Make java/string-concatenation-in-loop high precision

Go: Make models-as-data package column match any version without "$ANYVERSION"

.bazelrc

@@ -10,8 +10,19 @@ common --override_module=semmle_code=%workspace%/misc/bazel/semmle_code_stub
 
 build --repo_env=CC=clang --repo_env=CXX=clang++
 
-build:linux --cxxopt=-std=c++20
-build:macos --cxxopt=-std=c++20 --cpu=darwin_x86_64
-build:windows --cxxopt=/std:c++20 --cxxopt=/Zc:preprocessor
+# we use transitions that break builds of `...`, so for `test` to work with that we need the following
+test --build_tests_only
+
+# this requires developer mode, but is required to have pack installer functioning
+startup --windows_enable_symlinks
+common --enable_runfiles
+
+# with the above, we can avoid building python zips which is the default on windows as that's expensive
+build --nobuild_python_zip
+
+common --registry=file:///%workspace%/misc/bazel/registry
+common --registry=https://bcr.bazel.build
+
+common --@rules_dotnet//dotnet/settings:strict_deps=false
 
 try-import %workspace%/local.bazelrc

.bazelrc.internal

@@ -0,0 +1,10 @@
+# this file should contain bazel settings required to build things from `semmle-code`
+
+common --registry=file:///%workspace%/ql/misc/bazel/registry
+common --registry=https://bcr.bazel.build
+
+# See bazelbuild/rules_dotnet#413: strict_deps in C# also appliy to 3rd-party deps, and when we pull
+# in (for example) the xunit package, there's no code in this at all, it just depends transitively on
+# its implementation packages without providing any code itself.
+# We either can depend on internal implementation details, or turn of strict deps.
+common --@rules_dotnet//dotnet/settings:strict_deps=false

.bazelversion

@@ -1 +1 @@
-7.1.0
+7.2.1

.gitattributes

@@ -50,31 +50,40 @@
 *.dll -text
 *.pdb -text
 
-java/ql/test/stubs/**/*.java linguist-generated=true
-java/ql/test/experimental/stubs/**/*.java linguist-generated=true
+/java/ql/test/stubs/**/*.java linguist-generated=true
+/java/ql/test/experimental/stubs/**/*.java linguist-generated=true
+/java/kotlin-extractor/deps/*.jar filter=lfs diff=lfs merge=lfs -text
 
 # Force git not to modify line endings for go or html files under the go/ql directory
-go/ql/**/*.go -text
-go/ql/**/*.html -text
+/go/ql/**/*.go -text
+/go/ql/**/*.html -text
 # Force git not to modify line endings for go dbschemes
-go/*.dbscheme -text
+/go/*.dbscheme -text
 # Preserve unusual line ending from codeql-go merge
-go/extractor/opencsv/CSVReader.java -text
+/go/extractor/opencsv/CSVReader.java -text
 
 # For some languages, upgrade script testing references really old dbscheme
 # files from legacy upgrades that have CRLF line endings. Since upgrade
 # resolution relies on object hashes, we must suppress line ending conversion
 # for those testing dbscheme files.
-*/ql/lib/upgrades/initial/*.dbscheme -text
-
-# Generated test files - these are synced from the standard JavaScript libraries using
-# `javascript/ql/experimental/adaptivethreatmodeling/test/update_endpoint_test_files.py`.
-javascript/ql/experimental/adaptivethreatmodeling/test/endpoint_large_scale/autogenerated/**/*.js linguist-generated=true -merge
-javascript/ql/experimental/adaptivethreatmodeling/test/endpoint_large_scale/autogenerated/**/*.ts linguist-generated=true -merge
+/*/ql/lib/upgrades/initial/*.dbscheme -text
 
 # Auto-generated modeling for Python
-python/ql/lib/semmle/python/frameworks/data/internal/subclass-capture/*.yml linguist-generated=true
+/python/ql/lib/semmle/python/frameworks/data/internal/subclass-capture/*.yml linguist-generated=true
 
 # auto-generated bazel lock file
-ruby/extractor/cargo-bazel-lock.json linguist-generated=true
-ruby/extractor/cargo-bazel-lock.json -merge
+/ruby/extractor/cargo-bazel-lock.json linguist-generated=true
+/ruby/extractor/cargo-bazel-lock.json -merge
+
+# auto-generated files for the C# build
+/csharp/paket.lock linguist-generated=true
+# needs eol=crlf, as `paket` touches this file and saves it as crlf
+/csharp/.paket/Paket.Restore.targets linguist-generated=true eol=crlf
+/csharp/paket.main.bzl linguist-generated=true
+/csharp/paket.main_extension.bzl linguist-generated=true
+
+# ripunzip tool
+/misc/ripunzip/ripunzip-* filter=lfs diff=lfs merge=lfs -text
+
+# swift prebuilt resources
+/swift/third_party/resource-dir/*.zip filter=lfs diff=lfs merge=lfs -text

.github/labeler.yml

@@ -15,7 +15,7 @@ Java:
   - change-notes/**/*java.*
 
 JS:
-  - any: [ 'javascript/**/*', '!javascript/ql/experimental/adaptivethreatmodeling/**/*' ]
+  - any: [ 'javascript/**/*' ]
   - change-notes/**/*javascript*
 
 Kotlin:
@@ -46,6 +46,3 @@ documentation:
 # Since these are all shared files that need to be synced, just pick _one_ copy of each.
 "DataFlow Library":
   - "shared/dataflow/**/*"
-
-"ATM":
-  - javascript/ql/experimental/adaptivethreatmodeling/**/*

.github/workflows/build-ripunzip.yml

@@ -0,0 +1,74 @@
+name: Build runzip
+
+on:
+  workflow_dispatch:
+    inputs:
+      ripunzip-version:
+        description: "what reference to checktout from google/runzip"
+        required: false
+        default: v1.2.1
+      openssl-version:
+        description: "what reference to checkout from openssl/openssl for Linux"
+        required: false
+        default: openssl-3.3.0
+
+jobs:
+  build:
+    strategy:
+      fail-fast: false
+      matrix:
+        os: [ubuntu-20.04, macos-12, windows-2019]
+    runs-on: ${{ matrix.os }}
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          repository: google/ripunzip
+          ref: ${{ inputs.ripunzip-version }}
+      # we need to avoid ripunzip dynamically linking into libssl
+      # see https://github.com/sfackler/rust-openssl/issues/183
+      - if: runner.os == 'Linux'
+        name: checkout openssl
+        uses: actions/checkout@v4
+        with:
+          repository: openssl/openssl
+          path: openssl
+          ref: ${{ inputs.openssl-version }}
+      - if: runner.os == 'Linux'
+        name: build and install openssl with fPIC
+        shell: bash
+        working-directory: openssl
+        run: |
+          ./config -fPIC --prefix=$HOME/.local --openssldir=$HOME/.local/ssl
+          make -j $(nproc)
+          make install_sw -j $(nproc)
+      - if: runner.os == 'Linux'
+        name: build (linux)
+        shell: bash
+        run: |
+          env OPENSSL_LIB_DIR=$HOME/.local/lib64 OPENSSL_INCLUDE_DIR=$HOME/.local/include OPENSSL_STATIC=yes cargo build --release
+          mv target/release/ripunzip ripunzip-linux
+      - if: runner.os == 'Windows'
+        name: build (windows)
+        shell: bash
+        run: |
+          cargo build --release
+          mv target/release/ripunzip ripunzip-windows
+      - name: build (macOS)
+        if: runner.os == 'macOS'
+        shell: bash
+        run: |
+          rustup target install x86_64-apple-darwin
+          rustup target install aarch64-apple-darwin
+          cargo build --target x86_64-apple-darwin --release
+          cargo build --target aarch64-apple-darwin --release
+          lipo -create -output ripunzip-macos \
+            -arch x86_64 target/x86_64-apple-darwin/release/ripunzip \
+            -arch arm64 target/aarch64-apple-darwin/release/ripunzip
+      - uses: actions/upload-artifact@v4
+        with:
+          name: ripunzip-${{ runner.os }}
+          path: ripunzip-*
+      - name: Check built binary
+        shell: bash
+        run: |
+          ./ripunzip-* --version

.github/workflows/buildifier.yml

@@ -0,0 +1,28 @@
+name: Check bazel formatting
+
+on:
+  pull_request:
+    paths:
+      - "**.bazel"
+      - "**.bzl"
+    branches:
+      - main
+      - "rc/*"
+
+permissions:
+  contents: read
+
+jobs:
+  check:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v4
+      - name: Check bazel formatting
+        uses: pre-commit/action@646c83fcd040023954eafda54b4db0192ce70507
+        with:
+          extra_args: >
+            buildifier --all-files 2>&1 ||
+            (
+              echo -e "In order to format all bazel files, please run:\n  bazel run //misc/bazel:buildifier"; exit 1
+            )

.github/workflows/codeql-analysis.yml

@@ -56,7 +56,9 @@ jobs:
     #    uses a compiled language
 
     - run: |
-       dotnet build csharp
+       cd csharp
+       dotnet tool restore
+       dotnet build .
 
     - name: Perform CodeQL Analysis
       uses: github/codeql-action/analyze@main

.github/workflows/csharp-qltest.yml

@@ -65,7 +65,7 @@ jobs:
           key: csharp-qltest-${{ matrix.slice }}
       - name: Run QL tests
         run: |
-          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 }}"
+          codeql test run --threads=0 --ram 50000 --slice ${{ matrix.slice }} --search-path "${{ github.workspace }}" --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: ${{ github.token }}
   unit-tests:
@@ -81,10 +81,11 @@ jobs:
           dotnet-version: 8.0.101
       - name: Extractor unit tests
         run: |
+          dotnet tool restore
           dotnet test -p:RuntimeFrameworkVersion=8.0.1 extractor/Semmle.Util.Tests
           dotnet test -p:RuntimeFrameworkVersion=8.0.1 extractor/Semmle.Extraction.Tests
           dotnet test -p:RuntimeFrameworkVersion=8.0.1 autobuilder/Semmle.Autobuild.CSharp.Tests
-          dotnet test -p:RuntimeFrameworkVersion=8.0.1 "${{ github.workspace }}/cpp/autobuilder/Semmle.Autobuild.Cpp.Tests"
+          dotnet test -p:RuntimeFrameworkVersion=8.0.1 autobuilder/Semmle.Autobuild.Cpp.Tests
         shell: bash
   stubgentest:
     runs-on: ubuntu-latest
@@ -100,6 +101,6 @@ jobs:
           # Update existing stubs in the repo with the freshly generated ones
           mv "$STUBS_PATH/output/stubs/_frameworks" ql/test/resources/stubs/
           git status
-          codeql test run --threads=0 --search-path extractor-pack --check-databases --check-undefined-labels --check-repeated-labels --check-redefined-labels --consistency-queries ql/consistency-queries -- ql/test/library-tests/dataflow/flowsources/aspremote
+          codeql test run --threads=0 --search-path "${{ github.workspace }}" --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/go-tests-other-os.yml

@@ -7,8 +7,9 @@ on:
       - .github/workflows/go-tests-other-os.yml
       - .github/actions/**
       - codeql-workspace.yml
-env:
-  GO_VERSION: '~1.22.0'
+      - MODULE.bazel
+      - .bazelrc
+      - misc/bazel/**
 
 permissions:
   contents: read
@@ -18,72 +19,17 @@ jobs:
     name: Test MacOS
     runs-on: macos-latest
     steps:
-      - name: Set up Go ${{ env.GO_VERSION }}
-        uses: actions/setup-go@v5
-        with:
-          go-version: ${{ env.GO_VERSION }}
-          cache: false
-        id: go
-
       - name: Check out code
         uses: actions/checkout@v4
-
-      - name: Set up CodeQL CLI
-        uses: ./.github/actions/fetch-codeql
-
-      - name: Enable problem matchers in repository
-        shell: bash
-        run: 'find .github/problem-matchers -name \*.json -exec echo "::add-matcher::{}" \;'
-
-      - name: Build
-        run: |
-          cd go
-          make
-
-      - name: Cache compilation cache
-        id: query-cache
-        uses: ./.github/actions/cache-query-compilation
-        with:
-          key: go-qltest
-      - name: Test
-        run: |
-          cd go
-          make test cache="${{ steps.query-cache.outputs.cache-dir }}"
+      - name: Run tests
+        uses: ./go/actions/test
 
   test-win:
     if: github.repository_owner == 'github'
     name: Test Windows
     runs-on: windows-latest-xl
     steps:
-      - name: Set up Go ${{ env.GO_VERSION }}
-        uses: actions/setup-go@v5
-        with:
-          go-version: ${{ env.GO_VERSION }}
-          cache: false
-        id: go
-
       - name: Check out code
         uses: actions/checkout@v4
-
-      - name: Set up CodeQL CLI
-        uses: ./.github/actions/fetch-codeql
-
-      - name: Enable problem matchers in repository
-        shell: bash
-        run: 'find .github/problem-matchers -name \*.json -exec echo "::add-matcher::{}" \;'
-
-      - name: Build
-        run: |
-          cd go
-          make
-
-      - name: Cache compilation cache
-        id: query-cache
-        uses: ./.github/actions/cache-query-compilation
-        with:
-          key: go-qltest
-
-      - name: Test
-        run: |
-          cd go
-          make test cache="${{ steps.query-cache.outputs.cache-dir }}"
+      - name: Run tests
+        uses: ./go/actions/test

.github/workflows/go-tests.yml

@@ -15,9 +15,9 @@ on:
       - .github/workflows/go-tests.yml
       - .github/actions/**
       - codeql-workspace.yml
-
-env:
-  GO_VERSION: '~1.22.0'
+      - MODULE.bazel
+      - .bazelrc
+      - misc/bazel/**
 
 permissions:
   contents: read
@@ -28,51 +28,9 @@ jobs:
     name: Test Linux (Ubuntu)
     runs-on: ubuntu-latest-xl
     steps:
-      - name: Set up Go ${{ env.GO_VERSION }}
-        uses: actions/setup-go@v5
-        with:
-          go-version: ${{ env.GO_VERSION }}
-          cache: false
-        id: go
-
       - name: Check out code
         uses: actions/checkout@v4
-
-      - name: Set up CodeQL CLI
-        uses: ./.github/actions/fetch-codeql
-
-      - name: Enable problem matchers in repository
-        shell: bash
-        run: 'find .github/problem-matchers -name \*.json -exec echo "::add-matcher::{}" \;'
-
-      - name: Build
-        run: |
-          cd go
-          make
-
-      - name: Check that all Go code is autoformatted
-        run: |
-          cd go
-          make check-formatting
-
-      - name: Compile qhelp files to markdown
-        run: |
-          cd go
-          env QHELP_OUT_DIR=qhelp-out make qhelp-to-markdown
-
-      - name: Upload qhelp markdown
-        uses: actions/upload-artifact@v3
+      - name: Run tests
+        uses: ./go/actions/test
         with:
-          name: qhelp-markdown
-          path: go/qhelp-out/**/*.md
-
-      - name: Cache compilation cache
-        id: query-cache
-        uses: ./.github/actions/cache-query-compilation
-        with:
-          key: go-qltest
-
-      - name: Test
-        run: |
-          cd go
-          make test cache="${{ steps.query-cache.outputs.cache-dir }}"
+          run-code-checks: true

.github/workflows/kotlin-build.yml

@@ -0,0 +1,28 @@
+name: "Kotlin Build"
+
+on:
+  pull_request:
+    paths:
+      - "java/kotlin-extractor/**"
+      - "misc/bazel/**"
+      - "misc/codegen/**"
+      - "*.bazel*"
+      - .github/workflows/kotlin-build.yml
+    branches:
+      - main
+      - rc/*
+      - codeql-cli-*
+
+permissions:
+  contents: read
+
+jobs:
+  build:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - run: |
+          bazel query //java/kotlin-extractor/...
+          # only build the default version as a quick check that we can build from `codeql`
+          # the full official build will be checked by QLucie
+          bazel build //java/kotlin-extractor

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

@@ -49,20 +49,20 @@ jobs:
           key: ${{ runner.os }}-${{ steps.os_version.outputs.version }}-rust-cargo-${{ hashFiles('ql/**/Cargo.lock') }}
       - name: Release build
         if: steps.cache-extractor.outputs.cache-hit != 'true'
-        run: cd ql; ./scripts/create-extractor-pack.sh       
+        run: cd ql; ./scripts/create-extractor-pack.sh
         env:
-          GH_TOKEN: ${{ github.token }}   
+          GH_TOKEN: ${{ github.token }}
       - name: Cache compilation cache
         id: query-cache
         uses: ./.github/actions/cache-query-compilation
-        with: 
+        with:
           key: run-ql-for-ql
       - name: Make database and analyze
         run: |
           ./ql/target/release/buramu | tee deprecated.blame # Add a blame file for the extractor to parse.
-          ${CODEQL} database create -l=ql --search-path ql/extractor-pack ${DB}
+          ${CODEQL} database create -l=ql ${DB} --search-path "${{ github.workspace }}"
           ${CODEQL} database analyze -j0 --format=sarif-latest --output=ql-for-ql.sarif ${DB} ql/ql/src/codeql-suites/ql-code-scanning.qls  --compilation-cache "${{ steps.query-cache.outputs.cache-dir }}"
-        env: 
+        env:
           CODEQL: ${{ steps.find-codeql.outputs.codeql-path }}
           DB: ${{ runner.temp }}/DB
           LGTM_INDEX_FILTERS: |

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

@@ -53,8 +53,8 @@ jobs:
       - name: Create database
         run: |
           "${CODEQL}" database create \
-            --search-path "ql/extractor-pack" \
-            --threads 4 \
+          --search-path "${{ github.workspace }}"
+          --threads 4 \
             --language ql --source-root "${{ github.workspace }}/repo" \
             "${{ runner.temp }}/database"
         env:

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

@@ -49,15 +49,15 @@ jobs:
       - name: Cache compilation cache
         id: query-cache
         uses: ./.github/actions/cache-query-compilation
-        with: 
+        with:
           key: ql-for-ql-tests
       - name: Run QL tests
         run: |
-          "${CODEQL}" test run --check-databases --check-unused-labels --check-repeated-labels --check-redefined-labels --check-use-before-definition --search-path "${{ github.workspace }}/ql/extractor-pack" --consistency-queries ql/ql/consistency-queries --compilation-cache "${{ steps.query-cache.outputs.cache-dir }}" ql/ql/test
+          "${CODEQL}" test run --check-databases --check-unused-labels --check-repeated-labels --check-redefined-labels --check-use-before-definition --search-path "${{ github.workspace }}" --consistency-queries ql/ql/consistency-queries --compilation-cache "${{ steps.query-cache.outputs.cache-dir }}" ql/ql/test
         env:
           CODEQL: ${{ steps.find-codeql.outputs.codeql-path }}
 
-  other-os: 
+  other-os:
     strategy:
       matrix:
         os: [macos-latest, windows-latest]
@@ -65,7 +65,7 @@ jobs:
     runs-on: ${{ matrix.os }}
     steps:
       - uses: actions/checkout@v4
-      - name: Install GNU tar 
+      - name: Install GNU tar
         if: runner.os == 'macOS'
         run: |
           brew install gnu-tar
@@ -100,7 +100,7 @@ jobs:
       - name: Run a single QL tests - Unix
         if: runner.os != 'Windows'
         run: |
-          "${CODEQL}" test run --check-databases --search-path "${{ github.workspace }}/ql/extractor-pack" ql/ql/test/queries/style/DeadCode/DeadCode.qlref
+          "${CODEQL}" test run --check-databases --search-path "${{ github.workspace }}" ql/ql/test/queries/style/DeadCode/DeadCode.qlref
         env:
           CODEQL: ${{ steps.find-codeql.outputs.codeql-path }}
       - name: Run a single QL tests - Windows
@@ -108,5 +108,4 @@ jobs:
         shell: pwsh
         run: |
           $Env:PATH += ";$(dirname ${{ steps.find-codeql.outputs.codeql-path }})"
-          codeql test run --check-databases --search-path "${{ github.workspace }}/ql/extractor-pack" ql/ql/test/queries/style/DeadCode/DeadCode.qlref
-      
+          codeql test run --check-databases --search-path "${{ github.workspace }}" ql/ql/test/queries/style/DeadCode/DeadCode.qlref

.github/workflows/ruby-build.yml

@@ -7,6 +7,7 @@ on:
       - .github/workflows/ruby-build.yml
       - .github/actions/fetch-codeql/action.yml
       - codeql-workspace.yml
+      - "shared/tree-sitter-extractor/**"
     branches:
       - main
       - "rc/*"
@@ -16,6 +17,7 @@ on:
       - .github/workflows/ruby-build.yml
       - .github/actions/fetch-codeql/action.yml
       - codeql-workspace.yml
+      - "shared/tree-sitter-extractor/**"
     branches:
       - main
       - "rc/*"

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

@@ -44,7 +44,7 @@ jobs:
       - name: Create database
         run: |
           codeql database create \
-            --search-path "${{ github.workspace }}/ruby/extractor-pack" \
+            --search-path "${{ github.workspace }}" \
             --threads 4 \
             --language ruby --source-root "${{ github.workspace }}/repo" \
             "${{ runner.temp }}/database"

.github/workflows/ruby-qltest.yml

@@ -64,10 +64,10 @@ jobs:
       - name: Cache compilation cache
         id: query-cache
         uses: ./.github/actions/cache-query-compilation
-        with: 
+        with:
           key: ruby-qltest
       - name: Run QL tests
         run: |
-          codeql test run --threads=0 --ram 50000 --search-path "${{ github.workspace }}/ruby/extractor-pack" --check-databases --check-undefined-labels --check-unused-labels --check-repeated-labels --check-redefined-labels --check-use-before-definition --consistency-queries ql/consistency-queries ql/test --compilation-cache "${{ steps.query-cache.outputs.cache-dir }}"
+          codeql test run --threads=0 --ram 50000 --search-path "${{ github.workspace }}" --check-databases --check-undefined-labels --check-unused-labels --check-repeated-labels --check-redefined-labels --check-use-before-definition --consistency-queries ql/consistency-queries ql/test --compilation-cache "${{ steps.query-cache.outputs.cache-dir }}"
         env:
           GITHUB_TOKEN: ${{ github.token }}

.github/workflows/swift.yml

@@ -6,6 +6,7 @@ on:
       - "swift/**"
       - "misc/bazel/**"
       - "misc/codegen/**"
+      - "shared/**"
       - "*.bazel*"
       - .github/workflows/swift.yml
       - .github/actions/**
@@ -22,10 +23,12 @@ on:
       - "swift/**"
       - "misc/bazel/**"
       - "misc/codegen/**"
+      - "shared/**"
       - "*.bazel*"
       - .github/workflows/swift.yml
       - .github/actions/**
       - codeql-workspace.yml
+      - .pre-commit-config.yaml
       - "!**/*.md"
       - "!**/*.qhelp"
     branches:
@@ -65,21 +68,6 @@ jobs:
     steps:
       - uses: actions/checkout@v4
       - uses: ./swift/actions/run-ql-tests
-  integration-tests-linux:
-    if: github.repository_owner == 'github'
-    needs: build-and-test-linux
-    runs-on: ubuntu-latest-xl
-    steps:
-      - uses: actions/checkout@v4
-      - uses: ./swift/actions/run-integration-tests
-  integration-tests-macos:
-    if: ${{ github.repository_owner == 'github' && github.event_name == 'pull_request' }}
-    needs: build-and-test-macos
-    runs-on: macos-12-xl
-    timeout-minutes: 60
-    steps:
-      - uses: actions/checkout@v4
-      - uses: ./swift/actions/run-integration-tests
   clang-format:
     if : ${{ github.event_name == 'pull_request' }}
     runs-on: ubuntu-latest

.github/workflows/zipmerge-test.yml

@@ -0,0 +1,23 @@
+name: "Test zipmerge code"
+
+on:
+  pull_request:
+    paths:
+      - "misc/bazel/internal/zipmerge/**"
+      - "MODULE.bazel"
+      - ".bazelrc*"
+    branches:
+      - main
+      - "rc/*"
+
+permissions:
+  contents: read
+
+jobs:
+  test:
+    runs-on: ubuntu-latest
+
+    steps:
+      - uses: actions/checkout@v4
+      - run: |
+          bazel test //misc/bazel/internal/zipmerge:test --test_output=all

