets the call to a non-default `operator delete` that deallocates storage, if any.
-   *
-   * This will only be present when the type being deleted has a custom `operator delete`.
-   */
-  FunctionCall getAllocatorCall() { result = this.getChild(0) }
-
-  /**
-   * Gets the object being deleted.
-   */
-  Expr getExpr() { result = this.getChild(3) or result = this.getChild(1).getChild(-1) }
 }
 
 /**
@@ -1005,13 +1037,11 @@ class DeleteExpr extends Expr, @delete_expr {
  * delete[] arr;
  * ```
  */
-class DeleteArrayExpr extends Expr, @delete_array_expr {
+class DeleteArrayExpr extends DeleteOrDeleteArrayExpr, @delete_array_expr {
   override string toString() { result = "delete[]" }
 
   override string getAPrimaryQlClass() { result = "DeleteArrayExpr" }
 
-  override int getPrecedence() { result = 16 }
-
   /**
    * Gets the element type of the array being deleted.
    */
@@ -1024,58 +1054,6 @@ class DeleteArrayExpr extends Expr, @delete_array_expr {
           .(PointerType)
           .getBaseType()
   }
-
-  /**
-   * Gets the call to a destructor that occurs prior to the array's memory being deallocated, if any.
-   *
-   * At runtime, the destructor will be called once for each element in the array, but the
-   * destructor call only exists once in the AST.
-   */
-  DestructorCall getDestructorCall() { result = this.getChild(1) }
-
-  /**
-   * Gets the destructor to be called to destroy each element in the array, if any.
-   */
-  Destructor getDestructor() { result = this.getDestructorCall().getTarget() }
-
-  /**
-   * Gets the `operator delete[]` that deallocates storage.
-   */
-  Function getDeallocator() {
-    expr_deallocator(underlyingElement(this), unresolveElement(result), _)
-  }
-
-  /**
-   * Holds if the deallocation function expects a size argument.
-   */
-  predicate hasSizedDeallocation() {
-    exists(int form |
-      expr_deallocator(underlyingElement(this), _, form) and
-      form.bitAnd(1) != 0 // Bit zero is the "size" bit
-    )
-  }
-
-  /**
-   * Holds if the deallocation function expects an alignment argument.
-   */
-  predicate hasAlignedDeallocation() {
-    exists(int form |
-      expr_deallocator(underlyingElement(this), _, form) and
-      form.bitAnd(2) != 0 // Bit one is the "alignment" bit
-    )
-  }
-
-  /**
-   * Gets the call to a non-default `operator delete` that deallocates storage, if any.
-   *
-   * This will only be present when the type being deleted has a custom `operator delete`.
-   */
-  FunctionCall getAllocatorCall() { result = this.getChild(0) }
-
-  /**
-   * Gets the array being deleted.
-   */
-  Expr getExpr() { result = this.getChild(3) or result = this.getChild(1).getChild(-1) }
 }
 
 /**

cpp/ql/lib/semmle/code/cpp/ir/PrintIR.qll

@@ -4,8 +4,8 @@
  * This file contains the actual implementation of `PrintIR.ql`. For test cases and very small
  * databases, `PrintIR.ql` can be run directly to dump the IR for the entire database. For most
  * uses, however, it is better to write a query that imports `PrintIR.qll`, extends
- * `PrintIRConfiguration`, and overrides `shouldPrintFunction()` to select a subset of functions to
- * dump.
+ * `PrintIRConfiguration`, and overrides `shouldPrintDeclaration()` to select a subset of declarations
+ * to dump.
  */
 
 import implementation.aliased_ssa.PrintIR

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

@@ -22,6 +22,8 @@
 import cpp
 
 module DataFlow {
-  import semmle.code.cpp.ir.dataflow.internal.DataFlow
+  private import semmle.code.cpp.ir.dataflow.internal.DataFlowImplSpecific
+  private import codeql.dataflow.DataFlow
+  import DataFlowMake<CppDataFlow>
   import semmle.code.cpp.ir.dataflow.internal.DataFlowImpl1
 }

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

@@ -19,6 +19,10 @@ import semmle.code.cpp.ir.dataflow.DataFlow
 import semmle.code.cpp.ir.dataflow.DataFlow2
 
 module TaintTracking {
-  import semmle.code.cpp.ir.dataflow.internal.tainttracking1.TaintTracking
+  import semmle.code.cpp.ir.dataflow.internal.tainttracking1.TaintTrackingParameter::Public
+  private import semmle.code.cpp.ir.dataflow.internal.DataFlowImplSpecific
+  private import semmle.code.cpp.ir.dataflow.internal.TaintTrackingImplSpecific
+  private import codeql.dataflow.TaintTracking
+  import TaintFlowMake<CppDataFlow, CppTaintTracking>
   import semmle.code.cpp.ir.dataflow.internal.tainttracking1.TaintTrackingImpl
 }

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

@@ -1,433 +0,0 @@
-/**
- * Provides an implementation of global (interprocedural) data flow. This file
- * re-exports the local (intraprocedural) data flow analysis from
- * `DataFlowImplSpecific::Public` and adds a global analysis, mainly exposed
- * through the `Global` and `GlobalWithState` modules.
- */
-
-private import DataFlowImplCommon
-private import DataFlowImplSpecific::Private
-import DataFlowImplSpecific::Public
-import DataFlowImplCommonPublic
-private import DataFlowImpl
-
-/** An input configuration for data flow. */
-signature module ConfigSig {
-  /**
-   * Holds if `source` is a relevant data flow source.
-   */
-  predicate isSource(Node source);
-
-  /**
-   * Holds if `sink` is a relevant data flow sink.
-   */
-  predicate isSink(Node sink);
-
-  /**
-   * Holds if data flow through `node` is prohibited. This completely removes
-   * `node` from the data flow graph.
-   */
-  default predicate isBarrier(Node node) { none() }
-
-  /** Holds if data flow into `node` is prohibited. */
-  default predicate isBarrierIn(Node node) { none() }
-
-  /** Holds if data flow out of `node` is prohibited. */
-  default predicate isBarrierOut(Node node) { none() }
-
-  /**
-   * Holds if data may flow from `node1` to `node2` in addition to the normal data-flow steps.
-   */
-  default predicate isAdditionalFlowStep(Node node1, Node node2) { none() }
-
-  /**
-   * Holds if an arbitrary number of implicit read steps of content `c` may be
-   * taken at `node`.
-   */
-  default predicate allowImplicitRead(Node node, ContentSet c) { none() }
-
-  /**
-   * Holds if `node` should never be skipped over in the `PathGraph` and in path
-   * explanations.
-   */
-  default predicate neverSkip(Node node) {
-    isAdditionalFlowStep(node, _) or isAdditionalFlowStep(_, node)
-  }
-
-  /**
-   * Gets the virtual dispatch branching limit when calculating field flow.
-   * This can be overridden to a smaller value to improve performance (a
-   * value of 0 disables field flow), or a larger value to get more results.
-   */
-  default int fieldFlowBranchLimit() { result = 2 }
-
-  /**
-   * Gets a data flow configuration feature to add restrictions to the set of
-   * valid flow paths.
-   *
-   * - `FeatureHasSourceCallContext`:
-   *    Assume that sources have some existing call context to disallow
-   *    conflicting return-flow directly following the source.
-   * - `FeatureHasSinkCallContext`:
-   *    Assume that sinks have some existing call context to disallow
-   *    conflicting argument-to-parameter flow directly preceding the sink.
-   * - `FeatureEqualSourceSinkCallContext`:
-   *    Implies both of the above and additionally ensures that the entire flow
-   *    path preserves the call context.
-   *
-   * These features are generally not relevant for typical end-to-end data flow
-   * queries, but should only be used for constructing paths that need to
-   * somehow be pluggable in another path context.
-   */
-  default FlowFeature getAFeature() { none() }
-
-  /** Holds if sources should be grouped in the result of `flowPath`. */
-  default predicate sourceGrouping(Node source, string sourceGroup) { none() }
-
-  /** Holds if sinks should be grouped in the result of `flowPath`. */
-  default predicate sinkGrouping(Node sink, string sinkGroup) { none() }
-
-  /**
-   * Holds if hidden nodes should be included in the data flow graph.
-   *
-   * This feature should only be used for debugging or when the data flow graph
-   * is not visualized (as it is in a `path-problem` query).
-   */
-  default predicate includeHiddenNodes() { none() }
-}
-
-/** An input configuration for data flow using flow state. */
-signature module StateConfigSig {
-  bindingset[this]
-  class FlowState;
-
-  /**
-   * Holds if `source` is a relevant data flow source with the given initial
-   * `state`.
-   */
-  predicate isSource(Node source, FlowState state);
-
-  /**
-   * Holds if `sink` is a relevant data flow sink accepting `state`.
-   */
-  predicate isSink(Node sink, FlowState state);
-
-  /**
-   * Holds if data flow through `node` is prohibited. This completely removes
-   * `node` from the data flow graph.
-   */
-  default predicate isBarrier(Node node) { none() }
-
-  /**
-   * Holds if data flow through `node` is prohibited when the flow state is
-   * `state`.
-   */
-  default predicate isBarrier(Node node, FlowState state) { none() }
-
-  /** Holds if data flow into `node` is prohibited. */
-  default predicate isBarrierIn(Node node) { none() }
-
-  /** Holds if data flow out of `node` is prohibited. */
-  default predicate isBarrierOut(Node node) { none() }
-
-  /**
-   * Holds if data may flow from `node1` to `node2` in addition to the normal data-flow steps.
-   */
-  default predicate isAdditionalFlowStep(Node node1, Node node2) { none() }
-
-  /**
-   * Holds if data may flow from `node1` to `node2` in addition to the normal data-flow steps.
-   * This step is only applicable in `state1` and updates the flow state to `state2`.
-   */
-  default predicate isAdditionalFlowStep(Node node1, FlowState state1, Node node2, FlowState state2) {
-    none()
-  }
-
-  /**
-   * Holds if an arbitrary number of implicit read steps of content `c` may be
-   * taken at `node`.
-   */
-  default predicate allowImplicitRead(Node node, ContentSet c) { none() }
-
-  /**
-   * Holds if `node` should never be skipped over in the `PathGraph` and in path
-   * explanations.
-   */
-  default predicate neverSkip(Node node) {
-    isAdditionalFlowStep(node, _) or
-    isAdditionalFlowStep(_, node) or
-    isAdditionalFlowStep(node, _, _, _) or
-    isAdditionalFlowStep(_, _, node, _)
-  }
-
-  /**
-   * Gets the virtual dispatch branching limit when calculating field flow.
-   * This can be overridden to a smaller value to improve performance (a
-   * value of 0 disables field flow), or a larger value to get more results.
-   */
-  default int fieldFlowBranchLimit() { result = 2 }
-
-  /**
-   * Gets a data flow configuration feature to add restrictions to the set of
-   * valid flow paths.
-   *
-   * - `FeatureHasSourceCallContext`:
-   *    Assume that sources have some existing call context to disallow
-   *    conflicting return-flow directly following the source.
-   * - `FeatureHasSinkCallContext`:
-   *    Assume that sinks have some existing call context to disallow
-   *    conflicting argument-to-parameter flow directly preceding the sink.
-   * - `FeatureEqualSourceSinkCallContext`:
-   *    Implies both of the above and additionally ensures that the entire flow
-   *    path preserves the call context.
-   *
-   * These features are generally not relevant for typical end-to-end data flow
-   * queries, but should only be used for constructing paths that need to
-   * somehow be pluggable in another path context.
-   */
-  default FlowFeature getAFeature() { none() }
-
-  /** Holds if sources should be grouped in the result of `flowPath`. */
-  default predicate sourceGrouping(Node source, string sourceGroup) { none() }
-
-  /** Holds if sinks should be grouped in the result of `flowPath`. */
-  default predicate sinkGrouping(Node sink, string sinkGroup) { none() }
-
-  /**
-   * Holds if hidden nodes should be included in the data flow graph.
-   *
-   * This feature should only be used for debugging or when the data flow graph
-   * is not visualized (as it is in a `path-problem` query).
-   */
-  default predicate includeHiddenNodes() { none() }
-}
-
-/**
- * Gets the exploration limit for `partialFlow` and `partialFlowRev`
- * measured in approximate number of interprocedural steps.
- */
-signature int explorationLimitSig();
-
-/**
- * The output of a global data flow computation.
- */
-signature module GlobalFlowSig {
-  /**
-   * A `Node` augmented with a call context (except for sinks) and an access path.
-   * Only those `PathNode`s that are reachable from a source, and which can reach a sink, are generated.
-   */
-  class PathNode;
-
-  /**
-   * Holds if data can flow from `source` to `sink`.
-   *
-   * The corresponding paths are generated from the end-points and the graph
-   * included in the module `PathGraph`.
-   */
-  predicate flowPath(PathNode source, PathNode sink);
-
-  /**
-   * Holds if data can flow from `source` to `sink`.