.gitignore

@@ -62,3 +62,6 @@ node_modules/
 
 # Temporary folders for working with generated models
 .model-temp
+
+# bazel-built in-tree extractor packs
+/*/extractor-pack

.lfsconfig

@@ -0,0 +1,7 @@
+[lfs]
+# codeql is publicly forked by many users, and we don't want any LFS file polluting their working
+# copies. We therefore exclude everything by default.
+# For files required by bazel builds, use rules in `misc/bazel/lfs.bzl` to download them on demand.
+# we go for `fetchinclude` to something not exsiting rather than `fetchexclude = *` because the
+# former is easier to override (with `git -c` or a local git config) to fetch something specific
+fetchinclude = /nothing

.pre-commit-config.yaml

@@ -20,13 +20,23 @@ repos:
       - id: autopep8
         files: ^misc/codegen/.*\.py
 
-  - repo: https://github.com/warchant/pre-commit-buildifier
-    rev: 0.0.2
-    hooks:
-      - id: buildifier
-
   - repo: local
     hooks:
+      - id: buildifier
+        name: Format bazel files
+        files: \.(bazel|bzl)
+        language: system
+        entry: bazel run //misc/bazel:buildifier
+        pass_filenames: false
+
+#      DISABLED: can be enabled by copying this config and installing `pre-commit` with `--config` on the copy
+#      - id: go-gen
+#        name: Check checked in generated files in go
+#        files: ^go/.*
+#        language: system
+#        entry: bazel run //go:gen
+#        pass_filenames: false
+
       - id: codeql-format
         name: Fix QL file formatting
         files: \.qll?$

BUILD.bazel

@@ -0,0 +1 @@
+exports_files(["LICENSE"])

CODEOWNERS

@@ -1,6 +1,7 @@
 /cpp/ @github/codeql-c-analysis
-/cpp/autobuilder/ @github/codeql-c-extractor
 /csharp/ @github/codeql-csharp
+/csharp/autobuilder/Semmle.Autobuild.Cpp @github/codeql-c-extractor
+/csharp/autobuilder/Semmle.Autobuild.Cpp.Tests @github/codeql-c-extractor
 /go/ @github/codeql-go
 /java/ @github/codeql-java
 /javascript/ @github/codeql-javascript
@@ -12,9 +13,6 @@
 /java/ql/test-kotlin1/ @github/codeql-kotlin
 /java/ql/test-kotlin2/ @github/codeql-kotlin
 
-# ML-powered queries
-/javascript/ql/experimental/adaptivethreatmodeling/ @github/codeql-ml-powered-queries-reviewers
-
 # CodeQL tools and associated docs
 /docs/codeql/codeql-cli/ @github/codeql-cli-reviewers
 /docs/codeql/codeql-for-visual-studio-code/ @github/codeql-vscode-reviewers
@@ -37,9 +35,7 @@ MODULE.bazel @github/codeql-ci-reviewers
 
 # Workflows
 /.github/workflows/ @github/codeql-ci-reviewers
-/.github/workflows/atm-* @github/codeql-ml-powered-queries-reviewers
 /.github/workflows/go-* @github/codeql-go
-/.github/workflows/js-ml-tests.yml @github/codeql-ml-powered-queries-reviewers
 /.github/workflows/ql-for-ql-* @github/codeql-ql-for-ql-reviewers
 /.github/workflows/ruby-* @github/codeql-ruby
 /.github/workflows/swift.yml @github/codeql-swift

CONTRIBUTING.md

@@ -4,6 +4,8 @@ We welcome contributions to our CodeQL libraries and queries. Got an idea for a
 
 There is lots of useful documentation to help you write queries, ranging from information about query file structure to tutorials for specific target languages. For more information on the documentation available, see [CodeQL queries](https://codeql.github.com/docs/writing-codeql-queries/codeql-queries) on [codeql.github.com](https://codeql.github.com).
 
+Note that the CodeQL for Visual Studio Code documentation has been migrated to https://docs.github.com/en/code-security/codeql-for-vs-code/, but you can still contribute to it via a different repository. For more information, see [Contributing to GitHub Docs documentation](https://docs.github.com/en/contributing)."
+
 ## Change notes
 
 Any nontrivial user-visible change to a query pack or library pack should have a change note. For details on how to add a change note for your change, see [this guide](docs/change-notes.md).
@@ -43,7 +45,7 @@ If you have an idea for a query that you would like to share with other CodeQL u
 
 3. **Formatting**
 
-    - The queries and libraries must be autoformatted, for example using the "Format Document" command in [CodeQL for Visual Studio Code](https://codeql.github.com/docs/codeql-for-visual-studio-code/about-codeql-for-visual-studio-code).
+    - The queries and libraries must be autoformatted, for example using the "Format Document" command in [CodeQL for Visual Studio Code](https://docs.github.com/en/code-security/codeql-for-vs-code/).
 
     If you prefer, you can either:
     1. install the [pre-commit framework](https://pre-commit.com/) and install the configured hooks on this repo via `pre-commit install`, or

MODULE.bazel

@@ -13,14 +13,53 @@ local_path_override(
 
 # see https://registry.bazel.build/ for a list of available packages
 
-bazel_dep(name = "platforms", version = "0.0.8")
-bazel_dep(name = "rules_pkg", version = "0.9.1")
-bazel_dep(name = "rules_nodejs", version = "6.0.3")
-bazel_dep(name = "rules_python", version = "0.31.0")
-bazel_dep(name = "bazel_skylib", version = "1.5.0")
+bazel_dep(name = "platforms", version = "0.0.10")
+bazel_dep(name = "rules_go", version = "0.48.0")
+bazel_dep(name = "rules_pkg", version = "0.10.1")
+bazel_dep(name = "rules_nodejs", version = "6.2.0-codeql.1")
+bazel_dep(name = "rules_python", version = "0.32.2")
+bazel_dep(name = "bazel_skylib", version = "1.6.1")
 bazel_dep(name = "abseil-cpp", version = "20240116.0", repo_name = "absl")
 bazel_dep(name = "nlohmann_json", version = "3.11.3", repo_name = "json")
 bazel_dep(name = "fmt", version = "10.0.0")
+bazel_dep(name = "rules_kotlin", version = "1.9.4-codeql.1")
+bazel_dep(name = "gazelle", version = "0.37.0")
+bazel_dep(name = "rules_dotnet", version = "0.15.1")
+bazel_dep(name = "googletest", version = "1.14.0.bcr.1")
+bazel_dep(name = "rules_rust", version = "0.46.0")
+
+bazel_dep(name = "buildifier_prebuilt", version = "6.4.0", dev_dependency = True)
+
+crate = use_extension(
+    "@rules_rust//crate_universe:extension.bzl",
+    "crate",
+)
+crate.from_cargo(
+    name = "py_deps",
+    cargo_lockfile = "//python/extractor/tsg-python:Cargo.lock",
+    manifests = [
+        "//python/extractor/tsg-python:Cargo.toml",
+        "//python/extractor/tsg-python/tsp:Cargo.toml",
+    ],
+)
+crate.from_cargo(
+    name = "ruby_deps",
+    cargo_lockfile = "//ruby/extractor:Cargo.lock",
+    manifests = [
+        "//ruby/extractor:Cargo.toml",
+        "//ruby/extractor/codeql-extractor-fake-crate:Cargo.toml",
+    ],
+)
+use_repo(crate, "py_deps", "ruby_deps")
+
+dotnet = use_extension("@rules_dotnet//dotnet:extensions.bzl", "dotnet")
+dotnet.toolchain(dotnet_version = "8.0.101")
+use_repo(dotnet, "dotnet_toolchains")
+
+register_toolchains("@dotnet_toolchains//:all")
+
+csharp_main_extension = use_extension("//csharp:paket.main_extension.bzl", "main_extension")
+use_repo(csharp_main_extension, "paket.main")
 
 pip = use_extension("@rules_python//python/extensions:pip.bzl", "pip")
 pip.parse(
@@ -31,6 +70,8 @@ pip.parse(
 use_repo(pip, "codegen_deps")
 
 swift_deps = use_extension("//swift/third_party:load.bzl", "swift_deps")
+
+# following list can be kept in sync with `bazel mod tidy`
 use_repo(
     swift_deps,
     "binlog",
@@ -44,10 +85,92 @@ use_repo(
 node = use_extension("@rules_nodejs//nodejs:extensions.bzl", "node")
 node.toolchain(
     name = "nodejs",
+    node_urls = [
+        "https://nodejs.org/dist/v{version}/{filename}",
+        "https://mirrors.dotsrc.org/nodejs/release/v{version}/{filename}",
+    ],
     node_version = "18.15.0",
 )
 use_repo(node, "nodejs", "nodejs_toolchains")
 
+kotlin_extractor_deps = use_extension("//java/kotlin-extractor:deps.bzl", "kotlin_extractor_deps")
+
+# following list can be kept in sync by running `bazel mod tidy` in `codeql`
+use_repo(
+    kotlin_extractor_deps,
+    "codeql_kotlin_defaults",
+    "codeql_kotlin_embeddable",
+    "kotlin-compiler-1.5.0",
+    "kotlin-compiler-1.5.10",
+    "kotlin-compiler-1.5.20",
+    "kotlin-compiler-1.5.30",
+    "kotlin-compiler-1.6.0",
+    "kotlin-compiler-1.6.20",
+    "kotlin-compiler-1.7.0",
+    "kotlin-compiler-1.7.20",
+    "kotlin-compiler-1.8.0",
+    "kotlin-compiler-1.9.0-Beta",
+    "kotlin-compiler-1.9.20-Beta",
+    "kotlin-compiler-2.0.0-RC1",
+    "kotlin-compiler-embeddable-1.5.0",
+    "kotlin-compiler-embeddable-1.5.10",
+    "kotlin-compiler-embeddable-1.5.20",
+    "kotlin-compiler-embeddable-1.5.30",
+    "kotlin-compiler-embeddable-1.6.0",
+    "kotlin-compiler-embeddable-1.6.20",
+    "kotlin-compiler-embeddable-1.7.0",
+    "kotlin-compiler-embeddable-1.7.20",
+    "kotlin-compiler-embeddable-1.8.0",
+    "kotlin-compiler-embeddable-1.9.0-Beta",
+    "kotlin-compiler-embeddable-1.9.20-Beta",
+    "kotlin-compiler-embeddable-2.0.0-RC1",
+    "kotlin-stdlib-1.5.0",
+    "kotlin-stdlib-1.5.10",
+    "kotlin-stdlib-1.5.20",
+    "kotlin-stdlib-1.5.30",
+    "kotlin-stdlib-1.6.0",
+    "kotlin-stdlib-1.6.20",
+    "kotlin-stdlib-1.7.0",
+    "kotlin-stdlib-1.7.20",
+    "kotlin-stdlib-1.8.0",
+    "kotlin-stdlib-1.9.0-Beta",
+    "kotlin-stdlib-1.9.20-Beta",
+    "kotlin-stdlib-2.0.0-RC1",
+)
+
+go_sdk = use_extension("@rules_go//go:extensions.bzl", "go_sdk")
+go_sdk.download(version = "1.22.2")
+
+lfs_files = use_repo_rule("//misc/bazel:lfs.bzl", "lfs_files")
+
+lfs_files(
+    name = "ripunzip-linux",
+    srcs = ["//misc/ripunzip:ripunzip-linux"],
+    executable = True,
+)
+
+lfs_files(
+    name = "ripunzip-windows",
+    srcs = ["//misc/ripunzip:ripunzip-windows.exe"],
+    executable = True,
+)
+
+lfs_files(
+    name = "ripunzip-macos",
+    srcs = ["//misc/ripunzip:ripunzip-macos"],
+    executable = True,
+)
+
+lfs_files(
+    name = "swift-resource-dir-linux",
+    srcs = ["//swift/third_party/resource-dir:resource-dir-linux.zip"],
+)
+
+lfs_files(
+    name = "swift-resource-dir-macos",
+    srcs = ["//swift/third_party/resource-dir:resource-dir-macos.zip"],
+)
+
 register_toolchains(
     "@nodejs_toolchains//:all",
 )

README.md

@@ -4,7 +4,7 @@ This open source repository contains the standard CodeQL libraries and queries t
 
 ## How do I learn CodeQL and run queries?
 
-There is [extensive documentation](https://codeql.github.com/docs/) on getting started with writing CodeQL using the [CodeQL extension for Visual Studio Code](https://codeql.github.com/docs/codeql-for-visual-studio-code/) and the [CodeQL CLI](https://codeql.github.com/docs/codeql-cli/).
+There is extensive documentation about the [CodeQL language](https://codeql.github.com/docs/), writing CodeQL using the [CodeQL extension for Visual Studio Code](https://docs.github.com/en/code-security/codeql-for-vs-code/) and using the [CodeQL CLI](https://docs.github.com/en/code-security/codeql-cli).
 
 ## Contributing
 

codeql-workspace.yml

@@ -6,29 +6,16 @@ provide:
   - "*/ql/consistency-queries/qlpack.yml"
   - "*/ql/automodel/src/qlpack.yml"
   - "*/ql/automodel/test/qlpack.yml"
+  - "*/extractor-pack/codeql-extractor.yml"
   - "python/extractor/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"
-  - "go/build/codeql-extractor-go/codeql-extractor.yml"
-  - "javascript/ql/experimental/adaptivethreatmodeling/lib/qlpack.yml"
-  # This pack is explicitly excluded from the workspace since most users
-  # will want to use a version of this pack from the package cache. Internal
-  # users can uncomment the following line and place a custom ML model
-  # in the corresponding pack to test a custom ML model within their local
-  # checkout.
-  # - "javascript/ql/experimental/adaptivethreatmodeling/model/qlpack.yml"
-  - "javascript/ql/experimental/adaptivethreatmodeling/modelbuilding/qlpack.yml"
-  - "javascript/ql/experimental/adaptivethreatmodeling/src/qlpack.yml"
-  - "javascript/ql/experimental/adaptivethreatmodeling/test/qlpack.yml"
   - "csharp/ql/campaigns/Solorigate/lib/qlpack.yml"
   - "csharp/ql/campaigns/Solorigate/src/qlpack.yml"
   - "csharp/ql/campaigns/Solorigate/test/qlpack.yml"
   - "misc/legacy-support/*/qlpack.yml"
   - "misc/suite-helpers/qlpack.yml"
-  - "ruby/extractor-pack/codeql-extractor.yml"
-  - "swift/extractor-pack/codeql-extractor.yml"
-  - "ql/extractor-pack/codeql-extractor.yml"
   - ".github/codeql/extensions/**/codeql-pack.yml"
 
 versionPolicies:

config/dbscheme-fragments.json

@@ -28,6 +28,7 @@
     "/*- Yaml dbscheme -*/",
     "/*- Blame dbscheme -*/",
     "/*- JSON dbscheme -*/",
-    "/*- Python dbscheme -*/"
+    "/*- Python dbscheme -*/",
+    "/*- Empty location -*/"
   ]
 }

config/identical-files.json

@@ -61,10 +61,6 @@
     "java/ql/src/utils/modelgenerator/internal/CaptureModels.qll",
     "csharp/ql/src/utils/modelgenerator/internal/CaptureModels.qll"
   ],
-  "Model as Data Generation Java/C# - CaptureModelsPrinting": [
-    "java/ql/src/utils/modelgenerator/internal/CaptureModelsPrinting.qll",
-    "csharp/ql/src/utils/modelgenerator/internal/CaptureModelsPrinting.qll"
-  ],
   "Sign Java/C#": [
     "java/ql/lib/semmle/code/java/dataflow/internal/rangeanalysis/Sign.qll",
     "csharp/ql/lib/semmle/code/csharp/dataflow/internal/rangeanalysis/Sign.qll"
@@ -185,11 +181,6 @@
     "cpp/ql/lib/semmle/code/cpp/ir/implementation/unaliased_ssa/internal/AliasAnalysisImports.qll",
     "cpp/ql/lib/semmle/code/cpp/ir/implementation/aliased_ssa/internal/AliasAnalysisImports.qll"
   ],
-  "C++ IR ValueNumberingImports": [
-    "cpp/ql/lib/semmle/code/cpp/ir/implementation/raw/gvn/internal/ValueNumberingImports.qll",
-    "cpp/ql/lib/semmle/code/cpp/ir/implementation/unaliased_ssa/gvn/internal/ValueNumberingImports.qll",
-    "cpp/ql/lib/semmle/code/cpp/ir/implementation/aliased_ssa/gvn/internal/ValueNumberingImports.qll"
-  ],
   "IR SSA SSAConstruction": [
     "cpp/ql/lib/semmle/code/cpp/ir/implementation/unaliased_ssa/internal/SSAConstruction.qll",
     "cpp/ql/lib/semmle/code/cpp/ir/implementation/aliased_ssa/internal/SSAConstruction.qll"
@@ -251,12 +242,6 @@
     "cpp/ql/src/Security/CWE/CWE-020/SafeExternalAPIFunction.qll",
     "cpp/ql/src/Security/CWE/CWE-020/ir/SafeExternalAPIFunction.qll"
   ],
-  "XML": [
-    "cpp/ql/lib/semmle/code/cpp/XML.qll",
-    "csharp/ql/lib/semmle/code/csharp/XML.qll",
-    "java/ql/lib/semmle/code/xml/XML.qll",
-    "python/ql/lib/semmle/python/xml/XML.qll"
-  ],
   "DuplicationProblems.inc.qhelp": [
     "cpp/ql/src/Metrics/Files/DuplicationProblems.inc.qhelp",
     "javascript/ql/src/Metrics/DuplicationProblems.inc.qhelp",
@@ -368,7 +353,11 @@
     "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"
+    "python/ql/test/library-tests/dataflow/model-summaries/InlineTaintTest.ext.yml",
+    "python/ql/test/library-tests/dataflow/model-summaries/NormalDataflowTest.ext.yml"
+  ],
+  "shared tree-sitter extractor cargo.toml": [
+    "shared/tree-sitter-extractor/Cargo.toml",
+    "ruby/extractor/codeql-extractor-fake-crate/Cargo.toml"
   ]
-}
+}

cpp/BUILD.bazel

@@ -1,4 +1,4 @@
-load("@rules_pkg//:mappings.bzl", "pkg_filegroup")
+load("@rules_pkg//pkg:mappings.bzl", "pkg_filegroup")
 
 package(default_visibility = ["//visibility:public"])
 

cpp/autobuilder/.gitignore

@@ -1,13 +0,0 @@
-obj/
-TestResults/
-*.manifest
-*.pdb
-*.suo
-*.mdb
-*.vsmdi
-csharp.log
-**/bin/Debug
-**/bin/Release
-*.tlog
-.vs
-*.user

cpp/autobuilder/README.md

@@ -0,0 +1 @@
+The Windows autobuilder that used to live in this directory moved to `csharp/autobuilder/Semmle.Autobuild.Cpp`.

cpp/autobuilder/Semmle.Autobuild.Cpp.Tests/Semmle.Autobuild.Cpp.Tests.csproj

@@ -1,26 +0,0 @@
-<Project Sdk="Microsoft.NET.Sdk">
-
-  <PropertyGroup>
-    <OutputType>Exe</OutputType>
-    <TargetFramework>net8.0</TargetFramework>
-    <GenerateAssemblyInfo>false</GenerateAssemblyInfo>
-    <RuntimeIdentifiers>win-x64;linux-x64;osx-x64</RuntimeIdentifiers>
-    <Nullable>enable</Nullable>
-  </PropertyGroup>
-
-  <ItemGroup>
-    <PackageReference Include="System.IO.FileSystem" Version="4.3.0" />
-    <PackageReference Include="System.IO.FileSystem.Primitives" Version="4.3.0" />
-    <PackageReference Include="xunit" Version="2.6.2" />
-    <PackageReference Include="xunit.runner.visualstudio" Version="2.5.4">
-      <PrivateAssets>all</PrivateAssets>
-      <IncludeAssets>runtime; build; native; contentfiles; analyzers</IncludeAssets>
-    </PackageReference>
-    <PackageReference Include="Microsoft.NET.Test.Sdk" Version="17.8.0" />
-  </ItemGroup>
-
-  <ItemGroup>
-    <ProjectReference Include="..\Semmle.Autobuild.Cpp\Semmle.Autobuild.Cpp.csproj" />
-    <ProjectReference Include="..\..\..\csharp\autobuilder\Semmle.Autobuild.Shared\Semmle.Autobuild.Shared.csproj" />
-  </ItemGroup>
-</Project>

cpp/autobuilder/Semmle.Autobuild.Cpp/Properties/AssemblyInfo.cs

@@ -1,32 +0,0 @@
-using System.Reflection;
-using System.Runtime.InteropServices;
-
-// General Information about an assembly is controlled through the following
-// set of attributes. Change these attribute values to modify the information
-// associated with an assembly.
-[assembly: AssemblyTitle("Semmle.Autobuild.Cpp")]
-[assembly: AssemblyDescription("")]
-[assembly: AssemblyConfiguration("")]
-[assembly: AssemblyCompany("GitHub")]
-[assembly: AssemblyProduct("CodeQL autobuilder for C++")]
-[assembly: AssemblyCopyright("Copyright © GitHub 2020")]
-[assembly: AssemblyTrademark("")]
-[assembly: AssemblyCulture("")]
-
-// Setting ComVisible to false makes the types in this assembly not visible
-// to COM components.  If you need to access a type in this assembly from
-// COM, set the ComVisible attribute to true on that type.
-[assembly: ComVisible(false)]
-
-// Version information for an assembly consists of the following four values:
-//
-//      Major Version
-//      Minor Version
-//      Build Number
-//      Revision
-//
-// You can specify all the values or you can default the Build and Revision Numbers
-// by using the '*' as shown below:
-// [assembly: AssemblyVersion("1.0.*")]
-[assembly: AssemblyVersion("1.0.0.0")]
-[assembly: AssemblyFileVersion("1.0.0.0")]

cpp/autobuilder/Semmle.Autobuild.Cpp/Semmle.Autobuild.Cpp.csproj

@@ -1,28 +0,0 @@
-<Project Sdk="Microsoft.NET.Sdk">
-
-  <PropertyGroup>
-    <TargetFramework>net8.0</TargetFramework>
-    <AssemblyName>Semmle.Autobuild.Cpp</AssemblyName>
-    <RootNamespace>Semmle.Autobuild.Cpp</RootNamespace>
-    <ApplicationIcon />
-    <OutputType>Exe</OutputType>
-    <StartupObject />
-    <GenerateAssemblyInfo>false</GenerateAssemblyInfo>
-    <RuntimeIdentifiers>win-x64;linux-x64;osx-x64</RuntimeIdentifiers>
-    <Nullable>enable</Nullable>
-  </PropertyGroup>
-
-  <ItemGroup>
-    <Folder Include="Properties\" />
-  </ItemGroup>
-
-  <ItemGroup>
-    <PackageReference Include="Microsoft.Build" Version="17.8.3" />
-  </ItemGroup>
-
-  <ItemGroup>
-    <ProjectReference Include="..\..\..\csharp\extractor\Semmle.Util\Semmle.Util.csproj" />
-    <ProjectReference Include="..\..\..\csharp\autobuilder\Semmle.Autobuild.Shared\Semmle.Autobuild.Shared.csproj" />
-  </ItemGroup>
-
-</Project>

cpp/downgrades/19887dbd33327fb07d54251786e0cb2578539775/upgrade.properties

@@ -1,4 +1,4 @@
 description: Revert support for repeated initializers, which are allowed in C with designated initializers.
 compatibility: full
-aggregate_field_init.rel: reorder aggregate_field_init.rel (int aggregate, int initializer, int field, int position) aggregate initializer field
-aggregate_array_init.rel: reorder aggregate_array_init.rel (int aggregate, int initializer, int element_index, int position) aggregate initializer element_index
+aggregate_field_init.rel: reorder aggregate_field_init.rel (@aggregateliteral aggregate, @expr initializer, @membervariable field, int position) aggregate initializer field
+aggregate_array_init.rel: reorder aggregate_array_init.rel (@aggregateliteral aggregate, @expr initializer, int element_index, int position) aggregate initializer element_index

cpp/downgrades/BUILD.bazel

@@ -1,4 +1,4 @@
-load("@rules_pkg//:mappings.bzl", "pkg_files", "strip_prefix")
+load("@rules_pkg//pkg:mappings.bzl", "pkg_files", "strip_prefix")
 
 pkg_files(
     name = "downgrades",
@@ -6,7 +6,7 @@ pkg_files(
         ["**"],
         exclude = ["BUILD.bazel"],
     ),
-    prefix = "cpp/downgrades",
+    prefix = "downgrades",
     strip_prefix = strip_prefix.from_pkg(),
     visibility = ["//cpp:__pkg__"],
 )

cpp/ql/lib/BUILD.bazel

@@ -1,15 +1,13 @@
-load("@rules_pkg//:mappings.bzl", "pkg_files")
+load("@rules_pkg//pkg:mappings.bzl", "pkg_files")
 
 package(default_visibility = ["//cpp:__pkg__"])
 
 pkg_files(
     name = "dbscheme",
     srcs = ["semmlecode.cpp.dbscheme"],
-    prefix = "cpp",
 )
 
 pkg_files(
     name = "dbscheme-stats",
     srcs = ["semmlecode.cpp.dbscheme.stats"],
-    prefix = "cpp",
 )

cpp/ql/lib/CHANGELOG.md

@@ -1,3 +1,44 @@
+## 1.1.0
+
+### New Features
+
+* Data models can now be added with data extensions. In this way source, sink and summary models can be added in extension `.model.yml` files, rather than by writing classes in QL code. New models should be added in the `lib/ext` folder.
+
+### Minor Analysis Improvements
+
+* A partial model for the `Boost.Asio` network library has been added. This includes sources, sinks and summaries for certain functions in `Boost.Asio`, such as `read_until` and `write`.
+
+## 1.0.0
+
+### Breaking Changes
+
+* CodeQL package management is now generally available, and all GitHub-produced CodeQL packages have had their version numbers increased to 1.0.0.
+
+## 0.13.1
+
+No user-facing changes.
+
+## 0.13.0
+
+### Breaking Changes
+
+* Deleted the deprecated `GlobalValueNumberingImpl.qll` implementation.
+
+### New Features
+
+* Models-as-Data support has been added for C/C++. This feature allows flow sources, sinks and summaries to be expressed in compact strings as an alternative to modelling each source / sink / summary with explicit QL. See `dataflow/ExternalFlow.qll` for documentation and specification of the model format, and `models/implementations/ZMQ.qll` for a simple example of models. Importing models from `.yml` is not yet supported.
+
+### Minor Analysis Improvements
+
+* Source models have been added for the standard library function `getc` (and variations).
+* Source, sink and flow models for the ZeroMQ (ZMQ) networking library have been added.
+* Parameters of functions without definitions now have `ParameterNode`s.
+* The alias analysis used internally by various libraries has been improved to answer alias questions more conservatively. As a result, some queries may report fewer false positives.
+
+## 0.12.11
+
+No user-facing changes.
+
 ## 0.12.10
 
 ### New Features

cpp/ql/lib/change-notes/2024-06-10-builtin-expect.md

@@ -0,0 +1,4 @@
+---
+category: minorAnalysis
+---
+* The "Guards" library (`semmle.code.cpp.controlflow.Guards`) now also infers guards from calls to the builtin operation `__builtin_expect`. As a result, some queries may produce fewer false positives.

cpp/ql/lib/change-notes/2024-06-13-double-free.md

@@ -0,0 +1,4 @@
+---
+category: minorAnalysis
+---
+* The queries "Potential double free" (`cpp/double-free`) and "Potential use after free" (`cpp/use-after-free`) now produce fewer false positives.

cpp/ql/lib/change-notes/2024-06-20-extensible-allocation-deallocation.md

@@ -0,0 +1,4 @@
+---
+category: feature
+---
+* It is now possible to extend the classes `AllocationFunction` and `DeallocationFunction` via data extensions. Extensions of these classes should be added to the `lib/ext/allocation` and `lib/ext/deallocation` directories respectively.

cpp/ql/lib/change-notes/released/0.12.11.md

@@ -0,0 +1,3 @@
+## 0.12.11
+
+No user-facing changes.

cpp/ql/lib/change-notes/released/0.13.0.md

@@ -0,0 +1,16 @@
+## 0.13.0
+
+### Breaking Changes
+
+* Deleted the deprecated `GlobalValueNumberingImpl.qll` implementation.
+
+### New Features
+
+* Models-as-Data support has been added for C/C++. This feature allows flow sources, sinks and summaries to be expressed in compact strings as an alternative to modelling each source / sink / summary with explicit QL. See `dataflow/ExternalFlow.qll` for documentation and specification of the model format, and `models/implementations/ZMQ.qll` for a simple example of models. Importing models from `.yml` is not yet supported.
+
+### Minor Analysis Improvements
+
+* Source models have been added for the standard library function `getc` (and variations).
+* Source, sink and flow models for the ZeroMQ (ZMQ) networking library have been added.
+* Parameters of functions without definitions now have `ParameterNode`s.
+* The alias analysis used internally by various libraries has been improved to answer alias questions more conservatively. As a result, some queries may report fewer false positives.

cpp/ql/lib/change-notes/released/0.13.1.md

@@ -0,0 +1,3 @@
+## 0.13.1
+
+No user-facing changes.

cpp/ql/lib/change-notes/released/1.0.0.md

@@ -0,0 +1,5 @@
+## 1.0.0
+
+### Breaking Changes
+
+* CodeQL package management is now generally available, and all GitHub-produced CodeQL packages have had their version numbers increased to 1.0.0.

cpp/ql/lib/change-notes/released/1.1.0.md

@@ -0,0 +1,9 @@
+## 1.1.0
+
+### New Features
+
+* Data models can now be added with data extensions. In this way source, sink and summary models can be added in extension `.model.yml` files, rather than by writing classes in QL code. New models should be added in the `lib/ext` folder.
+
+### Minor Analysis Improvements
+
+* A partial model for the `Boost.Asio` network library has been added. This includes sources, sinks and summaries for certain functions in `Boost.Asio`, such as `read_until` and `write`.

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

@@ -1,2 +1,2 @@
 ---
-lastReleaseVersion: 0.12.10
+lastReleaseVersion: 1.1.0

cpp/ql/lib/ext/Boost.Asio.model.yml

@@ -0,0 +1,26 @@
+extensions:
+  # partial model of the Boost::Asio network library
+extensions:
+  - addsTo:
+      pack: codeql/cpp-all
+      extensible: sourceModel
+    data: # namespace, type, subtypes, name, signature, ext, output, kind, provenance
+      - ["boost::asio", "", False, "read", "", "", "Argument[*1]", "remote", "manual"]
+      - ["boost::asio", "", False, "read_at", "", "", "Argument[*2]", "remote", "manual"]
+      - ["boost::asio", "", False, "read_until", "", "", "Argument[*1]", "remote", "manual"]
+      - ["boost::asio", "", False, "async_read", "", "", "Argument[*1]", "remote", "manual"]
+      - ["boost::asio", "", False, "async_read_at", "", "", "Argument[*2]", "remote", "manual"]
+      - ["boost::asio", "", False, "async_read_until", "", "", "Argument[*1]", "remote", "manual"]
+  - addsTo:
+      pack: codeql/cpp-all
+      extensible: sinkModel
+    data: # namespace, type, subtypes, name, signature, ext, input, kind, provenance
+      - ["boost::asio", "", False, "write", "", "", "Argument[*1]", "remote-sink", "manual"]
+      - ["boost::asio", "", False, "write_at", "", "", "Argument[*2]", "remote-sink", "manual"]
+      - ["boost::asio", "", False, "async_write", "", "", "Argument[*1]", "remote-sink", "manual"]
+      - ["boost::asio", "", False, "async_write_at", "", "", "Argument[*2]", "remote-sink", "manual"]
+  - addsTo:
+      pack: codeql/cpp-all
+      extensible: summaryModel
+    data: # namespace, type, subtypes, name, signature, ext, input, output, kind, provenance
+      - ["boost::asio", "", False, "buffer", "", "", "Argument[*0]", "ReturnValue", "taint", "manual"]

cpp/ql/lib/ext/allocation/Bsd.allocation.model.yml

@@ -0,0 +1,7 @@
+extensions:
+  - addsTo:
+      pack: codeql/cpp-all
+      extensible: allocationFunctionModel
+    data:
+      - ["", "", False, "kmem_alloc", "0", "", "", True]
+      - ["", "", False, "kmem_zalloc", "0", "", "", True]

cpp/ql/lib/ext/allocation/Glibc.allocation.model.yml

@@ -0,0 +1,7 @@
+extensions:
+  - addsTo:
+      pack: codeql/cpp-all
+      extensible: allocationFunctionModel
+    data:
+      - ["", "", False, "g_malloc", "0", "", "", True]
+      - ["", "", False, "g_try_malloc", "0", "", "", True]

cpp/ql/lib/ext/allocation/OpenSSL.allocation.model.yml

@@ -0,0 +1,10 @@
+extensions:
+  - addsTo:
+      pack: codeql/cpp-all
+      extensible: allocationFunctionModel
+    data:
+      - ["", "", False, "CRYPTO_malloc", "0", "", "", True]
+      - ["", "", False, "CRYPTO_zalloc", "0", "", "", True]
+      - ["", "", False, "CRYPTO_secure_malloc", "0", "", "", True]
+      - ["", "", False, "CRYPTO_secure_zalloc", "0", "", "", True]
+

cpp/ql/lib/ext/allocation/Std.allocation.model.yml

@@ -0,0 +1,15 @@
+extensions:
+  - addsTo:
+      pack: codeql/cpp-all
+      extensible: allocationFunctionModel
+    data:
+      - ["", "", False, "malloc", "0", "", "", True]
+      - ["std", "", False, "malloc", "0", "", "", True]
+      - ["bsl", "", False, "malloc", "0", "", "", True]
+      - ["", "", False, "alloca", "0", "", "", False]
+      - ["", "", False, "__builtin_alloca", "0", "", "", False]
+      - ["", "", False, "_alloca", "0", "", "", False]
+      - ["", "", False, "_malloca", "0", "", "", False]
+      - ["", "", False, "calloc", "1", "0", "", True]
+      - ["std", "", False, "calloc", "1", "0", "", True]
+      - ["bsl", "", False, "calloc", "1", "0", "", True]

cpp/ql/lib/ext/allocation/Windows.allocation.model.yml

@@ -0,0 +1,29 @@
+extensions:
+  - addsTo:
+      pack: codeql/cpp-all
+      extensible: allocationFunctionModel
+    data:
+      - ["", "", False, "MmAllocateContiguousMemory", "0", "", "", True]
+      - ["", "", False, "MmAllocateContiguousNodeMemory", "0", "", "", True]
+      - ["", "", False, "MmAllocateContiguousMemorySpecifyCache", "0", "", "", True]
+      - ["", "", False, "MmAllocateContiguousMemorySpecifyCacheNode", "0", "", "", True]
+      - ["", "", False, "MmAllocateNonCachedMemory", "0", "", "", True]
+      - ["", "", False, "MmAllocateMappingAddress", "0", "", "", True]
+      - ["", "", False, "CoTaskMemAlloc", "0", "", "", True]
+      - ["", "", False, "ExAllocatePool", "1", "", "", True]
+      - ["", "", False, "ExAllocatePool2", "1", "", "", True]
+      - ["", "", False, "ExAllocatePool3", "1", "", "", True]
+      - ["", "", False, "ExAllocatePoolWithTag", "1", "", "", True]
+      - ["", "", False, "ExAllocatePoolWithTagPriority", "1", "", "", True]
+      - ["", "", False, "ExAllocatePoolWithQuota", "1", "", "", True]
+      - ["", "", False, "ExAllocatePoolWithQuotaTag", "1", "", "", True]
+      - ["", "", False, "ExAllocatePoolZero", "1", "", "", True]
+      - ["", "", False, "IoAllocateMdl", "1", "", "", True]
+      - ["", "", False, "IoAllocateErrorLogEntry", "1", "", "", True]
+      - ["", "", False, "LocalAlloc", "1", "", "", True]
+      - ["", "", False, "GlobalAlloc", "1", "", "", True]
+      - ["", "", False, "VirtualAlloc", "1", "", "", True]
+      - ["", "", False, "HeapAlloc", "2", "", "", True]
+      - ["", "", False, "MmAllocatePagesForMdl", "3", "", "", True]
+      - ["", "", False, "MmAllocatePagesForMdlEx", "3", "", "", True]
+      - ["", "", False, "MmAllocateNodePagesForMdlEx", "3", "", "", True]

cpp/ql/lib/ext/allocation/empty.allocation.model.yml

@@ -0,0 +1,5 @@
+extensions:
+  - addsTo:
+      pack: codeql/cpp-all
+      extensible: allocationFunctionModel
+    data: []

cpp/ql/lib/ext/deallocation/Bsd.deallocation.model.yml

@@ -0,0 +1,8 @@
+extensions:
+  - addsTo:
+      pack: codeql/cpp-all
+      extensible: deallocationFunctionModel
+    data:
+      - ["", "", False, "pool_put", "1"]
+      - ["", "", False, "pool_cache_put", "1"]
+      - ["", "", False, "kmem_free", "0"]

cpp/ql/lib/ext/deallocation/Std.deallocation.model.yml

@@ -0,0 +1,42 @@
+extensions:
+  - addsTo:
+      pack: codeql/cpp-all
+      extensible: deallocationFunctionModel
+    data:
+      - ["", "", False, "free", "0"]
+      - ["std", "", False, "free", "0"]
+      - ["bsl", "", False, "free", "0"]
+      - ["", "", False, "realloc", "0"]
+      - ["std", "", False, "realloc", "0"]
+      - ["bsl", "", False, "realloc", "0"]
+      - ["", "", False, "CRYPTO_free", "0"]
+      - ["", "", False, "CRYPTO_secure_free", "0"]
+      - ["", "", False, "g_free", "0"]
+      - ["", "", False, "ExFreePool", "0"]
+      - ["", "", False, "ExFreePoolWithTag", "0"]
+      - ["", "", False, "ExDeleteTimer", "0"]
+      - ["", "", False, "IoFreeIrp", "0"]
+      - ["", "", False, "IoFreeMdl", "0"]
+      - ["", "", False, "IoFreeErrorLogEntry", "0"]
+      - ["", "", False, "IoFreeWorkItem", "0"]
+      - ["", "", False, "MmFreeContiguousMemory", "0"]
+      - ["", "", False, "MmFreeContiguousMemorySpecifyCache", "0"]
+      - ["", "", False, "MmFreeNonCachedMemory", "0"]
+      - ["", "", False, "MmFreeMappingAddress", "0"]
+      - ["", "", False, "MmFreePagesFromMdl", "0"]
+      - ["", "", False, "MmUnmapReservedMapping", "0"]
+      - ["", "", False, "MmUnmapLockedPages", "0"]
+      - ["", "", False, "NdisFreeGenericObject", "0"]
+      - ["", "", False, "NdisFreeMemory", "0"]
+      - ["", "", False, "NdisFreeMemoryWithTag", "0"]
+      - ["", "", False, "NdisFreeMdl", "0"]
+      - ["", "", False, "NdisFreeNetBufferListPool", "0"]
+      - ["", "", False, "NdisFreeNetBufferPool", "0"]
+      - ["", "", False, "LocalFree", "0"]
+      - ["", "", False, "GlobalFree", "0"]
+      - ["", "", False, "LocalReAlloc", "0"]
+      - ["", "", False, "GlobalReAlloc", "0"]
+      - ["", "", False, "VirtualFree", "0"]
+      - ["", "", False, "CoTaskMemFree", "0"]
+      - ["", "", False, "CoTaskMemRealloc", "0"]
+      - ["", "", False, "SysFreeString", "0"]

cpp/ql/lib/ext/deallocation/Windows.deallocation.model.yml

@@ -0,0 +1,41 @@
+extensions:
+  - addsTo:
+      pack: codeql/cpp-all
+      extensible: deallocationFunctionModel
+    data:
+      - ["", "", False, "ExFreePool", "0"]
+      - ["", "", False, "ExFreePoolWithTag", "0"]
+      - ["", "", False, "ExDeleteTimer", "0"]
+      - ["", "", False, "IoFreeIrp", "0"]
+      - ["", "", False, "IoFreeMdl", "0"]
+      - ["", "", False, "IoFreeErrorLogEntry", "0"]
+      - ["", "", False, "IoFreeWorkItem", "0"]
+      - ["", "", False, "MmFreeContiguousMemory", "0"]
+      - ["", "", False, "MmFreeContiguousMemorySpecifyCache", "0"]
+      - ["", "", False, "MmFreeNonCachedMemory", "0"]
+      - ["", "", False, "MmFreeMappingAddress", "0"]
+      - ["", "", False, "MmFreePagesFromMdl", "0"]
+      - ["", "", False, "MmUnmapReservedMapping", "0"]
+      - ["", "", False, "MmUnmapLockedPages", "0"]
+      - ["", "", False, "NdisFreeGenericObject", "0"]
+      - ["", "", False, "NdisFreeMemory", "0"]
+      - ["", "", False, "NdisFreeMemoryWithTag", "0"]
+      - ["", "", False, "NdisFreeMdl", "0"]
+      - ["", "", False, "NdisFreeNetBufferListPool", "0"]
+      - ["", "", False, "NdisFreeNetBufferPool", "0"]
+      - ["", "", False, "LocalFree", "0"]
+      - ["", "", False, "GlobalFree", "0"]
+      - ["", "", False, "LocalReAlloc", "0"]
+      - ["", "", False, "GlobalReAlloc", "0"]
+      - ["", "", False, "VirtualFree", "0"]
+      - ["", "", False, "CoTaskMemFree", "0"]
+      - ["", "", False, "CoTaskMemRealloc", "0"]
+      - ["", "", False, "SysFreeString", "0"]
+      - ["", "", False, "ExFreeToLookasideListEx", "1"]
+      - ["", "", False, "ExFreeToPagedLookasideList", "1"]
+      - ["", "", False, "ExFreeToNPagedLookasideList", "1"]
+      - ["", "", False, "NdisFreeMemoryWithTagPriority", "1"]
+      - ["", "", False, "StorPortFreeMdl", "1"]
+      - ["", "", False, "StorPortFreePool", "1"]
+      - ["", "", False, "HeapFree", "2"]
+      - ["", "", False, "HeapReAlloc", "2"]

cpp/ql/lib/ext/deallocation/empty.deallocation.model.yml

@@ -0,0 +1,5 @@
+extensions:
+  - addsTo:
+      pack: codeql/cpp-all
+      extensible: deallocationFunctionModel
+    data: []

cpp/ql/lib/ext/empty.model.yml

@@ -0,0 +1,15 @@
+extensions:
+  # Make sure that the extensible model predicates have at least one definition
+  # to avoid errors about undefined extensionals.
+  - addsTo:
+      pack: codeql/cpp-all
+      extensible: sourceModel
+    data: []
+  - addsTo:
+      pack: codeql/cpp-all
+      extensible: sinkModel
+    data: []
+  - addsTo:
+      pack: codeql/cpp-all
+      extensible: summaryModel
+    data: []

cpp/ql/lib/qlpack.yml

@@ -1,5 +1,5 @@
 name: codeql/cpp-all
-version: 0.12.10
+version: 1.1.1-dev
 groups: cpp
 dbscheme: semmlecode.cpp.dbscheme
 extractor: cpp
@@ -7,8 +7,15 @@ library: true
 upgrades: upgrades
 dependencies:
   codeql/dataflow: ${workspace}
+  codeql/mad: ${workspace}
   codeql/rangeanalysis: ${workspace}
   codeql/ssa: ${workspace}
+  codeql/typeflow: ${workspace}
   codeql/tutorial: ${workspace}
   codeql/util: ${workspace}
+  codeql/xml: ${workspace}
+dataExtensions:
+  - ext/*.model.yml
+  - ext/deallocation/*.model.yml
+  - ext/allocation/*.model.yml
 warnOnImplicitThis: true

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

@@ -364,6 +364,8 @@ class ConversionNode extends ExprNode {
     childIndex = 0 and
     result.getAst() = conv.getExpr() and
     conv.getExpr() instanceof Conversion
+    or
+    result.getAst() = expr.getImplicitDestructorCall(childIndex - 1)
   }
 }
 
@@ -461,6 +463,25 @@ class StmtNode extends AstNode {
   }
 }
 
+/**
+ * A node representing a child of a `Stmt` that is itself a `Stmt`.
+ */
+class ChildStmtNode extends StmtNode {
+  Stmt childStmt;
+
+  ChildStmtNode() { exists(Stmt parent | parent.getAChild() = childStmt and childStmt = ast) }
+
+  override BaseAstNode getChildInternal(int childIndex) {
+    result = super.getChildInternal(childIndex)
+    or
+    exists(int destructorIndex |
+      result.getAst() = childStmt.getImplicitDestructorCall(destructorIndex) and
+      childIndex =
+        destructorIndex + max(int index | exists(childStmt.getChild(index)) or index = 0) + 1
+    )
+  }
+}
+
 /**
  * A node representing a `DeclStmt`.
  */
@@ -672,6 +693,13 @@ class FunctionNode extends FunctionOrGlobalOrNamespaceVariableNode {
 private string getChildAccessorWithoutConversions(Locatable parent, Element child) {
   shouldPrintDeclaration(getAnEnclosingDeclaration(parent)) and
   (
+    exists(Stmt s, int i | s.getChild(i) = parent |
+      exists(int n |
+        s.getChild(i).(Stmt).getImplicitDestructorCall(n) = child and
+        result = "getImplicitDestructorCall(" + n + ")"
+      )
+    )
+    or
     exists(Stmt s | s = parent |
       namedStmtChildPredicates(s, child, result)
       or
@@ -862,7 +890,7 @@ private predicate namedExprChildPredicates(Expr expr, Element ele, string pred)
     or
     expr.(DeleteOrDeleteArrayExpr).getDestructorCall() = ele and pred = "getDestructorCall()"
     or
-    expr.(DeleteOrDeleteArrayExpr).getExpr() = ele and pred = "getExpr()"
+    expr.(DeleteOrDeleteArrayExpr).getExprWithReuse() = ele and pred = "getExprWithReuse()"
     or
     expr.(DestructorFieldDestruction).getExpr() = ele and pred = "getExpr()"
     or

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

@@ -410,6 +410,10 @@ class LocalVariable extends LocalScopeVariable, @localvariable {
     or
     orphaned_variables(underlyingElement(this), unresolveElement(result))
   }
+
+  override predicate isStatic() {
+    super.isStatic() or orphaned_variables(underlyingElement(this), _)
+  }
 }
 
 /**

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

@@ -3,305 +3,67 @@
  */
 
 import semmle.files.FileSystem
+private import codeql.xml.Xml
 
-private class TXmlLocatable =
-  @xmldtd or @xmlelement or @xmlattribute or @xmlnamespace or @xmlcomment or @xmlcharacters;
+private module Input implements InputSig<File, Location> {
+  class XmlLocatableBase = @xmllocatable or @xmlnamespaceable;
 
-/** An XML element that has a location. */
-class XmlLocatable extends @xmllocatable, TXmlLocatable {
-  /** Gets the source location for this element. */
-  Location getLocation() { xmllocations(this, result) }
+  predicate xmllocations_(XmlLocatableBase e, Location loc) { xmllocations(e, loc) }
 
-  /**
-   * 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
+  class XmlParentBase = @xmlparent;
+
+  class XmlNamespaceableBase = @xmlnamespaceable;
+
+  class XmlElementBase = @xmlelement;
+
+  class XmlFileBase = File;
+
+  predicate xmlEncoding_(XmlFileBase f, string enc) { xmlEncoding(f, enc) }
+
+  class XmlDtdBase = @xmldtd;
+
+  predicate xmlDTDs_(XmlDtdBase e, string root, string publicId, string systemId, XmlFileBase file) {
+    xmlDTDs(e, root, publicId, systemId, file)
+  }
+
+  predicate xmlElements_(
+    XmlElementBase e, string name, XmlParentBase parent, int idx, XmlFileBase file
   ) {
-    exists(File f, Location l | l = this.getLocation() |
-      locations_default(l, f, startline, startcolumn, endline, endcolumn) and
-      filepath = f.getAbsolutePath()
-    )
+    xmlElements(e, name, parent, idx, file)
   }
 