-   */
-  predicate flow(Node source, Node sink);
-
-  /**
-   * Holds if data can flow from some source to `sink`.
-   */
-  predicate flowTo(Node sink);
-
-  /**
-   * Holds if data can flow from some source to `sink`.
-   */
-  predicate flowToExpr(DataFlowExpr sink);
-}
-
-/**
- * Constructs a global data flow computation.
- */
-module Global<ConfigSig Config> implements GlobalFlowSig {
-  private module C implements FullStateConfigSig {
-    import DefaultState<Config>
-    import Config
-  }
-
-  import Impl<C>
-}
-
-/** DEPRECATED: Use `Global` instead. */
-deprecated module Make<ConfigSig Config> implements GlobalFlowSig {
-  import Global<Config>
-}
-
-/**
- * Constructs a global data flow computation using flow state.
- */
-module GlobalWithState<StateConfigSig Config> implements GlobalFlowSig {
-  private module C implements FullStateConfigSig {
-    import Config
-  }
-
-  import Impl<C>
-}
-
-/** DEPRECATED: Use `GlobalWithState` instead. */
-deprecated module MakeWithState<StateConfigSig Config> implements GlobalFlowSig {
-  import GlobalWithState<Config>
-}
-
-signature class PathNodeSig {
-  /** Gets a textual representation of this element. */
-  string toString();
-
-  /**
-   * Holds if this element is at the specified location.
-   * The location spans column `startcolumn` of line `startline` to
-   * column `endcolumn` of line `endline` in file `filepath`.
-   * For more information, see
-   * [Locations](https://codeql.github.com/docs/writing-codeql-queries/providing-locations-in-codeql-queries/).
-   */
-  predicate hasLocationInfo(
-    string filepath, int startline, int startcolumn, int endline, int endcolumn
-  );
-
-  /** Gets the underlying `Node`. */
-  Node getNode();
-}
-
-signature module PathGraphSig<PathNodeSig PathNode> {
-  /** Holds if `(a,b)` is an edge in the graph of data flow path explanations. */
-  predicate edges(PathNode a, PathNode b);
-
-  /** Holds if `n` is a node in the graph of data flow path explanations. */
-  predicate nodes(PathNode n, string key, string val);
-
-  /**
-   * Holds if `(arg, par, ret, out)` forms a subpath-tuple, that is, flow through
-   * a subpath between `par` and `ret` with the connecting edges `arg -> par` and
-   * `ret -> out` is summarized as the edge `arg -> out`.
-   */
-  predicate subpaths(PathNode arg, PathNode par, PathNode ret, PathNode out);
-}
-
-/**
- * Constructs a `PathGraph` from two `PathGraph`s by disjoint union.
- */
-module MergePathGraph<
-  PathNodeSig PathNode1, PathNodeSig PathNode2, PathGraphSig<PathNode1> Graph1,
-  PathGraphSig<PathNode2> Graph2>
-{
-  private newtype TPathNode =
-    TPathNode1(PathNode1 p) or
-    TPathNode2(PathNode2 p)
-
-  /** A node in a graph of path explanations that is formed by disjoint union of the two given graphs. */
-  class PathNode extends TPathNode {
-    /** Gets this as a projection on the first given `PathGraph`. */
-    PathNode1 asPathNode1() { this = TPathNode1(result) }
-
-    /** Gets this as a projection on the second given `PathGraph`. */
-    PathNode2 asPathNode2() { this = TPathNode2(result) }
-
-    /** Gets a textual representation of this element. */
-    string toString() {
-      result = this.asPathNode1().toString() or
-      result = this.asPathNode2().toString()
-    }
-
-    /**
-     * Holds if this element is at the specified location.
-     * The location spans column `startcolumn` of line `startline` to
-     * column `endcolumn` of line `endline` in file `filepath`.
-     * For more information, see
-     * [Locations](https://codeql.github.com/docs/writing-codeql-queries/providing-locations-in-codeql-queries/).
-     */
-    predicate hasLocationInfo(
-      string filepath, int startline, int startcolumn, int endline, int endcolumn
-    ) {
-      this.asPathNode1().hasLocationInfo(filepath, startline, startcolumn, endline, endcolumn) or
-      this.asPathNode2().hasLocationInfo(filepath, startline, startcolumn, endline, endcolumn)
-    }
-
-    /** Gets the underlying `Node`. */
-    Node getNode() {
-      result = this.asPathNode1().getNode() or
-      result = this.asPathNode2().getNode()
-    }
-  }
-
-  /**
-   * Provides the query predicates needed to include a graph in a path-problem query.
-   */
-  module PathGraph implements PathGraphSig<PathNode> {
-    /** Holds if `(a,b)` is an edge in the graph of data flow path explanations. */
-    query predicate edges(PathNode a, PathNode b) {
-      Graph1::edges(a.asPathNode1(), b.asPathNode1()) or
-      Graph2::edges(a.asPathNode2(), b.asPathNode2())
-    }
-
-    /** Holds if `n` is a node in the graph of data flow path explanations. */
-    query predicate nodes(PathNode n, string key, string val) {
-      Graph1::nodes(n.asPathNode1(), key, val) or
-      Graph2::nodes(n.asPathNode2(), key, val)
-    }
-
-    /**
-     * Holds if `(arg, par, ret, out)` forms a subpath-tuple, that is, flow through
-     * a subpath between `par` and `ret` with the connecting edges `arg -> par` and
-     * `ret -> out` is summarized as the edge `arg -> out`.
-     */
-    query predicate subpaths(PathNode arg, PathNode par, PathNode ret, PathNode out) {
-      Graph1::subpaths(arg.asPathNode1(), par.asPathNode1(), ret.asPathNode1(), out.asPathNode1()) or
-      Graph2::subpaths(arg.asPathNode2(), par.asPathNode2(), ret.asPathNode2(), out.asPathNode2())
-    }
-  }
-}
-
-/**
- * Constructs a `PathGraph` from three `PathGraph`s by disjoint union.
- */
-module MergePathGraph3<
-  PathNodeSig PathNode1, PathNodeSig PathNode2, PathNodeSig PathNode3,
-  PathGraphSig<PathNode1> Graph1, PathGraphSig<PathNode2> Graph2, PathGraphSig<PathNode3> Graph3>
-{
-  private module MergedInner = MergePathGraph<PathNode1, PathNode2, Graph1, Graph2>;
-
-  private module Merged =
-    MergePathGraph<MergedInner::PathNode, PathNode3, MergedInner::PathGraph, Graph3>;
-
-  /** A node in a graph of path explanations that is formed by disjoint union of the three given graphs. */
-  class PathNode instanceof Merged::PathNode {
-    /** Gets this as a projection on the first given `PathGraph`. */
-    PathNode1 asPathNode1() { result = super.asPathNode1().asPathNode1() }
-
-    /** Gets this as a projection on the second given `PathGraph`. */
-    PathNode2 asPathNode2() { result = super.asPathNode1().asPathNode2() }
-
-    /** Gets this as a projection on the third given `PathGraph`. */
-    PathNode3 asPathNode3() { result = super.asPathNode2() }
-
-    /** Gets a textual representation of this element. */
-    string toString() { result = super.toString() }
-
-    /**
-     * Holds if this element is at the specified location.
-     * The location spans column `startcolumn` of line `startline` to
-     * column `endcolumn` of line `endline` in file `filepath`.
-     * For more information, see
-     * [Locations](https://codeql.github.com/docs/writing-codeql-queries/providing-locations-in-codeql-queries/).
-     */
-    predicate hasLocationInfo(
-      string filepath, int startline, int startcolumn, int endline, int endcolumn
-    ) {
-      super.hasLocationInfo(filepath, startline, startcolumn, endline, endcolumn)
-    }
-
-    /** Gets the underlying `Node`. */
-    Node getNode() { result = super.getNode() }
-  }
-
-  /**
-   * Provides the query predicates needed to include a graph in a path-problem query.
-   */
-  module PathGraph = Merged::PathGraph;
-}

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

@@ -9,7 +9,7 @@ private import DataFlowImplCommon as DataFlowImplCommon
  * Gets a function that might be called by `call`.
  */
 cached
-Function viableCallable(CallInstruction call) {
+DataFlowCallable viableCallable(DataFlowCall call) {
   DataFlowImplCommon::forceCachingInSameStage() and
   result = call.getStaticCallTarget()
   or
@@ -235,7 +235,7 @@ private predicate functionSignature(Function f, string qualifiedName, int nparam
  * Holds if the set of viable implementations that can be called by `call`
  * might be improved by knowing the call context.
  */
-predicate mayBenefitFromCallContext(CallInstruction call, Function f) {
+predicate mayBenefitFromCallContext(DataFlowCall call, DataFlowCallable f) {
   mayBenefitFromCallContext(call, f, _)
 }
 
@@ -259,7 +259,7 @@ private predicate mayBenefitFromCallContext(
  * Gets a viable dispatch target of `call` in the context `ctx`. This is
  * restricted to those `call`s for which a context might make a difference.
  */
-Function viableImplInCallContext(CallInstruction call, CallInstruction ctx) {
+DataFlowCallable viableImplInCallContext(DataFlowCall call, DataFlowCall ctx) {
   result = viableCallable(call) and
   exists(int i, Function f |
     mayBenefitFromCallContext(pragma[only_bind_into](call), f, i) and
@@ -271,13 +271,3 @@ Function viableImplInCallContext(CallInstruction call, CallInstruction ctx) {
 /** Holds if arguments at position `apos` match parameters at position `ppos`. */
 pragma[inline]
 predicate parameterMatch(ParameterPosition ppos, ArgumentPosition apos) { ppos = apos }
-
-/**
- * Holds if flow from `call`'s argument `arg` to parameter `p` is permissible.
- *
- * This is a temporary hook to support technical debt in the Go language; do not use.
- */
-pragma[inline]
-predicate golangSpecificParamArgFilter(DataFlowCall call, ParameterNode p, ArgumentNode arg) {
-  any()
-}

cpp/ql/lib/semmle/code/cpp/ir/dataflow/internal/DataFlowImpl1.qll

@@ -276,6 +276,8 @@ private module Config implements FullStateConfigSig {
     getConfig(state).isSource(source) and getState(state) instanceof FlowStateEmpty
   }
 
+  predicate isSink(Node sink) { none() }
+
   predicate isSink(Node sink, FlowState state) {
     getConfig(state).isSink(sink, getState(state))
     or
@@ -295,6 +297,10 @@ private module Config implements FullStateConfigSig {
 
   predicate isBarrierOut(Node node) { any(Configuration config).isBarrierOut(node) }
 
+  predicate isBarrierIn(Node node, FlowState state) { none() }
+
+  predicate isBarrierOut(Node node, FlowState state) { none() }
+
   predicate isAdditionalFlowStep(Node node1, Node node2) {
     singleConfiguration() and
     any(Configuration config).isAdditionalFlowStep(node1, node2)

cpp/ql/lib/semmle/code/cpp/ir/dataflow/internal/DataFlowImpl2.qll

@@ -276,6 +276,8 @@ private module Config implements FullStateConfigSig {
     getConfig(state).isSource(source) and getState(state) instanceof FlowStateEmpty
   }
 
+  predicate isSink(Node sink) { none() }
+
   predicate isSink(Node sink, FlowState state) {
     getConfig(state).isSink(sink, getState(state))
     or
@@ -295,6 +297,10 @@ private module Config implements FullStateConfigSig {
 
   predicate isBarrierOut(Node node) { any(Configuration config).isBarrierOut(node) }
 
+  predicate isBarrierIn(Node node, FlowState state) { none() }
+
+  predicate isBarrierOut(Node node, FlowState state) { none() }
+
   predicate isAdditionalFlowStep(Node node1, Node node2) {
     singleConfiguration() and
     any(Configuration config).isAdditionalFlowStep(node1, node2)

cpp/ql/lib/semmle/code/cpp/ir/dataflow/internal/DataFlowImpl3.qll

@@ -276,6 +276,8 @@ private module Config implements FullStateConfigSig {
     getConfig(state).isSource(source) and getState(state) instanceof FlowStateEmpty
   }
 
+  predicate isSink(Node sink) { none() }
+
   predicate isSink(Node sink, FlowState state) {
     getConfig(state).isSink(sink, getState(state))
     or
@@ -295,6 +297,10 @@ private module Config implements FullStateConfigSig {
 
   predicate isBarrierOut(Node node) { any(Configuration config).isBarrierOut(node) }
 
+  predicate isBarrierIn(Node node, FlowState state) { none() }
+
+  predicate isBarrierOut(Node node, FlowState state) { none() }
+
   predicate isAdditionalFlowStep(Node node1, Node node2) {
     singleConfiguration() and
     any(Configuration config).isAdditionalFlowStep(node1, node2)

cpp/ql/lib/semmle/code/cpp/ir/dataflow/internal/DataFlowImpl4.qll

@@ -276,6 +276,8 @@ private module Config implements FullStateConfigSig {
     getConfig(state).isSource(source) and getState(state) instanceof FlowStateEmpty
   }
 
+  predicate isSink(Node sink) { none() }
+
   predicate isSink(Node sink, FlowState state) {
     getConfig(state).isSink(sink, getState(state))
     or
@@ -295,6 +297,10 @@ private module Config implements FullStateConfigSig {
 
   predicate isBarrierOut(Node node) { any(Configuration config).isBarrierOut(node) }
 
+  predicate isBarrierIn(Node node, FlowState state) { none() }
+
+  predicate isBarrierOut(Node node, FlowState state) { none() }
+
   predicate isAdditionalFlowStep(Node node1, Node node2) {
     singleConfiguration() and
     any(Configuration config).isAdditionalFlowStep(node1, node2)

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

@@ -3,297 +3,17 @@
  * data-flow classes and predicates.