-  /** Gets a textual representation of this element. */
-  string toString() { none() } // overridden in subclasses
-}
+  class XmlAttributeBase = @xmlattribute;
 
-/**
- * An `XmlParent` is either an `XmlElement` or an `XmlFile`,
- * both of which can contain other elements.
- */
-class XmlParent extends @xmlparent {
-  XmlParent() {
-    // explicitly restrict `this` to be either an `XmlElement` or an `XmlFile`;
-    // the type `@xmlparent` currently also includes non-XML files
-    this instanceof @xmlelement or xmlEncoding(this, _)
+  predicate xmlAttrs_(
+    XmlAttributeBase e, XmlElementBase elementid, string name, string value, int idx,
+    XmlFileBase file
+  ) {
+    xmlAttrs(e, elementid, name, value, idx, file)
   }
 
-  /**
-   * Gets a printable representation of this XML parent.
-   * (Intended to be overridden in subclasses.)
-   */
-  string getName() { none() } // overridden in subclasses
+  class XmlNamespaceBase = @xmlnamespace;
 
-  /** Gets the file to which this XML parent belongs. */
-  XmlFile getFile() { result = this or xmlElements(this, _, _, _, result) }
-
-  /** Gets the child element at a specified index of this XML parent. */
-  XmlElement getChild(int index) { xmlElements(result, _, this, index, _) }
-
-  /** Gets a child element of this XML parent. */
-  XmlElement getAChild() { xmlElements(result, _, this, _, _) }
-
-  /** Gets a child element of this XML parent with the given `name`. */
-  XmlElement getAChild(string name) { xmlElements(result, _, this, _, _) and result.hasName(name) }
-
-  /** Gets a comment that is a child of this XML parent. */
-  XmlComment getAComment() { xmlComments(result, _, this, _) }
-
-  /** Gets a character sequence that is a child of this XML parent. */
-  XmlCharacters getACharactersSet() { xmlChars(result, _, this, _, _, _) }
-
-  /** Gets the depth in the tree. (Overridden in XmlElement.) */
-  int getDepth() { result = 0 }
-
-  /** Gets the number of child XML elements of this XML parent. */
-  int getNumberOfChildren() { result = count(XmlElement e | xmlElements(e, _, this, _, _)) }
-
-  /** Gets the number of places in the body of this XML parent where text occurs. */
-  int getNumberOfCharacterSets() { result = count(int pos | xmlChars(_, _, this, pos, _, _)) }
-
-  /**
-   * Gets the result of appending all the character sequences of this XML parent from
-   * left to right, separated by a space.
-   */
-  string allCharactersString() {
-    result =
-      concat(string chars, int pos | xmlChars(_, chars, this, pos, _, _) | chars, " " order by pos)
+  predicate xmlNs_(XmlNamespaceBase e, string prefixName, string uri, XmlFileBase file) {
+    xmlNs(e, prefixName, uri, file)
   }
 
-  /** Gets the text value contained in this XML parent. */
-  string getTextValue() { result = this.allCharactersString() }
+  predicate xmlHasNs_(XmlNamespaceableBase e, XmlNamespaceBase ns, XmlFileBase file) {
+    xmlHasNs(e, ns, file)
+  }
 
-  /** Gets a printable representation of this XML parent. */
-  string toString() { result = this.getName() }
-}
+  class XmlCommentBase = @xmlcomment;
 
-/** An XML file. */
-class XmlFile extends XmlParent, File {
-  XmlFile() { xmlEncoding(this, _) }
+  predicate xmlComments_(XmlCommentBase e, string text, XmlParentBase parent, XmlFileBase file) {
+    xmlComments(e, text, parent, file)
+  }
 
-  /** Gets a printable representation of this XML file. */
-  override string toString() { result = this.getName() }
+  class XmlCharactersBase = @xmlcharacters;
 
-  /** Gets the name of this XML file. */
-  override string getName() { result = File.super.getAbsolutePath() }
-
-  /** Gets the encoding of this XML file. */
-  string getEncoding() { xmlEncoding(this, result) }
-
-  /** Gets the XML file itself. */
-  override XmlFile getFile() { result = this }
-
-  /** Gets a top-most element in an XML file. */
-  XmlElement getARootElement() { result = this.getAChild() }
-
-  /** Gets a DTD associated with this XML file. */
-  XmlDtd getADtd() { xmlDTDs(result, _, _, _, this) }
-}
-
-/**
- * An XML document type definition (DTD).
- *
- * Example:
- *
- * ```
- * <!ELEMENT person (firstName, lastName?)>
- * <!ELEMENT firstName (#PCDATA)>
- * <!ELEMENT lastName (#PCDATA)>
- * ```
- */
-class XmlDtd extends XmlLocatable, @xmldtd {
-  /** Gets the name of the root element of this DTD. */
-  string getRoot() { xmlDTDs(this, result, _, _, _) }
-
-  /** Gets the public ID of this DTD. */
-  string getPublicId() { xmlDTDs(this, _, result, _, _) }
-
-  /** Gets the system ID of this DTD. */
-  string getSystemId() { xmlDTDs(this, _, _, result, _) }
-
-  /** Holds if this DTD is public. */
-  predicate isPublic() { not xmlDTDs(this, _, "", _, _) }
-
-  /** Gets the parent of this DTD. */
-  XmlParent getParent() { xmlDTDs(this, _, _, _, result) }
-
-  override string toString() {
-    this.isPublic() and
-    result = this.getRoot() + " PUBLIC '" + this.getPublicId() + "' '" + this.getSystemId() + "'"
-    or
-    not this.isPublic() and
-    result = this.getRoot() + " SYSTEM '" + this.getSystemId() + "'"
+  predicate xmlChars_(
+    XmlCharactersBase e, string text, XmlParentBase parent, int idx, int isCDATA, XmlFileBase file
+  ) {
+    xmlChars(e, text, parent, idx, isCDATA, file)
   }
 }
 
-/**
- * An XML element in an XML file.
- *
- * Example:
- *
- * ```
- * <manifest xmlns:android="http://schemas.android.com/apk/res/android"
- *           package="com.example.exampleapp" android:versionCode="1">
- * </manifest>
- * ```
- */
-class XmlElement extends @xmlelement, XmlParent, XmlLocatable {
-  /** Holds if this XML element has the given `name`. */
-  predicate hasName(string name) { name = this.getName() }
-
-  /** Gets the name of this XML element. */
-  override string getName() { xmlElements(this, result, _, _, _) }
-
-  /** Gets the XML file in which this XML element occurs. */
-  override XmlFile getFile() { xmlElements(this, _, _, _, result) }
-
-  /** Gets the parent of this XML element. */
-  XmlParent getParent() { xmlElements(this, _, result, _, _) }
-
-  /** Gets the index of this XML element among its parent's children. */
-  int getIndex() { xmlElements(this, _, _, result, _) }
-
-  /** Holds if this XML element has a namespace. */
-  predicate hasNamespace() { xmlHasNs(this, _, _) }
-
-  /** Gets the namespace of this XML element, if any. */
-  XmlNamespace getNamespace() { xmlHasNs(this, result, _) }
-
-  /** Gets the index of this XML element among its parent's children. */
-  int getElementPositionIndex() { xmlElements(this, _, _, result, _) }
-
-  /** Gets the depth of this element within the XML file tree structure. */
-  override int getDepth() { result = this.getParent().getDepth() + 1 }
-
-  /** Gets an XML attribute of this XML element. */
-  XmlAttribute getAnAttribute() { result.getElement() = this }
-
-  /** Gets the attribute with the specified `name`, if any. */
-  XmlAttribute getAttribute(string name) { result.getElement() = this and result.getName() = name }
-
-  /** Holds if this XML element has an attribute with the specified `name`. */
-  predicate hasAttribute(string name) { exists(this.getAttribute(name)) }
-
-  /** Gets the value of the attribute with the specified `name`, if any. */
-  string getAttributeValue(string name) { result = this.getAttribute(name).getValue() }
-
-  /** Gets a printable representation of this XML element. */
-  override string toString() { result = this.getName() }
-}
-
-/**
- * An attribute that occurs inside an XML element.
- *
- * Examples:
- *
- * ```
- * package="com.example.exampleapp"
- * android:versionCode="1"
- * ```
- */
-class XmlAttribute extends @xmlattribute, XmlLocatable {
-  /** Gets the name of this attribute. */
-  string getName() { xmlAttrs(this, _, result, _, _, _) }
-
-  /** Gets the XML element to which this attribute belongs. */
-  XmlElement getElement() { xmlAttrs(this, result, _, _, _, _) }
-
-  /** Holds if this attribute has a namespace. */
-  predicate hasNamespace() { xmlHasNs(this, _, _) }
-
-  /** Gets the namespace of this attribute, if any. */
-  XmlNamespace getNamespace() { xmlHasNs(this, result, _) }
-
-  /** Gets the value of this attribute. */
-  string getValue() { xmlAttrs(this, _, _, result, _, _) }
-
-  /** Gets a printable representation of this XML attribute. */
-  override string toString() { result = this.getName() + "=" + this.getValue() }
-}
-
-/**
- * A namespace used in an XML file.
- *
- * Example:
- *
- * ```
- * xmlns:android="http://schemas.android.com/apk/res/android"
- * ```
- */
-class XmlNamespace extends XmlLocatable, @xmlnamespace {
-  /** Gets the prefix of this namespace. */
-  string getPrefix() { xmlNs(this, result, _, _) }
-
-  /** Gets the URI of this namespace. */
-  string getUri() { xmlNs(this, _, result, _) }
-
-  /** Holds if this namespace has no prefix. */
-  predicate isDefault() { this.getPrefix() = "" }
-
-  override string toString() {
-    this.isDefault() and result = this.getUri()
-    or
-    not this.isDefault() and result = this.getPrefix() + ":" + this.getUri()
-  }
-}
-
-/**
- * A comment in an XML file.
- *
- * Example:
- *
- * ```
- * <!-- This is a comment. -->
- * ```
- */
-class XmlComment extends @xmlcomment, XmlLocatable {
-  /** Gets the text content of this XML comment. */
-  string getText() { xmlComments(this, result, _, _) }
-
-  /** Gets the parent of this XML comment. */
-  XmlParent getParent() { xmlComments(this, _, result, _) }
-
-  /** Gets a printable representation of this XML comment. */
-  override string toString() { result = this.getText() }
-}
-
-/**
- * A sequence of characters that occurs between opening and
- * closing tags of an XML element, excluding other elements.
- *
- * Example:
- *
- * ```
- * <content>This is a sequence of characters.</content>
- * ```
- */
-class XmlCharacters extends @xmlcharacters, XmlLocatable {
-  /** Gets the content of this character sequence. */
-  string getCharacters() { xmlChars(this, result, _, _, _, _) }
-
-  /** Gets the parent of this character sequence. */
-  XmlParent getParent() { xmlChars(this, _, result, _, _, _) }
-
-  /** Holds if this character sequence is CDATA. */
-  predicate isCDATA() { xmlChars(this, _, _, _, 1, _) }
-
-  /** Gets a printable representation of this XML character sequence. */
-  override string toString() { result = this.getCharacters() }
-}
+import Make<File, Location, Input>

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

@@ -375,6 +375,33 @@ cached
 class IRGuardCondition extends Instruction {
   Instruction branch;
 
+  /*
+   * An `IRGuardCondition` supports reasoning about four different kinds of
+   * relations:
+   * 1. A unary equality relation of the form `e == k`
+   * 2. A binary equality relation of the form `e1 == e2 + k`
+   * 3. A unary inequality relation of the form `e < k`
+   * 4. A binary inequality relation of the form `e1 < e2 + k`
+   *
+   * where `k` is a constant.
+   *
+   * Furthermore, the unary relations (i.e., case 1 and case 3) are also
+   * inferred from `switch` statement guards: equality relations are inferred
+   * from the unique `case` statement, if any, and inequality relations are
+   * inferred from the [case range](https://gcc.gnu.org/onlinedocs/gcc/Case-Ranges.html)
+   * gcc extension.
+   *
+   * The implementation of all four follows the same structure: Each relation
+   * has a cached user-facing predicate that. For example,
+   * `GuardCondition::comparesEq` calls `compares_eq`. This predicate has
+   * several cases that recursively decompose the relation to bring it to a
+   * canonical form (i.e., a relation of the form `e1 == e2 + k`). The base
+   * case for this relation (i.e., `simple_comparison_eq`) handles
+   * `CompareEQInstruction`s and `CompareNEInstruction`, and recursive
+   * predicates (e.g., `complex_eq`) rewrites larger expressions such as
+   * `e1 + k1 == e2 + k2` into canonical the form `e1 == e2 + (k2 - k1)`.
+   */
+
   cached
   IRGuardCondition() { branch = getBranchForCondition(this) }
 
@@ -565,7 +592,7 @@ class IRGuardCondition extends Instruction {
   /** Holds if (determined by this guard) `op == k` evaluates to `areEqual` if this expression evaluates to `value`. */
   cached
   predicate comparesEq(Operand op, int k, boolean areEqual, AbstractValue value) {
-    compares_eq(this, op, k, areEqual, value)
+    unary_compares_eq(this, op, k, areEqual, false, value)
   }
 
   /**
@@ -586,7 +613,7 @@ class IRGuardCondition extends Instruction {
   cached
   predicate ensuresEq(Operand op, int k, IRBlock block, boolean areEqual) {
     exists(AbstractValue value |
-      compares_eq(this, op, k, areEqual, value) and this.valueControls(block, value)
+      unary_compares_eq(this, op, k, areEqual, false, value) and this.valueControls(block, value)
     )
   }
 
@@ -611,7 +638,7 @@ class IRGuardCondition extends Instruction {
   cached
   predicate ensuresEqEdge(Operand op, int k, IRBlock pred, IRBlock succ, boolean areEqual) {
     exists(AbstractValue value |
-      compares_eq(this, op, k, areEqual, value) and
+      unary_compares_eq(this, op, k, areEqual, false, value) and
       this.valueControlsEdge(pred, succ, value)
     )
   }
@@ -735,33 +762,80 @@ private predicate compares_eq(
   exists(AbstractValue dual | value = dual.getDualValue() |
     compares_eq(test.(LogicalNotInstruction).getUnary(), left, right, k, areEqual, dual)
   )
+  or
+  compares_eq(test.(BuiltinExpectCallInstruction).getCondition(), left, right, k, areEqual, value)
 }
 
-/** Holds if `op == k` is `areEqual` given that `test` is equal to `value`. */
-private predicate compares_eq(
-  Instruction test, Operand op, int k, boolean areEqual, AbstractValue value
+/**
+ * Holds if `op == k` is `areEqual` given that `test` is equal to `value`.
+ *
+ * Many internal predicates in this file have a `inNonZeroCase` column.
+ * Ideally, the `k` column would be a type such as `Option<int>::Option`, to
+ * represent whether we have a concrete value `k` such that `op == k`, or whether
+ * we only know that `op != 0`.
+ * However, cannot instantiate `Option` with an infinite type. Thus the boolean
+ * `inNonZeroCase` is used to distinquish the `Some` (where we have a concrete
+ * value `k`) and `None` cases (where we only know that `op != 0`).
+ *
+ * Thus, if `inNonZeroCase = true` then `op != 0` and the value of `k` is
+ * meaningless.
+ *
+ * To see why `inNonZeroCase` is needed consider the following C program:
+ * ```c
+ * char* p = ...;
+ * if(p) {
+ *   use(p);
+ * }
+ * ```
+ * in C++ there would be an int-to-bool conversion on `p`. However, since C
+ * does not have booleans there is no conversion. We want to be able to
+ * conclude that `p` is non-zero in the true branch, so we need to give `k`
+ * some value. However, simply setting `k = 1` would make the rest of the
+ * analysis think that `k == 1` holds inside the branch. So we distinquish
+ * between the above case and
+ * ```c
+ * if(p == 1) {
+ *   use(p)
+ * }
+ * ```
+ * by setting `inNonZeroCase` to `true` in the former case, but not in the
+ * latter.
+ */
+private predicate unary_compares_eq(
+  Instruction test, Operand op, int k, boolean areEqual, boolean inNonZeroCase, AbstractValue value
 ) {
   /* The simple case where the test *is* the comparison so areEqual = testIsTrue xor eq. */
-  exists(AbstractValue v | simple_comparison_eq(test, op, k, v) |
+  exists(AbstractValue v |
+    unary_simple_comparison_eq(test, k, inNonZeroCase, v) and op.getDef() = test
+  |
     areEqual = true and value = v
     or
     areEqual = false and value = v.getDualValue()
   )
   or
-  complex_eq(test, op, k, areEqual, value)
+  unary_complex_eq(test, op, k, areEqual, inNonZeroCase, value)
   or
   /* (x is true => (op == k)) => (!x is false => (op == k)) */
-  exists(AbstractValue dual | value = dual.getDualValue() |
-    compares_eq(test.(LogicalNotInstruction).getUnary(), op, k, areEqual, dual)
+  exists(AbstractValue dual, boolean inNonZeroCase0 |
+    value = dual.getDualValue() and
+    unary_compares_eq(test.(LogicalNotInstruction).getUnary(), op, k, inNonZeroCase0, areEqual, dual)
+  |
+    k = 0 and inNonZeroCase = inNonZeroCase0
+    or
+    k != 0 and inNonZeroCase = true
   )
   or
   // ((test is `areEqual` => op == const + k2) and const == `k1`) =>
   // test is `areEqual` => op == k1 + k2
+  inNonZeroCase = false and
   exists(int k1, int k2, ConstantInstruction const |
     compares_eq(test, op, const.getAUse(), k2, areEqual, value) and
     int_value(const) = k1 and
     k = k1 + k2
   )
+  or
+  unary_compares_eq(test.(BuiltinExpectCallInstruction).getCondition(), op, k, areEqual,
+    inNonZeroCase, value)
 }
 
 /** Rearrange various simple comparisons into `left == right + k` form. */
@@ -781,14 +855,96 @@ private predicate simple_comparison_eq(
   value.(BooleanValue).getValue() = false
 }
 
-/** Rearrange various simple comparisons into `op == k` form. */
-private predicate simple_comparison_eq(Instruction test, Operand op, int k, AbstractValue value) {
+/**
+ * Rearrange various simple comparisons into `op == k` form.
+ */
+private predicate unary_simple_comparison_eq(
+  Instruction test, int k, boolean inNonZeroCase, AbstractValue value
+) {
   exists(SwitchInstruction switch, CaseEdge case |
     test = switch.getExpression() and
-    op.getDef() = test and
     case = value.(MatchValue).getCase() and
     exists(switch.getSuccessor(case)) and
-    case.getValue().toInt() = k
+    case.getValue().toInt() = k and
+    inNonZeroCase = false
+  )
+  or
+  // Any instruction with an integral type could potentially be part of a
+  // check for nullness when used in a guard. So we include all integral
+  // typed instructions here. However, since some of these instructions are
+  // already included as guards in other cases, we exclude those here.
+  // These are instructions that compute a binary equality or inequality
+  // relation. For example, the following:
+  // ```cpp
+  // if(a == b + 42) { ... }
+  // ```
+  // generates the following IR:
+  // ```
+  // r1(glval<int>) = VariableAddress[a]     :
+  // r2(int)        = Load[a]                : &:r1, m1
+  // r3(glval<int>) = VariableAddress[b]     :
+  // r4(int)        = Load[b]                : &:r3, m2
+  // r5(int)        = Constant[42]           :
+  // r6(int)        = Add                    : r4, r5
+  // r7(bool)       = CompareEQ              : r2, r6
+  // v1(void)       = ConditionalBranch      : r7
+  // ```
+  // and since `r7` is an integral typed instruction this predicate could
+  // include a case for when `r7` evaluates to true (in which case we would
+  // infer that `r6` was non-zero, and a case for when `r7` evaluates to false
+  // (in which case we would infer that `r6` was zero).
+  // However, since `a == b + 42` is already supported when reasoning about
+  // binary equalities we exclude those cases here.
+  not test.isGLValue() and
+  not simple_comparison_eq(test, _, _, _, _) and
+  not simple_comparison_lt(test, _, _, _) and
+  not test = any(SwitchInstruction switch).getExpression() and
+  (
+    test.getResultIRType() instanceof IRAddressType or
+    test.getResultIRType() instanceof IRIntegerType or
+    test.getResultIRType() instanceof IRBooleanType
+  ) and
+  (
+    k = 1 and
+    value.(BooleanValue).getValue() = true and
+    inNonZeroCase = true
+    or
+    k = 0 and
+    value.(BooleanValue).getValue() = false and
+    inNonZeroCase = false
+  )
+}
+
+/** A call to the builtin operation `__builtin_expect`. */
+private class BuiltinExpectCallInstruction extends CallInstruction {
+  BuiltinExpectCallInstruction() { this.getStaticCallTarget().hasName("__builtin_expect") }
+
+  /** Gets the condition of this call. */
+  Instruction getCondition() {
+    // The first parameter of `__builtin_expect` has type `long`. So we skip
+    // the conversion when inferring guards.
+    result = this.getArgument(0).(ConvertInstruction).getUnary()
+  }
+}
+
+/**
+ * Holds if `left == right + k` is `areEqual` if `cmp` evaluates to `value`,
+ * and `cmp` is an instruction that compares the value of
+ * `__builtin_expect(left == right + k, _)` to `0`.
+ */
+private predicate builtin_expect_eq(
+  CompareInstruction cmp, Operand left, Operand right, int k, boolean areEqual, AbstractValue value
+) {
+  exists(BuiltinExpectCallInstruction call, Instruction const, AbstractValue innerValue |
+    int_value(const) = 0 and
+    cmp.hasOperands(call.getAUse(), const.getAUse()) and
+    compares_eq(call.getCondition(), left, right, k, areEqual, innerValue)
+  |
+    cmp instanceof CompareNEInstruction and
+    value = innerValue
+    or
+    cmp instanceof CompareEQInstruction and
+    value.getDualValue() = innerValue
   )
 }
 
@@ -798,14 +954,39 @@ private predicate complex_eq(
   sub_eq(cmp, left, right, k, areEqual, value)
   or
   add_eq(cmp, left, right, k, areEqual, value)
+  or
+  builtin_expect_eq(cmp, left, right, k, areEqual, value)
 }
 
-private predicate complex_eq(
-  Instruction test, Operand op, int k, boolean areEqual, AbstractValue value
+/**
+ * Holds if `op == k` is `areEqual` if `cmp` evaluates to `value`, and `cmp` is
+ * an instruction that compares the value of `__builtin_expect(op == k, _)` to `0`.
+ */
+private predicate unary_builtin_expect_eq(
+  CompareInstruction cmp, Operand op, int k, boolean areEqual, boolean inNonZeroCase,
+  AbstractValue value
 ) {
-  sub_eq(test, op, k, areEqual, value)
+  exists(BuiltinExpectCallInstruction call, Instruction const, AbstractValue innerValue |
+    int_value(const) = 0 and
+    cmp.hasOperands(call.getAUse(), const.getAUse()) and
+    unary_compares_eq(call.getCondition(), op, k, areEqual, inNonZeroCase, innerValue)
+  |
+    cmp instanceof CompareNEInstruction and
+    value = innerValue
+    or
+    cmp instanceof CompareEQInstruction and
+    value.getDualValue() = innerValue
+  )
+}
+
+private predicate unary_complex_eq(
+  Instruction test, Operand op, int k, boolean areEqual, boolean inNonZeroCase, AbstractValue value
+) {
+  unary_sub_eq(test, op, k, areEqual, inNonZeroCase, value)
   or
-  add_eq(test, op, k, areEqual, value)
+  unary_add_eq(test, op, k, areEqual, inNonZeroCase, value)
+  or
+  unary_builtin_expect_eq(test, op, k, areEqual, inNonZeroCase, value)
 }
 