  */
 
-private import DataFlowImplSpecific::Private
-private import DataFlowImplSpecific::Public
-private import tainttracking1.TaintTrackingParameter::Private
-private import tainttracking1.TaintTrackingParameter::Public
+private import cpp
+private import DataFlowImplSpecific
+private import TaintTrackingImplSpecific
+private import codeql.dataflow.internal.DataFlowImplConsistency
 
-module Consistency {
-  private newtype TConsistencyConfiguration = MkConsistencyConfiguration()
-
-  /** A class for configuring the consistency queries. */
-  class ConsistencyConfiguration extends TConsistencyConfiguration {
-    string toString() { none() }
-
-    /** Holds if `n` should be excluded from the consistency test `uniqueEnclosingCallable`. */
-    predicate uniqueEnclosingCallableExclude(Node n) { none() }
-
-    /** Holds if `call` should be excluded from the consistency test `uniqueCallEnclosingCallable`. */
-    predicate uniqueCallEnclosingCallableExclude(DataFlowCall call) { none() }
-
-    /** Holds if `n` should be excluded from the consistency test `uniqueNodeLocation`. */
-    predicate uniqueNodeLocationExclude(Node n) { none() }
-
-    /** Holds if `n` should be excluded from the consistency test `missingLocation`. */
-    predicate missingLocationExclude(Node n) { none() }
-
-    /** Holds if `n` should be excluded from the consistency test `postWithInFlow`. */
-    predicate postWithInFlowExclude(Node n) { none() }
-
-    /** Holds if `n` should be excluded from the consistency test `argHasPostUpdate`. */
-    predicate argHasPostUpdateExclude(ArgumentNode n) { none() }
-
-    /** Holds if `n` should be excluded from the consistency test `reverseRead`. */
-    predicate reverseReadExclude(Node n) { none() }
-
-    /** Holds if `n` should be excluded from the consistency test `postHasUniquePre`. */
-    predicate postHasUniquePreExclude(PostUpdateNode n) { none() }
-
-    /** Holds if `n` should be excluded from the consistency test `uniquePostUpdate`. */
-    predicate uniquePostUpdateExclude(Node n) { none() }
-
-    /** Holds if `(call, ctx)` should be excluded from the consistency test `viableImplInCallContextTooLargeExclude`. */
-    predicate viableImplInCallContextTooLargeExclude(
-      DataFlowCall call, DataFlowCall ctx, DataFlowCallable callable
-    ) {
-      none()
-    }
-
-    /** Holds if `(c, pos, p)` should be excluded from the consistency test `uniqueParameterNodeAtPosition`. */
-    predicate uniqueParameterNodeAtPositionExclude(DataFlowCallable c, ParameterPosition pos, Node p) {
-      none()
-    }
-
-    /** Holds if `(c, pos, p)` should be excluded from the consistency test `uniqueParameterNodePosition`. */
-    predicate uniqueParameterNodePositionExclude(DataFlowCallable c, ParameterPosition pos, Node p) {
-      none()
-    }
-
-    /** Holds if `n` should be excluded from the consistency test `identityLocalStep`. */
-    predicate identityLocalStepExclude(Node n) { none() }
-  }
-
-  private class RelevantNode extends Node {
-    RelevantNode() {
-      this instanceof ArgumentNode or
-      this instanceof ParameterNode or
-      this instanceof ReturnNode or
-      this = getAnOutNode(_, _) or
-      simpleLocalFlowStep(this, _) or
-      simpleLocalFlowStep(_, this) or
-      jumpStep(this, _) or
-      jumpStep(_, this) or
-      storeStep(this, _, _) or
-      storeStep(_, _, this) or
-      readStep(this, _, _) or
-      readStep(_, _, this) or
-      defaultAdditionalTaintStep(this, _) or
-      defaultAdditionalTaintStep(_, this)
-    }
-  }
-
-  query predicate uniqueEnclosingCallable(Node n, string msg) {
-    exists(int c |
-      n instanceof RelevantNode and
-      c = count(nodeGetEnclosingCallable(n)) and
-      c != 1 and
-      not any(ConsistencyConfiguration conf).uniqueEnclosingCallableExclude(n) and
-      msg = "Node should have one enclosing callable but has " + c + "."
-    )
-  }
-
-  query predicate uniqueCallEnclosingCallable(DataFlowCall call, string msg) {
-    exists(int c |
-      c = count(call.getEnclosingCallable()) and
-      c != 1 and
-      not any(ConsistencyConfiguration conf).uniqueCallEnclosingCallableExclude(call) and
-      msg = "Call should have one enclosing callable but has " + c + "."
-    )
-  }
-
-  query predicate uniqueType(Node n, string msg) {
-    exists(int c |
-      n instanceof RelevantNode and
-      c = count(getNodeType(n)) and
-      c != 1 and
-      msg = "Node should have one type but has " + c + "."
-    )
-  }
-
-  query predicate uniqueNodeLocation(Node n, string msg) {
-    exists(int c |
-      c =
-        count(string filepath, int startline, int startcolumn, int endline, int endcolumn |
-          n.hasLocationInfo(filepath, startline, startcolumn, endline, endcolumn)
-        ) and
-      c != 1 and
-      not any(ConsistencyConfiguration conf).uniqueNodeLocationExclude(n) and
-      msg = "Node should have one location but has " + c + "."
-    )
-  }
-
-  query predicate missingLocation(string msg) {
-    exists(int c |
-      c =
-        strictcount(Node n |
-          not n.hasLocationInfo(_, _, _, _, _) and
-          not any(ConsistencyConfiguration conf).missingLocationExclude(n)
-        ) and
-      msg = "Nodes without location: " + c
-    )
-  }
-
-  query predicate uniqueNodeToString(Node n, string msg) {
-    exists(int c |
-      c = count(n.toString()) and
-      c != 1 and
-      msg = "Node should have one toString but has " + c + "."
-    )
-  }
-
-  query predicate missingToString(string msg) {
-    exists(int c |
-      c = strictcount(Node n | not exists(n.toString())) and
-      msg = "Nodes without toString: " + c
-    )
-  }
-
-  query predicate parameterCallable(ParameterNode p, string msg) {
-    exists(DataFlowCallable c | isParameterNode(p, c, _) and c != nodeGetEnclosingCallable(p)) and
-    msg = "Callable mismatch for parameter."
-  }
-
-  query predicate localFlowIsLocal(Node n1, Node n2, string msg) {
-    simpleLocalFlowStep(n1, n2) and
-    nodeGetEnclosingCallable(n1) != nodeGetEnclosingCallable(n2) and
-    msg = "Local flow step does not preserve enclosing callable."
-  }
-
-  query predicate readStepIsLocal(Node n1, Node n2, string msg) {
-    readStep(n1, _, n2) and
-    nodeGetEnclosingCallable(n1) != nodeGetEnclosingCallable(n2) and
-    msg = "Read step does not preserve enclosing callable."
-  }
-
-  query predicate storeStepIsLocal(Node n1, Node n2, string msg) {
-    storeStep(n1, _, n2) and
-    nodeGetEnclosingCallable(n1) != nodeGetEnclosingCallable(n2) and
-    msg = "Store step does not preserve enclosing callable."
-  }
-
-  private DataFlowType typeRepr() { result = getNodeType(_) }
-
-  query predicate compatibleTypesReflexive(DataFlowType t, string msg) {
-    t = typeRepr() and
-    not compatibleTypes(t, t) and
-    msg = "Type compatibility predicate is not reflexive."
-  }
-
-  query predicate unreachableNodeCCtx(Node n, DataFlowCall call, string msg) {
-    isUnreachableInCall(n, call) and
-    exists(DataFlowCallable c |
-      c = nodeGetEnclosingCallable(n) and
-      not viableCallable(call) = c
-    ) and
-    msg = "Call context for isUnreachableInCall is inconsistent with call graph."
-  }
-
-  query predicate localCallNodes(DataFlowCall call, Node n, string msg) {
-    (
-      n = getAnOutNode(call, _) and
-      msg = "OutNode and call does not share enclosing callable."
-      or
-      n.(ArgumentNode).argumentOf(call, _) and
-      msg = "ArgumentNode and call does not share enclosing callable."
-    ) and
-    nodeGetEnclosingCallable(n) != call.getEnclosingCallable()
-  }
-
-  // This predicate helps the compiler forget that in some languages
-  // it is impossible for a result of `getPreUpdateNode` to be an
-  // instance of `PostUpdateNode`.
-  private Node getPre(PostUpdateNode n) {
-    result = n.getPreUpdateNode()
-    or
-    none()
-  }
-
-  query predicate postIsNotPre(PostUpdateNode n, string msg) {
-    getPre(n) = n and
-    msg = "PostUpdateNode should not equal its pre-update node."
-  }
-
-  query predicate postHasUniquePre(PostUpdateNode n, string msg) {
-    not any(ConsistencyConfiguration conf).postHasUniquePreExclude(n) and
-    exists(int c |
-      c = count(n.getPreUpdateNode()) and
-      c != 1 and
-      msg = "PostUpdateNode should have one pre-update node but has " + c + "."
-    )
-  }
-
-  query predicate uniquePostUpdate(Node n, string msg) {
-    not any(ConsistencyConfiguration conf).uniquePostUpdateExclude(n) and
-    1 < strictcount(PostUpdateNode post | post.getPreUpdateNode() = n) and
-    msg = "Node has multiple PostUpdateNodes."
-  }
-
-  query predicate postIsInSameCallable(PostUpdateNode n, string msg) {
-    nodeGetEnclosingCallable(n) != nodeGetEnclosingCallable(n.getPreUpdateNode()) and
-    msg = "PostUpdateNode does not share callable with its pre-update node."
-  }
-
-  private predicate hasPost(Node n) { exists(PostUpdateNode post | post.getPreUpdateNode() = n) }
-
-  query predicate reverseRead(Node n, string msg) {
-    exists(Node n2 | readStep(n, _, n2) and hasPost(n2) and not hasPost(n)) and
-    not any(ConsistencyConfiguration conf).reverseReadExclude(n) and
-    msg = "Origin of readStep is missing a PostUpdateNode."
-  }
-
-  query predicate argHasPostUpdate(ArgumentNode n, string msg) {
-    not hasPost(n) and
-    not any(ConsistencyConfiguration c).argHasPostUpdateExclude(n) and
-    msg = "ArgumentNode is missing PostUpdateNode."
-  }
-
-  // This predicate helps the compiler forget that in some languages
-  // it is impossible for a `PostUpdateNode` to be the target of
-  // `simpleLocalFlowStep`.
-  private predicate isPostUpdateNode(Node n) { n instanceof PostUpdateNode or none() }
-
-  query predicate postWithInFlow(Node n, string msg) {
-    isPostUpdateNode(n) and
-    not clearsContent(n, _) and
-    simpleLocalFlowStep(_, n) and
-    not any(ConsistencyConfiguration c).postWithInFlowExclude(n) and
-    msg = "PostUpdateNode should not be the target of local flow."
-  }
-
-  query predicate viableImplInCallContextTooLarge(
-    DataFlowCall call, DataFlowCall ctx, DataFlowCallable callable
-  ) {
-    callable = viableImplInCallContext(call, ctx) and
-    not callable = viableCallable(call) and
-    not any(ConsistencyConfiguration c).viableImplInCallContextTooLargeExclude(call, ctx, callable)
-  }
-
-  query predicate uniqueParameterNodeAtPosition(
-    DataFlowCallable c, ParameterPosition pos, Node p, string msg
-  ) {
-    not any(ConsistencyConfiguration conf).uniqueParameterNodeAtPositionExclude(c, pos, p) and
-    isParameterNode(p, c, pos) and
-    not exists(unique(Node p0 | isParameterNode(p0, c, pos))) and
-    msg = "Parameters with overlapping positions."
-  }
-
-  query predicate uniqueParameterNodePosition(
-    DataFlowCallable c, ParameterPosition pos, Node p, string msg
-  ) {
-    not any(ConsistencyConfiguration conf).uniqueParameterNodePositionExclude(c, pos, p) and
-    isParameterNode(p, c, pos) and
-    not exists(unique(ParameterPosition pos0 | isParameterNode(p, c, pos0))) and
-    msg = "Parameter node with multiple positions."