 /*
@@ -834,7 +1015,8 @@ private predicate compares_lt(
 
 /** Holds if `op < k` evaluates to `isLt` given that `test` evaluates to `value`. */
 private predicate compares_lt(Instruction test, Operand op, int k, boolean isLt, AbstractValue value) {
-  simple_comparison_lt(test, op, k, isLt, value)
+  unary_simple_comparison_lt(test, k, isLt, value) and
+  op.getDef() = test
   or
   complex_lt(test, op, k, isLt, value)
   or
@@ -881,12 +1063,11 @@ private predicate simple_comparison_lt(CompareInstruction cmp, Operand left, Ope
 }
 
 /** Rearrange various simple comparisons into `op < k` form. */
-private predicate simple_comparison_lt(
-  Instruction test, Operand op, int k, boolean isLt, AbstractValue value
+private predicate unary_simple_comparison_lt(
+  Instruction test, int k, boolean isLt, AbstractValue value
 ) {
   exists(SwitchInstruction switch, CaseEdge case |
     test = switch.getExpression() and
-    op.getDef() = test and
     case = value.(MatchValue).getCase() and
     exists(switch.getSuccessor(case)) and
     case.getMaxValue() > case.getMinValue()
@@ -1069,16 +1250,20 @@ private predicate sub_eq(
 }
 
 // op - x == c => op == (c+x)
-private predicate sub_eq(Instruction test, Operand op, int k, boolean areEqual, AbstractValue value) {
+private predicate unary_sub_eq(
+  Instruction test, Operand op, int k, boolean areEqual, boolean inNonZeroCase, AbstractValue value
+) {
+  inNonZeroCase = false and
   exists(SubInstruction sub, int c, int x |
-    compares_eq(test, sub.getAUse(), c, areEqual, value) and
+    unary_compares_eq(test, sub.getAUse(), c, areEqual, inNonZeroCase, value) and
     op = sub.getLeftOperand() and
     x = int_value(sub.getRight()) and
     k = c + x
   )
   or
+  inNonZeroCase = false and
   exists(PointerSubInstruction sub, int c, int x |
-    compares_eq(test, sub.getAUse(), c, areEqual, value) and
+    unary_compares_eq(test, sub.getAUse(), c, areEqual, inNonZeroCase, value) and
     op = sub.getLeftOperand() and
     x = int_value(sub.getRight()) and
     k = c + x
@@ -1132,11 +1317,13 @@ private predicate add_eq(
 }
 
 // left + x == right + c => left == right + (c-x)
-private predicate add_eq(
-  Instruction test, Operand left, int k, boolean areEqual, AbstractValue value
+private predicate unary_add_eq(
+  Instruction test, Operand left, int k, boolean areEqual, boolean inNonZeroCase,
+  AbstractValue value
 ) {
+  inNonZeroCase = false and
   exists(AddInstruction lhs, int c, int x |
-    compares_eq(test, lhs.getAUse(), c, areEqual, value) and
+    unary_compares_eq(test, lhs.getAUse(), c, areEqual, inNonZeroCase, value) and
     (
       left = lhs.getLeftOperand() and x = int_value(lhs.getRight())
       or
@@ -1145,8 +1332,9 @@ private predicate add_eq(
     k = c - x
   )
   or
+  inNonZeroCase = false and
   exists(PointerAddInstruction lhs, int c, int x |
-    compares_eq(test, lhs.getAUse(), c, areEqual, value) and
+    unary_compares_eq(test, lhs.getAUse(), c, areEqual, inNonZeroCase, value) and
     (
       left = lhs.getLeftOperand() and x = int_value(lhs.getRight())
       or
@@ -1156,5 +1344,14 @@ private predicate add_eq(
   )
 }
 
+private class IntegerOrPointerConstantInstruction extends ConstantInstruction {
+  IntegerOrPointerConstantInstruction() {
+    this instanceof IntegerConstantInstruction or
+    this instanceof PointerConstantInstruction
+  }
+}
+
 /** The int value of integer constant expression. */
-private int int_value(Instruction i) { result = i.(IntegerConstantInstruction).getValue().toInt() }
+private int int_value(Instruction i) {
+  result = i.(IntegerOrPointerConstantInstruction).getValue().toInt()
+}

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

@@ -0,0 +1,567 @@
+/**
+ * INTERNAL use only. This is an experimental API subject to change without notice.
+ *
+ * Provides classes and predicates for dealing with flow models specified in CSV format.
+ *
+ * The CSV specification has the following columns:
+ * - Sources:
+ *   `namespace; type; subtypes; name; signature; ext; output; kind`
+ * - Sinks:
+ *   `namespace; type; subtypes; name; signature; ext; input; kind`
+ * - Summaries:
+ *   `namespace; type; subtypes; name; signature; ext; input; output; kind`
+ *
+ * The interpretation of a row is similar to API-graphs with a left-to-right
+ * reading.
+ * 1. The `namespace` column selects a namespace.
+ * 2. The `type` column selects a type within that namespace.
+ * 3. The `subtypes` is a boolean that indicates whether to jump to an
+ *    arbitrary subtype of that type. Set this to `false` if leaving the `type`
+ *    blank (for example, a free function).
+ * 4. The `name` column optionally selects a specific named member of the type.
+ * 5. The `signature` column optionally restricts the named member. If
+ *    `signature` is blank then no such filtering is done. The format of the
+ *    signature is a comma-separated list of types enclosed in parentheses. The
+ *    types can be short names or fully qualified names (mixing these two options
+ *    is not allowed within a single signature).
+ * 6. The `ext` column specifies additional API-graph-like edges. Currently
+ *    there is only one valid value: "".
+ * 7. The `input` column specifies how data enters the element selected by the
+ *    first 6 columns, and the `output` column specifies how data leaves the
+ *    element selected by the first 6 columns. An `input` can be either:
+ *    - "": Selects a write to the selected element in case this is a field.
+ *    - "Argument[n]": Selects an argument in a call to the selected element.
+ *      The arguments are zero-indexed, and `-1` specifies the qualifier object,
+ *      that is, `*this`.
+ *      - one or more "*" can be added in front of the argument index to indicate
+ *        indirection, for example, `Argument[*0]` indicates the first indirection
+ *        of the 0th argument.
+ *      - `n1..n2` syntax can be used to indicate a range of arguments, inclusive
+ *        at both ends. One or more "*"s can be added in front of the whole range
+ *        to indicate that every argument in the range is indirect, for example
+ *        `*0..1` is the first indirection of both arguments 0 and 1.
+ *    - "ReturnValue": Selects a value being returned by the selected element.
+ *      One or more "*" can be added as an argument to indicate indirection, for
+ *      example, "ReturnValue[*]" indicates the first indirection of the return
+ *      value.
+ *
+ *    An `output` can be either:
+ *    - "": Selects a read of a selected field.
+ *    - "Argument[n]": Selects the post-update value of an argument in a call to
+ *      the selected element. That is, the value of the argument after the call
+ *      returns. The arguments are zero-indexed, and `-1` specifies the qualifier
+ *      object, that is, `*this`.
+ *      - one or more "*" can be added in front of the argument index to indicate
+ *        indirection, for example, `Argument[*0]` indicates the first indirection
+ *        of the 0th argument.
+ *      - `n1..n2` syntax can be used to indicate a range of arguments, inclusive
+ *        at both ends. One or more "*"s can be added in front of the whole range
+ *        to indicate that every argument in the range is indirect, for example
+ *        `*0..1` is the first indirection of both arguments 0 and 1.
+ *    - "Parameter[n]": Selects the value of a parameter of the selected element.
+ *      The syntax is the same as for "Argument", for example "Parameter[0]",
+ *      "Parameter[*0]", "Parameter[0..2]" etc.
+ *    - "ReturnValue": Selects a value being returned by the selected element.
+ *      One or more "*" can be added as an argument to indicate indirection, for
+ *      example, "ReturnValue[*]" indicates the first indirection of the return
+ *      value.
+ * 8. The `kind` column is a tag that can be referenced from QL to determine to
+ *    which classes the interpreted elements should be added. For example, for
+ *    sources "remote" indicates a default remote flow source, and for summaries
+ *    "taint" indicates a default additional taint step and "value" indicates a
+ *    globally applicable value-preserving step.
+ */
+
+import cpp
+private import new.DataFlow
+private import internal.FlowSummaryImpl
+private import internal.FlowSummaryImpl::Public
+private import internal.FlowSummaryImpl::Private
+private import internal.FlowSummaryImpl::Private::External
+private import internal.ExternalFlowExtensions as Extensions
+private import codeql.mad.ModelValidation as SharedModelVal
+private import codeql.util.Unit
+
+/**
+ * A unit class for adding additional source model rows.
+ *
+ * Extend this class to add additional source definitions.
+ */
+class SourceModelCsv extends Unit {
+  /** Holds if `row` specifies a source definition. */
+  abstract predicate row(string row);
+}
+
+/**
+ * A unit class for adding additional sink model rows.
+ *
+ * Extend this class to add additional sink definitions.
+ */
+class SinkModelCsv extends Unit {
+  /** Holds if `row` specifies a sink definition. */
+  abstract predicate row(string row);
+}
+
+/**
+ * A unit class for adding additional summary model rows.
+ *
+ * Extend this class to add additional flow summary definitions.
+ */
+class SummaryModelCsv extends Unit {
+  /** Holds if `row` specifies a summary definition. */
+  abstract predicate row(string row);
+}
+
+/** Holds if `row` is a source model. */
+predicate sourceModel(string row) { any(SourceModelCsv s).row(row) }
+
+/** Holds if `row` is a sink model. */
+predicate sinkModel(string row) { any(SinkModelCsv s).row(row) }
+
+/** Holds if `row` is a summary model. */
+predicate summaryModel(string row) { any(SummaryModelCsv s).row(row) }
+
+/** Holds if a source model exists for the given parameters. */
+predicate sourceModel(
+  string namespace, string type, boolean subtypes, string name, string signature, string ext,
+  string output, string kind, string provenance
+) {
+  exists(string row |
+    sourceModel(row) and
+    row.splitAt(";", 0) = namespace and
+    row.splitAt(";", 1) = type and
+    row.splitAt(";", 2) = subtypes.toString() and
+    subtypes = [true, false] and
+    row.splitAt(";", 3) = name and
+    row.splitAt(";", 4) = signature and
+    row.splitAt(";", 5) = ext and
+    row.splitAt(";", 6) = output and
+    row.splitAt(";", 7) = kind
+  ) and
+  provenance = "manual"
+  or
+  Extensions::sourceModel(namespace, type, subtypes, name, signature, ext, output, kind, provenance,
+    _)
+}
+
+/** Holds if a sink model exists for the given parameters. */
+predicate sinkModel(
+  string namespace, string type, boolean subtypes, string name, string signature, string ext,
+  string input, string kind, string provenance
+) {
+  exists(string row |
+    sinkModel(row) and
+    row.splitAt(";", 0) = namespace and
+    row.splitAt(";", 1) = type and
+    row.splitAt(";", 2) = subtypes.toString() and
+    subtypes = [true, false] and
+    row.splitAt(";", 3) = name and
+    row.splitAt(";", 4) = signature and
+    row.splitAt(";", 5) = ext and
+    row.splitAt(";", 6) = input and
+    row.splitAt(";", 7) = kind
+  ) and
+  provenance = "manual"
+  or
+  Extensions::sinkModel(namespace, type, subtypes, name, signature, ext, input, kind, provenance, _)
+}
+
+/** Holds if a summary model exists for the given parameters. */
+predicate summaryModel(
+  string namespace, string type, boolean subtypes, string name, string signature, string ext,
+  string input, string output, string kind, string provenance
+) {
+  exists(string row |
+    summaryModel(row) and
+    row.splitAt(";", 0) = namespace and
+    row.splitAt(";", 1) = type and
+    row.splitAt(";", 2) = subtypes.toString() and
+    subtypes = [true, false] and
+    row.splitAt(";", 3) = name and
+    row.splitAt(";", 4) = signature and
+    row.splitAt(";", 5) = ext and
+    row.splitAt(";", 6) = input and
+    row.splitAt(";", 7) = output and
+    row.splitAt(";", 8) = kind
+  ) and
+  provenance = "manual"
+  or
+  Extensions::summaryModel(namespace, type, subtypes, name, signature, ext, input, output, kind,
+    provenance, _)
+}
+
+private predicate relevantNamespace(string namespace) {
+  sourceModel(namespace, _, _, _, _, _, _, _, _) or
+  sinkModel(namespace, _, _, _, _, _, _, _, _) or
+  summaryModel(namespace, _, _, _, _, _, _, _, _, _)
+}
+
+private predicate namespaceLink(string shortns, string longns) {
+  relevantNamespace(shortns) and
+  relevantNamespace(longns) and
+  longns.prefix(longns.indexOf("::")) = shortns
+}
+
+private predicate canonicalNamespace(string namespace) {
+  relevantNamespace(namespace) and not namespaceLink(_, namespace)
+}
+
+private predicate canonicalNamespaceLink(string namespace, string subns) {
+  canonicalNamespace(namespace) and
+  (subns = namespace or namespaceLink(namespace, subns))
+}
+
+/**
+ * Holds if MaD framework coverage of `namespace` is `n` api endpoints of the
+ * kind `(kind, part)`, and `namespaces` is the number of subnamespaces of
+ * `namespace` which have MaD framework coverage (including `namespace`
+ * itself).
+ */
+predicate modelCoverage(string namespace, int namespaces, string kind, string part, int n) {
+  namespaces = strictcount(string subns | canonicalNamespaceLink(namespace, subns)) and
+  (
+    part = "source" and
+    n =
+      strictcount(string subns, string type, boolean subtypes, string name, string signature,
+        string ext, string output, string provenance |
+        canonicalNamespaceLink(namespace, subns) and
+        sourceModel(subns, type, subtypes, name, signature, ext, output, kind, provenance)
+      )
+    or
+    part = "sink" and
+    n =
+      strictcount(string subns, string type, boolean subtypes, string name, string signature,
+        string ext, string input, string provenance |
+        canonicalNamespaceLink(namespace, subns) and
+        sinkModel(subns, type, subtypes, name, signature, ext, input, kind, provenance)
+      )
+    or
+    part = "summary" and
+    n =
+      strictcount(string subns, string type, boolean subtypes, string name, string signature,
+        string ext, string input, string output, string provenance |
+        canonicalNamespaceLink(namespace, subns) and
+        summaryModel(subns, type, subtypes, name, signature, ext, input, output, kind, provenance)
+      )
+  )
+}
+
+/** Provides a query predicate to check the CSV data for validation errors. */
+module CsvValidation {
+  private string getInvalidModelInput() {
+    exists(string pred, AccessPath input, string part |
+      sinkModel(_, _, _, _, _, _, input, _, _) and pred = "sink"
+      or
+      summaryModel(_, _, _, _, _, _, input, _, _, _) and pred = "summary"
+    |
+      (
+        invalidSpecComponent(input, part) and
+        not part = "" and
+        not (part = "Argument" and pred = "sink") and
+        not parseArg(part, _)
+        or
+        part = input.getToken(_) and
+        parseParam(part, _)
+      ) and
+      result = "Unrecognized input specification \"" + part + "\" in " + pred + " model."
+    )
+  }
+
+  private string getInvalidModelOutput() {
+    exists(string pred, string output, string part |
+      sourceModel(_, _, _, _, _, _, output, _, _) and pred = "source"
+      or
+      summaryModel(_, _, _, _, _, _, _, output, _, _) and pred = "summary"
+    |
+      invalidSpecComponent(output, part) and
+      not part = "" and
+      not (part = ["Argument", "Parameter"] and pred = "source") and
+      result = "Unrecognized output specification \"" + part + "\" in " + pred + " model."
+    )
+  }
+
+  private module KindValConfig implements SharedModelVal::KindValidationConfigSig {
+    predicate summaryKind(string kind) { summaryModel(_, _, _, _, _, _, _, _, kind, _) }
+
+    predicate sinkKind(string kind) { sinkModel(_, _, _, _, _, _, _, kind, _) }
+
+    predicate sourceKind(string kind) { sourceModel(_, _, _, _, _, _, _, kind, _) }
+  }
+
+  private module KindVal = SharedModelVal::KindValidation<KindValConfig>;
+
+  private string getInvalidModelSubtype() {
+    exists(string pred, string row |
+      sourceModel(row) and pred = "source"
+      or
+      sinkModel(row) and pred = "sink"
+      or
+      summaryModel(row) and pred = "summary"
+    |
+      exists(string b |
+        b = row.splitAt(";", 2) and
+        not b = ["true", "false"] and
+        result = "Invalid boolean \"" + b + "\" in " + pred + " model."
+      )
+    )
+  }
+
+  private string getInvalidModelColumnCount() {
+    exists(string pred, string row, int expect |
+      sourceModel(row) and expect = 8 and pred = "source"
+      or
+      sinkModel(row) and expect = 8 and pred = "sink"
+      or
+      summaryModel(row) and expect = 9 and pred = "summary"
+    |
+      exists(int cols |
+        cols = 1 + max(int n | exists(row.splitAt(";", n))) and
+        cols != expect and
+        result =
+          "Wrong number of columns in " + pred + " model row, expected " + expect + ", got " + cols +
+            "."
+      )
+    )
+  }
+
+  private string getInvalidModelSignature() {
+    exists(string pred, string namespace, string type, string name, string signature, string ext |
+      sourceModel(namespace, type, _, name, signature, ext, _, _, _) and pred = "source"
+      or
+      sinkModel(namespace, type, _, name, signature, ext, _, _, _) and pred = "sink"
+      or
+      summaryModel(namespace, type, _, name, signature, ext, _, _, _, _) and pred = "summary"
+    |
+      not namespace.regexpMatch("[a-zA-Z0-9_\\.:]*") and
+      result = "Dubious namespace \"" + namespace + "\" in " + pred + " model."
+      or
+      not type.regexpMatch("[a-zA-Z0-9_<>,\\+]*") and
+      result = "Dubious type \"" + type + "\" in " + pred + " model."
+      or
+      not name.regexpMatch("[a-zA-Z0-9_<>,]*") and
+      result = "Dubious member name \"" + name + "\" in " + pred + " model."
+      or
+      not signature.regexpMatch("|\\([a-zA-Z0-9_<>\\.\\+\\*,\\[\\]]*\\)") and
+      result = "Dubious signature \"" + signature + "\" in " + pred + " model."
+      or
+      not ext.regexpMatch("|Attribute") and
+      result = "Unrecognized extra API graph element \"" + ext + "\" in " + pred + " model."
+    )
+  }
+
+  /** Holds if some row in a CSV-based flow model appears to contain typos. */
+  query predicate invalidModelRow(string msg) {
+    msg =
+      [
+        getInvalidModelSignature(), getInvalidModelInput(), getInvalidModelOutput(),
+        getInvalidModelSubtype(), getInvalidModelColumnCount(), KindVal::getInvalidModelKind()
+      ]
+  }
+}
+
+private predicate elementSpec(
+  string namespace, string type, boolean subtypes, string name, string signature, string ext
+) {
+  sourceModel(namespace, type, subtypes, name, signature, ext, _, _, _) or
+  sinkModel(namespace, type, subtypes, name, signature, ext, _, _, _) or
+  summaryModel(namespace, type, subtypes, name, signature, ext, _, _, _, _)
+}
+
+private string paramsStringPart(Function c, int i) {
+  i = -1 and result = "(" and exists(c)
+  or
+  exists(int n, string p | c.getParameter(n).getType().toString() = p |
+    i = 2 * n and result = p
+    or
+    i = 2 * n - 1 and result = "," and n != 0
+  )
+  or
+  i = 2 * c.getNumberOfParameters() and result = ")"
+}
+
+/**
+ * Gets a parenthesized string containing all parameter types of this callable, separated by a comma.
+ *
+ * Returns the empty string if the callable has no parameters.
+ * Parameter types are represented by their type erasure.
+ */
+cached
+private string paramsString(Function c) {
+  result = concat(int i | | paramsStringPart(c, i) order by i)
+}
+
+bindingset[func]
+private predicate matchesSignature(Function func, string signature) {
+  signature = "" or
+  paramsString(func) = signature
+}
+
+/**
+ * Gets the element in module `namespace` that satisfies the following properties:
+ * 1. If the element is a member of a class-like type, then the class-like type has name `type`
+ * 2. If `subtypes = true` and the element is a member of a class-like type, then overrides of the element
+ *    are also returned.
+ * 3. The element has name `name`
+ * 4. If `signature` is non-empty, then the element has a list of parameter types described by `signature`.
+ *
+ * NOTE: `namespace` is currently not used (since we don't properly extract modules yet).
+ */
+pragma[nomagic]
+private Element interpretElement0(
+  string namespace, string type, boolean subtypes, string name, string signature
+) {
+  elementSpec(namespace, type, subtypes, name, signature, _) and
+  (
+    // Non-member functions
+    exists(Function func |
+      func.hasQualifiedName(namespace, name) and
+      type = "" and
+      matchesSignature(func, signature) and
+      subtypes = false and
+      not exists(func.getDeclaringType()) and
+      result = func
+    )
+    or
+    // Member functions
+    exists(Class namedClass, Class classWithMethod, Function method |
+      classWithMethod = method.getClassAndName(name) and
+      namedClass.hasQualifiedName(namespace, type) and
+      matchesSignature(method, signature) and
+      result = method
+    |
+      // member declared in the named type or a subtype of it
+      subtypes = true and
+      classWithMethod = namedClass.getADerivedClass*()
+      or
+      // member declared directly in the named type
+      subtypes = false and
+      classWithMethod = namedClass
+    )
+    or
+    // Member variables
+    signature = "" and
+    exists(Class namedClass, Class classWithMember, MemberVariable member |
+      member.getName() = name and
+      member = classWithMember.getAMember() and
+      namedClass.hasQualifiedName(namespace, type) and
+      result = member
+    |
+      // field declared in the named type or a subtype of it (or an extension of any)
+      subtypes = true and
+      classWithMember = namedClass.getADerivedClass*()
+      or
+      // field declared directly in the named type (or an extension of it)
+      subtypes = false and
+      classWithMember = namedClass
+    )
+    or
+    // Global or namespace variables
+    signature = "" and
+    type = "" and
+    subtypes = false and
+    result = any(GlobalOrNamespaceVariable v | v.hasQualifiedName(namespace, name))
+  )
+}
+
+/** Gets the source/sink/summary element corresponding to the supplied parameters. */
+Element interpretElement(
+  string namespace, string type, boolean subtypes, string name, string signature, string ext
+) {
+  elementSpec(namespace, type, subtypes, name, signature, ext) and
+  exists(Element e | e = interpretElement0(namespace, type, subtypes, name, signature) |
+    ext = "" and result = e
+  )
+}
+
+cached
+private module Cached {
+  /**
+   * Holds if `node` is specified as a source with the given kind in a CSV flow
+   * model.
+   */
+  cached
+  predicate sourceNode(DataFlow::Node node, string kind) {
+    exists(SourceSinkInterpretationInput::InterpretNode n |
+      isSourceNode(n, kind, _) and n.asNode() = node // TODO
+    )
+  }
+
+  /**
+   * Holds if `node` is specified as a sink with the given kind in a CSV flow
+   * model.
+   */
+  cached
+  predicate sinkNode(DataFlow::Node node, string kind) {
+    exists(SourceSinkInterpretationInput::InterpretNode n |
+      isSinkNode(n, kind, _) and n.asNode() = node // TODO
+    )
+  }
+}
+
+import Cached
+
+private predicate interpretSummary(
+  Function f, string input, string output, string kind, string provenance
+) {
+  exists(
+    string namespace, string type, boolean subtypes, string name, string signature, string ext
+  |
+    summaryModel(namespace, type, subtypes, name, signature, ext, input, output, kind, provenance) and
+    f = interpretElement(namespace, type, subtypes, name, signature, ext)
+  )
+}
+
+// adapter class for converting Mad summaries to `SummarizedCallable`s
+private class SummarizedCallableAdapter extends SummarizedCallable {
+  SummarizedCallableAdapter() { interpretSummary(this, _, _, _, _) }
+
+  private predicate relevantSummaryElementManual(string input, string output, string kind) {
+    exists(Provenance provenance |
+      interpretSummary(this, input, output, kind, provenance) and
+      provenance.isManual()
+    )
+  }
+
+  private predicate relevantSummaryElementGenerated(string input, string output, string kind) {
+    exists(Provenance provenance |
+      interpretSummary(this, input, output, kind, provenance) and
+      provenance.isGenerated()
+    )
+  }
+
+  override predicate propagatesFlow(
+    string input, string output, boolean preservesValue, string model
+  ) {
+    exists(string kind |
+      this.relevantSummaryElementManual(input, output, kind)
+      or
+      not this.relevantSummaryElementManual(_, _, _) and
+      this.relevantSummaryElementGenerated(input, output, kind)
+    |
+      if kind = "value" then preservesValue = true else preservesValue = false
+    ) and
+    model = "" // TODO
+  }
+
+  override predicate hasProvenance(Provenance provenance) {
+    interpretSummary(this, _, _, _, provenance)
+  }
+}
+
+// adapter class for converting Mad neutrals to `NeutralCallable`s
+private class NeutralCallableAdapter extends NeutralCallable {
+  string kind;
+  string provenance_;
+
+  NeutralCallableAdapter() {
+    // Neutral models have not been implemented for CPP.
+    none() and
+    exists(this) and
+    exists(kind) and
+    exists(provenance_)
+  }
+
+  override string getKind() { result = kind }
+
+  override predicate hasProvenance(Provenance provenance) { provenance = provenance_ }
+}