-  }
-
-  query predicate uniqueContentApprox(Content c, string msg) {
-    not exists(unique(ContentApprox approx | approx = getContentApprox(c))) and
-    msg = "Non-unique content approximation."
-  }
-
-  query predicate identityLocalStep(Node n, string msg) {
-    simpleLocalFlowStep(n, n) and
-    not any(ConsistencyConfiguration c).identityLocalStepExclude(n) and
-    msg = "Node steps to itself"
+private module Input implements InputSig<CppDataFlow> {
+  predicate argHasPostUpdateExclude(Private::ArgumentNode n) {
+    // The rules for whether an IR argument gets a post-update node are too
+    // complex to model here.
+    any()
   }
 }
+
+module Consistency = MakeConsistency<CppDataFlow, CppTaintTracking, Input>;

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

@@ -1,6 +1,9 @@
 /**
  * Provides IR-specific definitions for use in the data flow library.
  */
+
+private import codeql.dataflow.DataFlow
+
 module Private {
   import DataFlowPrivate
   import DataFlowDispatch
@@ -9,3 +12,12 @@ module Private {
 module Public {
   import DataFlowUtil
 }
+
+module CppDataFlow implements InputSig {
+  import Private
+  import Public
+
+  Node exprNode(DataFlowExpr e) { result = Public::exprNode(e) }
+
+  predicate getAdditionalFlowIntoCallNodeTerm = Private::getAdditionalFlowIntoCallNodeTerm/2;
+}

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

@@ -2,7 +2,6 @@ private import cpp as Cpp
 private import DataFlowUtil
 private import semmle.code.cpp.ir.IR
 private import DataFlowDispatch
-private import DataFlowImplConsistency
 private import semmle.code.cpp.ir.internal.IRCppLanguage
 private import SsaInternals as Ssa
 private import DataFlowImplCommon as DataFlowImplCommon
@@ -220,9 +219,10 @@ private module IndirectOperands {
     int indirectionIndex;
 
     IndirectOperandFromIRRepr() {
-      exists(Operand repr |
-        repr = Ssa::getIRRepresentationOfIndirectOperand(operand, indirectionIndex) and
-        nodeHasOperand(this, repr, indirectionIndex - 1)
+      exists(Operand repr, int indirectionIndexRepr |
+        Ssa::hasIRRepresentationOfIndirectOperand(operand, indirectionIndex, repr,
+          indirectionIndexRepr) and
+        nodeHasOperand(this, repr, indirectionIndexRepr)
       )
     }
 
@@ -262,9 +262,10 @@ private module IndirectInstructions {
     int indirectionIndex;
 
     IndirectInstructionFromIRRepr() {
-      exists(Instruction repr |
-        repr = Ssa::getIRRepresentationOfIndirectInstruction(instr, indirectionIndex) and
-        nodeHasInstruction(this, repr, indirectionIndex - 1)
+      exists(Instruction repr, int indirectionIndexRepr |
+        Ssa::hasIRRepresentationOfIndirectInstruction(instr, indirectionIndex, repr,
+          indirectionIndexRepr) and
+        nodeHasInstruction(this, repr, indirectionIndexRepr)
       )
     }
 
@@ -681,9 +682,7 @@ predicate storeStepImpl(Node node1, Content c, PostFieldUpdateNode node2, boolea
  * Thus, `node2` references an object with a field `f` that contains the
  * value of `node1`.
  */
-predicate storeStep(Node node1, Content c, PostFieldUpdateNode node2) {
-  storeStepImpl(node1, c, node2, _)
-}
+predicate storeStep(Node node1, ContentSet c, Node node2) { storeStepImpl(node1, c, node2, _) }
 
 /**
  * Holds if `operandFrom` flows to `operandTo` using a sequence of conversion-like
@@ -692,7 +691,7 @@ predicate storeStep(Node node1, Content c, PostFieldUpdateNode node2) {
 private predicate numberOfLoadsFromOperandRec(
   Operand operandFrom, Operand operandTo, int ind, boolean certain
 ) {
-  exists(Instruction load | Ssa::isDereference(load, operandFrom) |
+  exists(Instruction load | Ssa::isDereference(load, operandFrom, _) |
     operandTo = operandFrom and ind = 0 and certain = true
     or
     numberOfLoadsFromOperand(load.getAUse(), operandTo, ind - 1, certain)
@@ -716,7 +715,7 @@ private predicate numberOfLoadsFromOperand(
 ) {
   numberOfLoadsFromOperandRec(operandFrom, operandTo, n, certain)
   or
-  not Ssa::isDereference(_, operandFrom) and
+  not Ssa::isDereference(_, operandFrom, _) and
   not conversionFlow(operandFrom, _, _, _) and
   operandFrom = operandTo and
   n = 0 and
@@ -744,7 +743,7 @@ predicate nodeHasInstruction(Node node, Instruction instr, int indirectionIndex)
  * Thus, `node1` references an object with a field `f` whose value ends up in
  * `node2`.
  */
-predicate readStep(Node node1, Content c, Node node2) {
+predicate readStep(Node node1, ContentSet c, Node node2) {
   exists(FieldAddress fa1, Operand operand, int numberOfLoads, int indirectionIndex2 |
     nodeHasOperand(node2, operand, indirectionIndex2) and
     // The `1` here matches the `node2.getIndirectionIndex() = 1` conjunct
@@ -767,7 +766,7 @@ predicate readStep(Node node1, Content c, Node node2) {
 /**
  * Holds if values stored inside content `c` are cleared at node `n`.
  */
-predicate clearsContent(Node n, Content c) {
+predicate clearsContent(Node n, ContentSet c) {
   n =
     any(PostUpdateNode pun, Content d | d.impliesClearOf(c) and storeStepImpl(_, d, pun, true) | pun)
         .getPreUpdateNode() and
@@ -792,7 +791,7 @@ predicate clearsContent(Node n, Content c) {
       storeStepImpl(_, d, pun, true) and
       pun.getPreUpdateNode() = n
     |
-      c.getIndirectionIndex() = d.getIndirectionIndex()
+      c.(Content).getIndirectionIndex() = d.getIndirectionIndex()
     )
   )
 }
@@ -805,6 +804,8 @@ predicate expectsContent(Node n, ContentSet c) { none() }
 
 predicate typeStrongerThan(DataFlowType t1, DataFlowType t2) { none() }
 
+predicate localMustFlowStep(Node node1, Node node2) { none() }
+
 /** Gets the type of `n` used for type pruning. */
 DataFlowType getNodeType(Node n) {
   suppressUnusedNode(n) and
@@ -833,12 +834,6 @@ class CastNode extends Node {
   CastNode() { none() } // stub implementation
 }
 
-/**
- * Holds if `n` should never be skipped over in the `PathGraph` and in path
- * explanations.
- */
-predicate neverSkipInPathGraph(Node n) { none() }
-
 /**
  * A function that may contain code or a variable that may contain itself. When
  * flow crosses from one _enclosing callable_ to another, the interprocedural
@@ -853,7 +848,7 @@ class DataFlowType = Type;
 
 /** A function call relevant for data flow. */
 class DataFlowCall extends CallInstruction {
-  Function getEnclosingCallable() { result = this.getEnclosingFunction() }
+  DataFlowCallable getEnclosingCallable() { result = this.getEnclosingFunction() }
 }
 
 module IsUnreachableInCall {
@@ -924,8 +919,6 @@ module IsUnreachableInCall {
 
 import IsUnreachableInCall
 
-int accessPathLimit() { result = 5 }
-
 /**
  * Holds if access paths with `c` at their head always should be tracked at high
  * precision. This disables adaptive access path precision for such access paths.
@@ -1021,14 +1014,6 @@ ContentApprox getContentApprox(Content c) {
   )
 }
 
-private class MyConsistencyConfiguration extends Consistency::ConsistencyConfiguration {
-  override predicate argHasPostUpdateExclude(ArgumentNode n) {
-    // The rules for whether an IR argument gets a post-update node are too
-    // complex to model here.
-    any()
-  }
-}
-
 /**
  * A local flow relation that includes both local steps, read steps and
  * argument-to-return flow through summarized functions.
@@ -1088,7 +1073,7 @@ private IRVariable getIRVariableForParameterNode(ParameterNode p) {
 
 /** Holds if `v` is the source variable corresponding to the parameter represented by `p`. */
 pragma[nomagic]
-private predicate parameterNodeHasSourceVariable(ParameterNode p, Ssa::SourceIRVariable v) {
+private predicate parameterNodeHasSourceVariable(ParameterNode p, Ssa::SourceVariable v) {
   v.getIRVariable() = getIRVariableForParameterNode(p) and
   exists(Position pos | p.isParameterOf(_, pos) |
     pos instanceof DirectPosition and

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

@@ -193,13 +193,23 @@ class Node extends TIRDataFlowNode {
    * a `Conversion`, then the result is the underlying non-`Conversion` base
    * expression.
    */
-  Expr asExpr() { result = this.(ExprNode).getExpr() }
+  Expr asExpr() { result = this.asExpr(_) }
+
+  /**
+   * INTERNAL: Do not use.
+   */
+  Expr asExpr(int n) { result = this.(ExprNode).getExpr(n) }
+
+  /**
+   * INTERNAL: Do not use.
+   */
+  Expr asIndirectExpr(int n, int index) { result = this.(IndirectExprNode).getExpr(n, index) }
 
   /**
    * Gets the non-conversion expression that's indirectly tracked by this node
    * under `index` number of indirections.
    */
-  Expr asIndirectExpr(int index) { result = this.(IndirectExprNode).getExpr(index) }
+  Expr asIndirectExpr(int index) { result = this.asIndirectExpr(_, index) }
 
   /**
    * Gets the non-conversion expression that's indirectly tracked by this node
@@ -211,15 +221,26 @@ class Node extends TIRDataFlowNode {
    * Gets the expression corresponding to this node, if any. The returned
    * expression may be a `Conversion`.
    */
-  Expr asConvertedExpr() { result = this.(ExprNode).getConvertedExpr() }
+  Expr asConvertedExpr() { result = this.asConvertedExpr(_) }
+
+  /**
+   * Gets the expression corresponding to this node, if any. The returned
+   * expression may be a `Conversion`.
+   */
+  Expr asConvertedExpr(int n) { result = this.(ExprNode).getConvertedExpr(n) }
+
+  /**
+   * INTERNAL: Do not use.
+   */
+  Expr asIndirectConvertedExpr(int n, int index) {
+    result = this.(IndirectExprNode).getConvertedExpr(n, index)
+  }
 
   /**
    * Gets the expression that's indirectly tracked by this node
    * behind `index` number of indirections.
    */
-  Expr asIndirectConvertedExpr(int index) {
-    result = this.(IndirectExprNode).getConvertedExpr(index)
-  }
+  Expr asIndirectConvertedExpr(int index) { result = this.asIndirectConvertedExpr(_, index) }
 
   /**
    * Gets the expression that's indirectly tracked by this node behind a
@@ -254,9 +275,7 @@ class Node extends TIRDataFlowNode {
    * after the `f` has returned.
    */
   Expr asDefiningArgument(int index) {
-    // Subtract one because `DefinitionByReferenceNode` is defined to be in
-    // the range `[0 ... n - 1]` for some `n` instead of `[1 ... n]`.
-    this.(DefinitionByReferenceNode).getIndirectionIndex() = index - 1 and
+    this.(DefinitionByReferenceNode).getIndirectionIndex() = index and
     result = this.(DefinitionByReferenceNode).getArgument()
   }
 
@@ -393,9 +412,10 @@ class Node extends TIRDataFlowNode {
 }
 
 private string toExprString(Node n) {
-  result = n.asExpr().toString()
+  result = n.asExpr(0).toString()
   or
-  result = n.asIndirectExpr().toString() + " indirection"
+  not exists(n.asExpr()) and
+  result = n.asIndirectExpr(0, 1).toString() + " indirection"
 }
 
 /**
@@ -550,11 +570,14 @@ class SsaPhiNode extends Node, TSsaPhiNode {
    * `fromBackEdge` is true if data flows along a back-edge,
    * and `false` otherwise.
    */
+  cached
   final Node getAnInput(boolean fromBackEdge) {
     localFlowStep(result, this) and
-    if phi.getBasicBlock().dominates(result.getBasicBlock())
-    then fromBackEdge = true
-    else fromBackEdge = false
+    exists(IRBlock bPhi, IRBlock bResult |
+      bPhi = phi.getBasicBlock() and bResult = result.getBasicBlock()
+    |
+      if bPhi.dominates(bResult) then fromBackEdge = true else fromBackEdge = false
+    )
   }
 
   /** Gets a node that is used as input to this phi node. */
@@ -781,26 +804,12 @@ class IndirectArgumentOutNode extends Node, TIndirectArgumentOutNode, PartialDef
   override Expr getDefinedExpr() { result = operand.getDef().getUnconvertedResultExpression() }
 }
 
-pragma[nomagic]
-predicate indirectReturnOutNodeOperand0(CallInstruction call, Operand operand, int indirectionIndex) {
-  Ssa::hasRawIndirectInstruction(call, indirectionIndex) and
-  operandForFullyConvertedCall(operand, call)
-}
-
-pragma[nomagic]
-predicate indirectReturnOutNodeInstruction0(
-  CallInstruction call, Instruction instr, int indirectionIndex
-) {
-  Ssa::hasRawIndirectInstruction(call, indirectionIndex) and
-  instructionForFullyConvertedCall(instr, call)
-}
-
 /**
  * Holds if `node` is an indirect operand with columns `(operand, indirectionIndex)`, and
  * `operand` represents a use of the fully converted value of `call`.
  */
 private predicate hasOperand(Node node, CallInstruction call, int indirectionIndex, Operand operand) {
-  indirectReturnOutNodeOperand0(call, operand, indirectionIndex) and
+  operandForFullyConvertedCall(operand, call) and
   hasOperandAndIndex(node, operand, indirectionIndex)
 }
 
@@ -813,7 +822,7 @@ private predicate hasOperand(Node node, CallInstruction call, int indirectionInd
 private predicate hasInstruction(
   Node node, CallInstruction call, int indirectionIndex, Instruction instr
 ) {
-  indirectReturnOutNodeInstruction0(call, instr, indirectionIndex) and
+  instructionForFullyConvertedCall(instr, call) and
   hasInstructionAndIndex(node, instr, indirectionIndex)
 }
 
@@ -946,7 +955,7 @@ class RawIndirectOperand extends Node, TRawIndirectOperand {
   }
 
   override string toStringImpl() {
-    result = instructionNode(this.getOperand().getDef()).toStringImpl() + " indirection"
+    result = operandNode(this.getOperand()).toStringImpl() + " indirection"
   }
 }
 
@@ -1053,77 +1062,130 @@ class RawIndirectInstruction extends Node, TRawIndirectInstruction {
   }
 }
 
-/** Holds if `node` is an `OperandNode` that should map `node.asExpr()` to `e`. */
-predicate exprNodeShouldBeOperand(OperandNode node, Expr e) {
-  exists(Instruction def |
-    unique( | | getAUse(def)) = node.getOperand() and
-    e = def.getConvertedResultExpression()
-  )
+private module GetConvertedResultExpression {
+  private import semmle.code.cpp.ir.implementation.raw.internal.TranslatedExpr
+  private import semmle.code.cpp.ir.implementation.raw.internal.InstructionTag
+
+  private Operand getAnInitializeDynamicAllocationInstructionAddress() {
+    result = any(InitializeDynamicAllocationInstruction init).getAllocationAddressOperand()
+  }
+
+  /**
+   * Gets the expression that should be returned as the result expression from `instr`.
+   *
+   * Note that this predicate may return multiple results in cases where a conversion belongs to a
+   * different AST element than its operand.
+   */
+  Expr getConvertedResultExpression(Instruction instr, int n) {
+    // Only fully converted instructions have a result for `asConvertedExpr`
+    not conversionFlow(unique(Operand op |
+        // The address operand of a `InitializeDynamicAllocationInstruction` is
+        // special: we need to handle it during dataflow (since it's
+        // effectively a store to an indirection), but it doesn't appear in
+        // source syntax, so dataflow node <-> expression conversion shouldn't
+        // care about it.
+        op = getAUse(instr) and not op = getAnInitializeDynamicAllocationInstructionAddress()
+      |
+        op
+      ), _, false, false) and
+    result = getConvertedResultExpressionImpl(instr) and
+    n = 0
+    or
+    // If the conversion also has a result then we return multiple results
+    exists(Operand operand | conversionFlow(operand, instr, false, false) |
+      n = 1 and
+      result = getConvertedResultExpressionImpl(operand.getDef())
+      or
+      result = getConvertedResultExpression(operand.getDef(), n - 1)
+    )
+  }
+
+  private Expr getConvertedResultExpressionImpl0(Instruction instr) {
+    // For an expression such as `i += 2` we pretend that the generated
+    // `StoreInstruction` contains the result of the expression even though
+    // this isn't totally aligned with the C/C++ standard.
+    exists(TranslatedAssignOperation tao |
+      result = tao.getExpr() and
+      instr = tao.getInstruction(any(AssignmentStoreTag tag))
+    )
+    or
+    // Similarly for `i++` and `++i` we pretend that the generated
+    // `StoreInstruction` is contains the result of the expression even though
+    // this isn't totally aligned with the C/C++ standard.
+    exists(TranslatedCrementOperation tco |
+      result = tco.getExpr() and
+      instr = tco.getInstruction(any(CrementStoreTag tag))
+    )
+    or
+    // IR construction inserts an additional cast to a `size_t` on the extent
+    // of a `new[]` expression. The resulting `ConvertInstruction` doesn't have
+    // a result for `getConvertedResultExpression`. We remap this here so that
+    // this `ConvertInstruction` maps to the result of the expression that
+    // represents the extent.
+    exists(TranslatedNonConstantAllocationSize tas |
+      result = tas.getExtent().getExpr() and
+      instr = tas.getInstruction(any(AllocationExtentConvertTag tag))
+    )
+    or
+    // There's no instruction that returns `ParenthesisExpr`, but some queries
+    // expect this
+    exists(TranslatedTransparentConversion ttc |
+      result = ttc.getExpr().(ParenthesisExpr) and
+      instr = ttc.getResult()
+    )
+  }
+
+  private Expr getConvertedResultExpressionImpl(Instruction instr) {
+    result = getConvertedResultExpressionImpl0(instr)
+    or
+    not exists(getConvertedResultExpressionImpl0(instr)) and
+    result = instr.getConvertedResultExpression()
+  }
 }
 
-private predicate indirectExprNodeShouldBeIndirectOperand0(
-  VariableAddressInstruction instr, RawIndirectOperand node, Expr e
-) {
-  instr = node.getOperand().getDef() and
-  e = instr.getAst().(Expr).getUnconverted()
+private import GetConvertedResultExpression
+
+/** Holds if `node` is an `OperandNode` that should map `node.asExpr()` to `e`. */
+predicate exprNodeShouldBeOperand(OperandNode node, Expr e, int n) {
+  exists(Instruction def |
+    unique( | | getAUse(def)) = node.getOperand() and
+    e = getConvertedResultExpression(def, n)
+  )
 }
 
 /** Holds if `node` should be an `IndirectOperand` that maps `node.asIndirectExpr()` to `e`. */
-private predicate indirectExprNodeShouldBeIndirectOperand(RawIndirectOperand node, Expr e) {
-  exists(Instruction instr | instr = node.getOperand().getDef() |
-    exists(Expr e0 |
-      indirectExprNodeShouldBeIndirectOperand0(instr, node, e0) and
-      e = e0.getFullyConverted()
-    )
-    or
-    not indirectExprNodeShouldBeIndirectOperand0(_, node, _) and
-    e = instr.getConvertedResultExpression()
+private predicate indirectExprNodeShouldBeIndirectOperand(
+  IndirectOperand node, Expr e, int n, int indirectionIndex
+) {
+  exists(Instruction def |
+    node.hasOperandAndIndirectionIndex(unique( | | getAUse(def)), indirectionIndex) and
+    e = getConvertedResultExpression(def, n)
   )
 }
 
-private predicate exprNodeShouldBeIndirectOutNode(IndirectArgumentOutNode node, Expr e) {
+private predicate exprNodeShouldBeIndirectOutNode(IndirectArgumentOutNode node, Expr e, int n) {
   exists(CallInstruction call |
     call.getStaticCallTarget() instanceof Constructor and
-    e = call.getConvertedResultExpression() and
+    e = getConvertedResultExpression(call, n) and
     call.getThisArgumentOperand() = node.getAddressOperand()
   )
 }
 
 /** Holds if `node` should be an instruction node that maps `node.asExpr()` to `e`. */
-predicate exprNodeShouldBeInstruction(Node node, Expr e) {
-  not exprNodeShouldBeOperand(_, e) and
-  not exprNodeShouldBeIndirectOutNode(_, e) and
-  (
-    e = node.asInstruction().getConvertedResultExpression()
-    or
-    // The instruction that contains the result of an `AssignOperation` is
-    // the unloaded left operand (see the comments in `TranslatedAssignOperation`).
-    // That means that for cases like
-    // ```cpp
-    // int x = ...;
-    // x += 1;
-    // ```
-    // the result of `x += 1` is the `VariableAddressInstruction` that represents `x`. But
-    // that instruction doesn't receive the flow from this `AssignOperation`. So instead we
-    // map the operation to the `AddInstruction`.
-    node.asInstruction().getAst() = e.(AssignOperation)
-    or
-    // Same story for `CrementOperation`s (cf. the comments in the subclasses
-    // of `TranslatedCrementOperation`).
-    node.asInstruction().getAst() = e.(CrementOperation)
-  )
+predicate exprNodeShouldBeInstruction(Node node, Expr e, int n) {
+  not exprNodeShouldBeOperand(_, e, n) and
+  not exprNodeShouldBeIndirectOutNode(_, e, n) and
+  e = getConvertedResultExpression(node.asInstruction(), n)
 }
 
 /** Holds if `node` should be an `IndirectInstruction` that maps `node.asIndirectExpr()` to `e`. */
-predicate indirectExprNodeShouldBeIndirectInstruction(IndirectInstruction node, Expr e) {
+predicate indirectExprNodeShouldBeIndirectInstruction(
+  IndirectInstruction node, Expr e, int n, int indirectionIndex
+) {
+  not indirectExprNodeShouldBeIndirectOperand(_, e, n, indirectionIndex) and
   exists(Instruction instr |
-    node.hasInstructionAndIndirectionIndex(instr, _) and
-    not indirectExprNodeShouldBeIndirectOperand(_, e)
-  |
-    e = instr.(VariableAddressInstruction).getAst().(Expr).getFullyConverted()
-    or
-    not instr instanceof VariableAddressInstruction and
-    e = instr.getConvertedResultExpression()
+    node.hasInstructionAndIndirectionIndex(instr, indirectionIndex) and
+    e = getConvertedResultExpression(instr, n)
   )
 }
 
@@ -1132,30 +1194,32 @@ abstract private class ExprNodeBase extends Node {
    * Gets the expression corresponding to this node, if any. The returned
    * expression may be a `Conversion`.
    */
-  abstract Expr getConvertedExpr();
+  abstract Expr getConvertedExpr(int n);
 
   /** Gets the non-conversion expression corresponding to this node, if any. */
-  abstract Expr getExpr();
+  final Expr getExpr(int n) { result = this.getConvertedExpr(n).getUnconverted() }
 }
 
 private class InstructionExprNode extends ExprNodeBase, InstructionNode {
-  InstructionExprNode() { exprNodeShouldBeInstruction(this, _) }
+  InstructionExprNode() {
+    exists(Expr e, int n |
+      exprNodeShouldBeInstruction(this, e, n) and
+      not exprNodeShouldBeInstruction(_, e, n + 1)
+    )
+  }
 
-  final override Expr getConvertedExpr() { exprNodeShouldBeInstruction(this, result) }
-
-  final override Expr getExpr() { result = this.getConvertedExpr().getUnconverted() }
-
-  final override string toStringImpl() { result = this.getConvertedExpr().toString() }
+  final override Expr getConvertedExpr(int n) { exprNodeShouldBeInstruction(this, result, n) }
 }
 
 private class OperandExprNode extends ExprNodeBase, OperandNode {
-  OperandExprNode() { exprNodeShouldBeOperand(this, _) }
+  OperandExprNode() {
+    exists(Expr e, int n |
+      exprNodeShouldBeOperand(this, e, n) and
+      not exprNodeShouldBeOperand(_, e, n + 1)
+    )
+  }
 
-  final override Expr getConvertedExpr() { exprNodeShouldBeOperand(this, result) }
-
-  final override Expr getExpr() { result = this.getConvertedExpr().getUnconverted() }
-
-  final override string toStringImpl() { result = this.getConvertedExpr().toString() }
+  final override Expr getConvertedExpr(int n) { exprNodeShouldBeOperand(this, result, n) }
 }
 
 abstract private class IndirectExprNodeBase extends Node {
@@ -1163,67 +1227,75 @@ abstract private class IndirectExprNodeBase extends Node {
    * Gets the expression corresponding to this node, if any. The returned
    * expression may be a `Conversion`.