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

@@ -9,7 +9,7 @@ private import DataFlowUtil
 /**
  * Gets a function that might be called by `call`.
  */
-Function viableCallable(DataFlowCall call) {
+DataFlowCallable viableCallable(DataFlowCall call) {
   result = call.(Call).getTarget()
   or
   // If the target of the call does not have a body in the snapshot, it might

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

@@ -263,9 +263,10 @@ deprecated private module Config implements FullStateConfigSig {
 
   predicate isBarrierOut(Node node, FlowState state) { none() }
 
-  predicate isAdditionalFlowStep(Node node1, Node node2) {
+  predicate isAdditionalFlowStep(Node node1, Node node2, string model) {
     singleConfiguration() and
-    any(Configuration config).isAdditionalFlowStep(node1, node2)
+    any(Configuration config).isAdditionalFlowStep(node1, node2) and
+    model = ""
   }
 
   predicate isAdditionalFlowStep(Node node1, FlowState state1, Node node2, FlowState state2) {

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

@@ -263,9 +263,10 @@ deprecated private module Config implements FullStateConfigSig {
 
   predicate isBarrierOut(Node node, FlowState state) { none() }
 
-  predicate isAdditionalFlowStep(Node node1, Node node2) {
+  predicate isAdditionalFlowStep(Node node1, Node node2, string model) {
     singleConfiguration() and
-    any(Configuration config).isAdditionalFlowStep(node1, node2)
+    any(Configuration config).isAdditionalFlowStep(node1, node2) and
+    model = ""
   }
 
   predicate isAdditionalFlowStep(Node node1, FlowState state1, Node node2, FlowState state2) {

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

@@ -263,9 +263,10 @@ deprecated private module Config implements FullStateConfigSig {
 
   predicate isBarrierOut(Node node, FlowState state) { none() }
 
-  predicate isAdditionalFlowStep(Node node1, Node node2) {
+  predicate isAdditionalFlowStep(Node node1, Node node2, string model) {
     singleConfiguration() and
-    any(Configuration config).isAdditionalFlowStep(node1, node2)
+    any(Configuration config).isAdditionalFlowStep(node1, node2) and
+    model = ""
   }
 
   predicate isAdditionalFlowStep(Node node1, FlowState state1, Node node2, FlowState state2) {

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

@@ -263,9 +263,10 @@ deprecated private module Config implements FullStateConfigSig {
 
   predicate isBarrierOut(Node node, FlowState state) { none() }
 
-  predicate isAdditionalFlowStep(Node node1, Node node2) {
+  predicate isAdditionalFlowStep(Node node1, Node node2, string model) {
     singleConfiguration() and
-    any(Configuration config).isAdditionalFlowStep(node1, node2)
+    any(Configuration config).isAdditionalFlowStep(node1, node2) and
+    model = ""
   }
 
   predicate isAdditionalFlowStep(Node node1, FlowState state1, Node node2, FlowState state2) {

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

@@ -263,9 +263,10 @@ deprecated private module Config implements FullStateConfigSig {
 
   predicate isBarrierOut(Node node, FlowState state) { none() }
 
-  predicate isAdditionalFlowStep(Node node1, Node node2) {
+  predicate isAdditionalFlowStep(Node node1, Node node2, string model) {
     singleConfiguration() and
-    any(Configuration config).isAdditionalFlowStep(node1, node2)
+    any(Configuration config).isAdditionalFlowStep(node1, node2) and
+    model = ""
   }
 
   predicate isAdditionalFlowStep(Node node1, FlowState state1, Node node2, FlowState state2) {

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

@@ -216,7 +216,7 @@ predicate localMustFlowStep(Node node1, Node node2) { none() }
 
 /** Gets the type of `n` used for type pruning. */
 Type getNodeType(Node n) {
-  suppressUnusedNode(n) and
+  exists(n) and
   result instanceof VoidType // stub implementation
 }
 
@@ -227,13 +227,10 @@ string ppReprType(Type t) { none() } // stub implementation
  * Holds if `t1` and `t2` are compatible, that is, whether data can flow from
  * a node of type `t1` to a node of type `t2`.
  */
-pragma[inline]
 predicate compatibleTypes(Type t1, Type t2) {
-  any() // stub implementation
+  t1 instanceof VoidType and t2 instanceof VoidType // stub implementation
 }
 
-private predicate suppressUnusedNode(Node n) { any() }
-
 //////////////////////////////////////////////////////////////////////////////
 // Java QL library compatibility wrappers
 //////////////////////////////////////////////////////////////////////////////
@@ -242,7 +239,17 @@ class CastNode extends Node {
   CastNode() { none() } // stub implementation
 }
 
-class DataFlowCallable = Function;
+class DataFlowCallable extends Function {
+  /** Gets a best-effort total ordering. */
+  int totalorder() {
+    this =
+      rank[result](DataFlowCallable c, string file, int startline, int startcolumn |
+        c.getLocation().hasLocationInfo(file, startline, startcolumn, _, _)
+      |
+        c order by file, startline, startcolumn
+      )
+  }
+}
 
 class DataFlowExpr = Expr;
 
@@ -261,10 +268,28 @@ class DataFlowCall extends Expr instanceof Call {
   ExprNode getNode() { result.getExpr() = this }
 
   /** Gets the enclosing callable of this call. */
-  Function getEnclosingCallable() { result = this.getEnclosingFunction() }
+  DataFlowCallable getEnclosingCallable() { result = this.getEnclosingFunction() }
+
+  /** Gets a best-effort total ordering. */
+  int totalorder() {
+    this =
+      rank[result](DataFlowCall c, int startline, int startcolumn |
+        c.getLocation().hasLocationInfo(_, startline, startcolumn, _, _)
+      |
+        c order by startline, startcolumn
+      )
+  }
 }
 
-predicate isUnreachableInCall(Node n, DataFlowCall call) { none() } // stub implementation
+class NodeRegion instanceof Unit {
+  string toString() { result = "NodeRegion" }
+
+  predicate contains(Node n) { none() }
+
+  int totalOrder() { result = 1 }
+}
+
+predicate isUnreachableInCall(NodeRegion nr, DataFlowCall call) { none() } // stub implementation
 
 /**
  * Holds if access paths with `c` at their head always should be tracked at high
@@ -286,6 +311,12 @@ predicate lambdaCall(DataFlowCall call, LambdaCallKind kind, Node receiver) { no
 /** Extra data-flow steps needed for lambda flow analysis. */
 predicate additionalLambdaFlowStep(Node nodeFrom, Node nodeTo, boolean preservesValue) { none() }
 
+predicate knownSourceModel(Node source, string model) { none() }
+
+predicate knownSinkModel(Node sink, string model) { none() }
+
+class DataFlowSecondLevelScope = Unit;
+
 /**
  * Holds if flow is allowed to pass from parameter `p` and back to itself as a
  * side-effect, resulting in a summary from `p` to itself.

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

@@ -516,7 +516,7 @@ private module ThisFlow {
  */
 cached
 predicate localFlowStep(Node nodeFrom, Node nodeTo) {
-  simpleLocalFlowStep(nodeFrom, nodeTo)
+  simpleLocalFlowStep(nodeFrom, nodeTo, _)
   or
   // Field flow is not strictly a "step" but covers the whole function
   // transitively. There's no way to get a step-like relation out of the global
@@ -530,64 +530,67 @@ predicate localFlowStep(Node nodeFrom, Node nodeTo) {
  * This is the local flow predicate that's used as a building block in global
  * data flow. It may have less flow than the `localFlowStep` predicate.
  */
-predicate simpleLocalFlowStep(Node nodeFrom, Node nodeTo) {
-  // Expr -> Expr
-  exprToExprStep_nocfg(nodeFrom.asExpr(), nodeTo.asExpr())
-  or
-  // Assignment -> LValue post-update node
-  //
-  // This is used for assignments whose left-hand side is not a variable
-  // assignment or a storeStep but is still modeled by other means. It could be
-  // a call to `operator*` or `operator[]` where taint should flow to the
-  // post-update node of the qualifier.
-  exists(AssignExpr assign |
-    nodeFrom.asExpr() = assign and
-    nodeTo.(PostUpdateNode).getPreUpdateNode().asExpr() = assign.getLValue()
-  )
-  or
-  // Node -> FlowVar -> VariableAccess
-  exists(FlowVar var |
-    (
-      exprToVarStep(nodeFrom.asExpr(), var)
-      or
-      varSourceBaseCase(var, nodeFrom.asParameter())
-      or
-      varSourceBaseCase(var, nodeFrom.asUninitialized())
-      or
-      var.definedPartiallyAt(nodeFrom.asPartialDefinition())
-    ) and
-    varToNodeStep(var, nodeTo)
-  )
-  or
-  // Expr -> DefinitionByReferenceNode
-  exprToDefinitionByReferenceStep(nodeFrom.asExpr(), nodeTo.asDefiningArgument())
-  or
-  // `this` -> adjacent-`this`
-  ThisFlow::adjacentThisRefs(nodeFrom, nodeTo)
-  or
-  // post-update-`this` -> following-`this`-ref
-  ThisFlow::adjacentThisRefs(nodeFrom.(PostUpdateNode).getPreUpdateNode(), nodeTo)
-  or
-  // In `f(&x->a)`, this step provides the flow from post-`&` to post-`x->a`,
-  // from which there is field flow to `x` via reverse read.
-  exists(PartialDefinition def, Expr inner, Expr outer |
-    def.definesExpressions(inner, outer) and
-    inner = nodeTo.(InnerPartialDefinitionNode).getPreUpdateNode().asExpr() and
-    outer = nodeFrom.(PartialDefinitionNode).getPreUpdateNode().asExpr()
-  )
-  or
-  // Reverse flow: data that flows from the post-update node of a reference
-  // returned by a function call, back into the qualifier of that function.
-  // This allows data to flow 'in' through references returned by a modeled
-  // function such as `operator[]`.
-  exists(DataFlowFunction f, Call call, FunctionInput inModel, FunctionOutput outModel |
-    call.getTarget() = f and
-    inModel.isReturnValueDeref() and
-    outModel.isQualifierObject() and
-    f.hasDataFlow(inModel, outModel) and
-    nodeFrom.(PostUpdateNode).getPreUpdateNode().asExpr() = call and
-    nodeTo.asDefiningArgument() = call.getQualifier()
-  )
+predicate simpleLocalFlowStep(Node nodeFrom, Node nodeTo, string model) {
+  (
+    // Expr -> Expr
+    exprToExprStep_nocfg(nodeFrom.asExpr(), nodeTo.asExpr())
+    or
+    // Assignment -> LValue post-update node
+    //
+    // This is used for assignments whose left-hand side is not a variable
+    // assignment or a storeStep but is still modeled by other means. It could be
+    // a call to `operator*` or `operator[]` where taint should flow to the
+    // post-update node of the qualifier.
+    exists(AssignExpr assign |
+      nodeFrom.asExpr() = assign and
+      nodeTo.(PostUpdateNode).getPreUpdateNode().asExpr() = assign.getLValue()
+    )
+    or
+    // Node -> FlowVar -> VariableAccess
+    exists(FlowVar var |
+      (
+        exprToVarStep(nodeFrom.asExpr(), var)
+        or
+        varSourceBaseCase(var, nodeFrom.asParameter())
+        or
+        varSourceBaseCase(var, nodeFrom.asUninitialized())
+        or
+        var.definedPartiallyAt(nodeFrom.asPartialDefinition())
+      ) and
+      varToNodeStep(var, nodeTo)
+    )
+    or
+    // Expr -> DefinitionByReferenceNode
+    exprToDefinitionByReferenceStep(nodeFrom.asExpr(), nodeTo.asDefiningArgument())
+    or
+    // `this` -> adjacent-`this`
+    ThisFlow::adjacentThisRefs(nodeFrom, nodeTo)
+    or
+    // post-update-`this` -> following-`this`-ref
+    ThisFlow::adjacentThisRefs(nodeFrom.(PostUpdateNode).getPreUpdateNode(), nodeTo)
+    or
+    // In `f(&x->a)`, this step provides the flow from post-`&` to post-`x->a`,
+    // from which there is field flow to `x` via reverse read.
+    exists(PartialDefinition def, Expr inner, Expr outer |
+      def.definesExpressions(inner, outer) and
+      inner = nodeTo.(InnerPartialDefinitionNode).getPreUpdateNode().asExpr() and
+      outer = nodeFrom.(PartialDefinitionNode).getPreUpdateNode().asExpr()
+    )
+    or
+    // Reverse flow: data that flows from the post-update node of a reference
+    // returned by a function call, back into the qualifier of that function.
+    // This allows data to flow 'in' through references returned by a modeled
+    // function such as `operator[]`.
+    exists(DataFlowFunction f, Call call, FunctionInput inModel, FunctionOutput outModel |
+      call.getTarget() = f and
+      inModel.isReturnValueDeref() and
+      outModel.isQualifierObject() and
+      f.hasDataFlow(inModel, outModel) and
+      nodeFrom.(PostUpdateNode).getPreUpdateNode().asExpr() = call and
+      nodeTo.asDefiningArgument() = call.getQualifier()
+    )
+  ) and
+  model = ""
 }
 
 /**

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

@@ -0,0 +1,27 @@
+/**
+ * This module provides extensible predicates for defining MaD models.
+ */
+
+/**
+ * Holds if an external source model exists for the given parameters.
+ */
+extensible predicate sourceModel(
+  string namespace, string type, boolean subtypes, string name, string signature, string ext,
+  string output, string kind, string provenance, QlBuiltins::ExtensionId madId
+);
+
+/**
+ * Holds if an external sink model exists for the given parameters.
+ */
+extensible predicate sinkModel(
+  string namespace, string type, boolean subtypes, string name, string signature, string ext,
+  string input, string kind, string provenance, QlBuiltins::ExtensionId madId
+);
+
+/**
+ * Holds if an external summary model exists for the given parameters.
+ */
+extensible predicate summaryModel(
+  string namespace, string type, boolean subtypes, string name, string signature, string ext,
+  string input, string output, string kind, string provenance, QlBuiltins::ExtensionId madId
+);

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

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

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

@@ -32,8 +32,8 @@ predicate localTaintStep(DataFlow::Node src, DataFlow::Node sink) {
  * Holds if the additional step from `src` to `sink` should be included in all
  * global taint flow configurations.
  */
-predicate defaultAdditionalTaintStep(DataFlow::Node src, DataFlow::Node sink) {
-  localAdditionalTaintStep(src, sink)
+predicate defaultAdditionalTaintStep(DataFlow::Node src, DataFlow::Node sink, string model) {
+  localAdditionalTaintStep(src, sink) and model = ""
 }
 
 /**

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

@@ -127,7 +127,7 @@ abstract deprecated class Configuration extends DataFlow::Configuration {
 
   final override predicate isAdditionalFlowStep(DataFlow::Node node1, DataFlow::Node node2) {
     this.isAdditionalTaintStep(node1, node2) or
-    defaultAdditionalTaintStep(node1, node2)
+    defaultAdditionalTaintStep(node1, node2, _)
   }
 
   /**

cpp/ql/lib/semmle/code/cpp/dataflow/internal/tainttracking2/TaintTrackingImpl.qll

@@ -127,7 +127,7 @@ abstract deprecated class Configuration extends DataFlow::Configuration {
 
   final override predicate isAdditionalFlowStep(DataFlow::Node node1, DataFlow::Node node2) {
     this.isAdditionalTaintStep(node1, node2) or
-    defaultAdditionalTaintStep(node1, node2)
+    defaultAdditionalTaintStep(node1, node2, _)
   }
 
   /**

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

@@ -63,6 +63,12 @@ class Expr extends StmtParent, @expr {
    * order of destruction.
    */
   DestructorCall getImplicitDestructorCall(int n) {
+    exists(Expr e |
+      e = this.(TemporaryObjectExpr).getExpr() and
+      synthetic_destructor_call(e, max(int i | synthetic_destructor_call(e, i, _)) - n, result)
+    )
+    or
+    not this = any(TemporaryObjectExpr temp).getExpr() and
     synthetic_destructor_call(this, max(int i | synthetic_destructor_call(this, i, _)) - n, result)
   }
 
@@ -1015,8 +1021,33 @@ class DeleteOrDeleteArrayExpr extends Expr, TDeleteOrDeleteArrayExpr {
   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()
+    exists(Expr exprWithReuse | exprWithReuse = this.getExprWithReuse() |
+      if not exprWithReuse instanceof ReuseExpr
+      then result = exprWithReuse
+      else result = this.getDestructorCall().getQualifier()
+    )
   }
+
+  /**
+   * Gets the object or array being deleted, and gets a `ReuseExpr` when there
+   * is a destructor call and the object is also the qualifier of the call.
+   *
+   * For example, given:
+   * ```
+   * struct HasDestructor { ~HasDestructor(); };
+   * struct PlainOldData { int x, char y; };
+   *
+   * void f(HasDestructor* hasDestructor, PlainOldData* pod) {
+   *   delete hasDestructor;
+   *   delete pod;
+   * }
+   * ```
+   * This predicate yields a `ReuseExpr` for `delete hasDestructor`, as the
+   * the deleted expression has a destructor, and that expression is also
+   * the qualifier of the destructor call. In the case of `delete pod` the
+   * predicate does not yield a `ReuseExpr`, as there is no destructor call.
+   */
+  Expr getExprWithReuse() { result = this.getChild(3) }
 }
 
 /**
@@ -1307,6 +1338,24 @@ class CoAwaitExpr extends UnaryOperation, @co_await {
   override string getOperator() { result = "co_await" }
 
   override int getPrecedence() { result = 16 }
+
+  /**
+   * Gets the Boolean expression that is used to decide if the enclosing
+   * coroutine should be suspended.
+   */
+  Expr getAwaitReady() { result = this.getChild(1) }
+
+  /**
+   * Gets the expression that represents the resume point if the enclosing
+   * coroutine was suspended.
+   */
+  Expr getAwaitResume() { result = this.getChild(2) }
+
+  /**
+   * Gets the expression that is evaluated when the enclosing coroutine is
+   * suspended.
+   */
+  Expr getAwaitSuspend() { result = this.getChild(3) }
 }
 
 /**
@@ -1321,6 +1370,24 @@ class CoYieldExpr extends UnaryOperation, @co_yield {
   override string getOperator() { result = "co_yield" }
 
   override int getPrecedence() { result = 2 }
+
+  /**
+   * Gets the Boolean expression that is used to decide if the enclosing
+   * coroutine should be suspended.
+   */
+  Expr getAwaitReady() { result = this.getChild(1) }
+
+  /**
+   * Gets the expression that represents the resume point if the enclosing
+   * coroutine was suspended.
+   */
+  Expr getAwaitResume() { result = this.getChild(2) }
+
+  /**
+   * Gets the expression that is evaluated when the enclosing coroutine is
+   * suspended.
+   */
+  Expr getAwaitSuspend() { result = this.getChild(3) }
 }
 
 /**

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

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

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

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

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

@@ -263,9 +263,10 @@ deprecated private module Config implements FullStateConfigSig {
 
   predicate isBarrierOut(Node node, FlowState state) { none() }
 
-  predicate isAdditionalFlowStep(Node node1, Node node2) {
+  predicate isAdditionalFlowStep(Node node1, Node node2, string model) {
     singleConfiguration() and
-    any(Configuration config).isAdditionalFlowStep(node1, node2)
+    any(Configuration config).isAdditionalFlowStep(node1, node2) and
+    model = ""
   }
 
   predicate isAdditionalFlowStep(Node node1, FlowState state1, Node node2, FlowState state2) {

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

@@ -263,9 +263,10 @@ deprecated private module Config implements FullStateConfigSig {
 
   predicate isBarrierOut(Node node, FlowState state) { none() }
 
-  predicate isAdditionalFlowStep(Node node1, Node node2) {
+  predicate isAdditionalFlowStep(Node node1, Node node2, string model) {
     singleConfiguration() and
-    any(Configuration config).isAdditionalFlowStep(node1, node2)
+    any(Configuration config).isAdditionalFlowStep(node1, node2) and
+    model = ""
   }
 
   predicate isAdditionalFlowStep(Node node1, FlowState state1, Node node2, FlowState state2) {

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

@@ -263,9 +263,10 @@ deprecated private module Config implements FullStateConfigSig {
 
   predicate isBarrierOut(Node node, FlowState state) { none() }
 
-  predicate isAdditionalFlowStep(Node node1, Node node2) {
+  predicate isAdditionalFlowStep(Node node1, Node node2, string model) {
     singleConfiguration() and
-    any(Configuration config).isAdditionalFlowStep(node1, node2)
+    any(Configuration config).isAdditionalFlowStep(node1, node2) and
+    model = ""
   }
 
   predicate isAdditionalFlowStep(Node node1, FlowState state1, Node node2, FlowState state2) {

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

@@ -263,9 +263,10 @@ deprecated private module Config implements FullStateConfigSig {
 
   predicate isBarrierOut(Node node, FlowState state) { none() }
 
-  predicate isAdditionalFlowStep(Node node1, Node node2) {
+  predicate isAdditionalFlowStep(Node node1, Node node2, string model) {
     singleConfiguration() and
-    any(Configuration config).isAdditionalFlowStep(node1, node2)
+    any(Configuration config).isAdditionalFlowStep(node1, node2) and
+    model = ""
   }
 
   predicate isAdditionalFlowStep(Node node1, FlowState state1, Node node2, FlowState state2) {

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

@@ -22,9 +22,13 @@ module CppDataFlow implements InputSig<Location> {
 
   predicate getAdditionalFlowIntoCallNodeTerm = Private::getAdditionalFlowIntoCallNodeTerm/2;
 
+  predicate getSecondLevelScope = Private::getSecondLevelScope/1;
+
   predicate validParameterAliasStep = Private::validParameterAliasStep/2;
 
   predicate mayBenefitFromCallContext = Private::mayBenefitFromCallContext/1;
 
   predicate viableImplInCallContext = Private::viableImplInCallContext/2;
+
+  predicate neverSkipInPathGraph = Private::neverSkipInPathGraph/1;
 }