    */
-  abstract Expr getConvertedExpr(int indirectionIndex);
+  abstract Expr getConvertedExpr(int n, int indirectionIndex);
 
   /** Gets the non-conversion expression corresponding to this node, if any. */
-  abstract Expr getExpr(int indirectionIndex);
-}
-
-private class IndirectOperandIndirectExprNode extends IndirectExprNodeBase, RawIndirectOperand {
-  IndirectOperandIndirectExprNode() { indirectExprNodeShouldBeIndirectOperand(this, _) }
-
-  final override Expr getConvertedExpr(int index) {
-    this.getIndirectionIndex() = index and
-    indirectExprNodeShouldBeIndirectOperand(this, result)
-  }
-
-  final override Expr getExpr(int index) {
-    this.getIndirectionIndex() = index and
-    result = this.getConvertedExpr(index).getUnconverted()
+  final Expr getExpr(int n, int indirectionIndex) {
+    result = this.getConvertedExpr(n, indirectionIndex).getUnconverted()
   }
 }
 
-private class IndirectInstructionIndirectExprNode extends IndirectExprNodeBase,
-  RawIndirectInstruction
+private class IndirectOperandIndirectExprNode extends IndirectExprNodeBase instanceof IndirectOperand
 {
-  IndirectInstructionIndirectExprNode() { indirectExprNodeShouldBeIndirectInstruction(this, _) }
-
-  final override Expr getConvertedExpr(int index) {
-    this.getIndirectionIndex() = index and
-    indirectExprNodeShouldBeIndirectInstruction(this, result)
+  IndirectOperandIndirectExprNode() {
+    exists(Expr e, int n, int indirectionIndex |
+      indirectExprNodeShouldBeIndirectOperand(this, e, n, indirectionIndex) and
+      not indirectExprNodeShouldBeIndirectOperand(_, e, n + 1, indirectionIndex)
+    )
   }
 
-  final override Expr getExpr(int index) {
-    this.getIndirectionIndex() = index and
-    result = this.getConvertedExpr(index).getUnconverted()
+  final override Expr getConvertedExpr(int n, int index) {
+    indirectExprNodeShouldBeIndirectOperand(this, result, n, index)
+  }
+}
+
+private class IndirectInstructionIndirectExprNode extends IndirectExprNodeBase instanceof IndirectInstruction
+{
+  IndirectInstructionIndirectExprNode() {
+    exists(Expr e, int n, int indirectionIndex |
+      indirectExprNodeShouldBeIndirectInstruction(this, e, n, indirectionIndex) and
+      not indirectExprNodeShouldBeIndirectInstruction(_, e, n + 1, indirectionIndex)
+    )
+  }
+
+  final override Expr getConvertedExpr(int n, int index) {
+    indirectExprNodeShouldBeIndirectInstruction(this, result, n, index)
   }
 }
 
 private class IndirectArgumentOutExprNode extends ExprNodeBase, IndirectArgumentOutNode {
-  IndirectArgumentOutExprNode() { exprNodeShouldBeIndirectOutNode(this, _) }
+  IndirectArgumentOutExprNode() { exprNodeShouldBeIndirectOutNode(this, _, _) }
 
-  final override Expr getConvertedExpr() { exprNodeShouldBeIndirectOutNode(this, result) }
-
-  final override Expr getExpr() { result = this.getConvertedExpr() }
+  final override Expr getConvertedExpr(int n) { exprNodeShouldBeIndirectOutNode(this, result, n) }
 }
 
 /**
  * An expression, viewed as a node in a data flow graph.
  */
 class ExprNode extends Node instanceof ExprNodeBase {
+  /**
+   * INTERNAL: Do not use.
+   */
+  Expr getExpr(int n) { result = super.getExpr(n) }
+
   /**
    * Gets the non-conversion expression corresponding to this node, if any. If
    * this node strictly (in the sense of `getConvertedExpr`) corresponds to a
    * `Conversion`, then the result is that `Conversion`'s non-`Conversion` base
    * expression.
    */
-  Expr getExpr() { result = super.getExpr() }
+  final Expr getExpr() { result = this.getExpr(_) }
+
+  /**
+   * INTERNAL: Do not use.
+   */
+  Expr getConvertedExpr(int n) { result = super.getConvertedExpr(n) }
 
   /**
    * Gets the expression corresponding to this node, if any. The returned
    * expression may be a `Conversion`.
    */
-  Expr getConvertedExpr() { result = super.getConvertedExpr() }
+  final Expr getConvertedExpr() { result = this.getConvertedExpr(_) }
 }
 
 /**
@@ -1236,13 +1308,27 @@ class IndirectExprNode extends Node instanceof IndirectExprNodeBase {
    * `Conversion`, then the result is that `Conversion`'s non-`Conversion` base
    * expression.
    */
-  Expr getExpr(int indirectionIndex) { result = super.getExpr(indirectionIndex) }
+  final Expr getExpr(int indirectionIndex) { result = this.getExpr(_, indirectionIndex) }
+
+  /**
+   * INTERNAL: Do not use.
+   */
+  Expr getExpr(int n, int indirectionIndex) { result = super.getExpr(n, indirectionIndex) }
+
+  /**
+   * INTERNAL: Do not use.
+   */
+  Expr getConvertedExpr(int n, int indirectionIndex) {
+    result = super.getConvertedExpr(n, indirectionIndex)
+  }
 
   /**
    * Gets the expression corresponding to this node, if any. The returned
    * expression may be a `Conversion`.
    */
-  Expr getConvertedExpr(int indirectionIndex) { result = super.getConvertedExpr(indirectionIndex) }
+  Expr getConvertedExpr(int indirectionIndex) {
+    result = this.getConvertedExpr(_, indirectionIndex)
+  }
 }
 
 /**
@@ -1268,6 +1354,9 @@ class ParameterNode extends Node {
    * pointer-indirection parameters are at further negative positions.
    */
   predicate isParameterOf(Function f, ParameterPosition pos) { none() } // overridden by subclasses
+
+  /** Gets the `Parameter` associated with this node, if it exists. */
+  Parameter getParameter() { none() } // overridden by subclasses
 }
 
 /** An explicit positional parameter, including `this`, but not `...`. */
@@ -1290,10 +1379,9 @@ private class ExplicitParameterNode extends ParameterNode, DirectParameterNode {
     f.getParameter(pos.(DirectPosition).getIndex()) = instr.getParameter()
   }
 
-  /** Gets the `Parameter` associated with this node. */
-  Parameter getParameter() { result = instr.getParameter() }
-
   override string toStringImpl() { result = instr.getParameter().toString() }
+
+  override Parameter getParameter() { result = instr.getParameter() }
 }
 
 /** An implicit `this` parameter. */
@@ -1455,7 +1543,7 @@ OperandNode operandNode(Operand operand) { result.getOperand() = operand }
  * _out of_ an expression, like when an argument is passed by reference, use
  * `definitionByReferenceNodeFromArgument` instead.
  */
-ExprNode exprNode(Expr e) { result.getExpr() = e }
+ExprNode exprNode(Expr e) { result.getExpr(_) = e }
 
 /**
  * Gets the `Node` corresponding to the value of evaluating `e`. Here, `e` may
@@ -1463,7 +1551,7 @@ ExprNode exprNode(Expr e) { result.getExpr() = e }
  * argument is passed by reference, use
  * `definitionByReferenceNodeFromArgument` instead.
  */
-ExprNode convertedExprNode(Expr e) { result.getConvertedExpr() = e }
+ExprNode convertedExprNode(Expr e) { result.getConvertedExpr(_) = e }
 
 /**
  * Gets the `Node` corresponding to the value of `p` at function entry.
@@ -1534,6 +1622,25 @@ private module Cached {
     )
   }
 
+  /**
+   * Holds if `operand.getDef() = instr`, but there exists a `StoreInstruction` that
+   * writes to an address that is equivalent to the value computed by `instr` in
+   * between `instr` and `operand`, and therefore there should not be flow from `*instr`
+   * to `*operand`.
+   */
+  pragma[nomagic]
+  private predicate isStoredToBetween(Instruction instr, Operand operand) {
+    simpleOperandLocalFlowStep(pragma[only_bind_into](instr), pragma[only_bind_into](operand)) and
+    exists(StoreInstruction store, IRBlock block, int storeIndex, int instrIndex, int operandIndex |
+      store.getDestinationAddress() = instr and
+      block.getInstruction(storeIndex) = store and
+      block.getInstruction(instrIndex) = instr and
+      block.getInstruction(operandIndex) = operand.getUse() and
+      instrIndex < storeIndex and
+      storeIndex < operandIndex
+    )
+  }
+
   private predicate indirectionInstructionFlow(
     RawIndirectInstruction nodeFrom, IndirectOperand nodeTo
   ) {
@@ -1543,7 +1650,8 @@ private module Cached {
       simpleOperandLocalFlowStep(pragma[only_bind_into](instr), pragma[only_bind_into](operand))
     |
       hasOperandAndIndex(nodeTo, operand, pragma[only_bind_into](indirectionIndex)) and
-      hasInstructionAndIndex(nodeFrom, instr, pragma[only_bind_into](indirectionIndex))
+      hasInstructionAndIndex(nodeFrom, instr, pragma[only_bind_into](indirectionIndex)) and
+      not isStoredToBetween(instr, operand)
     )
   }
 

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

@@ -87,6 +87,30 @@ module ProductFlow {
      * dataflow graph.
      */
     default predicate isBarrierIn2(DataFlow::Node node) { none() }
+
+    /**
+     * Gets the virtual dispatch branching limit when calculating field flow in the first
+     * projection of the product dataflow graph.
+     *
+     * This can be overridden to a smaller value to improve performance (a
+     * value of 0 disables field flow), or a larger value to get more results.
+     */
+    default int fieldFlowBranchLimit1() {
+      // NOTE: This should be synchronized with the default value in the shared dataflow library
+      result = 2
+    }
+
+    /**
+     * Gets the virtual dispatch branching limit when calculating field flow in the second
+     * projection of the product dataflow graph.
+     *
+     * This can be overridden to a smaller value to improve performance (a
+     * value of 0 disables field flow), or a larger value to get more results.
+     */
+    default int fieldFlowBranchLimit2() {
+      // NOTE: This should be synchronized with the default value in the shared dataflow library
+      result = 2
+    }
   }
 
   /**
@@ -272,6 +296,30 @@ module ProductFlow {
      * dataflow graph.
      */
     default predicate isBarrierIn2(DataFlow::Node node) { none() }
+
+    /**
+     * Gets the virtual dispatch branching limit when calculating field flow in the first
+     * projection of the product dataflow graph.
+     *
+     * This can be overridden to a smaller value to improve performance (a
+     * value of 0 disables field flow), or a larger value to get more results.
+     */
+    default int fieldFlowBranchLimit1() {
+      // NOTE: This should be synchronized with the default value in the shared dataflow library
+      result = 2
+    }
+
+    /**
+     * Gets the virtual dispatch branching limit when calculating field flow in the second
+     * projection of the product dataflow graph.
+     *
+     * This can be overridden to a smaller value to improve performance (a
+     * value of 0 disables field flow), or a larger value to get more results.
+     */
+    default int fieldFlowBranchLimit2() {
+      // NOTE: This should be synchronized with the default value in the shared dataflow library
+      result = 2
+    }
   }
 
   /**
@@ -335,6 +383,8 @@ module ProductFlow {
       }
 
       predicate isBarrierIn(DataFlow::Node node) { Config::isBarrierIn1(node) }
+
+      int fieldFlowBranchLimit() { result = Config::fieldFlowBranchLimit1() }
     }
 
     private module Flow1 = DataFlow::GlobalWithState<Config1>;
@@ -367,6 +417,8 @@ module ProductFlow {
       }
 
       predicate isBarrierIn(DataFlow::Node node) { Config::isBarrierIn2(node) }
+
+      int fieldFlowBranchLimit() { result = Config::fieldFlowBranchLimit2() }
     }
 
     private module Flow2 = DataFlow::GlobalWithState<Config2>;

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

@@ -10,32 +10,35 @@ private import ssa0.SsaInternals as SsaInternals0
 import SsaInternalsCommon
 
 private module SourceVariables {
-  int getMaxIndirectionForIRVariable(IRVariable var) {
-    exists(Type type, boolean isGLValue |
-      var.getLanguageType().hasType(type, isGLValue) and
-      if isGLValue = true
-      then result = 1 + getMaxIndirectionsForType(type)
-      else result = getMaxIndirectionsForType(type)
-    )
-  }
-
   cached
   private newtype TSourceVariable =
-    TSourceIRVariable(BaseIRVariable baseVar, int ind) {
-      ind = [0 .. getMaxIndirectionForIRVariable(baseVar.getIRVariable())]
-    } or
-    TCallVariable(AllocationInstruction call, int ind) {
-      ind = [0 .. countIndirectionsForCppType(getResultLanguageType(call))]
+    TMkSourceVariable(SsaInternals0::SourceVariable base, int ind) {
+      ind = [0 .. countIndirectionsForCppType(base.getLanguageType()) + 1]
     }
 
-  abstract class SourceVariable extends TSourceVariable {
+  class SourceVariable extends TSourceVariable {
+    SsaInternals0::SourceVariable base;
     int ind;
 
-    bindingset[ind]
-    SourceVariable() { any() }
+    SourceVariable() { this = TMkSourceVariable(base, ind) }
+
+    /** Gets the IR variable associated with this `SourceVariable`, if any. */
+    IRVariable getIRVariable() { result = base.(BaseIRVariable).getIRVariable() }
+
+    /**
+     * Gets the base source variable (i.e., the variable without any
+     * indirections) of this source variable.
+     */
+    SsaInternals0::SourceVariable getBaseVariable() { result = base }
 
     /** Gets a textual representation of this element. */
-    abstract string toString();
+    string toString() {
+      ind = 0 and
+      result = this.getBaseVariable().toString()
+      or
+      ind > 0 and
+      result = this.getBaseVariable().toString() + " indirection"
+    }
 
     /**
      * Gets the number of loads performed on the base source variable
@@ -43,65 +46,19 @@ private module SourceVariables {
      */
     int getIndirection() { result = ind }
 
-    /**
-     * Gets the base source variable (i.e., the variable without any
-     * indirections) of this source variable.
-     */
-    abstract BaseSourceVariable getBaseVariable();
-
     /** Holds if this variable is a glvalue. */
-    predicate isGLValue() { none() }
+    predicate isGLValue() { ind = 0 }
 
     /**
      * Gets the type of this source variable. If `isGLValue()` holds, then
      * the type of this source variable should be thought of as "pointer
      * to `getType()`".
      */
-    abstract DataFlowType getType();
-  }
-
-  class SourceIRVariable extends SourceVariable, TSourceIRVariable {
-    BaseIRVariable var;
-
-    SourceIRVariable() { this = TSourceIRVariable(var, ind) }
-
-    IRVariable getIRVariable() { result = var.getIRVariable() }
-
-    override BaseIRVariable getBaseVariable() { result.getIRVariable() = this.getIRVariable() }
-
-    override string toString() {
-      ind = 0 and
-      result = this.getIRVariable().toString()
-      or
-      ind > 0 and
-      result = this.getIRVariable().toString() + " indirection"
+    DataFlowType getType() {
+      if this.isGLValue()
+      then result = base.getType()
+      else result = getTypeImpl(base.getType(), ind - 1)
     }
-
-    override predicate isGLValue() { ind = 0 }
-
-    override DataFlowType getType() {
-      if ind = 0 then result = var.getType() else result = getTypeImpl(var.getType(), ind - 1)
-    }
-  }
-
-  class CallVariable extends SourceVariable, TCallVariable {
-    AllocationInstruction call;
-
-    CallVariable() { this = TCallVariable(call, ind) }
-
-    AllocationInstruction getCall() { result = call }
-
-    override BaseCallVariable getBaseVariable() { result.getCallInstruction() = call }
-
-    override string toString() {
-      ind = 0 and
-      result = "Call"
-      or
-      ind > 0 and
-      result = "Call indirection"
-    }
-
-    override DataFlowType getType() { result = getTypeImpl(call.getResultType(), ind) }
   }
 }
 
@@ -117,7 +74,7 @@ predicate hasRawIndirectOperand(Operand op, int indirectionIndex) {
     type = getLanguageType(op) and
     m = countIndirectionsForCppType(type) and
     indirectionIndex = [1 .. m] and
-    not exists(getIRRepresentationOfIndirectOperand(op, indirectionIndex))
+    not hasIRRepresentationOfIndirectOperand(op, indirectionIndex, _, _)
   )
 }
 
@@ -131,14 +88,15 @@ predicate hasRawIndirectInstruction(Instruction instr, int indirectionIndex) {
     type = getResultLanguageType(instr) and
     m = countIndirectionsForCppType(type) and
     indirectionIndex = [1 .. m] and
-    not exists(getIRRepresentationOfIndirectInstruction(instr, indirectionIndex))
+    not hasIRRepresentationOfIndirectInstruction(instr, indirectionIndex, _, _)
   )
 }
 
 cached
 private newtype TDefOrUseImpl =
-  TDefImpl(Operand address, int indirectionIndex) {
-    exists(Instruction base | isDef(_, _, address, base, _, indirectionIndex) |
+  TDefImpl(BaseSourceVariableInstruction base, Operand address, int indirectionIndex) {
+    isDef(_, _, address, base, _, indirectionIndex) and
+    (
       // We only include the definition if the SSA pruning stage
       // concluded that the definition is live after the write.
       any(SsaInternals0::Def def).getAddressOperand() = address
@@ -148,9 +106,9 @@ private newtype TDefOrUseImpl =
       base.(VariableAddressInstruction).getAstVariable() instanceof GlobalLikeVariable
     )
   } or
-  TUseImpl(Operand operand, int indirectionIndex) {
-    isUse(_, operand, _, _, indirectionIndex) and
-    not isDef(_, _, operand, _, _, _)
+  TUseImpl(BaseSourceVariableInstruction base, Operand operand, int indirectionIndex) {
+    isUse(_, operand, base, _, indirectionIndex) and
+    not isDef(true, _, operand, _, _, _)
   } or
   TGlobalUse(GlobalLikeVariable v, IRFunction f, int indirectionIndex) {
     // Represents a final "use" of a global variable to ensure that
@@ -236,7 +194,7 @@ abstract private class DefOrUseImpl extends TDefOrUseImpl {
 
   /**
    * Gets the instruction that computes the base of this definition or use.
-   * This is always a `VariableAddressInstruction` or an `AllocationInstruction`.
+   * This is always a `VariableAddressInstruction` or an `CallInstruction`.
    */
   abstract BaseSourceVariableInstruction getBase();
 
@@ -308,15 +266,17 @@ abstract class DefImpl extends DefOrUseImpl {
 }
 
 private class DirectDef extends DefImpl, TDefImpl {
-  DirectDef() { this = TDefImpl(address, ind) }
+  BaseSourceVariableInstruction base;
 
-  override BaseSourceVariableInstruction getBase() { isDef(_, _, address, result, _, _) }
+  DirectDef() { this = TDefImpl(base, address, ind) }
 
-  override int getIndirection() { isDef(_, _, address, _, result, ind) }
+  override BaseSourceVariableInstruction getBase() { result = base }
 
-  override Node0Impl getValue() { isDef(_, result, address, _, _, _) }
+  override int getIndirection() { isDef(_, _, address, base, result, ind) }
 
-  override predicate isCertain() { isDef(true, _, address, _, _, ind) }
+  override Node0Impl getValue() { isDef(_, result, address, base, _, _) }
+
+  override predicate isCertain() { isDef(true, _, address, base, _, ind) }
 }
 
 private class IteratorDef extends DefImpl, TIteratorDef {
@@ -359,6 +319,7 @@ abstract class UseImpl extends DefOrUseImpl {
 
 abstract private class OperandBasedUse extends UseImpl {
   Operand operand;
+  BaseSourceVariableInstruction base;
 
   bindingset[ind]
   OperandBasedUse() { any() }
@@ -366,50 +327,44 @@ abstract private class OperandBasedUse extends UseImpl {
   final override predicate hasIndexInBlock(IRBlock block, int index) {
     // See the comment in `ssa0`'s `OperandBasedUse` for an explanation of this
     // predicate's implementation.
-    exists(BaseSourceVariableInstruction base | base = this.getBase() |
-      if base.getAst() = any(Cpp::PostfixCrementOperation c).getOperand()
-      then
-        exists(Operand op, int indirectionIndex, int indirection |
-          indirectionIndex = this.getIndirectionIndex() and
-          indirection = this.getIndirection() and
-          op =
-            min(Operand cand, int i |
-              isUse(_, cand, base, indirection, indirectionIndex) and
-              block.getInstruction(i) = cand.getUse()
-            |
-              cand order by i
-            ) and
-          block.getInstruction(index) = op.getUse()
-        )
-      else operand.getUse() = block.getInstruction(index)
-    )
+    if base.getAst() = any(Cpp::PostfixCrementOperation c).getOperand()
+    then
+      exists(Operand op, int indirectionIndex, int indirection |
+        indirectionIndex = this.getIndirectionIndex() and
+        indirection = this.getIndirection() and
+        op =
+          min(Operand cand, int i |
+            isUse(_, cand, base, indirection, indirectionIndex) and
+            block.getInstruction(i) = cand.getUse()
+          |
+            cand order by i
+          ) and
+        block.getInstruction(index) = op.getUse()
+      )
+    else operand.getUse() = block.getInstruction(index)
   }
 
+  final override BaseSourceVariableInstruction getBase() { result = base }
+
   final Operand getOperand() { result = operand }
 
   final override Cpp::Location getLocation() { result = operand.getLocation() }
 }
 
 private class DirectUse extends OperandBasedUse, TUseImpl {
-  DirectUse() { this = TUseImpl(operand, ind) }
+  DirectUse() { this = TUseImpl(base, operand, ind) }
 
-  override int getIndirection() { isUse(_, operand, _, result, ind) }
+  override int getIndirection() { isUse(_, operand, base, result, ind) }
 
-  override BaseSourceVariableInstruction getBase() { isUse(_, operand, result, _, ind) }
-
-  override predicate isCertain() { isUse(true, operand, _, _, ind) }
+  override predicate isCertain() { isUse(true, operand, base, _, ind) }
 
   override Node getNode() { nodeHasOperand(result, operand, ind) }
 }
 
 private class IteratorUse extends OperandBasedUse, TIteratorUse {
-  BaseSourceVariableInstruction container;
+  IteratorUse() { this = TIteratorUse(operand, base, ind) }
 
-  IteratorUse() { this = TIteratorUse(operand, container, ind) }
-
-  override int getIndirection() { isIteratorUse(container, operand, result, ind) }
-
-  override BaseSourceVariableInstruction getBase() { result = container }
+  override int getIndirection() { isIteratorUse(base, operand, result, ind) }
 
   override predicate isCertain() { none() }
 
@@ -492,9 +447,16 @@ class GlobalUse extends UseImpl, TGlobalUse {
   IRFunction getIRFunction() { result = f }
 
   final override predicate hasIndexInBlock(IRBlock block, int index) {
-    exists(ExitFunctionInstruction exit |
-      exit = f.getExitFunctionInstruction() and
-      block.getInstruction(index) = exit
+    // Similar to the `FinalParameterUse` case, we want to generate flow out of
+    // globals at any exit so that we can flow out of non-returning functions.
+    // Obviously this isn't correct as we can't actually flow but the global flow
+    // requires this if we want to flow into children.
+    exists(Instruction return |
+      return instanceof ReturnInstruction or
+      return instanceof UnreachedInstruction
+    |
+      block.getInstruction(index) = return and
+      return.getEnclosingIRFunction() = f
     )
   }
 
@@ -655,7 +617,7 @@ private predicate indirectConversionFlowStep(Node nFrom, Node nTo) {
       hasOperandAndIndex(nFrom, op1, pragma[only_bind_into](indirectionIndex)) and
       hasOperandAndIndex(nTo, op2, indirectionIndex - 1) and
       instr = op2.getDef() and
-      isDereference(instr, op1)
+      isDereference(instr, op1, _)
     )
   )
 }
@@ -683,12 +645,24 @@ private predicate adjustForPointerArith(PostUpdateNode pun, UseOrPhi use) {
   )
 }
 
+/**
+ * Holds if `nodeFrom` flows to `nodeTo` because there is `def-use` or
+ * `use-use` flow from `defOrUse` to `use`.
+ *
+ * `uncertain` is `true` if the `defOrUse` is an uncertain definition.
+ */
+private predicate localSsaFlow(
+  SsaDefOrUse defOrUse, Node nodeFrom, UseOrPhi use, Node nodeTo, boolean uncertain
+) {
+  nodeToDefOrUse(nodeFrom, defOrUse, uncertain) and
+  adjacentDefRead(defOrUse, use) and
+  useToNode(use, nodeTo) and
+  nodeFrom != nodeTo
+}
+
 private predicate ssaFlowImpl(SsaDefOrUse defOrUse, Node nodeFrom, Node nodeTo, boolean uncertain) {
   exists(UseOrPhi use |
-    nodeToDefOrUse(nodeFrom, defOrUse, uncertain) and
-    adjacentDefRead(defOrUse, use) and
-    useToNode(use, nodeTo) and
-    nodeFrom != nodeTo
+    localSsaFlow(defOrUse, nodeFrom, use, nodeTo, uncertain)
     or
     // Initial global variable value to a first use
     nodeFrom.(InitialGlobalValue).getGlobalDef() = defOrUse and
@@ -729,19 +703,99 @@ predicate ssaFlow(Node nodeFrom, Node nodeTo) {
   )
 }
 
-private predicate isArgumentOfCallable(DataFlowCall call, ArgumentNode arg) {
-  arg.argumentOf(call, _)
+private predicate isArgumentOfCallableInstruction(DataFlowCall call, Instruction instr) {
+  isArgumentOfCallableOperand(call, unique( | | getAUse(instr)))
 }
 
-/** Holds if there is def-use or use-use flow from `pun` to `nodeTo`. */
-predicate postUpdateFlow(PostUpdateNode pun, Node nodeTo) {
-  exists(UseOrPhi use, Node preUpdate |
+private predicate isArgumentOfCallableOperand(DataFlowCall call, Operand operand) {
+  operand.(ArgumentOperand).getCall() = call
+  or
+  exists(FieldAddressInstruction fai |
+    fai.getObjectAddressOperand() = operand and
+    isArgumentOfCallableInstruction(call, fai)
+  )
+  or
+  exists(Instruction deref |
+    isArgumentOfCallableInstruction(call, deref) and
+    isDereference(deref, operand, _)
+  )
+  or
+  exists(Instruction instr |
+    isArgumentOfCallableInstruction(call, instr) and
+    conversionFlow(operand, instr, _, _)
+  )
+}
+
+private predicate isArgumentOfCallable(DataFlowCall call, Node n) {
+  isArgumentOfCallableOperand(call, n.asOperand())
+  or
+  exists(Operand op |
+    n.(IndirectOperand).hasOperandAndIndirectionIndex(op, _) and
+    isArgumentOfCallableOperand(call, op)
+  )
+  or
+  exists(Instruction instr |
+    n.(IndirectInstruction).hasInstructionAndIndirectionIndex(instr, _) and
+    isArgumentOfCallableInstruction(call, instr)
+  )
+}
+
+/**
+ * Holds if there is use-use flow from `pun`'s pre-update node to `n`.