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

@@ -0,0 +1,518 @@
+/**
+ * Provides the classes `ExprNode` and `IndirectExprNode` for converting between `Expr` and `Node`.
+ */
+
+private import cpp
+private import semmle.code.cpp.ir.IR
+private import DataFlowUtil
+private import DataFlowPrivate
+private import semmle.code.cpp.ir.implementation.raw.internal.TranslatedExpr
+private import semmle.code.cpp.ir.implementation.raw.internal.InstructionTag
+
+cached
+private module Cached {
+  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.
+   */
+  private 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) {
+    // 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(AllocationExtentConvertTag())
+    )
+    or
+    // There's no instruction that returns `ParenthesisExpr`, but some queries
+    // expect this
+    exists(TranslatedTransparentConversion ttc |
+      result = ttc.getExpr().(ParenthesisExpr) and
+      instr = ttc.getResult()
+    )
+    or
+    // Certain expressions generate `CopyValueInstruction`s only when they
+    // are needed. Examples of this include crement operations and compound
+    // assignment operations. For example:
+    // ```cpp
+    // int x = ...
+    // int y = x++;
+    // ```
+    // this generate IR like:
+    // ```
+    // r1(glval<int>) = VariableAddress[x] :
+    // r2(int)        = Constant[0]        :
+    // m3(int)        = Store[x]           : &:r1, r2
+    // r4(glval<int>) = VariableAddress[y] :
+    // r5(glval<int>) = VariableAddress[x] :
+    // r6(int)        = Load[x]            : &:r5, m3
+    // r7(int)        = Constant[1]        :
+    // r8(int)        = Add                : r6, r7
+    // m9(int)        = Store[x]           : &:r5, r8
+    // r11(int)       = CopyValue         : r6
+    // m12(int)       = Store[y]          : &:r4, r11
+    // ```
+    // When the `CopyValueInstruction` is not generated there is no instruction
+    // whose `getConvertedResultExpression` maps back to the expression. When
+    // such an instruction doesn't exist it means that the old value is not
+    // needed, and in that case the only value that will propagate forward in
+    // the program is the value that's been updated. So in those cases we just
+    // use the result of `node.asDefinition()` as the result of `node.asExpr()`.
+    exists(TranslatedCoreExpr tco |
+      tco.getInstruction(_) = instr and
+      tco.producesExprResult() and
+      result = asDefinitionImpl0(instr)
+    )
+  }
+
+  private Expr getConvertedResultExpressionImpl(Instruction instr) {
+    result = getConvertedResultExpressionImpl0(instr)
+    or
+    not exists(getConvertedResultExpressionImpl0(instr)) and
+    result = instr.getConvertedResultExpression()
+  }
+
+  /**
+   * Gets the result for `node.asDefinition()` (when `node` is the instruction
+   * node that wraps `store`) in the cases where `store.getAst()` should not be
+   * used to define the result of `node.asDefinition()`.
+   */
+  private Expr asDefinitionImpl0(StoreInstruction store) {
+    // 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 |
+      store = tao.getInstruction(AssignmentStoreTag()) and
+      result = tao.getExpr()
+    )
+    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 |
+      store = tco.getInstruction(CrementStoreTag()) and
+      result = tco.getExpr()
+    )
+  }
+
+  /**
+   * Holds if the expression returned by `store.getAst()` should not be
+   * returned as the result of `node.asDefinition()` when `node` is the
+   * instruction node that wraps `store`.
+   */
+  private predicate excludeAsDefinitionResult(StoreInstruction store) {
+    // Exclude the store to the temporary generated by a ternary expression.
+    exists(TranslatedConditionalExpr tce |
+      store = tce.getInstruction(ConditionValueFalseStoreTag())
+      or
+      store = tce.getInstruction(ConditionValueTrueStoreTag())
+    )
+  }
+
+  /**
+   * Gets the expression that represents the result of `StoreInstruction` for
+   * dataflow purposes.
+   *
+   * For example, consider the following example
+   * ```cpp
+   * int x = 42;     // 1
+   * x = 34;         // 2
+   * ++x;            // 3
+   * x++;            // 4
+   * x += 1;         // 5
+   * int y = x += 2; // 6
+   * ```
+   * For (1) the result is `42`.
+   * For (2) the result is `x = 34`.
+   * For (3) the result is `++x`.
+   * For (4) the result is `x++`.
+   * For (5) the result is `x += 1`.
+   * For (6) there are two results:
+   *   - For the `StoreInstruction` generated by `x += 2` the result
+   *     is `x += 2`
+   *   - For the `StoreInstruction` generated by `int y = ...` the result
+   *     is also `x += 2`
+   */
+  cached
+  Expr asDefinitionImpl(StoreInstruction store) {
+    not exists(asDefinitionImpl0(store)) and
+    not excludeAsDefinitionResult(store) and
+    result = store.getAst().(Expr).getUnconverted()
+    or
+    result = asDefinitionImpl0(store)
+  }
+
+  /** Holds if `node` is an `OperandNode` that should map `node.asExpr()` to `e`. */
+  private predicate exprNodeShouldBeOperand(OperandNode node, Expr e, int n) {
+    not exprNodeShouldBeIndirectOperand(_, e, n) and
+    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(
+    IndirectOperand node, Expr e, int n, int indirectionIndex
+  ) {
+    exists(Instruction def |
+      node.hasOperandAndIndirectionIndex(unique( | | getAUse(def)), indirectionIndex) and
+      e = getConvertedResultExpression(def, n)
+    )
+  }
+
+  /** Holds if `operand`'s definition is a `VariableAddressInstruction` whose variable is a temporary */
+  private predicate isIRTempVariable(Operand operand) {
+    operand.getDef().(VariableAddressInstruction).getIRVariable() instanceof IRTempVariable
+  }
+
+  /**
+   * Holds if `node` is an indirect operand whose operand is an argument, and
+   * the `n`'th expression associated with the operand is `e`.
+   */
+  private predicate isIndirectOperandOfArgument(
+    IndirectOperand node, ArgumentOperand operand, Expr e, int n
+  ) {
+    node.hasOperandAndIndirectionIndex(operand, 1) and
+    e = getConvertedResultExpression(operand.getDef(), n)
+  }
+
+  /**
+   * Holds if `opFrom` is an operand to a conversion, and `opTo` is the unique
+   * use of the conversion.
+   */
+  private predicate isConversionStep(Operand opFrom, Operand opTo) {
+    exists(Instruction mid |
+      conversionFlow(opFrom, mid, false, false) and
+      opTo = unique( | | getAUse(mid))
+    )
+  }
+
+  /**
+   * Holds if an operand that satisfies `isIRTempVariable` flows to `op`
+   * through a (possibly empty) sequence of conversions.
+   */
+  private predicate irTempOperandConversionFlows(Operand op) {
+    isIRTempVariable(op)
+    or
+    exists(Operand mid |
+      irTempOperandConversionFlows(mid) and
+      isConversionStep(mid, op)
+    )
+  }
+
+  /** Holds if `node` should be an `IndirectOperand` that maps `node.asExpr()` to `e`. */
+  private predicate exprNodeShouldBeIndirectOperand(IndirectOperand node, Expr e, int n) {
+    exists(ArgumentOperand operand |
+      // When an argument (qualifier or positional) is a prvalue and the
+      // parameter (qualifier or positional) is a (const) reference, IR
+      // construction introduces a temporary `IRVariable`. The `VariableAddress`
+      // instruction has the argument as its `getConvertedResultExpression`
+      // result. However, the instruction actually represents the _address_ of
+      // the argument. So to fix this mismatch, we have the indirection of the
+      // `VariableAddressInstruction` map to the expression.
+      isIndirectOperandOfArgument(node, operand, e, n) and
+      irTempOperandConversionFlows(operand)
+    )
+  }
+
+  private predicate exprNodeShouldBeIndirectOutNode(IndirectArgumentOutNode node, Expr e, int n) {
+    exists(CallInstruction call |
+      call.getStaticCallTarget() instanceof Constructor and
+      e = getConvertedResultExpression(call, n) and
+      call.getThisArgumentOperand() = node.getAddressOperand()
+    )
+  }
+
+  /** Holds if `node` should be an instruction node that maps `node.asExpr()` to `e`. */
+  private predicate exprNodeShouldBeInstruction(Node node, Expr e, int n) {
+    not exprNodeShouldBeOperand(_, e, n) and
+    not exprNodeShouldBeIndirectOutNode(_, e, n) and
+    not exprNodeShouldBeIndirectOperand(_, e, n) and
+    e = getConvertedResultExpression(node.asInstruction(), n)
+  }
+
+  /** Holds if `node` should be an `IndirectInstruction` that maps `node.asIndirectExpr()` to `e`. */
+  private predicate indirectExprNodeShouldBeIndirectInstruction(
+    IndirectInstruction node, Expr e, int n, int indirectionIndex
+  ) {
+    not indirectExprNodeShouldBeIndirectOperand(_, e, n, indirectionIndex) and
+    exists(Instruction instr |
+      node.hasInstructionAndIndirectionIndex(instr, indirectionIndex) and
+      e = getConvertedResultExpression(instr, n)
+    )
+  }
+
+  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(int n);
+
+    /** Gets the non-conversion expression corresponding to this node, if any. */
+    final Expr getExpr(int n) { result = this.getConvertedExpr(n).getUnconverted() }
+  }
+
+  /**
+   * Holds if there exists a dataflow node whose `asExpr(n)` should evaluate
+   * to `e`.
+   */
+  private predicate exprNodeShouldBe(Expr e, int n) {
+    exprNodeShouldBeInstruction(_, e, n) or
+    exprNodeShouldBeOperand(_, e, n) or
+    exprNodeShouldBeIndirectOutNode(_, e, n) or
+    exprNodeShouldBeIndirectOperand(_, e, n)
+  }
+
+  private class InstructionExprNode extends ExprNodeBase, InstructionNode {
+    InstructionExprNode() {
+      exists(Expr e, int n |
+        exprNodeShouldBeInstruction(this, e, n) and
+        not exists(Expr conv |
+          exprNodeShouldBe(conv, n + 1) and
+          conv.getUnconverted() = e.getUnconverted()
+        )
+      )
+    }
+
+    final override Expr getConvertedExpr(int n) { exprNodeShouldBeInstruction(this, result, n) }
+  }
+
+  private class OperandExprNode extends ExprNodeBase, OperandNode {
+    OperandExprNode() {
+      exists(Expr e, int n |
+        exprNodeShouldBeOperand(this, e, n) and
+        not exists(Expr conv |
+          exprNodeShouldBe(conv, n + 1) and
+          conv.getUnconverted() = e.getUnconverted()
+        )
+      )
+    }
+
+    final override Expr getConvertedExpr(int n) { exprNodeShouldBeOperand(this, result, n) }
+  }
+
+  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 n, int indirectionIndex);
+
+    /** Gets the non-conversion expression corresponding to this node, if any. */
+    final Expr getExpr(int n, int indirectionIndex) {
+      result = this.getConvertedExpr(n, indirectionIndex).getUnconverted()
+    }
+  }
+
+  /** A signature for converting an indirect node to an expression. */
+  private signature module IndirectNodeToIndirectExprSig {
+    /** The indirect node class to be converted to an expression */
+    class IndirectNode;
+
+    /**
+     * Holds if the indirect expression at indirection index `indirectionIndex`
+     * of `node` is `e`. The integer `n` specifies how many conversions has been
+     * applied to `node`.
+     */
+    predicate indirectNodeHasIndirectExpr(IndirectNode node, Expr e, int n, int indirectionIndex);
+  }
+
+  /**
+   * A module that implements the logic for deciding whether an indirect node
+   * should be an `IndirectExprNode`.
+   */
+  private module IndirectNodeToIndirectExpr<IndirectNodeToIndirectExprSig Sig> {
+    import Sig
+
+    /**
+     * This predicate shifts the indirection index by one when `conv` is a
+     * `ReferenceDereferenceExpr`.
+     *
+     * This is necessary because `ReferenceDereferenceExpr` is a conversion
+     * in the AST, but appears as a `LoadInstruction` in the IR.
+     */
+    bindingset[e, indirectionIndex]
+    private predicate adjustForReference(
+      Expr e, int indirectionIndex, Expr conv, int adjustedIndirectionIndex
+    ) {
+      conv.(ReferenceDereferenceExpr).getExpr() = e and
+      adjustedIndirectionIndex = indirectionIndex - 1
+      or
+      not conv instanceof ReferenceDereferenceExpr and
+      conv = e and
+      adjustedIndirectionIndex = indirectionIndex
+    }
+
+    /** Holds if `node` should be an `IndirectExprNode`. */
+    predicate charpred(IndirectNode node) {
+      exists(Expr e, int n, int indirectionIndex |
+        indirectNodeHasIndirectExpr(node, e, n, indirectionIndex) and
+        not exists(Expr conv, int adjustedIndirectionIndex |
+          adjustForReference(e, indirectionIndex, conv, adjustedIndirectionIndex) and
+          indirectExprNodeShouldBe(conv, n + 1, adjustedIndirectionIndex)
+        )
+      )
+    }
+  }
+
+  private predicate indirectExprNodeShouldBe(Expr e, int n, int indirectionIndex) {
+    indirectExprNodeShouldBeIndirectOperand(_, e, n, indirectionIndex) or
+    indirectExprNodeShouldBeIndirectInstruction(_, e, n, indirectionIndex)
+  }
+
+  private module IndirectOperandIndirectExprNodeImpl implements IndirectNodeToIndirectExprSig {
+    class IndirectNode = IndirectOperand;
+
+    predicate indirectNodeHasIndirectExpr = indirectExprNodeShouldBeIndirectOperand/4;
+  }
+
+  module IndirectOperandToIndirectExpr =
+    IndirectNodeToIndirectExpr<IndirectOperandIndirectExprNodeImpl>;
+
+  private class IndirectOperandIndirectExprNode extends IndirectExprNodeBase instanceof IndirectOperand
+  {
+    IndirectOperandIndirectExprNode() { IndirectOperandToIndirectExpr::charpred(this) }
+
+    final override Expr getConvertedExpr(int n, int index) {
+      IndirectOperandToIndirectExpr::indirectNodeHasIndirectExpr(this, result, n, index)
+    }
+  }
+
+  private module IndirectInstructionIndirectExprNodeImpl implements IndirectNodeToIndirectExprSig {
+    class IndirectNode = IndirectInstruction;
+
+    predicate indirectNodeHasIndirectExpr = indirectExprNodeShouldBeIndirectInstruction/4;
+  }
+
+  module IndirectInstructionToIndirectExpr =
+    IndirectNodeToIndirectExpr<IndirectInstructionIndirectExprNodeImpl>;
+
+  private class IndirectInstructionIndirectExprNode extends IndirectExprNodeBase instanceof IndirectInstruction
+  {
+    IndirectInstructionIndirectExprNode() { IndirectInstructionToIndirectExpr::charpred(this) }
+
+    final override Expr getConvertedExpr(int n, int index) {
+      IndirectInstructionToIndirectExpr::indirectNodeHasIndirectExpr(this, result, n, index)
+    }
+  }
+
+  private class IndirectArgumentOutExprNode extends ExprNodeBase, IndirectArgumentOutNode {
+    IndirectArgumentOutExprNode() { exprNodeShouldBeIndirectOutNode(this, _, _) }
+
+    final override Expr getConvertedExpr(int n) { exprNodeShouldBeIndirectOutNode(this, result, n) }
+  }
+
+  private class IndirectOperandExprNode extends ExprNodeBase instanceof IndirectOperand {
+    IndirectOperandExprNode() { exprNodeShouldBeIndirectOperand(this, _, _) }
+
+    final override Expr getConvertedExpr(int n) { exprNodeShouldBeIndirectOperand(this, result, n) }
+  }
+
+  /**
+   * An expression, viewed as a node in a data flow graph.
+   */
+  cached
+  class ExprNode extends Node instanceof ExprNodeBase {
+    /**
+     * INTERNAL: Do not use.
+     */
+    cached
+    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.
+     */
+    cached
+    final Expr getExpr() { result = this.getExpr(_) }
+
+    /**
+     * INTERNAL: Do not use.
+     */
+    cached
+    Expr getConvertedExpr(int n) { result = super.getConvertedExpr(n) }
+
+    /**
+     * Gets the expression corresponding to this node, if any. The returned
+     * expression may be a `Conversion`.
+     */
+    cached
+    final Expr getConvertedExpr() { result = this.getConvertedExpr(_) }
+  }
+
+  /**
+   * An indirect expression, viewed as a node in a data flow graph.
+   */
+  cached
+  class IndirectExprNode extends Node instanceof IndirectExprNodeBase {
+    /**
+     * 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.
+     */
+    cached
+    final Expr getExpr(int indirectionIndex) { result = this.getExpr(_, indirectionIndex) }
+
+    /**
+     * INTERNAL: Do not use.
+     */
+    cached
+    Expr getExpr(int n, int indirectionIndex) { result = super.getExpr(n, indirectionIndex) }
+
+    /**
+     * INTERNAL: Do not use.
+     */
+    cached
+    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`.
+     */
+    cached
+    Expr getConvertedExpr(int indirectionIndex) {
+      result = this.getConvertedExpr(_, indirectionIndex)
+    }
+  }
+}
+
+import Cached

cpp/ql/lib/semmle/code/cpp/ir/dataflow/internal/NormalNode0ToString.qll

@@ -3,12 +3,26 @@
  * `toString` for `Instruction` and `Operand` dataflow nodes.
  */
 
+private import cpp
 private import semmle.code.cpp.ir.IR
 private import codeql.util.Unit
 private import Node0ToString
 private import DataFlowUtil
 private import DataFlowPrivate
 
+/**
+ * Gets the string representation of the unconverted expression `loc` if
+ * `loc` is an `Expression`.
+ *
+ * Otherwise, this gets the string representation of `loc`.
+ */
+private string unconvertedAstToString(Locatable loc) {
+  result = loc.(Expr).getUnconverted().toString()
+  or
+  not loc instanceof Expr and
+  result = loc.toString()
+}
+
 private class NormalNode0ToString extends Node0ToString {
   NormalNode0ToString() {
     // Silence warning about `this` not being bound.
@@ -18,14 +32,10 @@ private class NormalNode0ToString extends Node0ToString {
   override string instructionToString(Instruction i) {
     if i.(InitializeParameterInstruction).getIRVariable() instanceof IRThisVariable
     then result = "this"
-    else result = i.getAst().toString()
+    else result = unconvertedAstToString(i.getAst())
   }
 
-  override string operandToString(Operand op) {
-    if op.getDef().(InitializeParameterInstruction).getIRVariable() instanceof IRThisVariable
-    then result = "this"
-    else result = op.getDef().getAst().toString()
-  }
+  override string operandToString(Operand op) { result = this.instructionToString(op.getDef()) }
 
   override string toExprString(Node n) {
     result = n.asExpr(0).toString()

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

@@ -10,7 +10,7 @@ private import PrintIRUtilities
  */
 private string getFromFlow(Node node2, int order1, int order2) {
   exists(Node node1 |
-    simpleLocalFlowStep(node1, node2) and
+    simpleLocalFlowStep(node1, node2, _) and
     result = nodeId(node1, order1, order2)
   )
 }
@@ -20,7 +20,7 @@ private string getFromFlow(Node node2, int order1, int order2) {
  */
 private string getToFlow(Node node1, int order1, int order2) {
   exists(Node node2 |
-    simpleLocalFlowStep(node1, node2) and
+    simpleLocalFlowStep(node1, node2, _) and
     result = nodeId(node2, order1, order2)
   )
 }

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

@@ -546,7 +546,7 @@ module ProductFlow {
       Flow1::PathGraph::edges(pred1, succ1, _, _) and
       exists(ReturnKindExt returnKind |
         succ1.getNode() = returnKind.getAnOutNode(call) and
-        pred1.getNode().(ReturnNodeExt).getKind() = returnKind
+        paramReturnNode(_, pred1.asParameterReturnNode(), _, returnKind)
       )
     }
 
@@ -574,7 +574,7 @@ module ProductFlow {
       Flow2::PathGraph::edges(pred2, succ2, _, _) and
       exists(ReturnKindExt returnKind |
         succ2.getNode() = returnKind.getAnOutNode(call) and
-        pred2.getNode().(ReturnNodeExt).getKind() = returnKind
+        paramReturnNode(_, pred2.asParameterReturnNode(), _, returnKind)
       )
     }
 

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 TypeFlow
 private import semmle.code.cpp.ir.ValueNumbering
 
 /**
@@ -245,14 +246,6 @@ private module IteratorIndirections {
       baseType = super.getValueType()
     }
 
-    override predicate isAdditionalDereference(Instruction deref, Operand address) {
-      exists(CallInstruction call |
-        operandForFullyConvertedCall(getAUse(deref), call) and
-        this = call.getStaticCallTarget().getClassAndName("operator*") and
-        address = call.getThisArgumentOperand()
-      )
-    }
-
     override predicate isAdditionalWrite(Node0Impl value, Operand address, boolean certain) {
       exists(CallInstruction call | call.getArgumentOperand(0) = value.asOperand() |
         this = call.getStaticCallTarget().getClassAndName("operator=") and
@@ -261,16 +254,6 @@ private module IteratorIndirections {
       )
     }
 