+ */
+private predicate postUpdateNodeToFirstUse(PostUpdateNode pun, Node n) {
+  exists(UseOrPhi use |
     adjustForPointerArith(pun, use) and
-    useToNode(use, nodeTo) and
+    useToNode(use, n)
+  )
+}
+
+private predicate stepUntilNotInCall(DataFlowCall call, Node n1, Node n2) {
+  isArgumentOfCallable(call, n1) and
+  exists(Node mid | localSsaFlow(_, n1, _, mid, _) |
+    isArgumentOfCallable(call, mid) and
+    stepUntilNotInCall(call, mid, n2)
+    or
+    not isArgumentOfCallable(call, mid) and
+    mid = n2
+  )
+}
+
+bindingset[n1, n2]
+pragma[inline_late]
+private predicate isArgumentOfSameCall(DataFlowCall call, Node n1, Node n2) {
+  isArgumentOfCallable(call, n1) and isArgumentOfCallable(call, n2)
+}
+
+/**
+ * Holds if there is def-use or use-use flow from `pun` to `nodeTo`.
+ *
+ * Note: This is more complex than it sounds. Consider a call such as:
+ * ```cpp
+ * write_first_argument(x, x);
+ * sink(x);
+ * ```
+ * Assume flow comes out of the first argument to `write_first_argument`. We
+ * don't want flow to go to the `x` that's also an argument to
+ * `write_first_argument` (because we just flowed out of that function, and we
+ * don't want to flow back into it again).
+ *
+ * We do, however, want flow from the output argument to `x` on the next line, and
+ * similarly we want flow from the second argument of `write_first_argument` to `x`
+ * on the next line.
+ */
+predicate postUpdateFlow(PostUpdateNode pun, Node nodeTo) {
+  exists(Node preUpdate, Node mid |
     preUpdate = pun.getPreUpdateNode() and
-    not exists(DataFlowCall call |
-      isArgumentOfCallable(call, preUpdate) and isArgumentOfCallable(call, nodeTo)
+    postUpdateNodeToFirstUse(pun, mid)
+  |
+    exists(DataFlowCall call |
+      isArgumentOfSameCall(call, preUpdate, mid) and
+      stepUntilNotInCall(call, mid, nodeTo)
     )
+    or
+    not isArgumentOfSameCall(_, preUpdate, mid) and
+    nodeTo = mid
   )
 }
 
@@ -771,7 +825,7 @@ predicate fromPhiNode(SsaPhiNode nodeFrom, Node nodeTo) {
     or
     exists(PhiNode phiTo |
       phi != phiTo and
-      lastRefRedefExt(phi, _, _, phiTo) and
+      lastRefRedefExt(phi, bb1, i1, phiTo) and
       nodeTo.(SsaPhiNode).getPhiNode() = phiTo
     )
   )

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

@@ -6,6 +6,7 @@ private import DataFlowImplCommon as DataFlowImplCommon
 private import DataFlowUtil
 private import semmle.code.cpp.models.interfaces.PointerWrapper
 private import DataFlowPrivate
+private import semmle.code.cpp.ir.ValueNumbering
 
 /**
  * Holds if `operand` is an operand that is not used by the dataflow library.
@@ -146,14 +147,6 @@ int countIndirectionsForCppType(LanguageType langType) {
   )
 }
 
-/**
- * A `CallInstruction` that calls an allocation function such
- * as `malloc` or `operator new`.
- */
-class AllocationInstruction extends CallInstruction {
-  AllocationInstruction() { this.getStaticCallTarget() instanceof Cpp::AllocationFunction }
-}
-
 private predicate isIndirectionType(Type t) { t instanceof Indirection }
 
 private predicate hasUnspecifiedBaseType(Indirection t, Type base) {
@@ -327,10 +320,20 @@ private module IteratorIndirections {
   }
 }
 
-predicate isDereference(Instruction deref, Operand address) {
-  any(Indirection ind).isAdditionalDereference(deref, address)
+/**
+ * Holds if `deref` is the result of loading the value at the address
+ * represented by `address`.
+ *
+ * If `additional = true` then the dereference comes from an `Indirection`
+ * class (such as a call to an iterator's `operator*`), and if
+ * `additional = false` the dereference is a `LoadInstruction`.
+ */
+predicate isDereference(Instruction deref, Operand address, boolean additional) {
+  any(Indirection ind).isAdditionalDereference(deref, address) and
+  additional = true
   or
-  deref.(LoadInstruction).getSourceAddressOperand() = address
+  deref.(LoadInstruction).getSourceAddressOperand() = address and
+  additional = false
 }
 
 predicate isWrite(Node0Impl value, Operand address, boolean certain) {
@@ -368,17 +371,22 @@ newtype TBaseSourceVariable =
   // Each IR variable gets its own source variable
   TBaseIRVariable(IRVariable var) or
   // Each allocation gets its own source variable
-  TBaseCallVariable(AllocationInstruction call)
+  TBaseCallVariable(CallInstruction call) { not call.getResultIRType() instanceof IRVoidType }
 
-abstract class BaseSourceVariable extends TBaseSourceVariable {
+abstract private class AbstractBaseSourceVariable extends TBaseSourceVariable {
   /** Gets a textual representation of this element. */
   abstract string toString();
 
   /** Gets the type of this base source variable. */
-  abstract DataFlowType getType();
+  final DataFlowType getType() { this.getLanguageType().hasUnspecifiedType(result, _) }
+
+  /** Gets the `CppType` of this base source variable. */
+  abstract CppType getLanguageType();
 }
 
-class BaseIRVariable extends BaseSourceVariable, TBaseIRVariable {
+final class BaseSourceVariable = AbstractBaseSourceVariable;
+
+class BaseIRVariable extends AbstractBaseSourceVariable, TBaseIRVariable {
   IRVariable var;
 
   IRVariable getIRVariable() { result = var }
@@ -387,19 +395,19 @@ class BaseIRVariable extends BaseSourceVariable, TBaseIRVariable {
 
   override string toString() { result = var.toString() }
 
-  override DataFlowType getType() { result = var.getType() }
+  override CppType getLanguageType() { result = var.getLanguageType() }
 }
 
-class BaseCallVariable extends BaseSourceVariable, TBaseCallVariable {
-  AllocationInstruction call;
+class BaseCallVariable extends AbstractBaseSourceVariable, TBaseCallVariable {
+  CallInstruction call;
 
   BaseCallVariable() { this = TBaseCallVariable(call) }
 
-  AllocationInstruction getCallInstruction() { result = call }
+  CallInstruction getCallInstruction() { result = call }
 
   override string toString() { result = call.toString() }
 
-  override DataFlowType getType() { result = call.getResultType() }
+  override CppType getLanguageType() { result = getResultLanguageType(call) }
 }
 
 /**
@@ -499,8 +507,7 @@ private class BaseIRVariableInstruction extends BaseSourceVariableInstruction,
   override BaseIRVariable getBaseSourceVariable() { result.getIRVariable() = this.getIRVariable() }
 }
 
-private class BaseAllocationInstruction extends BaseSourceVariableInstruction, AllocationInstruction
-{
+private class BaseCallInstruction extends BaseSourceVariableInstruction, CallInstruction {
   override BaseCallVariable getBaseSourceVariable() { result.getCallInstruction() = this }
 }
 
@@ -548,7 +555,7 @@ private module Cached {
       isDef(_, value, iteratorDerefAddress, iteratorBase, numberOfLoads + 2, 0) and
       isUse(_, iteratorAddress, iteratorBase, numberOfLoads + 1, 0) and
       iteratorBase.getResultType() instanceof Interfaces::Iterator and
-      isDereference(iteratorAddress.getDef(), read.getArgumentDef().getAUse()) and
+      isDereference(iteratorAddress.getDef(), read.getArgumentDef().getAUse(), _) and
       memory = read.getSideEffectOperand().getAnyDef()
     )
   }
@@ -784,11 +791,14 @@ private module Cached {
    * instead associated with the operand returned by this predicate.
    */
   cached
-  Operand getIRRepresentationOfIndirectOperand(Operand operand, int indirectionIndex) {
+  predicate hasIRRepresentationOfIndirectOperand(
+    Operand operand, int indirectionIndex, Operand operandRepr, int indirectionIndexRepr
+  ) {
+    indirectionIndex = [1 .. countIndirectionsForCppType(getLanguageType(operand))] and
     exists(Instruction load |
-      isDereference(load, operand) and
-      result = unique( | | getAUse(load)) and
-      isUseImpl(operand, _, indirectionIndex - 1)
+      isDereference(load, operand, false) and
+      operandRepr = unique( | | getAUse(load)) and
+      indirectionIndexRepr = indirectionIndex - 1
     )
   }
 
@@ -800,12 +810,15 @@ private module Cached {
    * instead associated with the instruction returned by this predicate.
    */
   cached
-  Instruction getIRRepresentationOfIndirectInstruction(Instruction instr, int indirectionIndex) {
+  predicate hasIRRepresentationOfIndirectInstruction(
+    Instruction instr, int indirectionIndex, Instruction instrRepr, int indirectionIndexRepr
+  ) {
+    indirectionIndex = [1 .. countIndirectionsForCppType(getResultLanguageType(instr))] and
     exists(Instruction load, Operand address |
       address.getDef() = instr and
-      isDereference(load, address) and
-      isUseImpl(address, _, indirectionIndex - 1) and
-      result = load
+      isDereference(load, address, false) and
+      instrRepr = load and
+      indirectionIndexRepr = indirectionIndex - 1
     )
   }
 
@@ -826,7 +839,7 @@ private module Cached {
     or
     exists(int ind0 |
       exists(Operand address |
-        isDereference(operand.getDef(), address) and
+        isDereference(operand.getDef(), address, _) and
         isUseImpl(address, base, ind0)
       )
       or
@@ -868,7 +881,7 @@ private module Cached {
    * to a specific address.
    */
   private predicate isCertainAddress(Operand operand) {
-    operand.getDef() instanceof VariableAddressInstruction
+    valueNumberOfOperand(operand).getAnInstruction() instanceof VariableAddressInstruction
     or
     operand.getType() instanceof Cpp::ReferenceType
   }
@@ -896,7 +909,7 @@ private module Cached {
     )
     or
     exists(Operand address, boolean certain0 |
-      isDereference(operand.getDef(), address) and
+      isDereference(operand.getDef(), address, _) and
       isDefImpl(address, base, ind - 1, certain0)
     |
       if isCertainAddress(operand) then certain = certain0 else certain = false

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

@@ -0,0 +1,10 @@
+/**
+ * Provides C++-specific definitions for use in the taint tracking library.
+ */
+
+private import codeql.dataflow.TaintTracking
+private import DataFlowImplSpecific
+
+module CppTaintTracking implements InputSig<CppDataFlow> {
+  import TaintTrackingUtil
+}

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

@@ -57,7 +57,7 @@ private predicate operandToInstructionTaintStep(Operand opFrom, Instruction inst
   )
   or
   // Taint flow from an address to its dereference.
-  Ssa::isDereference(instrTo, opFrom)
+  Ssa::isDereference(instrTo, opFrom, _)
   or
   // Unary instructions tend to preserve enough information in practice that we
   // want taint to flow through.
@@ -112,7 +112,7 @@ predicate defaultAdditionalTaintStep(DataFlow::Node src, DataFlow::Node sink) {
  * of `c` at sinks and inputs to additional taint steps.
  */
 bindingset[node]
-predicate defaultImplicitTaintRead(DataFlow::Node node, DataFlow::Content c) { none() }
+predicate defaultImplicitTaintRead(DataFlow::Node node, DataFlow::ContentSet c) { none() }
 
 /**
  * Holds if `node` should be a sanitizer in all global taint flow configurations