-    override predicate isAdditionalTaintStep(Node node1, Node node2) {
-      exists(CallInstruction call |
-        // Taint through `operator+=` and `operator-=` on iterators.
-        call.getStaticCallTarget() instanceof Iterator::IteratorAssignArithmeticOperator and
-        node2.(IndirectArgumentOutNode).getPreUpdateNode() = node1 and
-        node1.(IndirectOperand).hasOperandAndIndirectionIndex(call.getArgumentOperand(0), _) and
-        node1.getType().getUnspecifiedType() = this
-      )
-    }
-
     override predicate isAdditionalConversionFlow(Operand opFrom, Instruction instrTo) {
       // This is a bit annoying: Consider the following snippet:
       // ```
@@ -588,230 +571,6 @@ private class BaseCallInstruction extends BaseSourceVariableInstruction, CallIns
 
 cached
 private module Cached {
-  private import semmle.code.cpp.models.interfaces.Iterator as Interfaces
-  private import semmle.code.cpp.models.implementations.Iterator as Iterator
-  private import semmle.code.cpp.models.interfaces.FunctionInputsAndOutputs as IO
-
-  /**
-   * Holds if `next` is a instruction with a memory result that potentially
-   * updates the memory produced by `prev`.
-   */
-  private predicate memorySucc(Instruction prev, Instruction next) {
-    prev = next.(ChiInstruction).getTotal()
-    or
-    // Phi inputs can be inexact.
-    prev = next.(PhiInstruction).getAnInputOperand().getAnyDef()
-    or
-    prev = next.(CopyInstruction).getSourceValue()
-    or
-    exists(ReadSideEffectInstruction read |
-      next = read.getPrimaryInstruction() and
-      isAdditionalConversionFlow(_, next) and
-      prev = read.getSideEffectOperand().getAnyDef()
-    )
-  }
-
-  /**
-   * Holds if `iteratorDerefAddress` is an address of an iterator dereference (i.e., `*it`)
-   * that is used for a write operation that writes the value `value`. The `memory` instruction
-   * represents the memory that the IR's SSA analysis determined was read by the call to `operator*`.
-   *
-   * The `numberOfLoads` integer represents the number of dereferences this write corresponds to
-   * on the underlying container that produced the iterator.
-   */
-  private predicate isChiAfterIteratorDef(
-    Instruction memory, Operand iteratorDerefAddress, Node0Impl value, int numberOfLoads
-  ) {
-    exists(
-      BaseSourceVariableInstruction iteratorBase, ReadSideEffectInstruction read,
-      Operand iteratorAddress
-    |
-      numberOfLoads >= 0 and
-      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
-      memory = read.getSideEffectOperand().getAnyDef()
-    )
-  }
-
-  private predicate isSource(Instruction instr, Operand iteratorAddress, int numberOfLoads) {
-    getAUse(instr) = iteratorAddress and
-    exists(BaseSourceVariableInstruction iteratorBase |
-      iteratorBase.getResultType() instanceof Interfaces::Iterator and
-      not iteratorBase.getResultType() instanceof Cpp::PointerType and
-      isUse(_, iteratorAddress, iteratorBase, numberOfLoads - 1, 0)
-    )
-  }
-
-  private predicate isSink(Instruction instr, CallInstruction call) {
-    getAUse(instr).(ArgumentOperand).getCall() = call and
-    // Only include operations that may modify the object that the iterator points to.
-    // The following is a non-exhaustive list of things that may modify the value of the
-    // iterator, but never the value of what the iterator points to.
-    // The more things we can exclude here, the faster the small dataflow-like analysis
-    // done by `convertsIntoArgument` will converge.
-    not exists(Function f | f = call.getStaticCallTarget() |
-      f instanceof Iterator::IteratorCrementOperator or
-      f instanceof Iterator::IteratorBinaryArithmeticOperator or
-      f instanceof Iterator::IteratorAssignArithmeticOperator or
-      f instanceof Iterator::IteratorCrementMemberOperator or
-      f instanceof Iterator::IteratorBinaryArithmeticMemberOperator or
-      f instanceof Iterator::IteratorAssignArithmeticMemberOperator or
-      f instanceof Iterator::IteratorAssignmentMemberOperator
-    )
-  }
-
-  private predicate convertsIntoArgumentFwd(Instruction instr) {
-    isSource(instr, _, _)
-    or
-    exists(Instruction prev | convertsIntoArgumentFwd(prev) |
-      conversionFlow(unique( | | getAUse(prev)), instr, false, _)
-    )
-  }
-
-  private predicate convertsIntoArgumentRev(Instruction instr) {
-    convertsIntoArgumentFwd(instr) and
-    (
-      isSink(instr, _)
-      or
-      exists(Instruction next | convertsIntoArgumentRev(next) |
-        conversionFlow(unique( | | getAUse(instr)), next, false, _)
-      )
-    )
-  }
-
-  private predicate convertsIntoArgument(
-    Operand iteratorAddress, CallInstruction call, int numberOfLoads
-  ) {
-    exists(Instruction iteratorAddressDef |
-      isSource(iteratorAddressDef, iteratorAddress, numberOfLoads) and
-      isSink(iteratorAddressDef, call) and
-      convertsIntoArgumentRev(pragma[only_bind_into](iteratorAddressDef))
-    )
-  }
-
-  private predicate isChiAfterIteratorArgument(
-    Instruction memory, Operand iteratorAddress, int numberOfLoads
-  ) {
-    // Ideally, `iteratorAddress` would be an `ArgumentOperand`, but there might be
-    // various conversions applied to it before it becomes an argument.
-    // So we do a small amount of flow to find the call that the iterator is passed to.
-    exists(CallInstruction call | convertsIntoArgument(iteratorAddress, call, numberOfLoads) |
-      exists(ReadSideEffectInstruction read |
-        read.getPrimaryInstruction() = call and
-        read.getSideEffectOperand().getAnyDef() = memory
-      )
-      or
-      exists(LoadInstruction load |
-        iteratorAddress.getDef() = load and
-        memory = load.getSourceValueOperand().getAnyDef()
-      )
-    )
-  }
-
-  /**
-   * Holds if `iterator` is a `StoreInstruction` that stores the result of some function
-   * returning an iterator into an address computed started at `containerBase`.
-   *
-   * For example, given a declaration like `std::vector<int>::iterator it = v.begin()`,
-   * the `iterator` will be the `StoreInstruction` generated by the write to `it`, and
-   * `containerBase` will be the address of `v`.
-   */
-  private predicate isChiAfterBegin(
-    BaseSourceVariableInstruction containerBase, StoreInstruction iterator
-  ) {
-    exists(
-      CallInstruction getIterator, Iterator::GetIteratorFunction getIteratorFunction,
-      IO::FunctionInput input, int i
-    |
-      getIterator = iterator.getSourceValue() and
-      getIteratorFunction = getIterator.getStaticCallTarget() and
-      getIteratorFunction.getsIterator(input, _) and
-      isDef(_, any(Node0Impl n | n.asInstruction() = iterator), _, _, 1, 0) and
-      input.isParameterDerefOrQualifierObject(i) and
-      isUse(_, getIterator.getArgumentOperand(i), containerBase, 0, 0)
-    )
-  }
-
-  /**
-   * Holds if `iteratorAddress` is an address of an iterator that is used for
-   * a read operation. The `memory` instruction represents the memory that
-   * the IR's SSA analysis determined was read by the call to `operator*`.
-   *
-   * Finally, the `numberOfLoads` integer represents the number of dereferences
-   * this read corresponds to on the underlying container that produced the iterator.
-   */
-  private predicate isChiBeforeIteratorUse(
-    Operand iteratorAddress, Instruction memory, int numberOfLoads
-  ) {
-    exists(
-      BaseSourceVariableInstruction iteratorBase, LoadInstruction load,
-      ReadSideEffectInstruction read, Operand iteratorDerefAddress
-    |
-      numberOfLoads >= 0 and
-      isUse(_, iteratorAddress, iteratorBase, numberOfLoads + 1, 0) and
-      isUse(_, iteratorDerefAddress, iteratorBase, numberOfLoads + 2, 0) and
-      iteratorBase.getResultType() instanceof Interfaces::Iterator and
-      load.getSourceAddressOperand() = iteratorDerefAddress and
-      read.getPrimaryInstruction() = load.getSourceAddress() and
-      memory = read.getSideEffectOperand().getAnyDef()
-    )
-  }
-
-  /**
-   * Holds if `iteratorDerefAddress` is an address of an iterator dereference (i.e., `*it`)
-   * that is used for a write operation that writes the value `value` to a container that
-   * created the iterator. `container` represents the base of the address of the container
-   * that was used to create the iterator.
-   */
-  cached
-  predicate isIteratorDef(
-    BaseSourceVariableInstruction container, Operand iteratorDerefAddress, Node0Impl value,
-    int numberOfLoads, int indirectionIndex
-  ) {
-    exists(Instruction memory, Instruction begin, int upper, int ind |
-      isChiAfterIteratorDef(memory, iteratorDerefAddress, value, numberOfLoads) and
-      memorySucc*(begin, memory) and
-      isChiAfterBegin(container, begin) and
-      upper = countIndirectionsForCppType(getResultLanguageType(container)) and
-      ind = numberOfLoads + [1 .. upper] and
-      indirectionIndex = ind - (numberOfLoads + 1)
-    )
-  }
-
-  /**
-   * Holds if `iteratorAddress` is an address of an iterator that is used for a
-   * read operation to read a value from a container that created the iterator.
-   * `container` represents the base of the address of the container that was used
-   * to create the iterator.
-   */
-  cached
-  predicate isIteratorUse(
-    BaseSourceVariableInstruction container, Operand iteratorAddress, int numberOfLoads,
-    int indirectionIndex
-  ) {
-    // Direct use
-    exists(Instruction begin, Instruction memory, int upper, int ind |
-      isChiBeforeIteratorUse(iteratorAddress, memory, numberOfLoads) and
-      memorySucc*(begin, memory) and
-      isChiAfterBegin(container, begin) and
-      upper = countIndirectionsForCppType(getResultLanguageType(container)) and
-      ind = numberOfLoads + [1 .. upper] and
-      indirectionIndex = ind - (numberOfLoads + 1)
-    )
-    or
-    // Use through function output
-    exists(Instruction memory, Instruction begin, int upper, int ind |
-      isChiAfterIteratorArgument(memory, iteratorAddress, numberOfLoads) and
-      memorySucc*(begin, memory) and
-      isChiAfterBegin(container, begin) and
-      upper = countIndirectionsForCppType(getResultLanguageType(container)) and
-      ind = numberOfLoads + [1 .. upper] and
-      indirectionIndex = ind - (numberOfLoads - 1)
-    )
-  }
-
   /** Holds if `op` is the only use of its defining instruction, and that op is used in a conversation */
   private predicate isConversion(Operand op) {
     exists(Instruction def, Operand use |
@@ -955,11 +714,7 @@ private module Cached {
    * Holds if the address computed by `operand` is guaranteed to write
    * to a specific address.
    */
-  private predicate isCertainAddress(Operand operand) {
-    valueNumberOfOperand(operand).getAnInstruction() instanceof VariableAddressInstruction
-    or
-    operand.getType() instanceof Cpp::ReferenceType
-  }
+  private predicate isCertainAddress(Operand operand) { isPointerToSingleObject(operand.getDef()) }
 
   /**
    * Holds if `address` is a use of an SSA variable rooted at `base`, and the

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

@@ -6,16 +6,26 @@ private import semmle.code.cpp.models.interfaces.SideEffect
 private import DataFlowUtil
 private import DataFlowPrivate
 private import SsaInternals as Ssa
+private import semmle.code.cpp.dataflow.internal.FlowSummaryImpl as FlowSummaryImpl
 private import semmle.code.cpp.ir.dataflow.FlowSteps
 
 /**
  * Holds if taint propagates from `nodeFrom` to `nodeTo` in exactly one local
- * (intra-procedural) step.
+ * (intra-procedural) step. This relation is only used for local taint flow
+ * (for example `TaintTracking::localTaint(source, sink)`) so it may contain
+ * special cases that should only apply to local taint flow.
  */
 predicate localTaintStep(DataFlow::Node nodeFrom, DataFlow::Node nodeTo) {
+  // dataflow step
   DataFlow::localFlowStep(nodeFrom, nodeTo)
   or
-  localAdditionalTaintStep(nodeFrom, nodeTo)
+  // taint flow step
+  localAdditionalTaintStep(nodeFrom, nodeTo, _)
+  or
+  // models-as-data summarized flow for local data flow (i.e. special case for flow
+  // through calls to modeled functions, without relying on global dataflow to join
+  // the dots).
+  FlowSummaryImpl::Private::Steps::summaryThroughStepTaint(nodeFrom, nodeTo, _)
 }
 
 /**
@@ -24,10 +34,11 @@ predicate localTaintStep(DataFlow::Node nodeFrom, DataFlow::Node nodeTo) {
  * different objects.
  */
 cached
-predicate localAdditionalTaintStep(DataFlow::Node nodeFrom, DataFlow::Node nodeTo) {
-  operandToInstructionTaintStep(nodeFrom.asOperand(), nodeTo.asInstruction())
+predicate localAdditionalTaintStep(DataFlow::Node nodeFrom, DataFlow::Node nodeTo, string model) {
+  operandToInstructionTaintStep(nodeFrom.asOperand(), nodeTo.asInstruction()) and
+  model = ""
   or
-  modeledTaintStep(nodeFrom, nodeTo)
+  modeledTaintStep(nodeFrom, nodeTo, model)
   or
   // Flow from (the indirection of) an operand of a pointer arithmetic instruction to the
   // indirection of the pointer arithmetic instruction. This provides flow from `source`
@@ -35,15 +46,22 @@ predicate localAdditionalTaintStep(DataFlow::Node nodeFrom, DataFlow::Node nodeT
   exists(PointerArithmeticInstruction pai, int indirectionIndex |
     nodeHasOperand(nodeFrom, pai.getAnOperand(), pragma[only_bind_into](indirectionIndex)) and
     hasInstructionAndIndex(nodeTo, pai, indirectionIndex + 1)
-  )
+  ) and
+  model = ""
   or
-  any(Ssa::Indirection ind).isAdditionalTaintStep(nodeFrom, nodeTo)
+  any(Ssa::Indirection ind).isAdditionalTaintStep(nodeFrom, nodeTo) and
+  model = ""
+  or
+  // models-as-data summarized flow
+  FlowSummaryImpl::Private::Steps::summaryLocalStep(nodeFrom.(FlowSummaryNode).getSummaryNode(),
+    nodeTo.(FlowSummaryNode).getSummaryNode(), false, model)
   or
   // object->field conflation for content that is a `TaintInheritingContent`.
   exists(DataFlow::ContentSet f |
     readStep(nodeFrom, f, nodeTo) and
     f.getAReadContent() instanceof TaintInheritingContent
-  )
+  ) and
+  model = ""
 }
 
 /**
@@ -120,8 +138,8 @@ predicate localExprTaint(Expr e1, Expr e2) {
  * Holds if the additional step from `src` to `sink` should be included in all
  * global taint flow configurations.
  */
-predicate defaultAdditionalTaintStep(DataFlow::Node src, DataFlow::Node sink) {
-  localAdditionalTaintStep(src, sink)
+predicate defaultAdditionalTaintStep(DataFlow::Node src, DataFlow::Node sink, string model) {
+  localAdditionalTaintStep(src, sink, model)
 }
 
 /**
@@ -141,7 +159,7 @@ predicate defaultTaintSanitizer(DataFlow::Node node) { none() }
  * Holds if taint can flow from `nodeIn` to `nodeOut` through a call to a
  * modeled function.
  */
-predicate modeledTaintStep(DataFlow::Node nodeIn, DataFlow::Node nodeOut) {
+predicate modeledTaintStep(DataFlow::Node nodeIn, DataFlow::Node nodeOut, string model) {
   // Normal taint steps
   exists(CallInstruction call, TaintFunction func, FunctionInput modelIn, FunctionOutput modelOut |
     call.getStaticCallTarget() = func and
@@ -150,7 +168,8 @@ predicate modeledTaintStep(DataFlow::Node nodeIn, DataFlow::Node nodeOut) {
     nodeIn = callInput(call, modelIn) and nodeOut = callOutput(call, modelOut)
     or
     exists(int d | nodeIn = callInput(call, modelIn, d) and nodeOut = callOutput(call, modelOut, d))
-  )
+  ) and
+  model = "TaintFunction"
   or
   // Taint flow from one argument to another and data flow from an argument to a
   // return value. This happens in functions like `strcat` and `memcpy`. We
@@ -167,7 +186,8 @@ predicate modeledTaintStep(DataFlow::Node nodeIn, DataFlow::Node nodeOut) {
     func.(TaintFunction).hasTaintFlow(modelIn, modelMidOut) and
     func.(DataFlowFunction).hasDataFlow(modelMidIn, modelOut) and
     modelMidOut.isParameterDeref(indexMid) and
-    modelMidIn.isParameter(indexMid)
+    modelMidIn.isParameter(indexMid) and
+    model = "TaintFunction"
   )
   or
   // Taint flow from a pointer argument to an output, when the model specifies flow from the deref
@@ -180,9 +200,11 @@ predicate modeledTaintStep(DataFlow::Node nodeIn, DataFlow::Node nodeOut) {
     indirectArgument.hasAddressOperandAndIndirectionIndex(nodeIn.asOperand(), _) and
     call.getStaticCallTarget() = func and
     (
-      func.(DataFlowFunction).hasDataFlow(modelIn, modelOut)
+      func.(DataFlowFunction).hasDataFlow(modelIn, modelOut) and
+      model = "DataFlowFunction"
       or
-      func.(TaintFunction).hasTaintFlow(modelIn, modelOut)
+      func.(TaintFunction).hasTaintFlow(modelIn, modelOut) and
+      model = "TaintFunction"
     ) and
     nodeOut = callOutput(call, modelOut)
   )

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

@@ -0,0 +1,278 @@
+private import cpp
+private import semmle.code.cpp.ir.IR
+private import codeql.typeflow.TypeFlow
+
+private module Input implements TypeFlowInput<Location> {
+  /** Holds if `alloc` dynamically allocates a single object. */
+  private predicate isSingleObjectAllocation(AllocationExpr alloc) {
+    // i.e., `new int`;
+    alloc instanceof NewExpr
+    or
+    // i.e., `malloc(sizeof(int))`
+    exists(SizeofTypeOperator sizeOf | sizeOf = alloc.getSizeExpr() |
+      not sizeOf.getTypeOperand().getUnspecifiedType() instanceof ArrayType
+    )
+  }
+
+  /**
+   * Holds if `i` is the result of a dynamic allocation.
+   *
+   * `isObject` is `true` if the allocation allocated a single object,
+   * and `false` otherwise.
+   */
+  private predicate isAllocation(Instruction i, boolean isObject) {
+    exists(AllocationExpr alloc | alloc = i.getUnconvertedResultExpression() |
+      if isSingleObjectAllocation(alloc) then isObject = true else isObject = false
+    )
+  }
+
+  private predicate hasExactSingleType(Instruction i) {
+    // The address of a variable is always a single object
+    i instanceof VariableAddressInstruction
+    or
+    // A reference always points to a single object
+    i.getResultLanguageType().hasUnspecifiedType(any(ReferenceType rt), false)
+    or
+    // `this` is never an array
+    i instanceof InitializeThisInstruction
+    or
+    // An allocation of a non-array object
+    isAllocation(i, true)
+  }
+
+  private predicate hasExactBufferType(Instruction i) {
+    // Anything with an array type is a buffer
+    i.getResultLanguageType().hasUnspecifiedType(any(ArrayType at), false)
+    or
+    // An allocation expression that we couldn't conclude allocated a single
+    // expression is assigned a buffer type.
+    isAllocation(i, false)
+  }
+
+  private newtype TTypeFlowNode =
+    TInstructionNode(Instruction i) or
+    TFunctionNode(IRFunction func)
+
+  abstract class TypeFlowNode extends TTypeFlowNode {
+    /** Gets a textual representation of this node. */
+    abstract string toString();
+
+    /**
+     * Gets the type of this node. This type may not be the most precise
+     * possible type, but will be used as a starting point of the analysis.
+     */
+    abstract Type getType();
+
+    /** Gets the location of this node. */
+    abstract Location getLocation();
+
+    /** Gets the underlying `Instruction` of this node, if any. */
+    Instruction asInstruction() { none() }
+
+    /** Gets the underlying `IRFunction` of this node, if any. */
+    IRFunction asFunction() { none() }
+
+    /** Holds if the value of this node is always null. */
+    abstract predicate isNullValue();
+  }
+
+  private class InstructionNode extends TypeFlowNode, TInstructionNode {
+    Instruction instr;
+
+    InstructionNode() { this = TInstructionNode(instr) }
+
+    override string toString() { result = instr.toString() }
+
+    override Type getType() {
+      if hasExactSingleType(instr) then result.isSingle() else result.isBuffer()
+    }
+
+    override Location getLocation() { result = instr.getLocation() }
+
+    override Instruction asInstruction() { result = instr }
+
+    override predicate isNullValue() {
+      instr.(ConstantInstruction).getValue() = "0" and
+      instr.getResultIRType() instanceof IRAddressType
+    }
+  }
+
+  /** Gets the `TypeFlowNode` corresponding to `i`. */
+  additional InstructionNode instructionNode(Instruction i) { result.asInstruction() = i }
+
+  private class FunctionNode extends TypeFlowNode, TFunctionNode {
+    IRFunction func;
+
+    FunctionNode() { this = TFunctionNode(func) }
+
+    override string toString() { result = func.toString() }
+
+    Instruction getReturnValueInstruction() {
+      result = func.getReturnInstruction().(ReturnValueInstruction).getReturnValue()
+    }
+
+    override Type getType() { result = instructionNode(this.getReturnValueInstruction()).getType() }
+
+    override Location getLocation() { result = func.getLocation() }
+
+    override IRFunction asFunction() { result = func }
+
+    override predicate isNullValue() {
+      instructionNode(this.getReturnValueInstruction()).isNullValue()
+    }
+  }
+
+  /**
+   * Gets an ultimiate definition of `phi`. That is, an input to `phi` that is
+   * not itself a `PhiInstruction`.
+   */
+  private Instruction getAnUltimateLocalDefinition(PhiInstruction phi) {
+    result = phi.getAnInput*() and not result instanceof PhiInstruction
+  }
+
+  /**
+   * Holds if this function is private (i.e., cannot be accessed outside its
+   * compilation unit). This means we can use a closed-world assumption about
+   * calls to this function.
+   */
+  private predicate isPrivate(Function func) {
+    // static functions have internal linkage
+    func.isStatic()
+    or
+    // anonymous namespaces have internal linkage
+    func.getNamespace().getParentNamespace*().isAnonymous()
+    or
+    // private member functions are only called internally from inside the class
+    func.(MemberFunction).isPrivate()
+  }
+
+  /**
+   * Holds if `arg` is an argument for the parameter `p` in a private callable.
+   */
+  pragma[nomagic]
+  private predicate privateParamArg(InitializeParameterInstruction p, Instruction arg) {
+    exists(CallInstruction call, int i, Function func |
+      call.getArgument(pragma[only_bind_into](i)) = arg and
+      func = call.getStaticCallTarget() and
+      func.getParameter(pragma[only_bind_into](i)) = p.getParameter() and
+      isPrivate(func)
+    )
+  }
+
+  predicate joinStep(TypeFlowNode n1, TypeFlowNode n2) {
+    // instruction -> phi
+    getAnUltimateLocalDefinition(n2.asInstruction()) = n1.asInstruction()
+    or
+    // return value -> function
+    n2.(FunctionNode).getReturnValueInstruction() = n1.asInstruction()
+    or
+    // function -> call
+    exists(Function func | func = n1.asFunction().getFunction() |
+      not func.isVirtual() and
+      n2.asInstruction().(CallInstruction).getStaticCallTarget() = func
+    )
+    or
+    // Argument -> parameter where the parameter's enclosing function
+    // is "private".
+    exists(Instruction arg, Instruction p |
+      privateParamArg(p, arg) and
+      n1.asInstruction() = arg and
+      n2.asInstruction() = p
+    )
+  }
+
+  /**
+   * Holds if knowing whether `i1` points to a single object or buffer implies
+   * knowing whether `i2` points to a single object or buffer.
+   */
+  private predicate instructionStep(Instruction i1, Instruction i2) {
+    i2.(CopyInstruction).getSourceValue() = i1
+    or
+    i2.(CopyValueInstruction).getSourceValue() = i1
+    or
+    i2.(ConvertInstruction).getUnary() = i1
+    or
+    i2.(CheckedConvertOrNullInstruction).getUnary() = i1
+    or
+    i2.(InheritanceConversionInstruction).getUnary() = i1
+    or
+    i2.(PointerArithmeticInstruction).getLeft() = i1
+  }
+
+  predicate step(TypeFlowNode n1, TypeFlowNode n2) {
+    instructionStep(n1.asInstruction(), n2.asInstruction())
+  }
+
+  predicate isNullValue(TypeFlowNode n) { n.isNullValue() }
+
+  private newtype TType =
+    TSingle() or
+    TBuffer()
+
+  class Type extends TType {
+    string toString() {
+      this.isSingle() and
+      result = "Single"
+      or
+      this.isBuffer() and
+      result = "Buffer"
+    }
+
+    /** Holds if this type is the type that represents a single object. */
+    predicate isSingle() { this = TSingle() }
+
+    /** Holds if this type is the type that represents a buffer. */
+    predicate isBuffer() { this = TBuffer() }
+
+    /**
+     * Gets a super type of this type, if any.
+     *
+     * The type relation is `Single <: Buffer`.
+     */
+    Type getASupertype() {
+      this.isSingle() and
+      result.isBuffer()
+    }
+  }
+
+  predicate exactTypeBase(TypeFlowNode n, Type t) {
+    exists(Instruction instr | instr = n.asInstruction() |
+      hasExactSingleType(instr) and t.isSingle()
+      or
+      hasExactBufferType(instr) and t.isBuffer()
+    )
+  }
+
+  pragma[nomagic]
+  private predicate upcastCand(TypeFlowNode n, Type t1, Type t2) {
+    exists(TypeFlowNode next |
+      step(n, next)
+      or
+      joinStep(n, next)
+    |
+      n.getType() = t1 and
+      next.getType() = t2 and
+      t1 != t2
+    )
+  }
+
+  private predicate upcast(TypeFlowNode n, Type t1) {
+    exists(Type t2 | upcastCand(n, t1, t2) |
+      // No need for transitive closure since the subtyping relation is just `Single <: Buffer`
+      t1.getASupertype() = t2
+    )
+  }
+
+  predicate typeFlowBaseCand(TypeFlowNode n, Type t) { upcast(n, t) }
+}
+
+private module TypeFlow = Make<Location, Input>;
+
+/**
+ * Holds if `i` is an instruction that computes an address that points to a
+ * single object (as opposed to pointing into a buffer).
+ */
+pragma[nomagic]
+predicate isPointerToSingleObject(Instruction i) {
+  TypeFlow::bestTypeFlow(Input::instructionNode(i), any(Input::Type t | t.isSingle()), _)
+}

cpp/ql/lib/semmle/code/cpp/ir/dataflow/internal/tainttracking1/TaintTrackingImpl.qll

@@ -127,7 +127,7 @@ abstract deprecated class Configuration extends DataFlow::Configuration {
 
   final override predicate isAdditionalFlowStep(DataFlow::Node node1, DataFlow::Node node2) {
     this.isAdditionalTaintStep(node1, node2) or
-    defaultAdditionalTaintStep(node1, node2)
+    defaultAdditionalTaintStep(node1, node2, _)
   }
 
   /**